CN118031246B - Combustion control optimization method and device for industrial boiler - Google Patents
Combustion control optimization method and device for industrial boiler Download PDFInfo
- Publication number
- CN118031246B CN118031246B CN202410424196.0A CN202410424196A CN118031246B CN 118031246 B CN118031246 B CN 118031246B CN 202410424196 A CN202410424196 A CN 202410424196A CN 118031246 B CN118031246 B CN 118031246B
- Authority
- CN
- China
- Prior art keywords
- boiler
- combustion
- data
- steam pressure
- oxygen content
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
- 238000002485 combustion reaction Methods 0.000 title claims abstract description 121
- 238000000034 method Methods 0.000 title claims abstract description 104
- 238000005457 optimization Methods 0.000 title claims abstract description 55
- QVGXLLKOCUKJST-UHFFFAOYSA-N atomic oxygen Chemical compound [O] QVGXLLKOCUKJST-UHFFFAOYSA-N 0.000 claims abstract description 45
- 229910052760 oxygen Inorganic materials 0.000 claims abstract description 45
- 239000001301 oxygen Substances 0.000 claims abstract description 45
- 239000007789 gas Substances 0.000 claims abstract description 27
- 238000003062 neural network model Methods 0.000 claims abstract description 20
- 238000004806 packaging method and process Methods 0.000 claims abstract description 14
- 239000002245 particle Substances 0.000 claims description 57
- 238000004422 calculation algorithm Methods 0.000 claims description 52
- 230000006870 function Effects 0.000 claims description 50
- 230000008569 process Effects 0.000 claims description 35
- 241000282461 Canis lupus Species 0.000 claims description 33
- 238000013528 artificial neural network Methods 0.000 claims description 26
- 238000012549 training Methods 0.000 claims description 20
- 238000012545 processing Methods 0.000 claims description 18
- 239000000779 smoke Substances 0.000 claims description 16
- 239000013598 vector Substances 0.000 claims description 15
- 230000002159 abnormal effect Effects 0.000 claims description 8
- 238000004590 computer program Methods 0.000 claims description 7
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 claims description 7
- 230000004913 activation Effects 0.000 claims description 6
- 230000001502 supplementing effect Effects 0.000 claims description 6
- 238000012795 verification Methods 0.000 claims description 6
- 208000025174 PANDAS Diseases 0.000 claims description 5
- 208000021155 Paediatric autoimmune neuropsychiatric disorders associated with streptococcal infection Diseases 0.000 claims description 5
- 235000016496 Panda oleosa Nutrition 0.000 claims description 5
- 230000008878 coupling Effects 0.000 claims description 4
- 238000010168 coupling process Methods 0.000 claims description 4
- 238000005859 coupling reaction Methods 0.000 claims description 4
- 210000002569 neuron Anatomy 0.000 claims description 3
- 240000000220 Panda oleosa Species 0.000 claims 1
- 239000000243 solution Substances 0.000 description 20
- 239000000446 fuel Substances 0.000 description 16
- 229910052799 carbon Inorganic materials 0.000 description 13
- OKTJSMMVPCPJKN-UHFFFAOYSA-N Carbon Chemical compound [C] OKTJSMMVPCPJKN-UHFFFAOYSA-N 0.000 description 11
- 238000003860 storage Methods 0.000 description 9
- 230000000694 effects Effects 0.000 description 8
- 230000008901 benefit Effects 0.000 description 6
- 230000001965 increasing effect Effects 0.000 description 5
- 238000004519 manufacturing process Methods 0.000 description 5
- CURLTUGMZLYLDI-UHFFFAOYSA-N Carbon dioxide Chemical compound O=C=O CURLTUGMZLYLDI-UHFFFAOYSA-N 0.000 description 4
- 240000004718 Panda Species 0.000 description 4
- 239000003344 environmental pollutant Substances 0.000 description 4
- 231100000719 pollutant Toxicity 0.000 description 4
- 230000003247 decreasing effect Effects 0.000 description 3
- 238000005265 energy consumption Methods 0.000 description 3
- 230000006872 improvement Effects 0.000 description 3
- 238000013507 mapping Methods 0.000 description 3
- 238000012544 monitoring process Methods 0.000 description 3
- 238000010606 normalization Methods 0.000 description 3
- 238000012856 packing Methods 0.000 description 3
- 238000010298 pulverizing process Methods 0.000 description 3
- 230000009467 reduction Effects 0.000 description 3
- 238000004458 analytical method Methods 0.000 description 2
- 229910002092 carbon dioxide Inorganic materials 0.000 description 2
- 239000001569 carbon dioxide Substances 0.000 description 2
- 230000008859 change Effects 0.000 description 2
- 238000010276 construction Methods 0.000 description 2
- 230000001276 controlling effect Effects 0.000 description 2
- 238000007405 data analysis Methods 0.000 description 2
- 238000011161 development Methods 0.000 description 2
- 238000009826 distribution Methods 0.000 description 2
- 230000008030 elimination Effects 0.000 description 2
- 238000003379 elimination reaction Methods 0.000 description 2
- 239000012535 impurity Substances 0.000 description 2
- 238000007689 inspection Methods 0.000 description 2
- 230000003993 interaction Effects 0.000 description 2
- 238000012804 iterative process Methods 0.000 description 2
- 238000012423 maintenance Methods 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 230000009466 transformation Effects 0.000 description 2
- 241000282421 Canidae Species 0.000 description 1
- UGFAIRIUMAVXCW-UHFFFAOYSA-N Carbon monoxide Chemical compound [O+]#[C-] UGFAIRIUMAVXCW-UHFFFAOYSA-N 0.000 description 1
- 238000012952 Resampling Methods 0.000 description 1
- 229910000831 Steel Inorganic materials 0.000 description 1
- 230000006978 adaptation Effects 0.000 description 1
- 238000013459 approach Methods 0.000 description 1
- 230000006399 behavior Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000008033 biological extinction Effects 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 230000015556 catabolic process Effects 0.000 description 1
- 238000006243 chemical reaction Methods 0.000 description 1
- 239000003245 coal Substances 0.000 description 1
- 230000008094 contradictory effect Effects 0.000 description 1
- 230000001808 coupling effect Effects 0.000 description 1
- 238000007418 data mining Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000006731 degradation reaction Methods 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 238000004134 energy conservation Methods 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 230000002708 enhancing effect Effects 0.000 description 1
- 230000007613 environmental effect Effects 0.000 description 1
- 230000005284 excitation Effects 0.000 description 1
- 230000008713 feedback mechanism Effects 0.000 description 1
- 239000003546 flue gas Substances 0.000 description 1
- 230000002431 foraging effect Effects 0.000 description 1
- 230000005484 gravity Effects 0.000 description 1
- 230000020169 heat generation Effects 0.000 description 1
- 238000002347 injection Methods 0.000 description 1
- 239000007924 injection Substances 0.000 description 1
- 238000013178 mathematical model Methods 0.000 description 1
- 238000013508 migration Methods 0.000 description 1
- 230000005012 migration Effects 0.000 description 1
- 238000002156 mixing Methods 0.000 description 1
- 238000006386 neutralization reaction Methods 0.000 description 1
- 238000007781 pre-processing Methods 0.000 description 1
- 230000001105 regulatory effect Effects 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 238000010845 search algorithm Methods 0.000 description 1
- 239000010959 steel Substances 0.000 description 1
- 239000000126 substance Substances 0.000 description 1
- 238000012546 transfer Methods 0.000 description 1
- 230000000007 visual effect Effects 0.000 description 1
Classifications
-
- F—MECHANICAL ENGINEERING; LIGHTING; HEATING; WEAPONS; BLASTING
- F23—COMBUSTION APPARATUS; COMBUSTION PROCESSES
- F23N—REGULATING OR CONTROLLING COMBUSTION
- F23N5/00—Systems for controlling combustion
-
- F—MECHANICAL ENGINEERING; LIGHTING; HEATING; WEAPONS; BLASTING
- F23—COMBUSTION APPARATUS; COMBUSTION PROCESSES
- F23N—REGULATING OR CONTROLLING COMBUSTION
- F23N2223/00—Signal processing; Details thereof
- F23N2223/10—Correlation
-
- F—MECHANICAL ENGINEERING; LIGHTING; HEATING; WEAPONS; BLASTING
- F23—COMBUSTION APPARATUS; COMBUSTION PROCESSES
- F23N—REGULATING OR CONTROLLING COMBUSTION
- F23N2223/00—Signal processing; Details thereof
- F23N2223/44—Optimum control
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02P—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN THE PRODUCTION OR PROCESSING OF GOODS
- Y02P90/00—Enabling technologies with a potential contribution to greenhouse gas [GHG] emissions mitigation
- Y02P90/02—Total factory control, e.g. smart factories, flexible manufacturing systems [FMS] or integrated manufacturing systems [IMS]
Landscapes
- Engineering & Computer Science (AREA)
- Chemical & Material Sciences (AREA)
- Combustion & Propulsion (AREA)
- Mechanical Engineering (AREA)
- General Engineering & Computer Science (AREA)
- Regulation And Control Of Combustion (AREA)
Abstract
The invention relates to a combustion control optimization method and a device for an industrial boiler, which firstly obtain related data of a boiler combustion process; the key characteristic parameters influencing the combustion control of the boiler are determined from the related data by a Fisher-Tropsch score method and a packaging method, the characteristic parameter data are used as input, the steam pressure and exhaust gas oxygen content prediction values are used as output, the steam pressure and exhaust gas oxygen content prediction is carried out based on a trained neural network model, the combustion process is optimally controlled by an adjusting function based on the prediction values, and the combustion process of the boiler can be optimally controlled by the method, so that the combustion efficiency and performance of the industrial boiler are improved.
Description
Technical Field
The invention relates to the technical field of industrial boiler combustion control, in particular to a combustion control optimization method of an industrial boiler.
Background
Industrial boilers are widely applied in industries such as heat supply, petrochemical industry, chemical industry, steel, colored, papermaking and the like and daily life, are important infrastructure for guaranteeing national economic development and people's life, and are also main energy consumption equipment and important atmospheric pollutants and carbon emission sources. The total carbon emission of the industrial boiler industry is high, which is about 15 percent of the total carbon emission of China. The boiler is an important energy conversion device and is also an important carbon dioxide emission source for energy consumption households.
The industrial boiler in China mainly uses fire coal, has large conservation quantity, wide distribution, high energy consumption and carbon emission, has a certain gap between the energy efficiency and the overall level of carbon dioxide emission control compared with foreign countries, and has great potential of energy conservation and emission reduction. Most of the prior large industrial boilers are provided with automatic control systems, so that automatic intelligent control on boiler operation can be realized, but enterprises monitor, manage and control the efficiency of the industrial boilers, the efficiency is low, and the traditional automatic control method of the boilers also needs to be improved, so that the operation working conditions of the boilers are optimized more efficiently.
Therefore, the energy-saving carbon reduction transformation of the boiler is implemented, and means such as automatic control, combustion optimization adjustment, heat exchange system transformation and the like are combined with actual adoption, so that the method has important significance for improving the energy efficiency level of the industrial boiler, optimizing the fuel structure and process, reducing the carbon emission, further improving the overall competitiveness of enterprises, realizing the green low-carbon high-quality development of the industrial boiler industry and realizing the carbon peak and carbon neutralization targets in China.
Disclosure of Invention
In order to make up the defects of the prior art, the invention provides the combustion control optimizing method and the device for the industrial boiler, and the combustion process of the industrial boiler is optimally controlled through an intelligent optimizing algorithm, so that the heat efficiency and the energy utilization efficiency of the boiler are improved, the carbon emission of the boiler is reduced, and the enterprise cost is reduced.
The first aspect of the invention provides a combustion optimization control method of an industrial boiler, which comprises the following steps:
S1, under different operation conditions, collecting data generated in the operation process of an industrial boiler by using a data acquisition module of a DCS digital control system, and acquiring related data of the combustion control process of the boiler by combining historical operation data of the industrial boiler;
s2, determining key characteristic parameters influencing boiler combustion control from the combustion control process related data through a Fisher fraction and a packaging method;
S3, eliminating abnormal values of data in the key characteristic parameters by adopting a Grabbs criterion; filling the data of each characteristic parameter by adopting Pandas interpolation filling method, so that the time labels of the data of all the characteristic parameters are matched, and carrying out standardized processing on the filled data to serve as training characteristic data;
s4, building a neural network prediction model by taking the characteristic data as input and taking a steam pressure and exhaust smoke oxygen content predicted value as output parameters;
S5, using a gray wolf optimization algorithm boundary strategy in a particle swarm optimization algorithm updating formula to obtain a performance-enhanced updating strategy, optimizing a neural network prediction model based on the strategy, and training the optimized neural network model by utilizing characteristic data;
s6, steam pressure and exhaust gas oxygen content prediction is carried out based on the trained neural network model;
S7, setting an adjusting function, determining a steam pressure value and a smoke exhaust oxygen content value corresponding to the optimal combustion condition according to the type of the industrial boiler and the actual working condition, combining the steam pressure and the smoke exhaust oxygen content predicted value of the model, selecting a target value of the minimum adjusting function as a parameter corresponding to the optimal combustion condition of the industrial boiler, and providing the parameter to control equipment so that the control equipment generates a control signal for optimizing the combustion of the boiler based on the parameter.
As a further improvement of the invention, for step S2, determining key characteristic parameters affecting boiler combustion control from the combustion control process related data by fischer-tropsch fraction and packing method; comprising the following steps: in the feature processing stage, firstly, features with strong coupling property are found out from the related data of the combustion control process through a Fisher score method, the features with low correlation of the combustion control process are removed, then important features are further extracted through a packaging method, the packaging method iterates through a feature selection model to select an optimal feature subset, wherein the iterating mode is that one feature is eliminated each time, the importance degree of the feature is updated through cross verification again according to a new feature subset, finally, the feature subset with highest prediction precision is selected as an optimal feature subset according to the number of final target feature subsets, and therefore a plurality of features selected through the Fisher score method are filtered again to select related features with optimal performance to be used as key feature parameters affecting the combustion control of a boiler.
As a further improvement of the invention, the key characteristic parameters affecting the combustion control of the boiler comprise the current steam pressure of the boiler, the current oxygen content of the discharged smoke, the water supplementing quantity of the boiler per hour, the pressure of the boiler hearth, the temperature of the boiler hearth, the rotating speed of an induced draft fan, the rotating speed of a blower, the rotating speed of a fire grate, the water supplementing variable quantity of the boiler every five minutes, the steam pressure variable value of the boiler every five minutes and the oxygen content variable value of the discharged smoke every five minutes.
As a further improvement of the present invention, the neural network model includes a 3-layer structure of: an input layer, an hidden layer and an output layer; wherein the activation function of the hidden layer neuron is a Gaussian function; the first layer is used for inputting the training characteristic data related to the boiler combustion, the second layer uses a Gaussian function as an activation function to carry out nonlinear processing on the input parameters, and the third layer is an output layer used for outputting the final steam pressure and exhaust gas oxygen content prediction result.
Compared with the prior art, the invention has the following beneficial effects: the invention provides a combustion optimization control method of an industrial boiler, which comprises the steps of determining key influence parameters of boiler combustion control from related data through a Fisher-Tropsch score method and a packaging method, taking characteristic data as input, taking predicted values of steam pressure and exhaust gas oxygen content as output, predicting the steam pressure and the exhaust gas oxygen content based on a trained neural network model, and optimally controlling a combustion process through an adjusting function; the complex nonlinear relation can be processed by adopting the neural network model, so that the prediction precision can be improved; the method has the advantages that the boundary strategy of the gray wolf optimization algorithm is used in the particle swarm optimization algorithm updating formula, the performance-enhanced updating strategy is obtained, the neural network model is optimized through the strategy, so that the neural network model is improved, the global searching capability can be improved through the combination, the problem that the later convergence performance of the particle swarm algorithm is reduced is solved, the boiler combustion process can be optimally controlled through the prediction method, and the efficiency and the performance of the industrial boiler combustion process can be improved. The method provided by the invention has strong practicability, can be widely applied to actual production, and helps a driller to better control the boiler combustion process, so that the energy efficiency level of an industrial boiler is improved, the fuel structure and process are optimized, the carbon emission is reduced, and the overall competitiveness of an enterprise is further improved.
Drawings
Fig. 1 is a flow chart of the steps of a combustion optimizing control method of an industrial boiler according to embodiment 1 of the present invention.
Detailed Description
Exemplary embodiments of the present invention will now be described with reference to the accompanying drawings, in which various details of the embodiments of the present invention are included to facilitate understanding, and are to be considered merely exemplary. Accordingly, those of ordinary skill in the art will recognize that various changes and modifications of the embodiments described herein can be made without departing from the scope and spirit of the invention. Also, descriptions of well-known functions and constructions are omitted in the following description for clarity and conciseness. It is noted that embodiments of the invention and features of the embodiments may be combined with each other without conflict.
Fig. 1 is a schematic flow chart of a combustion optimizing control method of an industrial boiler according to an embodiment of the present invention. As shown in fig. 1, a combustion optimization control method of an industrial boiler according to an embodiment of the present invention includes the following steps:
S1, under different operation conditions, collecting data generated in the operation process of an industrial boiler by using a data acquisition module of a DCS digital control system, and acquiring related data of the combustion control process of the boiler by combining historical operation data of the industrial boiler;
s2, determining key characteristic parameters influencing boiler combustion control from the combustion control process related data through a Fisher fraction and a packaging method;
S3, eliminating abnormal values of data in the key characteristic parameters by adopting a Grabbs criterion; filling the data of each characteristic parameter by adopting Pandas interpolation filling method, so that the time labels of the data of all the characteristic parameters are matched, and carrying out standardized processing on the filled data to serve as training characteristic data;
s4, building a neural network prediction model by taking the characteristic data as input and taking a steam pressure and exhaust smoke oxygen content predicted value as output parameters;
S5, using a gray wolf optimization algorithm boundary strategy in a particle swarm optimization algorithm updating formula to obtain a performance-enhanced updating strategy, optimizing a neural network prediction model based on the strategy, and training the optimized neural network model by utilizing characteristic data;
s6, steam pressure and exhaust gas oxygen content prediction is carried out based on the trained neural network model;
S7, setting an adjusting function, determining a steam pressure value and a smoke exhaust oxygen content value corresponding to the optimal combustion condition according to the type of the industrial boiler and the actual working condition, combining the steam pressure and the smoke exhaust oxygen content predicted value of the model, selecting a target value of the minimum adjusting function as a parameter corresponding to the optimal combustion condition of the industrial boiler, and providing the parameter to control equipment so that the control equipment generates a control signal for optimizing the combustion of the boiler based on the parameter.
The industrial boiler DCS digital control system consists of a data acquisition module, a server rear end and control equipment comprising a browser front end: the data acquisition module and the boiler are positioned in the same local area network and are responsible for the acquisition and transfer of data of a plurality of boiler sensors in the local area network; the rear end of the server realizes the functions of boiler operation optimization system and boiler parameter monitoring, and simultaneously realizes background operation in the user interaction system; the front end of the browser provides visual user interaction and control functions; for data acquisition in the step S1, the data used by the invention are acquired by a boiler site sensor, and the data acquisition module has the functions of measuring parameters such as boiler feedwater temperature, steam temperature, gas pressure, steam pressure, gas flow and steam flow, blower rotation speed, induced draft fan rotation speed and the like and mainly comprises a boiler feedwater temperature sensor, a steam temperature sensor, a gas pressure sensor, a steam pressure sensor, a gas flow sensor, a steam flow sensor, a tachometer and the like; the data part required by the modeling is directly DCS system data, and the part of the data is obtained after further processing.
The industrial boiler equipment is a very complex controlled object, and the combustion process of the industrial boiler equipment has the characteristics of high nonlinearity, strong coupling, multiple interference factors and the like. The prediction model adopted by the invention takes data and historical data acquired in real time as training samples, and sample input directly influences model precision, so that the selection of input variables becomes the key of data preprocessing. Therefore, the method firstly processes the data of the boiler DCS system through a data mining technology, modifies the time corresponding relation between the data, processes the data such as missing values, abnormal values and the like, and ensures that the subsequent analysis data is in a normal and reasonable operation range. And steady-state data is further extracted through a data resampling method, so that the expressive force of the data is improved.
In the step S2, in the feature processing stage, the redundant features are removed by using the fischer score, and then the feature latitude is further reduced by adopting a packaging method, so that the complexity of the subsequent modular operation is reduced. Aiming at the characteristics of multivariable, nonlinear and strong coupling in the boiler combustion process, firstly, the characteristic with strong coupling is found out through the Fisher score, the characteristic with low correlation with the combustion control is removed, and the important characteristic is further extracted through a packaging method, so that the calculation complexity of the model is reduced. The packing method is a method for selecting features according to a specific prediction model, and adopts a recursive feature elimination method (Recursive feature elimination, RFE), and the packing method is combined with a model selected by modeling to iteratively select an optimal feature subset. And (5) selecting the related features with the best performance from the multiple features selected by the regression model through a packaging method. The RFE algorithm processing flow is completed by three steps: firstly, 45 features to be screened are input into a corresponding learner as an initial feature subset, and the importance degree of different features is calculated by adopting a 5-fold cross verification mode. Then, eliminating one feature at a time in an iterative mode, and carrying out cross verification again according to the new feature subset to update the feature importance degree. And finally, selecting the feature subset with the highest prediction precision feature as the optimal feature subset according to the number of the final target feature subsets.
The operation of industrial boilers is a complex process involving many factors and parameters, which have different effects on combustion efficiency and efficiency. In order to better understand and optimize the combustion process of the boiler, we need to go through intensive discussion of these key factors and parameters. First, fuel is the basis of the combustion process. The quality is directly related to the combustion effect. High quality fuels typically contain higher calorific values, which help to improve combustion efficiency. In addition, the supply pressure of the fuel is also an important factor. Unstable pressure may lead to insufficient combustion. Secondly, the air supply plays a vital role in the combustion process. The air quantity and the air temperature have influence on the combustion efficiency. Insufficient air volume may result in incomplete combustion of the fuel, while excessive air volume may carry away excessive heat. Therefore, the air volume control requires precise adjustment. The temperature and pressure in the furnace are also key parameters. Too high or too low a furnace temperature may affect combustion efficiency, while too high or too low a pressure may pose a threat to safe operation of the boiler. The performance and stability of the pulverizing system also have an impact on the combustion process. Failure of the pulverizing system may cause unstable fuel supply, thereby affecting combustion. Therefore, maintaining a good operation state of the pulverizing system is important to ensure stable combustion of the boiler. Different types of burners have different effects on the combustion process. The design of the burner, the fuel injection method, the air mixing method, etc. affect the combustion effect of the fuel. Therefore, it is important to select a burner suitable for the type of boiler and the fuel characteristics. Finally, the skill and experience of the operator has a significant impact on the combustion effect. Proper operation and timely adjustment can ensure that the boiler operates in an optimal state, thereby improving combustion efficiency and reducing pollutant emission. In addition, there are other factors and parameters that affect the boiler combustion process. For example, impurities and ash content in fuels can affect combustion efficiency and pollutant emissions. If the fuel contains higher levels of impurities and ash, incomplete combustion may result, increasing smoke and harmful gas emissions. Therefore, fuel selection and handling are important elements in the combustion process of the boiler. At the same time, maintenance and servicing of the burner is also one of the key factors. Regular inspection and maintenance of the burner ensures its proper operation, increases service life and increases combustion efficiency. If the burner is not maintained for a long time, performance may be degraded, affecting combustion.
By combining the above, key characteristic parameters influencing boiler combustion control are screened out from parameters such as boiler feed water temperature, steam temperature, gas pressure, steam pressure, gas flow and steam flow, air blower rotating speed, induced draft fan rotating speed and the like, wherein the characteristic parameters comprise current boiler steam pressure, current exhaust gas oxygen content, boiler water supplementing quantity per hour, boiler furnace pressure, boiler furnace temperature, induced draft fan rotating speed, air blower rotating speed, fire grate rotating speed, boiler water supplementing variable quantity per five minutes, boiler steam pressure variable value per five minutes and exhaust gas oxygen content variable value per five minutes;
Because the acquired parameters comprise different dimension parameters such as pressure, flow, temperature, rotating speed and the like, the size difference between the characteristic parameters is too large, the influence of certain parameters with larger numerical values on the specific gravity is larger during model training, and the neural network model is sensitive to the input value and cannot be built into an accurate model. Normalization of the parameters is required before the neural network is used to build the predictive model. And selecting necessary variables according to the optimization target and combining with the actual combustion characteristics of the boiler to perform data standardization pretreatment, and taking the data standardization pretreatment as a model training sample.
In the step S3, eliminating the abnormal value of the data in the key characteristic parameters by adopting a Grabbs criterion; filling the data of each characteristic parameter by adopting Pandas interpolation filling method to enable the time labels of all the data to be matched, and carrying out standardized processing on the filled data to serve as training characteristic data; during data processing, we take a series of rigorous measures to ensure data quality and accuracy. Firstly, identifying and eliminating abnormal values in related influence parameters by using a Graibus criterion so as to eliminate interference of the abnormal values on data analysis. This step determines outliers based on the degree of deviation of the data from the average, and identifies and excludes outliers by comparing the absolute values of the deviation of each data point from the average.
Next, to ensure the consistency of the time labels of the data, pandas interpolation padding is used to pad the data of each parameter. The method generates new data points according to the trend and mode of the existing data points to compensate the missing part. By interpolation filling, a more complete and consistent data set can be obtained.
After filling the data, a normalization process is performed to eliminate the influence of the dimension and numerical differences of the original data on the data analysis. The normalization process is a process of converting data into a standard normal distribution, where the mean is 0 and the standard deviation is 1. This step helps to better reflect the inherent links between the data and improves the accuracy and generalization ability of model training. By removing abnormal values, filling missing data, performing standardized processing and the like, a group of high-quality training characteristic data is obtained, and the accuracy and generalization capability of model training can be improved.
The choice of output is a critical task in building neural network predictive models. The method not only determines the prediction precision of the model, but also influences the practical application value of the model. In order to ensure the applicability and accuracy of the model, careful consideration must be given to the combustion process and combustion efficiency of the industrial boiler and the relevant data characteristics. Steam pressure is an important monitoring parameter during operation of industrial boilers. It is not only related to the economy and safety of the boiler, but also affects the combustion process of the boiler. First, the variation in steam pressure reflects the combustion conditions of the boiler. When the steam pressure increases, this may mean that the combustion efficiency of the boiler is higher and the heat generation is higher. Conversely, if the steam pressure drops, it may indicate that the combustion efficiency is reduced, or that the external demand for steam is increased, resulting in the boiler failing to maintain the normal pressure. Secondly, stability of the steam pressure is also critical to stability of combustion. If the steam pressure fluctuates too much, instability of the combustion process may result, thereby affecting the operating state of the boiler. Therefore, operators need to pay close attention to the change of steam pressure and intervene in time to keep combustion stable. In addition, the steam pressure has a certain relationship with the efficiency of the boiler. At a certain steam pressure, the efficiency of the boiler increases with increasing steam flow. But if the steam pressure is too high or too low, it may cause a decrease in boiler efficiency. Therefore, operators need to adjust the steam pressure according to the actual situation to ensure efficient operation of the boiler. In summary, the steam pressure is very closely related to the boiler combustion. By monitoring the change of the steam pressure, operators can know the combustion working condition of the boiler in time and take corresponding measures to adjust so as to ensure the safe and economic operation of the boiler.
And the relation between the oxygen content of the exhaust gas and the combustion of the boiler is mainly reflected on the influence on the combustion efficiency and pollutant emission. Ideally, the lower the oxygen content in the exhaust gas, the higher the combustion efficiency, and the greater the likelihood that the fuel will be fully combusted. This is because the combustion process requires oxygen, and if the oxygen content in the exhaust gas is too high, it means that there may be an excess of oxygen in the combustion process, which may result in incomplete combustion of the fuel, thereby reducing combustion efficiency. However, the lower the oxygen content in the flue gas is not, the better. Theoretically, when the oxygen content in the exhaust smoke is reduced to a certain value, extinction of the fuel layer in the hearth may be caused, which not only results in reduction of combustion efficiency, but also may cause safety problems. Therefore, the control of the oxygen content of the exhaust smoke of the boiler combustion needs to be carried out according to the actual situation and the requirements of safety and environmental protection. In summary, the steam pressure and the exhaust oxygen content are determined as model outputs.
For step S4, since the aforementioned input and output are nonlinear, and fast prediction needs to be implemented, so that the on-site operator can control combustion in time, and the radial basis function neural network exhibits a powerful function when dealing with complex nonlinear problems. The nonlinear mapping system has nonlinear mapping capability, and can effectively map an input space to an output space to process various nonlinear relations. In addition, the radial basis function neural network also has the characteristic of global approximation, and can approximate any nonlinear function with any precision, so that the radial basis function neural network has remarkable advantages when dealing with complex nonlinear problems. The radial basis function neural network has the advantages that the learning speed is high, only one time of scanning is needed for training samples, repeated iteration is not needed, and the learning efficiency is greatly improved. Meanwhile, the network structure is relatively simple, easy to realize and deploy, and the application difficulty is reduced. The radial basis function neural network has strong generalization capability, can popularize rules in training samples into new samples, and effectively avoids the problem of overfitting.
The radial basis function neural network has no problem of local minimum value, and only searches for the globally optimal solution, thereby avoiding the influence of the local minimum value on the model performance. Meanwhile, the network can process high-dimension data, and the problem of dimension disasters of the high-dimension data is effectively avoided by adopting an implicit feature mapping mode. In summary, the radial basis function neural network has significant advantages in terms of processing non-linearity problems, improving learning efficiency, simplifying implementation and deployment processes, enhancing generalization capability, avoiding local minimum problems, processing high-dimensional data, and the like.
Therefore, the invention adopts the radial basis function neural network as a basic model, wherein the input vector of the radial basis function neural network isInput vector/>Middle/>For different characteristic data, T is a transpose, and the basis function is a nonlinear activation function/>：
Wherein j=1, 2 … m, j is an integer,
Is the center of the j-th node,/>Is the width of the j-th node,/>Is a sample/>And node center/>The Euclidean distance between m is the node number of the middle layer, and the output formula is obtained by weighting the output data of the middle layer:
n, wherein/> For output,/>The connection weight value from the middle layer to the output layer is n is the number of network outputs; the steam pressure and exhaust gas oxygen content prediction results can be obtained through the model.
Whereas parameters of radial basis function neural networks, e.g. node centresRadial base Width/>And weight/>It needs to be determined by iteration. In order to better optimize the parameters, the invention provides a hybrid optimization method for better optimizing the parameters.
In order to meet the real-time on-line prediction requirements, the optimized boiler operating control parameters must be predicted in a limited time to facilitate the driller's site adjustment, so a fast converging optimization algorithm is required. Based on the reasons, the particle swarm optimization algorithm is adopted as a predictive optimization algorithm, and the parameters which need to be manually set in the radial basis function neural network are initially optimized.
The particle swarm optimization algorithm is an optimization method, is inspired by artificial life research results, and is a global random search algorithm based on swarm intelligence, which is provided by simulating migration and swarm behaviors in the process of foraging of the bird swarm. The particle swarm optimization algorithm is used for searching an optimal solution after initializing a group of random particles and continuously iterating, and continuously updating is realized by tracking two extreme values, wherein each iteration process is a one-time updating process. The first extremum is the optimal solution found by the particles in the iterative process, which is called the individual extremum, and the other extremum is the optimal solution found by the whole population at present, which is called the global optimal solution. The particle swarm algorithm has strong capability of solving the nonlinear problem and the multimodal problem.
In the particle swarm, the selection of parameters has important influence on the performance of the particle swarm algorithm. The parameters involved are mainly the following, population numbers: the iteration speed of the particle swarm algorithm is high, and the influence of the number of the population is small, so that the value range of the initial population is usually 50-1000, the number setting of the initial population is not too large, and the iteration speed is influenced by the excessive number of the population; iteration number: the setting of the iteration times is to select a proper range, the obtained solution is unstable due to too few times, the algorithm running time is longer due to too many times, 100-4000 is usually selected according to experimental experience, and the setting of the iteration times can be correspondingly increased when the relatively complex problem is processed; inertial weight: the method is mainly used for reflecting the influence degree of the historical value condition of an individual on the current condition, and is usually 0.5-1; learning factors: the value of the learning factor takes the value range of the independent variable as the basis, and the independent learning factor and the group learning factor are two types, and the value range is usually between 0 and 4; speed limit: it is important to limit the particle velocity, and when the particle velocity is too high, the particle may directly jump over the optimal solution, and when the particle velocity is too low, the convergence speed may be slow. In the present invention, the position of the particles corresponds to the parameter values that the neural network needs to train. The particle fitness corresponds to the error magnitude of the prediction model. In the iterative process, each particle can correct and adjust the position of the speed of the particle through the found current individual extremum and the current optimal value of the particle group.
Provided that m particles fly continuously in n-dimensional space, they form a particle population The update formula of the speed and the position of the particles in the population is as follows:
；
；
Wherein the method comprises the steps of For the velocity component of the k+1st iteration,/>For the position component at iteration k+1,/>Is the individual optimal position for the current pass of the particle,/>Optimal position in the whole particle population searched for the whole particle population, and position component/>Is the position of the ith particle in space, the velocity component/>Is the flight speed of the ith particle in space,/>Is the weight coefficient of the middle layer, which represents the influence degree of the individual speed at the current moment on the individual speed at the next moment, k is the iteration number, j is the particle dimension sequence number, c 1 and c 2 are learning factors, c 1 is the step length for adjusting the flight direction of the particles close to the individual optimal position, and c 2 is the step length for adjusting the flight direction of the particles towards the individual global optimal position; and r 1 and r 2 are randomly assigned values between 0 and 1.
；
Where k represents the number of iterations so far and k max represents the maximum number of iterations. TerminologyAnd/>Representing the maximum weight and the minimum weight, respectively, are typically set to 0.9 and 0.4, respectively.
Particle swarms tend to aggregate in the final iteration, which reduces their searchability and causes the trouble of the predicted outcome falling into a local optimum. Furthermore, relying on global optimization alone to directly iterate parameters would result in slower subsequent iterations and reduce convergence capacity.
The present invention will therefore employ a gray wolf optimization algorithm to enhance the parameter iterative update formula and solve this problem. The gray wolf optimization algorithm is a novel group intelligent optimization algorithm, the algorithm has strong local searching capability, and the convergence speed is faster than that of other algorithms. The algorithm has the advantages of less required artificial setting parameters, simple structure and principle, easy realization and the like. The gray wolf optimization algorithm has stronger robustness in terms of optimization performance, and the information feedback mechanism of the search individual is arranged in the algorithm, so that the relation between the global search and the local search of the algorithm is effectively balanced, and the convergence capability is stronger. The gray wolf optimization algorithm adopts elite group suggestion, and three optimal elite individuals are selected, wherein the three optimal elite individuals are respectively: alpha wolf, beta wolf and delta wolf. The gray wolf optimization algorithm is introduced into the particle swarm optimization algorithm, so that the prediction error of which three persons in the elite swarm is minimum during iterative updating can be determined. The introduction of the wolf optimization algorithm improves the problem of late convergence performance degradation of the particle swarm algorithm, as other particles will surround three elite individuals rather than a single optimal individual.
The boundary search strategy is the most important search strategy in the wolf algorithm, and defines the population consisting of n wolves as follows: the mathematical model of the population approaching and surrounding the prey is as follows:
；
；
Wherein D represents the distance between the individual and the target, A, C is a coefficient vector, the current position of the wolf can be updated to other positions around the optimal solution by changing the values of the system vectors A and C, k represents the current iteration times, X p (k) is a position vector of the kth generation of hunting object, X (k) is the current position vector of the wolf individual, and X (k+1) is a position vector of the next movement of the wolf individual;
Assuming that the wolf is in position (X, Y), the prey is in position (X 1,Y1), the wolf will move to (X 1-X,Y1), different assignments of coefficients A, C will produce different boundary effects, 、/>Is a vector representation of the aforementioned coefficients, which can be expressed by the following formula:
；
；
Where r 1 and r 2 are random numbers within [0,1], convergence factor Is a control parameter, is linearly decreased along with the iteration times within the range of [0,2], and the decreasing formula is/>K max is the maximum number of iterations; the solution space around the optimal solution can be searched by adjusting the values of A and C, so that the local searching capability of the algorithm is ensured. Meanwhile, r 1 and r 2 are random numbers in [0,1], so that the gray wolf population can traverse the whole solution space, namely, the global searching capability of the algorithm is ensured.
According to the foregoing description, the gray wolf algorithm uses elite population suggestions to select the three best elite individuals, which are: the optimal solution alpha, the suboptimal solution beta and the third optimal solution delta are used for guiding particle parameter updating. Three best elite individuals will guide the grain according to the hunting boundary. The instruction strategy formula is as follows:
；
；
；
Wherein, in the above equation 、/>、/>Representing the distance between each particle and three elite individuals (α, β and δ), respectively; the following equation is the direction of movement of the particles to the three elite individuals;
；
；
；
And equation of
；
The movement direction of the particle swarm after being guided by the three elite individuals is used;
In the above ，/>，/>Is the current position of the optimal solution alpha wolf, the suboptimal solution beta wolf and the third optimal solution delta wolf,，/>，/>Alpha wolf, beta wolf and delta wolf are weighted/>, respectivelyThe direction vector of the next movement affected, C 1,C2,C3 is a random number between [0,2], A 1,A2,A3 is a random number between [ -2,2], formula/>The position updating formula of the common wolf is obtained.
In dealing with optimization problems, local searching and global searching may be two contradictory processes. In the global searching process, the gray wolf optimization algorithm obtains a new space in the searching space through a suddenly changed solution, so that a better regional space can be found, and the solution is prevented from falling into local optimum through the operation. The use of a gray-wolf algorithm boundary strategy in the velocity and position update formula of the particle swarm algorithm may enhance the performance of the particle swarm optimization algorithm. The formula optimized by the gray wolf algorithm is changed into:
；
；
Updating strategy of the gray wolf optimization algorithm according to the formula The particle swarm optimization algorithm is incorporated into a group optimal guiding strategy based on a position updating formula which keeps individual experience and the particle swarm optimization algorithm, so that the problem of the position updating formula of the particle swarm optimization algorithm can be relieved to a certain extent.
There is only one intermediate layer in the radial basis neural network used in the present invention. The number of nodes in the middle layer affects the predictive effect of the neural network model. This can be calculated from the following formula:
；
wherein n, h and m are the node numbers of the first layer, the second layer and the third layer respectively, Is a random number of [1,10 ].
Thus, the range of the number of intermediate layer nodes can be approximately determined as [3,30]. The training error tends to decrease as the number of nodes in this range increases, and is minimal when the number of nodes is set to 20. When the number of nodes is greater than 20, the error slowly increases. Thus, the number of nodes is determined to be 20.
The three parameters of initial weight, neuron center and radial width determine whether the neural network can converge to the minimum error and training speed in the training process. In the subsequent iterative optimization process, these three parameters will be continually modified and approach values that minimize global errors. The weights, neuron centers, and radial basis widths are initially set to random values between (0, 1).
Based on the above analysis, the radial basis neural network structure was determined herein to be 11-20-2. Furthermore, a gaussian function is used as the excitation function.
Each particle in the particle swarm optimization algorithm holds its parameter value. The position data used in the invention is the radial basis function neural network parameters which need to be adjusted. The particle number c=100 is selected. The dimension of the particle is the dimension of the solution space, and refers to the necessary information contained in the particle position, namely the weight, the neuron center and the radial basis width, taking d=60. The maximum number of iterations is T max = 200; highest velocity V max = 1; learning factor C 1=C2 =1.5; the inertial weights are updated according to the iterations. The termination condition is global optimal fitness, global precision requirements need to be met, MSE index is used as a particle fitness function, and iteration is stopped when particles exist in the particle iteration process to enable the global precision to meet the requirements.
After the construction of the optimized prediction model is completed, verification can be performed. The verification purpose is to verify the effect of the algorithm model on increasing the vapor pressure and decreasing the oxygen content of the exhaust.
Because the industrial boiler belongs to a pressure container, a certain pressure safety upper limit exists, and a reasonable range exists for the oxygen content of the exhaust gas, if the parameter corresponding to the predicted value of the steam pressure maximization and the exhaust gas oxygen content minimization is simply set as the optimization parameter, the control equipment controls the combustion process of the boiler by determining the control signal corresponding to the optimization parameter, and the boiler running operation parameter inconsistent with the safety production inspection standard can be calculated.
Therefore, according to different boiler types and actual running conditions, the invention sets an optimal value of the adjusting parameter in advance, and sets the adjusting function as the difference between the predicted value and the threshold value. The adjusting function is
；
Wherein the method comprises the steps ofTo adjust the function,/>For inputting vectors,/>Is the optimal value,/>In order to predict the model output value, in the present invention, set/>For input vector/>The steam pressure and exhaust oxygen content prediction output values obtained after the prediction model is input; the adjustment function is specifically:
；
The setting of the regulating function can be based on the production requirements of the factory site, for example, the optimal value of the operation of the industrial boiler is set to be that the steam pressure of y a is 7.0kg/cm 2 and the exhaust oxygen content of y b is 6.5%, wherein the steam pressure of the industrial boiler is directly related to the heat energy use and the production, and the importance degree is higher than that of the exhaust oxygen content, so that the optimal weight of the steam pressure is set 2, Then the adjustment function is:
；
When the sum of absolute values of the prediction errors is minimized, i.e. the target value corresponding to the minimum of the adjustment function is selected as the corresponding parameter when the combustion condition of the boiler is optimal, the target value is provided to the control device, so that the control device generates a set of control signals for controlling the combustion process of the boiler based on the target value.
The invention provides a combustion optimization control method of an industrial boiler, which is characterized in that key influence parameters of boiler combustion control are determined from related data through a Fisher-Tropsch score method and a packaging method, characteristic data are used as input, predicted values of steam pressure and exhaust gas oxygen content are used as output, prediction of the steam pressure and the exhaust gas oxygen content is performed based on a trained neural network model, the combustion process is optimally controlled through an adjusting function, a complex nonlinear relation can be processed by adopting the neural network model, and prediction accuracy can be improved; the method has the advantages that the boundary strategy of the gray wolf optimization algorithm is used in the particle swarm optimization algorithm updating formula, the performance-enhanced updating strategy is obtained, the neural network model is optimized through the strategy, so that the neural network model is improved, the global searching capability can be improved through the combination, the problem that the later convergence performance of the particle swarm algorithm is reduced is solved, the boiler combustion process can be optimally controlled through the prediction method, and the efficiency and the performance of the industrial boiler combustion process can be improved. The method provided by the invention has strong practicability, can be widely applied to actual production, and helps a driller to better control the boiler combustion process, so that the energy efficiency level of an industrial boiler is improved, the combustion structure and process are optimized, the carbon emission is reduced, and the overall competitiveness of an enterprise is further improved.
A second aspect of the present invention provides a combustion optimization control device for an industrial boiler, comprising a memory, a processor and a computer program stored on the memory and executable on the processor, wherein the processor executes the computer program and implements the combustion optimization control method for an industrial boiler as described in the foregoing embodiments.
For the device embodiments, reference is made to the description of the method embodiments for the relevant points, since they essentially correspond to the method embodiments. The apparatus embodiments described above are merely illustrative, wherein the elements illustrated as separate elements may or may not be physically separate, and the elements shown as elements may or may not be physical elements, may be located in one place, or may be distributed over a plurality of network elements. Some or all of the modules may be selected according to actual needs to achieve the purposes of the present application. Those of ordinary skill in the art will understand and implement the present application without undue burden.
Correspondingly, the application also provides a computer program product, comprising a computer program/instruction which realizes the combustion optimization control method of the industrial boiler when being executed by a processor.
Correspondingly, the application also provides electronic equipment, which comprises: one or more processors; a memory for storing one or more programs; when the one or more programs are executed by the one or more processors, the one or more processors are caused to implement the combustion optimization control method of the industrial boiler as described above.
Accordingly, the present application also provides a computer readable storage medium having stored thereon computer instructions which when executed by a processor perform the above-described method. The computer readable storage medium may be an internal storage unit, such as a hard disk or a memory, of any of the data processing enabled devices described in any of the previous embodiments. The computer readable storage medium may also be an external storage device, such as a plug-in hard disk, a smart memory card (SMART MEDIA CARD, SMC), an SD card, a flash memory card (FLASH CARD), etc. provided on the device. Further, the computer readable storage medium may include both internal storage units and external storage devices of any device having data processing capabilities. The computer readable storage medium is used for storing the computer program and other programs and data required by the arbitrary data processing apparatus, and may also be used for temporarily storing data that has been output or is to be output.
Other embodiments of the application will be apparent to those skilled in the art from consideration of the specification and practice of the disclosure herein. This application is intended to cover any variations, uses, or adaptations of the application following, in general, the principles of the application and including such departures from the present disclosure as come within known or customary practice within the art to which the application pertains.
It is to be understood that the application is not limited to the precise arrangements and instrumentalities shown in the drawings, which have been described above, and that various modifications and changes may be effected without departing from the scope thereof.
Claims (2)
1. The combustion optimizing control method for the industrial boiler is characterized by comprising the following steps of:
S1, under different operation conditions, collecting data generated in the operation process of an industrial boiler by using a data acquisition module of a DCS digital control system, and acquiring related data of the combustion control process of the boiler by combining historical operation data of the industrial boiler;
s2, determining key characteristic parameters influencing boiler combustion control from the combustion control process related data through a Fisher fraction and a packaging method;
For the step S2, determining key characteristic parameters influencing boiler combustion control from the combustion control process related data through a Fisher fraction and a packaging method; comprising the following steps: in the feature processing stage, firstly, finding out features with strong coupling from the related data of the combustion control process through a Fisher score method, removing features with low correlation of the combustion control process, then, further extracting important features through a packaging method, and iteratively selecting an optimal feature subset through a feature selection model by the packaging method, wherein the iterative mode is that one feature is eliminated each time, the importance degree of the updated feature is updated through cross verification again according to a new feature subset, and finally, the feature subset with highest prediction precision is selected as an optimal feature subset according to the number of final target feature subsets, so that a plurality of features selected by the Fisher score method are filtered again to select related features with optimal performance as key feature parameters affecting the combustion control of a boiler;
The key characteristic parameters affecting the combustion control of the boiler comprise the current steam pressure of the boiler, the current oxygen content of discharged smoke, the water supplementing amount of the boiler per hour, the pressure of a boiler hearth, the temperature of the boiler hearth, the rotating speed of an induced draft fan, the rotating speed of a blower, the rotating speed of a fire grate, the water supplementing variable amount of the boiler per five minutes, the steam pressure variable value of the boiler per five minutes and the oxygen content variable value of discharged smoke per five minutes;
S3, eliminating abnormal values of data in the key characteristic parameters by adopting a Grabbs criterion; filling the data of each characteristic parameter by adopting Pandas interpolation filling method, so that the time labels of the data of all the characteristic parameters are matched, and carrying out standardized processing on the filled data to serve as training characteristic data;
s4, building a neural network prediction model by taking the characteristic data as input and taking a steam pressure and exhaust smoke oxygen content predicted value as output parameters;
Wherein, neural network model includes 3 layer structure, is respectively: an input layer, an hidden layer and an output layer; wherein the activation function of the hidden layer neuron is a Gaussian function; the first layer is used for inputting training characteristic data related to boiler combustion, the second layer uses a Gaussian function as an activation function to carry out nonlinear processing on input parameters, and the third layer is an output layer and is used for outputting steam pressure and exhaust gas oxygen content prediction results;
wherein the neural network model is a radial basis neural network, and the input vector of the radial basis neural network is Input vector/>Middle/>For different characteristic data, T is a transpose, and the basis function is a nonlinear activation function/>：
Wherein j=1, 2 … m, m is the number of nodes,
Is the center of the j-th node,/>Is the width of the j-th node,/>The Euclidean distance between the sample and the node center, m is the node number of the middle layer, and the output formula is obtained by weighting the output data of the middle layer:
the connection weight value from the middle layer to the output layer is n is the number of network outputs; the steam pressure and exhaust gas oxygen content prediction result can be obtained through the model;
S5, using a gray wolf optimization algorithm boundary strategy in a particle swarm optimization algorithm updating formula to obtain a performance-enhanced updating strategy, optimizing a neural network prediction model based on the strategy, and training the optimized neural network model by utilizing characteristic data;
s6, steam pressure and exhaust gas oxygen content prediction is carried out based on the trained neural network model;
S7, setting an adjusting function, determining a steam pressure value and a smoke exhaust oxygen content value corresponding to the optimal combustion condition according to the type of the industrial boiler and the actual working condition, selecting a target value of the minimum adjusting function as a parameter corresponding to the optimal combustion condition of the industrial boiler by combining the steam pressure and the smoke exhaust oxygen content predicted value of the model, and providing the parameter to control equipment so that the control equipment generates a control signal for optimizing the combustion of the boiler based on the parameter;
Wherein the adjustment function is:
；
Wherein, To adjust the function,/>For inputting characteristic data vector,/>Is weight value/>Steam pressure and exhaust oxygen content optimum values of industrial boiler respectively,/>For input vector/>And inputting the steam pressure and exhaust oxygen content prediction output values obtained after the prediction model.
2. A combustion optimization control device of an industrial boiler, comprising a memory, a processor and a computer program stored on the memory and executable on the processor, wherein the processor executes the computer program and implements the combustion optimization control method of an industrial boiler as claimed in claim 1.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202410424196.0A CN118031246B (en) | 2024-04-10 | 2024-04-10 | Combustion control optimization method and device for industrial boiler |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202410424196.0A CN118031246B (en) | 2024-04-10 | 2024-04-10 | Combustion control optimization method and device for industrial boiler |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN118031246A CN118031246A (en) | 2024-05-14 |
| CN118031246B true CN118031246B (en) | 2024-06-07 |
Family
ID=90995292
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202410424196.0A Active CN118031246B (en) | 2024-04-10 | 2024-04-10 | Combustion control optimization method and device for industrial boiler |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN118031246B (en) |
Families Citing this family (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN118550199B (en) * | 2024-07-24 | 2024-10-01 | 咸阳新兴分布式能源有限公司 | Energy supply efficiency optimization method and system for gas steam boiler |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113137766A (en) * | 2021-04-22 | 2021-07-20 | 江西江右净达热能科技有限公司 | Boiler automation control system |
| CN115145152A (en) * | 2022-07-04 | 2022-10-04 | 浙江大学 | Boiler combustion and denitration process collaborative optimization control method |
| CN117311141A (en) * | 2023-08-23 | 2023-12-29 | 华北电力大学 | Modeling method for thermal power unit coordination control system integrating hierarchical parameter optimization and fuzzy neural network |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US7401577B2 (en) * | 2003-03-19 | 2008-07-22 | American Air Liquide, Inc. | Real time optimization and control of oxygen enhanced boilers |
-
2024
- 2024-04-10 CN CN202410424196.0A patent/CN118031246B/en active Active
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113137766A (en) * | 2021-04-22 | 2021-07-20 | 江西江右净达热能科技有限公司 | Boiler automation control system |
| CN115145152A (en) * | 2022-07-04 | 2022-10-04 | 浙江大学 | Boiler combustion and denitration process collaborative optimization control method |
| CN117311141A (en) * | 2023-08-23 | 2023-12-29 | 华北电力大学 | Modeling method for thermal power unit coordination control system integrating hierarchical parameter optimization and fuzzy neural network |
Non-Patent Citations (1)
| Title |
|---|
| 锅炉烟气含氧量预测混合模型及应用;刘闯;李素芬;陈贵军;;能源与环境;20080430(第02期);全文 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN118031246A (en) | 2024-05-14 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Zhan et al. | Deepthermal: Combustion optimization for thermal power generating units using offline reinforcement learning | |
| US7389151B2 (en) | Systems and methods for multi-level optimizing control systems for boilers | |
| CN107726358B (en) | Boiler Combustion Optimization System based on CFD numerical simulations and intelligent modeling and method | |
| Tuttle et al. | A systematic comparison of machine learning methods for modeling of dynamic processes applied to combustion emission rate modeling | |
| CN105020705B (en) | Burning in circulating fluid bed boiler performance method for real-time optimization control and system | |
| CN104763999A (en) | Power plant pulverized coal boiler combustion performance online optimizing method and system | |
| CN109508818B (en) | Online NOx prediction method based on LSSVM | |
| CN105864797B (en) | Real-time prediction system and method for boiler entering heat value of circulating fluidized bed household garbage incineration boiler | |
| CN118031246B (en) | Combustion control optimization method and device for industrial boiler | |
| CN107203687B (en) | Multi-target cooperative intelligent optimization control method for desulfurization process of absorption tower | |
| Nemitallah et al. | Artificial intelligence for control and optimization of boilers’ performance and emissions: A review | |
| Rossi et al. | Artificial neural networks and physical modeling for determination of baseline consumption of CHP plants | |
| Tang et al. | Adaptive nonlinear model predictive control of NOx emissions under load constraints in power plant boilers | |
| CN109325313A (en) | Based on improvement quantum telepotation boiler of power plant NOXPrediction model device | |
| CN110059821A (en) | Neural network optimization, device, computer equipment and storage medium | |
| CN116300755A (en) | Double-layer optimal scheduling method and device for heat storage-containing heating system based on MPC | |
| Yan et al. | A knowledge transfer online stochastic configuration network-based prediction model for furnace temperature in a municipal solid waste incineration process | |
| Li et al. | Neural networks and genetic algorithms can support human supervisory control to reduce fossil fuel power plant emissions | |
| Movahed et al. | Modeling and optimization of NO emission for a steam power plant by data‐driven methods | |
| CN109101683B (en) | Model updating method for pyrolysis kettle of coal quality-based utilization and clean pretreatment system | |
| CN117930776A (en) | Accurate steam supply pre-regulation method and system for white spirit production based on model prediction | |
| CN113887130B (en) | Industrial boiler operation optimization method based on ensemble learning | |
| CN116227350A (en) | Multi-target optimization method and device for boiler | |
| Sujatha et al. | Soft sensor with shape descriptors for flame quality prediction based on lstm regression | |
| CN114723159A (en) | Boiler combustion efficiency prediction method and system based on LLE dimension reduction algorithm |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant |