CN104574221B9 - A kind of photovoltaic plant running status discrimination method based on loss electricity characteristic parameter - Google Patents

Abstract

A method for identifying the operating state of a photovoltaic power station based on the characteristic parameters of the power loss. The method first uses the historical data of the energy management system (EMS) of the photovoltaic power station to calculate the power loss information of each power generation unit of the photovoltaic power station, and then establishes it through the support vector machine method. The operating state identification model of each photovoltaic power generation unit fits the mapping relationship between the characteristic parameters of the power loss information and the operating state, and finally uses these operating state identification models to identify the operating state of each photovoltaic power generation unit, and then obtains the entire photovoltaic power plant A collection of running states. The invention can reduce the monitoring workload of the photovoltaic power generation unit, help operators to discover component and system faults in time, and provide important scientific basis for the operation and maintenance of the photovoltaic power station.

Description

A photovoltaic power plant operating state identification method based on the characteristic parameters of power loss

technical field

The invention relates to a method for accurately identifying the operating state of a photovoltaic power station, which belongs to the field of power generation technology area.

Background technique

As an important part of renewable energy, photovoltaic power generation has developed rapidly in recent years, and grid-connected Photovoltaic power plants are currently the main form of utilization of photovoltaic power generation. Generally, a photovoltaic power station consists of several photovoltaic A photovoltaic power generation unit is composed of several photovoltaic modules connected in series and parallel. in actual operation In the process, in addition to being affected by external environmental factors, the power generation efficiency of photovoltaic power plants is also related to photovoltaic power generation units, The operating status of the inverter is closely related. Inverter faults, generator unit faults and unit surface Affected by factors such as dust accumulation, the actual output of photovoltaic power plants is always less than its theoretical output, which could have been The electricity delivered but not delivered is defined as the lost electricity.

The running state of the photovoltaic power station mainly refers to the set of running states of the inverter and the photovoltaic power generation unit. For the inverter of a photovoltaic power station, there are only two operating states, namely normal state and fault state. and for For the photovoltaic power generation unit of the photovoltaic power station, the operating status is generally divided into the following types: ① Normal state, that is, power generation The electrical unit is intact and the surface of the unit is clean and free of dust. ②The power generation unit fails and the surface of the unit is clean No dust accumulation. ③The power generation unit is intact but the surface of the unit is polluted and has dust accumulation. ④The power generation unit fails and The surface of the unit is contaminated with dust. Scientifically and accurately identify the status of photovoltaic power plants, not only can help power plants The personnel can detect and deal with the fault in time, prevent the fault from further spreading and cause more serious accidents, and put the power The loss of the station is minimized, and it is conducive to the stable and reliable operation of the photovoltaic power station, which is conducive to the promotion of photovoltaic The large-scale development of power generation is of great significance.

At present, most photovoltaic power plants use manual inspection to monitor scattered photovoltaic power generation units. This traditional monitoring method not only has a huge workload and poor real-time performance, it is difficult to detect system failures in time, and there is no This method ensures the stable and reliable operation of photovoltaic power plants.

Contents of the invention

The object of the present invention is to provide a method based on the characteristic parameter The operating status identification method of the photovoltaic power station can detect system faults in time, and provide a comprehensive solution for the operation and maintenance of photovoltaic power stations. Provide a scientific basis.

Problem described in the present invention is realized with following technical scheme:

A method for identifying the operating state of a photovoltaic power station based on the characteristic parameters of power loss, which is characterized in that the The above method first uses the energy management system (Energy Management System, EMS) of the photovoltaic power station The historical data is calculated to obtain the power loss information of each power generation unit of the photovoltaic power station, and then through the support vector machine (Support Vector Machine, SVM) method establishes the operating state identification model of each power generation unit, and fits The mapping relationship between the characteristic parameters of the unit’s power loss and the operating state, and finally use these operating states The identification model identifies the operating status of each power generation unit, and then obtains the operating status of the entire photovoltaic power station. State collection, described method comprises the following steps:

①Calculation of loss coefficient

a. For each photovoltaic power station, divide it into several areas according to the meteorological conditions, and install a Standard components; the installation methods, specifications, models and physical characteristics of the standard components are different from those of other groups in the same area. components are the same; in addition, power plant staff need to maintain the standard components frequently to ensure that their surfaces are clean and Work normally, so as to ensure that its output is always the theoretical output, and record its power output collection to EMS In the system; through the normalized output of the power generation unit (that is, the output of the power generation unit is converted into the standard photovoltaic Whether the ratio of the output under the same number of modules) to the output of standard photovoltaic modules in the same area is greater than a given threshold To judge whether the output of the unit is theoretical output;

The meteorological conditions refer to solar irradiance, ambient temperature, component temperature, wind speed, wind direction, air pressure, Meteorological environment data including humidity;

Among them, the judgment threshold is generally taken as 0.8, and its adjustment range is 0.7-0.9. The principle of adjustment is: if When the judgment threshold is 0.8, the theoretical output data screened is less, which cannot make the theoretical output calculation in b If the model is fully trained, the decision threshold should be appropriately reduced, otherwise, the decision threshold should be appropriately increased;

b. Obtain each power generation unit through the historical database of the EMS system, including solar irradiance, ambient temperature, Meteorological environment data, including module temperature, wind speed, wind direction, air pressure, humidity, etc., are communicated with each power generation unit output power data. Through the method described in a, the work required for theoretical output modeling is screened out from historical data rate data. With the filtered power data as the output and the meteorological environment data at the corresponding time as the input, Using statistical modeling methods to establish a mathematical mapping that reflects the relationship between input variables and output power of a power generation unit model, which is the theoretical output calculation model.

c. Taking the real-time operating condition information of photovoltaic power plants as input, including solar irradiance, ambient temperature, Component temperature, wind speed, wind direction, air pressure and humidity are calculated using the calculation model of theoretical output to obtain each The theoretical output value of each photovoltaic power generation unit at the sampling time;

d. Obtain the actual output of each photovoltaic power generation unit at each sampling moment from the EMS system and calculate the theoretical The difference between the output and the actual output value, and then respectively calculate the output value of each photovoltaic power generation unit in a specified period of time The difference between the theoretical output and the actual output value of each sampling point is multiplied by the sampling period and accumulated to obtain the The power loss of the electric unit during this time period, and at the same time, each photovoltaic power generation unit takes The theoretical output of the sample point is multiplied by the sampling period and then accumulated to obtain the output of each photovoltaic power generation unit in this period Theoretical power;

e. Divide the power loss of each photovoltaic power generation unit in the time period by the corresponding theoretical power, Obtain the loss coefficient of each photovoltaic power generation unit in this time period;

②Determine the identification parameter index set of photovoltaic module operating status

The characteristic parameters that characterize the operating state of photovoltaic power generation units include conventional characteristic parameters and custom characteristic parameters number:

a. Conventional characteristic parameters

The conventional characteristic parameters that characterize the change law of the operating state of the photovoltaic power generation unit include: photovoltaic power generation unit The maximum value, minimum value, average value, and variance of the loss coefficient within the specified time period;

b. Custom feature parameters

The self-defined characteristic parameters that characterize the operating state information of photovoltaic power generation units include correlation coefficient and discrete Difference,

The correlation coefficient r _Qq of a photovoltaic power generation unit in a specified period of time is defined as:

Among them, Q _i is the theoretical power of the i-th sampling period, that is, the product of the theoretical output of the i-th sampling point and the sampling period; q _i is the power loss of the i-th sampling period, that is, the theoretical output of the i-th sampling point The difference between the actual output value and the product of the sampling period, N is the number of sampling points;

The discrete difference LS is defined as:

③Establish a support vector machine operating state identification model

Use the variables in the identification parameter set of photovoltaic power generation unit operating status determined in step ② as support The input of the vector machine model, the operating status information of the photovoltaic power generation unit is used as the ideal support vector machine model Output, establish a support vector machine-based operating state identification model for each photovoltaic power generation unit;

④Training and verification of the support vector machine operating state identification model

According to the historical data of the operating state of the known photovoltaic power generation unit, the operating state identification parameters are calculated separately Each feature parameter in the index set, and then select a part of the data as the support vector machine identification model Type training samples to train the model, and the rest as verification data to check the identification effect of the model check;

⑤ Identify the operating status of the photovoltaic power station

For the time period Δt to be identified, calculate the operating state identification parameters of each photovoltaic power generation unit Each feature parameter in the index set, and then input the above parameter sequence into the trained support vector in step ④ machine identification model to obtain the operating status of each power generation unit within the time period, so as to obtain the A collection of operating states for a station.

In the above operating state identification method based on the characteristic parameters of power loss, the operating state of the photovoltaic power station The state also includes the running state of the inverter, and the running state of the inverter is read through the EMS system.

The above-mentioned operating state identification method based on the characteristic parameters of the power loss, in the operating state of the support vector machine During the training and verification process of the identification model and the identification process of the actual photovoltaic power plant operating status, the calculation The characteristic parameters of power loss in the obtained operating state identification parameter index set are normalized.

The above-mentioned identification method for the operating state of photovoltaic power plants based on the characteristic parameters of power loss The accuracy of the identification results of the operating status of the power station requires the calculation of the theoretical output of each power generation unit of the photovoltaic power station The model and the operating state identification model are updated online.

The present invention utilizes the characteristic parameters of the power loss to carry out online identification of the operating state of the photovoltaic power station, greatly It reduces the monitoring workload of the photovoltaic power generation unit, and can help the on-duty personnel to detect and fix system failures in time, It provides an important scientific basis for the operation and maintenance of the power station.

Description of drawings

The present invention will be described in further detail below in conjunction with the accompanying drawings.

Fig. 1 is the flow chart of the calculation steps of the power loss;

Figure 2 is a flow chart of running state identification.

The symbols in this paper are expressed as: r _Qq is the correlation coefficient; LS is the discrete difference; Q _i is the theoretical power of the i sampling period; q _i is the loss of power in the i sampling period; N is the number of sampling points; is the theoretical average value of electricity; is the average value of power loss.

detailed description

The principles and features of the present invention are described below in conjunction with the accompanying drawings, and examples are only used to explain the present invention. Invention is not intended to limit the protection scope of the present invention.

Taking 5 photovoltaic power generation units (numbered #1-#5) in a certain area in a photovoltaic power station as the research object For elephants, the historical operation data in 2011 is obtained through the EMS system, and the data sampling interval is 1 minute. Minute Don't take it as an example on September 3, 2011 10:00-14:00 Generation Unit One (#1), October 18, 2011 13:00-16:00 Generation Unit 5 (#5), Generation Unit 2 on November 7, 2011 9:13-11:00 (#2) and the actual operation status of No. The state is a test sample set.

The identification method includes the following steps:

Step 1: Obtain the output power data of standard modules in the area for the first 8 months from the EMS system. Ruoxiang At the same time, the ratio of the normalized output of other photovoltaic power generation units to the output of standard module units in the same area is greater than the specified The set judgment threshold is considered to be theoretical contribution. In this way, the power numbers required for theoretical output modeling are screened out according to. The judgment threshold is generally taken as 0.8, and its adjustment range is 0.7-0.9. The principle of adjustment is: calculation should be judged When the threshold is set to 0.8, the theoretical output data is obtained by screening. If the obtained theoretical output data is less, it cannot If the SVM theoretical output calculation model is fully trained, the judgment threshold can be appropriately reduced, and vice versa. When increasing the decision threshold.

Step 2: Use the above method to screen out the power output of 5 power generation units, and use it as the SVM Output, the irradiance, ambient temperature, relative humidity, wind speed, air pressure of each power generation unit at the corresponding time Meteorological and environmental parameters are used as the input of SVM, and the data of the previous 8 months are used as training samples to establish SVM calculation model of theoretical output.

Step 3: Calculate the power loss and theoretical power of each photovoltaic power generation unit respectively, and divide the power loss by the theoretical power respectively to obtain the loss coefficients λ ₁ , λ ₂ ,...λ _n of each photovoltaic power generation unit.

Step 4: Select the appropriate characteristic parameters of power loss according to the actual situation of the photovoltaic power plant, and determine Running state identification parameter index set. Here select {max value of loss coefficient, average value of loss coefficient, loss The minimum value of the loss coefficient, the variance of the loss coefficient, the correlation coefficient, and the discrete difference}, a total of 6 dimensional variables, as the operating state State identification parameter index set.

Step 5: Take the variables in the parameter index set of operating state identification as the input of the support vector machine model, The operating state information with complete information is used as the ideal output of the model, and the operating state identification of support vector machine is established Model. The operating states that need to be identified are mainly four types of operating states of photovoltaic modules (①normal state, namely The electrical unit is intact and the surface of the unit is clean and free of dust. ②The power generation unit fails and the surface of the unit is clean No dust accumulation. ③The power generation unit is intact but the surface of the unit is polluted and has dust accumulation. ④The power generation unit fails and The surface of the unit is contaminated with dust). When used as the output of the identification model, use integers 1, 2, 3, and 4 to represent table them.

Step 6: Use the historical data with complete operating information in the first 8 months as the support vector machine identification model training samples to train the model. Separately calculate the running status information and complete historical data corresponding to the running status The characteristic parameters of each power loss in the state identification parameter index set are normalized to support Training of the vector machine identification model.

Step 7: Read the operation of the inverter of power generation unit No. 1 during the time period of 10:00-14:00 on September 3 Status, the result shows normal. At the same time, calculate the operating state type identification of No. 1 photovoltaic module in this period of time. The characteristic parameters of each power loss in the knowledge parameter index set, the actual calculation results: #1-#5 these 5 power generation units The loss coefficients are 0.78, 0.14, 0.11, 0.08, 0.16 respectively, the maximum value of the loss coefficient is 0.78, and the minimum The value is 0.08, the average value is 0.254, the variance is 0.0874, and the correlation coefficients of each power generation unit are 0.26, 0.79, 0.82, 0.88, 0.83, the discrete difference is 3.55, 11.49, 18.98, 15.65, 16.44 respectively. Enter After performing normalization processing, input it into the support vector machine identification model, and the SVM outputs 2, which is converted into the corresponding time period The operating status of the internal photovoltaic power generation unit is ② the power generation unit fails and the surface of the unit is clean and free of dust. After the actual inspection and test, it was found that the power generation unit was indeed faulty and the surface was in a clean state, and the identification result was correct. indeed.

Step 8: Calculate other samples in the test sample set separately in the same way

For power generation unit No. 5 from 13:00 to 16:00 on October 18, 2011, the inverter was in normal state, The loss coefficients of the 5 generating units #1-#5 are 0.1, 0.12, 0.11, 0.15, 0.12 respectively. The maximum value is 0.15, the minimum value is 0.1, the average value is 0.12, and the variance is 0.00035. Each power generation unit The correlation coefficients are 0.85, 0.82, 0.82, 0.88, 0.83, and the dispersion differences are 12.55, 13.69, 15.98, 18.65, 16.44. After normalization processing, it is input into the support vector machine identification model, and the SVM input Out of 1, it is converted into the operating state of the photovoltaic power generation unit in the corresponding period of time is ① normal state, that is, the power generation unit It is intact and the surface of the unit is clean and free of dust. After actual testing, the power generation unit is indeed in a normal state. The recognition result is correct.

For power generation unit No. 2 from 9:13 to 11:00 on November 7, 2011, the inverter was in a fault state, The photovoltaic power generation unit is in a normal state, and the identification result of the test is correct.

For generating unit No. 4 (#4) from 12:00 to 14:00 on December 24, 2011, the inverter was in positive In the normal state, the loss coefficients of the five generating units #1-#5 are 0.45, 0.4, 0.42, 0.46, and 0.39 respectively. The maximum value of the loss coefficient is 0.46, the minimum value is 0.39, the average value is 0.424, and the variance is 0.00093. The correlation coefficients of each power generation unit are 0.65, 0.69, 0.67, 0.62, 0.7, and the dispersion difference is 6.66, 8.89, 7.58, 6.32, 9.56. After normalization processing, it is input into the support vector machine identification model, SVM Output 3, converted to the operating state of the photovoltaic power generation unit in the corresponding time period is ③The power generation unit is intact However, the surface of the unit is contaminated with dust, and the identification result is correct after actual testing.

The above specific embodiments are only preferred embodiments of the present invention, and are not intended to limit the present invention. appoint Any changes and replacement methods that can be easily imagined by those skilled in the art should be covered by the scope of the present invention. within the scope of protection.

Claims (4)

1. A photovoltaic power plant operating state identification method based on the characteristic parameters of the loss of electricity, characterized in that, The method first uses the historical data of the energy management system EMS of the photovoltaic power station to calculate and obtain the The power loss information of the power unit, and then establish the operating status of each power generation unit through the support vector machine method Identify the model, fit the mapping relationship between the characteristic parameters of the unit’s power loss and the operating state, and finally use Use these operating state identification models to identify the operating state of each power generation unit, and then get the entire light A set of operating states of a volt power station, the method may further comprise the steps of: ①Calculation of loss coefficient a. For each photovoltaic power station, divide it into several areas according to the meteorological conditions, and install a Standard components, and record their power output collection into the EMS system; through the normalization of each power generation unit The output of the unit is judged by whether the ratio of the output of the photovoltaic module to the output of the standard photovoltaic module in the same area is greater than the set threshold. Whether the force is a theoretical effort; b. Obtain each power generation unit through the historical database of the EMS system, including solar irradiance, ambient temperature, Meteorological environmental data including module temperature, wind speed, wind direction, air pressure, and humidity, and the output of each power generation unit Power data; through the method described in step a, filter out the theoretical output required for modeling the theoretical output from the historical data Power data; the filtered power data is used as the output, and the meteorological environment data at the corresponding time is used as the output input, using the statistical modeling method to establish a map that reflects the relationship between the input variables and output power of the power generation unit Mathematical model, which is the theoretical output calculation model; c. Taking the real-time operating condition information of photovoltaic power plants as input, and using the calculation model of theoretical output to calculate Obtain the theoretical output value of each photovoltaic power generation unit at each sampling moment; d. Obtain the actual output of each photovoltaic power generation unit at each sampling moment from the EMS system and calculate the theoretical The difference between the output and the actual output value, and then respectively calculate the output value of each photovoltaic power generation unit in a specified period of time The difference between the theoretical output and the actual output value of each sampling point is multiplied by the sampling period and accumulated to obtain the The power loss of the electric unit in this time period, and each photovoltaic power generation unit is sampled in this time period The theoretical output of each point is multiplied by the sampling period and accumulated to obtain the theoretical output of each photovoltaic power generation unit in this period of time. On electricity; e. Divide the power loss of each photovoltaic power generation unit in the time period by the corresponding theoretical power, Obtain the loss coefficient of each photovoltaic power generation unit in this time period; ②Determine the identification parameter index set of the operating state of the photovoltaic power generation unit The characteristic parameters that characterize the operating state of photovoltaic power generation units include conventional characteristic parameters and custom characteristic parameters number: a. Conventional characteristic parameters The conventional characteristic parameters that characterize the change law of the operating state of the photovoltaic power generation unit include: photovoltaic power generation unit The maximum value, minimum value, average value, variance and cumulative value of the loss coefficient within the specified time period; b. Define characteristic parameters The self-defined characteristic parameters that characterize the operating state information of photovoltaic power generation units include correlation coefficient and discrete Difference, The correlation coefficient r _Qq of a photovoltaic power generation unit in a specified period of time is defined as: <mrow><msub><mi>r</mi><mrow><mi>Q</mi><mi>q</mi></mrow></msub><mo>=</mo><mfrac><mrow><mfrac><mn>1</mn><mrow><mi>N</mi><mo>-</mo><mn>1</mn></mrow></mfrac><munderover><mo>&amp;Sigma;</mo><mrow><mi>i</mi><mo>=</mo><mn>1</mn></mrow><mi>N</mi></munderover><mrow><mo>(</mo><msub><mi>Q</mi><mi>i</mi></msub><mo>-</mo><mover><mi>Q</mi><mo>&amp;OverBar;</mo></mover><mo>)</mo></mrow><mrow><mo>(</mo><msub><mi>q</mi><mi>i</mi></msub><mo>-</mo><mover><mi>q</mi><mo>&amp;OverBar;</msub>mo></mover><mo>)</mo></mrow></mrow><mrow><msqrt><mrow><mfrac><mn>1</mn><mrow><mi>N</mi><mo>-</mo><mn>1</mn></mrow></mfrac><munderover><mo>&amp;Sigma;</mo><mrow><mi>i</mo>mi><mo>=</mo><mn>1</mn></mrow><mi>N</mi></munderover><msup><mrow><mo>(</mo><msub><mi>Q</mi><mi>i</mi></msub><mo>-</mo><mover><mi>Q</mi><mo>&amp;OverBar;</mo></mover><mo>)</mo></mrow><mn>2</mn></msup></mrow></msqrt><msqrt><mrow><mfrac><mn>1</mn><mrow><mi>N</mi><mo>-</mo><mn>1</mn></mrow></mfrac><munderover><mo>&amp;Sigma;</mo><mrow><mi>i</mi><mo>=</mo><mn>1</mn></mrow><mi>N</mi></munderover><msup><mrow><mo>(</mo><msub><mi>q</mi><mi>i</mi></msub><mo>-</mo><mover><mi>q</mi><mo>&amp;OverBar;</mo></mover><mo>)</mo></mrow><mn>2</mn></msup></mrow></msqrt></mrow></mfrac><mo>,</mo></mrow> Among them, Q _i is the theoretical power of the i-th sampling period, that is, the product of the theoretical output of the i-th sampling point and the sampling period; q _i is the power loss of the i-th sampling period, that is, the theoretical output of the i-th sampling point The difference between the actual output value and the product of the sampling period, N is the number of sampling points; is the theoretical average value of electricity; is the average value of power loss; The discrete difference LS is defined as: <mrow><mi>L</mi><mi>S</mi><mo>=</mo><msqrt><mrow><mfrac><mn>1</mn><mi>N</mi></mfrac><munderover><mo>&amp;Sigma;</mo><mrow><mi>i</mi><mo>=</mo><mn>1</mn></mrow><mi>N</mi></munderover><msup><mrow><mo>(</mo><msub><mi>Q</mi><mi>i</mi></msub><mo>-</mo><msub><mi>q</mi><mi>i</mi></msub><mo>)</mo></mrow><mn>2</mn></msup></mrow></msqrt><mo>,</mo></mrow> ③Establish a support vector machine operating state identification model Use the variables in the identification parameter set of photovoltaic power generation unit operating status determined in step ② as support The input of the vector machine model, the operating status information of the photovoltaic power generation unit is used as the ideal support vector machine model Output, establish a support vector machine-based operating state identification model for each photovoltaic power generation unit; ④Training and verification of the support vector machine operating state identification model According to the historical data of the operating state of known photovoltaic photovoltaic power generation units, the operating state identification is calculated separately Each feature parameter in the parameter index set, and then select a part of the data as the support vector machine to identify Identify the training samples of the model, train the model, and use the rest as verification data to determine the recognition effect of the model to verify; ⑤ Identify the operating state set of photovoltaic power plants For the time period Δt to be identified, calculate the operating state identification parameters of each photovoltaic power generation unit Each feature parameter in the index set, and then input the above parameter sequence into the trained support vector in step ④ machine identification model to obtain the operating status of each power generation unit within the time period, so as to obtain the A collection of operating states for a station. 2. The operating state identification of photovoltaic power plants based on the characteristic parameters of power loss according to claim 1 The method is characterized in that the operating state of the photovoltaic power plant also includes the operating state of the inverter, and the inverter The operating status of the machine is read through the EMS system. 3. According to claim 3, the identification of the operating state of the photovoltaic power plant based on the characteristic parameters of the power loss The method is characterized in that the training and verification process of the support vector machine running state identification model and the actual In the process of operating state identification of photovoltaic power plants, it is necessary to deal with the concentration of the calculated operating state identification parameter indicators. The characteristic parameters of power loss are normalized. 4. According to claim 4, the operating state identification of photovoltaic power plants based on the characteristic parameters of power loss The method is characterized in that, in order to improve the accuracy of the identification results of the operating state of the photovoltaic power station, it is necessary to The theoretical output calculation model and operating state identification model of each power generation unit in the station are updated online.