CN117784849B - Automatic control system of refrigeration constant temperature tank based on artificial intelligence - Google Patents
Automatic control system of refrigeration constant temperature tank based on artificial intelligence Download PDFInfo
- Publication number
- CN117784849B CN117784849B CN202410212561.1A CN202410212561A CN117784849B CN 117784849 B CN117784849 B CN 117784849B CN 202410212561 A CN202410212561 A CN 202410212561A CN 117784849 B CN117784849 B CN 117784849B
- Authority
- CN
- China
- Prior art keywords
- temperature
- unit
- module
- network
- value
- 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
- 238000005057 refrigeration Methods 0.000 title claims abstract description 35
- 238000013473 artificial intelligence Methods 0.000 title claims abstract description 12
- 238000013528 artificial neural network Methods 0.000 claims abstract description 52
- 230000009471 action Effects 0.000 claims abstract description 48
- 238000000354 decomposition reaction Methods 0.000 claims abstract description 24
- 238000012545 processing Methods 0.000 claims abstract description 18
- 238000012544 monitoring process Methods 0.000 claims abstract description 16
- 238000005457 optimization Methods 0.000 claims abstract description 10
- 238000006243 chemical reaction Methods 0.000 claims abstract description 8
- 230000006870 function Effects 0.000 claims description 99
- 238000012549 training Methods 0.000 claims description 25
- 230000002159 abnormal effect Effects 0.000 claims description 16
- 238000004364 calculation method Methods 0.000 claims description 16
- 238000000034 method Methods 0.000 claims description 14
- 238000000605 extraction Methods 0.000 claims description 9
- 230000008569 process Effects 0.000 claims description 8
- 230000007704 transition Effects 0.000 claims description 8
- 238000010276 construction Methods 0.000 claims description 7
- 210000002569 neuron Anatomy 0.000 claims description 6
- 238000007781 pre-processing Methods 0.000 claims description 5
- 238000013461 design Methods 0.000 claims description 4
- 238000010438 heat treatment Methods 0.000 claims description 4
- 238000010606 normalization Methods 0.000 claims description 4
- ORILYTVJVMAKLC-UHFFFAOYSA-N Adamantane Natural products C1C(C2)CC3CC1CC2C3 ORILYTVJVMAKLC-UHFFFAOYSA-N 0.000 claims description 3
- 238000002474 experimental method Methods 0.000 claims description 3
- 238000003062 neural network model Methods 0.000 claims description 3
- 230000004044 response Effects 0.000 claims description 3
- 230000035945 sensitivity Effects 0.000 claims description 3
- 239000012141 concentrate Substances 0.000 claims 1
- 230000008859 change Effects 0.000 description 11
- 230000000875 corresponding effect Effects 0.000 description 6
- 230000001276 controlling effect Effects 0.000 description 5
- 238000010586 diagram Methods 0.000 description 4
- 238000001514 detection method Methods 0.000 description 2
- 238000002759 z-score normalization Methods 0.000 description 2
- 230000004075 alteration Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000013480 data collection Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
Landscapes
- Devices That Are Associated With Refrigeration Equipment (AREA)
- Feedback Control In General (AREA)
Abstract
The invention discloses an automatic control system of a refrigeration constant temperature tank based on artificial intelligence, which comprises a temperature acquisition module, a temperature prediction module, a user interface setting module, an executor module, an optimization system module and a system monitoring module. The invention belongs to the technical field of data processing, in particular to an automatic control system of a refrigeration constant temperature tank based on artificial intelligence, which converts a residual signal between a predicted temperature and a measured temperature into a continuous time sequence, carries out multiple decomposition to obtain an intrinsic mode function component and a residual item, carries out mode aliasing processing, and improves the accuracy and the stability of temperature acquisition; constructing a temperature prediction model, and minimizing error adjustment connection weights between an actual temperature value and a neural network predicted value; initializing neural network parameters, performing action selection and noise addition, performing state conversion storage, calculating a loss function and a strategy gradient, updating a target network by using a soft updating mode, and improving system stability.
Description
Technical Field
The invention belongs to the technical field of data processing, and particularly relates to an automatic control system of a refrigeration constant temperature tank based on artificial intelligence.
Background
The automatic control system for the refrigerating thermostatic bath is a system for realizing automatic regulation and control of the refrigerating thermostatic bath through components such as a sensor, a controller and an actuator, wherein the sensor senses temperature change in the bath, the controller is used for data processing and decision making, the actuator is used for controlling the working state of a refrigerating and heating device, the temperature in the refrigerating thermostatic bath is kept constant within a preset range, the aim is to improve the working efficiency and stability, reduce manual intervention and accurately control the temperature. However, the existing automatic control system for the refrigeration constant temperature tank has the technical problems that temperature data acquisition is inaccurate, the control system cannot detect actual temperature change, and temperature fluctuation in the constant temperature tank is caused; the technical problems that the temperature prediction is inaccurate, so that the processing temperature deviates from the requirement and the product quality is affected exist; when the temperature changes dynamically, the system is difficult to adjust the decision in time according to the temperature changes.
Disclosure of Invention
Aiming at the technical problems that the temperature data in the constant temperature tank is not accurately acquired, the control system cannot detect the change of the actual temperature, and the fluctuation of the temperature in the constant temperature tank is caused, the temperature data in the constant temperature tank is periodically acquired, the data standardization is carried out, the residual signals between the predicted temperature and the measured temperature are converted into continuous time sequences, the multiple decomposition is carried out to obtain intrinsic mode function components and residual terms, the mode aliasing processing is carried out, the characteristic information in the signals is revealed, the influence of noise and interference on the accuracy of the data is reduced, and the accuracy and the stability of the temperature acquisition are improved; aiming at the technical problems that the temperature prediction is inaccurate, the processing temperature deviates from the requirement, and the product quality is affected, a temperature prediction model is built, a radial basis function neural network is built by using a Gaussian kernel function, a connection weight coefficient is initialized, the connection weight is adjusted by minimizing the error between an actual temperature value and a neural network prediction value, and the model is trained repeatedly to predict the actual temperature; aiming at the technical problem that when the temperature dynamic change exists, the system is difficult to adjust decisions in time according to the temperature change, the initialized neural network parameters are adopted, action selection and noise addition are carried out, state transition storage is carried out, a loss function and strategy gradient are calculated, the system is optimized, a target network is updated in a soft update mode, the system stability is improved, and decisions are adjusted in time according to the temperature change.
The invention provides an artificial intelligence-based automatic control system for a refrigeration constant temperature tank, which comprises a temperature acquisition module, a temperature prediction module, a user interface setting module, an executor module, an optimization system module and a system monitoring module; the temperature acquisition module is used for periodically acquiring temperature data in the constant temperature tank, carrying out data standardization, converting residual signals between the predicted temperature and the measured temperature into a continuous time sequence, carrying out multiple decomposition to obtain an intrinsic mode function component and a residual error item, and carrying out abnormal feature extraction and mode aliasing treatment; the temperature prediction module is specifically used for constructing a temperature prediction model, constructing a radial basis function neural network by using a Gaussian kernel function, initializing a connection weight coefficient, adjusting the connection weight by minimizing an error between an actual temperature value and a neural network predicted value, and repeating training the model to predict the actual temperature; the user interface setting module is specifically used for designing a user interface and carrying out mode selection, parameter adjustment and fault prompt; the executor module receives the real-time temperature data and the control signal to execute corresponding actions and feeds the actions back to the controller; the optimization system module is used for initializing neural network parameters, performing action selection and noise addition, performing state transition storage, calculating a loss function and a strategy gradient, and updating a target network by using a soft updating mode; the system monitoring module is specifically a remote monitoring system and is used for early warning faults.
Further, in the temperature acquisition module, a temperature data acquisition unit, a data standardization unit, an abnormal data characteristic extraction unit and a mode aliasing processing unit are provided, and the temperature acquisition module comprises the following contents:
The temperature data acquisition unit is used for periodically acquiring temperature data in the refrigeration constant temperature tank by using a temperature sensor and carrying out data preprocessing, including removing noise, filling missing values and improving the accuracy of abnormal temperature data detection;
The data normalization unit is used for performing Z-score normalization preprocessing, the data only changes in the direction of the data value without changing the size relation of the data, the influence of the value of the data on network training and feature selection is eliminated, the network learning speed is accelerated, the training weight is accurately distributed, and the formula is as follows: ; wherein x normal represents a standardized data value, x represents a sample value, namely temperature data in the refrigeration constant temperature tank, u represents a sample mean value, and delta represents a sample standard deviation, namely an average value of deviations between all sample values and the sample mean value;
the abnormal data feature extraction unit converts the residual signals between the predicted temperature and the measured temperature into a continuous time sequence, decomposes the temperature abnormal result layer by layer, generates a group of eigenvalue function components and a residual item by decomposition each time, takes the selected decomposed signals as new original signals, decomposes again, and iterates repeatedly until a satisfactory decomposition result is achieved, and better observes and understands the nature of abnormal temperature data; after the signal is decomposed, the value of the signal is calculated, and the formula is as follows: ; wherein x (t) represents the value of the signal, iz represents the index of the local extremum points, nz represents the total number of the local extremum points, c iz (t) represents the value of the envelope curve of the local extremum points, and r nz (t) represents the residual error term which contains the detail information and noise which cannot be described by the envelope curve in the signal;
The mode aliasing processing unit, the mode aliasing means that a certain eigenmode function component has similar characteristics with other components, can not fully reveal the characteristic information of the signal, adds white noise with even distribution of amplitude into the original signal, makes the signal concentrated and continuous, weakens the influence of instantaneous pulse on signal decomposition, reveals the characteristic information in the signal, reduces the influence of noise and interference on data accuracy, improves the accuracy and stability of temperature acquisition, and comprises the following contents: adding a random white noise signal to x (t) uses the formula: ; in the/> Is a signal added with random white noise, and n (t) is a random white noise signal; the signal after adding random white noise is decomposed into a series of eigenmode function components using the formula: ; in the/> Is the ith eigenmode function component obtained by the jth decomposition,/>Is the residual term of the jth decomposition, n1 is the total number of eigenmode function components, i is the index of eigenmode function components; different random white noise signals are added to x (t) to obtain a plurality of groups of eigenvalue function components and residual terms, and the total average value of the corresponding components is calculated as a final result, wherein the formula is as follows: /(I); Where c i represents the final result of the ith eigenmode function component and M represents the number of repetitions of the decomposition step.
Further, in the temperature prediction module, a temperature prediction model construction unit, an initialization unit, a forward propagation unit, an output calculation unit, a backward propagation unit and a training and application unit are provided, and the temperature prediction module comprises the following contents:
The temperature prediction model construction unit is used for constructing a radial basis function neural network by using a Gaussian kernel function, and determining the node number in the hidden layer, the center point and the width of each kernel function; the initialization unit randomly initializes the connection weight coefficient when constructing the neural network;
The forward propagation unit inputs the training sample into the neural network, calculates the hidden unit output of the radial basis function neural network, and uses the formula: ; where φ is the radial basis function, x is the training sample, c a is the a-th center point of the kernel, x p represents the p-th input sample,/> The width of the a-th center point of the kernel function is the width, the width of the center point of the kernel function controls the radial range of the radial basis function and is an important factor affecting the performance of the RBF neural network, when the width is too small, the boundary becomes fuzzy, the classification precision is reduced, the width is too large, and the coverage area of the basis function becomes relatively small, so that the generalization capability of the network is reduced;
The output calculation unit calculates the output of the neural network, and the formula is as follows: ; wherein y a represents the output of the RBF neural network, w cb represents the connection weight coefficient between the hidden layer neuron c and the output layer neuron b, and h represents the node number in the hidden layer;
The back propagation unit is used for updating the connection weight coefficient by using an error back propagation algorithm and minimizing the error between the actual temperature value and the temperature predicted value; and the training and application unit is used for repeating the training process until the prediction error of the neural network reaches a satisfactory degree, applying the trained neural network model to a temperature prediction task, inputting new temperature data, predicting through the neural network, evaluating and adjusting the prediction result, and improving the temperature prediction accuracy.
Further, in the user interface setting module, there are provided a design user interface unit, a mode selection unit, and a parameter adjustment unit, the user interface setting module including:
Designing a user interface unit, providing an intuitively friendly interface, and enabling a user to input a set temperature through the interface and check real-time temperature data; the mode selection unit is used for selecting different working modes by a user to meet the actual requirements of experiments, wherein the working modes comprise a constant temperature mode, a quick refrigeration mode and a quick heating mode; and the parameter adjusting unit is used for monitoring the working state of the refrigeration constant temperature tank and comprises a current temperature, a set temperature and a working time length, and adjusting the temperature tolerance, the response time and the control sensitivity of the refrigeration constant temperature tank.
Further, in the executor module, the real-time temperature data is transmitted to the executor module, the executor module executes corresponding operations including adjusting the working state and controlling the temperature of the refrigerating system, and the executor module feeds back the execution result to the controller, so that the controller dynamically adjusts according to the actual situation, and the internal temperature of the refrigerating thermostatic bath is ensured to be always maintained within the set range.
Further, in the optimizing system module, there are provided an action selection and noise adding unit, a state transition storage unit, a loss function calculating unit, a strategy gradient calculating unit and a target network parameter updating unit, the optimizing system module includes the following contents:
The action selecting and noise adding unit initializes the parameters of the neural network, selects an action according to the action strategy, adds noise N t into the action output by the strategy network, increases the exploration degree, transmits the exploration degree to the environment to execute the action a t, and calculates the executed action, wherein the formula is as follows: ; where a t denotes the action performed, s t denotes the current state, Representing the action of the policy network output, N t representing the added noise, θ μ representing the parameters of the policy network;
The state conversion storage unit returns rewards r t and new states s t+1 after the environment execution a t is finished, and the state conversion (s t,at,rt,st+1) stored by the intelligent agent is put into the playback memory to be used as a training set of the online neural network;
loss function calculation unit for updating parameters of Q-value network using Adam optimizer The loss function is calculated using the formula:
；
；
Where L represents the loss function, N represents the number of samples, i.e., the total number of samples used to calculate the loss function, t represents the index of the samples, y t represents the target value, Q represents the online Q-value function, Parameters representing Q-value network, r t representing instant rewards, gamma representing discount factors,/>Representing a target Q function, s t+1 representing the next state,/>Parameters representing the target policy network,/>Parameters representing a target Q-value network;
the strategy gradient calculation unit calculates the strategy gradient of the strategy network, and the formula is as follows: ; in the/> Is a policy gradient, representing a parameter/>, related to a policyTarget function/>N is the number of samples, s represents the state, a1 represents the action,Is a gradient of the Q value function related to action a1,/>Is the gradient of the policy network, s t represents the state of sample t,/>Is an action generated from policies and states;
The target network parameter updating unit is used for optimizing the target network parameter by using a soft updating mode and is used for smoothly updating the target network parameter, and the formula is as follows: ; wherein, tau is a soft update parameter, the value range is 0 to 1, and the weight of the original network parameter and the target network parameter is balanced by controlling the size of the soft update parameter tau, so that the update of the target network is more gradual and stable,/> Parameters representing Q-value network,/>Parameters representing policy networks,/>Parameters representing the target policy network,/>Parameters representing a target Q-value network.
Further, in the system monitoring module, data of the automatic control system of the refrigeration constant temperature tank are monitored in real time, and the system is monitored remotely and early-warned of faults, so that normal operation of the system is ensured.
By adopting the scheme, the beneficial effects obtained by the invention are as follows:
(1) Aiming at the technical problems that temperature data acquisition is inaccurate, a control system cannot detect actual temperature change and temperature fluctuation in a constant temperature tank is caused, temperature data in the constant temperature tank are periodically acquired, data standardization is carried out, residual signals between the predicted temperature and the measured temperature are converted into continuous time sequences, multiple decomposition is carried out to obtain eigenmode function components and residual terms, mode aliasing processing is carried out, characteristic information in the signals is revealed, influence of noise and interference on data accuracy is reduced, and accuracy and stability of temperature acquisition are improved;
(2) Aiming at the technical problems that the temperature prediction is inaccurate, the processing temperature deviates from the requirement, and the product quality is affected, a temperature prediction model is built, a radial basis function neural network is built by using a Gaussian kernel function, a connection weight coefficient is initialized, the connection weight is adjusted by minimizing the error between an actual temperature value and a neural network prediction value, and the model is trained repeatedly to predict the actual temperature;
(3) Aiming at the technical problem that when the temperature dynamic change exists, the system is difficult to adjust decisions in time according to the temperature change, the initialized neural network parameters are adopted, action selection and noise addition are carried out, state transition storage is carried out, a loss function and strategy gradient are calculated, the system is optimized, a target network is updated in a soft update mode, the system stability is improved, and decisions are adjusted in time according to the temperature change.
Drawings
FIG. 1 is a schematic diagram of an automatic control system of a refrigeration thermostat based on artificial intelligence;
FIG. 2 is a schematic diagram of a temperature acquisition module;
FIG. 3 is a schematic diagram of a temperature prediction module;
FIG. 4 is a schematic diagram of an optimization system module.
The accompanying drawings are included to provide a further understanding of the invention and are incorporated in and constitute a part of this specification, illustrate the invention and together with the embodiments of the invention, serve to explain the invention.
Detailed Description
The following description of the embodiments of the present invention will be made clearly and fully with reference to the accompanying drawings, in which it is evident that the embodiments described are only some, but not all embodiments of the invention; all other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention.
In the description of the present invention, it should be understood that the terms "upper," "lower," "front," "rear," "left," "right," "top," "bottom," "inner," "outer," and the like indicate orientation or positional relationships based on those shown in the drawings, merely to facilitate description of the invention and simplify the description, and do not indicate or imply that the devices or elements referred to must have a particular orientation, be constructed and operated in a particular orientation, and thus should not be construed as limiting the invention.
Referring to fig. 1, the automatic control system of the refrigeration thermostat provided by the invention comprises a temperature acquisition module, a temperature prediction module, a user interface setting module, an executor module, an optimization system module and a system monitoring module; the temperature acquisition module is used for periodically acquiring temperature data in the constant temperature tank, carrying out data standardization, converting residual signals between the predicted temperature and the measured temperature into a continuous time sequence, carrying out multiple decomposition to obtain an intrinsic mode function component and a residual error item, and carrying out abnormal feature extraction and mode aliasing treatment; the temperature prediction module is specifically used for constructing a temperature prediction model, constructing a radial basis function neural network by using a Gaussian kernel function, initializing a connection weight coefficient, adjusting the connection weight by minimizing an error between an actual temperature value and a neural network predicted value, and repeating training the model to be applied to a temperature prediction task; the user interface setting module is specifically used for designing a user interface and carrying out mode selection, parameter adjustment and fault prompt; the executor module receives the real-time temperature data and the control signal to execute corresponding actions and feeds the actions back to the controller; the optimization system module is used for initializing neural network parameters, performing action selection and noise addition, performing state transition storage, calculating a loss function and a strategy gradient, and updating a target network by using a soft updating mode; the system monitoring module is specifically used for monitoring data of the automatic control system of the refrigeration constant temperature tank in real time, diagnosing faults and carrying out early warning, and evaluating system performance.
In a second embodiment, referring to fig. 1 and2, the embodiment is based on the above embodiment, in a temperature acquisition module, there are provided a temperature data acquisition unit, a data normalization unit, an abnormal data feature extraction unit, and a mode aliasing processing unit, where the temperature acquisition module includes the following contents:
The temperature data acquisition unit is used for periodically acquiring temperature data in the refrigeration constant temperature tank by using a temperature sensor and carrying out data preprocessing, including removing noise, filling missing values and improving the accuracy of abnormal temperature data detection;
The data normalization unit is used for performing Z-score normalization preprocessing, the data only changes in the direction of the data value without changing the size relation of the data, the influence of the value of the data on network training and feature selection is eliminated, the network learning speed is accelerated, the training weight is accurately distributed, and the formula is as follows: ; wherein x normal represents a standardized data value, x represents a sample value, namely temperature data in the refrigeration constant temperature tank, u represents a sample mean value, and delta represents a sample standard deviation, namely an average value of deviations between all sample values and the sample mean value;
the abnormal data feature extraction unit converts the residual signals between the predicted temperature and the measured temperature into a continuous time sequence, decomposes the temperature abnormal result layer by layer, generates a group of eigenvalue function components and a residual item by decomposition each time, takes the selected decomposed signals as new original signals, decomposes again, and iterates repeatedly until a satisfactory decomposition result is achieved, and better observes and understands the nature of abnormal temperature data; after the signal is decomposed, the value of the signal is calculated, and the formula is as follows: ; wherein x (t) represents the value of the signal, iz represents the index of the local extremum points, nz represents the total number of the local extremum points, c iz (t) represents the value of the envelope curve of the local extremum points, and r nz (t) represents the residual error term which contains the detail information and noise which cannot be described by the envelope curve in the signal;
The mode aliasing processing unit, the mode aliasing means that a certain eigenmode function component has similar characteristics with other components, can not fully reveal the characteristic information of the signal, adds white noise with even distribution of amplitude into the original signal, makes the signal concentrated and continuous, weakens the influence of instantaneous pulse on signal decomposition, reveals the characteristic information in the signal, reduces the influence of noise and interference on data accuracy, improves the accuracy and stability of temperature acquisition, and comprises the following contents: adding a random white noise signal to x (t) uses the formula: ; in the/> Is a signal added with random white noise, and n (t) is a random white noise signal; the signal after adding random white noise is decomposed into a series of eigenmode function components using the formula: ; in the/> Is the ith eigenmode function component obtained by the jth decomposition,/>Is the residual term of the jth decomposition, n1 is the total number of eigenmode function components, i is the index of eigenmode function components; different random white noise signals are added to x (t) to obtain a plurality of groups of eigenvalue function components and residual terms, and the total average value of the corresponding components is calculated as a final result, wherein the formula is as follows: /(I); Where c i represents the final result of the ith eigenmode function component and M represents the number of repetitions of the decomposition step.
Through executing the operation, temperature data inside the constant temperature tank are collected regularly, data standardization is carried out, residual signals between the predicted temperature and the measured temperature are converted into continuous time sequences, the continuous time sequences are decomposed for multiple times to obtain intrinsic mode function components and residual error items, mode aliasing processing is carried out, characteristic information in the signals is revealed, influence of noise and interference on data accuracy is reduced, accuracy and stability of temperature collection are improved, the technical problem that temperature data collection is inaccurate, a control system cannot perceive change of actual temperature, and fluctuation of temperature in the constant temperature tank is caused is solved.
An embodiment III, referring to FIG. 1 and FIG. 3, is based on the above embodiment, in a temperature prediction module, there is provided a temperature prediction model construction unit, an initialization unit, a forward propagation unit, an output calculation unit, a backward propagation unit, and a training and application unit, the temperature prediction module includes:
The temperature prediction model construction unit is used for constructing a radial basis function neural network by using a Gaussian kernel function, and determining the node number in the hidden layer, the center point and the width of each kernel function; the initialization unit randomly initializes the connection weight coefficient when constructing the neural network;
The forward propagation unit inputs the training sample into the neural network, calculates the hidden unit output of the radial basis function neural network, and uses the formula: ; where φ is the radial basis function, x is the training sample, c a is the a-th center point of the kernel, x p represents the p-th input sample,/> The width of the a-th center point of the kernel function is the width, the width of the center point of the kernel function controls the radial range of the radial basis function and is an important factor affecting the performance of the RBF neural network, when the width is too small, the boundary becomes fuzzy, the classification precision is reduced, the width is too large, and the coverage area of the basis function becomes relatively small, so that the generalization capability of the network is reduced;
The output calculation unit calculates the output of the neural network, and the formula is as follows: ; wherein y a represents the output of the RBF neural network, w cb represents the connection weight coefficient between the hidden layer neuron c and the output layer neuron b, and h represents the node number in the hidden layer;
The back propagation unit is used for updating the connection weight coefficient by using an error back propagation algorithm and minimizing the error between the actual temperature value and the temperature predicted value; and the training and application unit is used for repeating the training process until the prediction error of the neural network reaches a satisfactory degree, applying the trained neural network model to a temperature prediction task, inputting new temperature data, predicting through the neural network, evaluating and adjusting the prediction result, and improving the temperature prediction accuracy.
By executing the operations, a temperature prediction model is built, a radial basis function neural network is built by using a Gaussian kernel function, a connection weight coefficient is initialized, the connection weight is adjusted by minimizing the error between an actual temperature value and a neural network predicted value, and the model is trained repeatedly to predict the actual temperature, so that the technical problems that the processing temperature deviates from the requirement and the product quality is affected due to inaccurate temperature prediction are solved.
Fourth embodiment, referring to fig. 1, the embodiment is based on the above embodiment, and in a user interface setting module, there are provided a design user interface unit, a mode selection unit, and a parameter adjustment unit, the user interface setting module includes:
Designing a user interface unit, providing an intuitively friendly interface, and enabling a user to input a set temperature through the interface and check real-time temperature data; the mode selection unit is used for selecting different working modes by a user to meet the actual requirements of experiments, wherein the working modes comprise a constant temperature mode, a quick refrigeration mode and a quick heating mode; and the parameter adjusting unit is used for monitoring the working state of the refrigeration constant temperature tank and comprises a current temperature, a set temperature and a working time length, and adjusting the temperature tolerance, the response time and the control sensitivity of the refrigeration constant temperature tank.
In the fifth embodiment, referring to fig. 1, the embodiment is based on the above embodiment, in the actuator module, real-time temperature data is transmitted to the actuator module, and the actuator module performs corresponding operations, including adjusting the working state of the refrigeration system and controlling the temperature, and the actuator module feeds back the execution result to the controller, so that the controller dynamically adjusts according to the actual situation, and the internal temperature of the refrigeration thermostat is ensured to be always maintained within the set range.
An embodiment six, referring to fig. 1 and fig. 4, is based on the above embodiment, and in an optimization system module, there are provided an action selection and noise adding unit, a state transition storage unit, a loss function calculation unit, a policy gradient calculation unit, and a target network parameter updating unit, where the optimization system module includes the following contents:
The action selecting and noise adding unit initializes the parameters of the neural network, selects an action according to the action strategy, adds noise N t into the action output by the strategy network, increases the exploration degree, transmits the exploration degree to the environment to execute the action a t, and calculates the executed action, wherein the formula is as follows: ; where a t denotes the action performed, s t denotes the current state, Representing the action of the policy network output, N t representing the added noise, θ μ representing the parameters of the policy network;
The state conversion storage unit returns rewards r t and new states s t+1 after the environment execution a t is finished, and the state conversion (s t,at,rt,st+1) stored by the intelligent agent is put into the playback memory to be used as a training set of the online neural network;
loss function calculation unit for updating parameters of Q-value network using Adam optimizer The loss function is calculated using the formula:
；
；
Where L represents the loss function, N represents the number of samples, i.e., the total number of samples used to calculate the loss function, t represents the index of the samples, y t represents the target value, Q represents the online Q-value function, Parameters representing Q-value network, r t representing instant rewards, gamma representing discount factors,/>Representing a target Q function, s t+1 representing the next state,/>Parameters representing the target policy network,/>Parameters representing a target Q-value network;
the strategy gradient calculation unit calculates the strategy gradient of the strategy network, and the formula is as follows: ; in the/> Is a policy gradient, representing a parameter/>, related to a policyTarget function/>N is the number of samples, s represents the state, a1 represents the action,Is a gradient of the Q value function related to action a1,/>Is the gradient of the policy network, s t represents the state of sample t,/>Is an action generated from policies and states;
The target network parameter updating unit is used for optimizing the target network parameter by using a soft updating mode and is used for smoothly updating the target network parameter, and the formula is as follows: ; wherein, tau is a soft update parameter, the value range is 0 to 1, and the weight of the original network parameter and the target network parameter is balanced by controlling the size of the soft update parameter tau, so that the update of the target network is more gradual and stable,/> Parameters representing Q-value network,/>Parameters representing policy networks,/>Parameters representing the target policy network,/>Parameters representing a target Q-value network.
By executing the operations, the method adopts the initialized neural network parameters to perform action selection and noise addition, performs state transition storage, calculates a loss function and a strategy gradient, optimizes the system, updates the target network by using a soft update mode, improves the stability of the system, timely adjusts decisions according to temperature changes, and solves the technical problem that the system is difficult to timely adjust decisions according to the temperature changes when the temperature changes dynamically.
In a seventh embodiment, referring to fig. 1, the embodiment is based on the above embodiment, in the system monitoring module, data of the automatic control system of the refrigeration thermostatic bath is monitored in real time, and the system is monitored remotely and early-warned of faults, so that normal operation of the system is ensured.
It is noted that relational terms such as first and second, and the like are used solely to distinguish one entity or action from another entity or action without necessarily requiring or implying any actual such relationship or order between such entities or actions. Moreover, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or apparatus.
Although embodiments of the present invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made hereto without departing from the spirit and principles of the present invention.
The invention and its embodiments have been described above with no limitation, and the actual construction is not limited to the embodiments of the invention as shown in the drawings. In summary, if one of ordinary skill in the art is informed by this disclosure, a structural manner and an embodiment similar to the technical solution should not be creatively devised without departing from the gist of the present invention.
Claims (5)
1. An artificial intelligence based automatic control system for a refrigeration constant temperature tank is characterized in that: the system comprises a temperature acquisition module, a temperature prediction module, a user interface setting module, an actuator module, an optimization system module and a system monitoring module;
The temperature acquisition module is used for normalizing temperature data in the constant temperature tank, converting residual signals between predicted temperature and measured temperature into continuous time sequences, decomposing for multiple times to obtain intrinsic mode function components and residual items, and carrying out abnormal feature extraction and mode aliasing treatment;
the temperature prediction module is specifically used for constructing a temperature prediction model, constructing a radial basis function neural network by using a Gaussian kernel function, initializing a connection weight coefficient, adjusting the connection weight by minimizing an error between an actual temperature value and a neural network predicted value, and repeatedly training the temperature prediction model to perform actual temperature prediction;
The user interface setting module is specifically a design user interface;
The executor module is used for receiving the real-time temperature data and the control signals and executing corresponding operations;
The optimization system module is specifically used for initializing neural network parameters, performing action selection and noise addition, performing state transition storage, calculating a loss function and a strategy gradient, and updating a target network by using a soft updating mode;
The system monitoring module is specifically a remote monitoring system and is used for early warning faults;
the temperature acquisition module is provided with a temperature data acquisition unit, a data standardization unit, an abnormal data characteristic extraction unit and a mode aliasing processing unit, and comprises the following contents:
The temperature data acquisition unit is used for periodically acquiring temperature data in the refrigeration constant temperature tank by using a temperature sensor and preprocessing the data;
The data normalization unit uses the following formula:
；
Wherein x normal represents a standardized data value, x represents a sample value, namely temperature data in the refrigeration constant temperature tank, u represents a sample mean value, and delta represents a sample standard deviation, namely an average value of deviations between all sample values and the sample mean value;
the abnormal data feature extraction unit converts the residual signals between the predicted temperature and the measured temperature into a continuous time sequence, decomposes the temperature abnormal result layer by layer, generates a group of intrinsic mode function components and a residual item by decomposition each time, takes the selected decomposed signals as new original signals, decomposes again, and iterates repeatedly until a satisfactory decomposition result is achieved;
after signal decomposition, the values of the signals are calculated using the following formula:
；
Wherein x (t) represents the value of the signal, iz represents the index of the local extremum points, nz represents the total number of the local extremum points, c iz (t) represents the value of the envelope curve of the local extremum points, and r nz (t) represents the residual error term which contains the detail information and noise which cannot be described by the envelope curve in the signal;
the mode aliasing processing unit adds white noise with even distribution of amplitude into the original signal to concentrate and continue the signal and weaken the influence of instantaneous pulse on signal decomposition, and the mode aliasing processing unit comprises the following contents:
the random white noise signal is added to x (t) using the following formula:
；
In the method, in the process of the invention, Is a signal added with random white noise, and n (t) is a random white noise signal;
the signal after adding random white noise is decomposed into a series of eigenmode function components using the following formula:
；
In the method, in the process of the invention, Is the ith eigenmode function component obtained by the jth decomposition,/>Is the residual term of the jth decomposition, n1 is the total number of eigenmode function components, i is the index of eigenmode function components;
different random white noise signals are added to x (t) to obtain a plurality of groups of eigenvalue function components and residual terms, and the total average value of the corresponding components is calculated as a final result, wherein the following formula is used:
；
Wherein c i represents the final result of the ith eigenmode function component, and M represents the number of repetitions of the decomposition step;
The optimizing system module is provided with an action selection and noise adding unit, a state conversion storage unit, a loss function calculation unit, a strategy gradient calculation unit and a target network parameter updating unit, and comprises the following contents:
The action selecting and noise adding unit initializes the parameters of the neural network, selects an action according to the action strategy, adds noise N t into the action output by the strategy network, increases the exploration degree, transmits the exploration degree to the environment to execute the action a t, and calculates the executed action, wherein the formula is as follows:
；
Where a t denotes the action performed, s t denotes the current state, Representing the action of the policy network output, N t representing the added noise, θ μ representing the parameters of the policy network;
The state conversion storage unit returns rewards r t and new states s t+1 after the environment execution a t is finished, and the state conversion (s t,at,rt,st+1) stored by the intelligent agent is put into the playback memory to be used as a training set of the online neural network;
loss function calculation unit for updating parameters of Q-value network using Adam optimizer The loss function is calculated using the following formula:
；
；
Where L represents the loss function, N represents the number of samples, i.e., the total number of samples used to calculate the loss function, t represents the index of the samples, y t represents the target value, Q represents the online Q-value function, Parameters representing Q-value network, r t representing instant rewards, gamma representing discount factors,/>Representing a target Q function, s t+1 representing the next state,/>Parameters representing the target policy network,/>Parameters representing a target Q-value network;
the strategy gradient calculation unit calculates the strategy gradient of the strategy network, and the formula is as follows:
；
In the method, in the process of the invention, Is a policy gradient, representing a parameter/>, related to a policyTarget function/>N is the number of samples, s represents the state, a1 represents the action,/>The gradient of the Q function with respect to action a1,Is the gradient of the policy network, s t represents the state of sample t,/>Is an action generated from policies and states;
the target network parameter updating unit is used for optimizing the target network parameter by using a soft updating mode and is used for smoothly updating the target network parameter, and the following formula is used:
；
where τ is a soft update parameter, by controlling the magnitude of the soft update parameter τ, balancing the weights of the original network parameter and the target network parameter, making the update of the target network more gradual and stable, Parameters representing Q-value network,/>Parameters representing policy networks,/>Parameters representing the target policy network,/>Parameters representing a target Q-value network.
2. An artificial intelligence based automatic control system for a refrigeration thermostat according to claim 1, wherein: the temperature prediction module is provided with a temperature prediction model construction unit, an initialization unit, a forward propagation unit, an output calculation unit, a backward propagation unit and a training and application unit, and comprises the following contents:
The temperature prediction model construction unit is used for constructing a radial basis function neural network by using a Gaussian kernel function, and determining the node number in the hidden layer, the center point and the width of each kernel function;
the initialization unit randomly initializes the connection weight coefficient when constructing the neural network;
The forward propagation unit inputs the training sample into the neural network, calculates the hidden unit output of the radial basis function neural network, and uses the following formula:
；
Where phi is the radial basis function, x is the training sample, c a is the a-th center point of the kernel, a is the index of the center point of the kernel, Is the width of the a-th center point of the kernel function, x p represents the p-th input sample;
the output calculation unit calculates the output of the neural network by the following formula:
；
wherein y a represents the output of the RBF neural network, w cb represents the connection weight coefficient between the hidden layer neuron c and the output layer neuron b, and h represents the total number of kernel function center points;
the back propagation unit is used for updating the connection weight coefficient by using an error back propagation algorithm and minimizing the error between the actual temperature value and the temperature predicted value;
And the training and application unit is used for repeating the training process until the prediction error of the neural network reaches a satisfactory degree, applying the trained neural network model to a temperature prediction task, inputting new temperature data, predicting through the neural network, and evaluating and adjusting the prediction result.
3. An artificial intelligence based automatic control system for a refrigeration thermostat according to claim 1, wherein: in the user interface setting module, a design user interface unit, a mode selection unit and a parameter adjustment unit are provided, and the user interface setting module comprises the following contents:
designing a user interface unit, wherein a user inputs a set temperature through an interface, and checks real-time temperature data;
The mode selection unit is used for selecting different working modes by a user to meet the actual requirements of experiments, wherein the working modes comprise a constant temperature mode, a quick refrigeration mode and a quick heating mode;
And the parameter adjusting unit is used for monitoring the working state of the refrigeration constant temperature tank and adjusting the temperature tolerance, response time and control sensitivity of the refrigeration constant temperature tank.
4. An artificial intelligence based automatic control system for a refrigeration thermostat according to claim 1, wherein: and the actuator module transmits the real-time temperature data to the actuator module, and the actuator module executes corresponding operations including adjusting the working state and controlling the temperature of the refrigerating system, and feeds back the execution result to the controller so that the controller can dynamically adjust according to the actual situation.
5. An artificial intelligence based automatic control system for a refrigeration thermostat according to claim 1, wherein: in the system monitoring module, data of the automatic control system of the refrigeration constant temperature tank are monitored in real time, and the system is monitored remotely and early-warned of faults, so that normal operation of the system is ensured.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202410212561.1A CN117784849B (en) | 2024-02-27 | 2024-02-27 | Automatic control system of refrigeration constant temperature tank based on artificial intelligence |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202410212561.1A CN117784849B (en) | 2024-02-27 | 2024-02-27 | Automatic control system of refrigeration constant temperature tank based on artificial intelligence |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN117784849A CN117784849A (en) | 2024-03-29 |
| CN117784849B true CN117784849B (en) | 2024-05-07 |
Family
ID=90389565
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202410212561.1A Active CN117784849B (en) | 2024-02-27 | 2024-02-27 | Automatic control system of refrigeration constant temperature tank based on artificial intelligence |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN117784849B (en) |
Families Citing this family (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN118443510A (en) * | 2024-06-24 | 2024-08-06 | 深圳市铭宇仪器设备有限公司 | Control method, device, equipment and storage medium for cold and hot impact testing machine |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107121218A (en) * | 2017-06-30 | 2017-09-01 | 国家海洋局第海洋研究所 | A kind of one-touch air-temperature sensor calibration system |
| CN207264226U (en) * | 2017-09-14 | 2018-04-20 | 美核电气（济南）股份有限公司 | Multichannel temperature controller intelligence calibrating installation |
| CN111240268A (en) * | 2020-01-14 | 2020-06-05 | 重庆大学 | Axle system thermal error modeling method and thermal error compensation system based on SLSTM neural network |
| CN111396032A (en) * | 2020-04-23 | 2020-07-10 | 西安石油大学 | Method and device for correcting temperature drift and computer storage medium |
| CN116560428A (en) * | 2023-06-13 | 2023-08-08 | 山东科技大学 | Poultry house temperature prediction control system and control method based on LMBP neural network |
| CN117331384A (en) * | 2023-10-25 | 2024-01-02 | 中冶赛迪信息技术(重庆)有限公司 | Coke oven temperature control method, system, equipment and storage medium |
Family Cites Families (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US10234848B2 (en) * | 2017-05-24 | 2019-03-19 | Relativity Space, Inc. | Real-time adaptive control of additive manufacturing processes using machine learning |
| US11568310B2 (en) * | 2019-06-04 | 2023-01-31 | Lg Electronics Inc. | Apparatus for generating temperature prediction model and method for providing simulation environment |
-
2024
- 2024-02-27 CN CN202410212561.1A patent/CN117784849B/en active Active
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107121218A (en) * | 2017-06-30 | 2017-09-01 | 国家海洋局第海洋研究所 | A kind of one-touch air-temperature sensor calibration system |
| CN207264226U (en) * | 2017-09-14 | 2018-04-20 | 美核电气（济南）股份有限公司 | Multichannel temperature controller intelligence calibrating installation |
| CN111240268A (en) * | 2020-01-14 | 2020-06-05 | 重庆大学 | Axle system thermal error modeling method and thermal error compensation system based on SLSTM neural network |
| CN111396032A (en) * | 2020-04-23 | 2020-07-10 | 西安石油大学 | Method and device for correcting temperature drift and computer storage medium |
| CN116560428A (en) * | 2023-06-13 | 2023-08-08 | 山东科技大学 | Poultry house temperature prediction control system and control method based on LMBP neural network |
| CN117331384A (en) * | 2023-10-25 | 2024-01-02 | 中冶赛迪信息技术(重庆)有限公司 | Coke oven temperature control method, system, equipment and storage medium |
Also Published As
| Publication number | Publication date |
|---|---|
| CN117784849A (en) | 2024-03-29 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN112906735B (en) | Domestic fungus environment big data detecting system | |
| CN117784849B (en) | Automatic control system of refrigeration constant temperature tank based on artificial intelligence | |
| CN112292642B (en) | Control device for controlling a technical system and method for configuring a control device | |
| CN109472105B (en) | Semiconductor product yield upper limit analysis method | |
| CN109472397B (en) | Polymerization process parameter adjusting method based on viscosity change | |
| CN113138555A (en) | GRNN electric spindle thermal error modeling method based on genetic algorithm optimization | |
| CN110147825A (en) | A kind of strawberry greenhouse temperature intelligent detection device based on empirical mode decomposition model | |
| CN115167102A (en) | Reinforced learning self-adaptive PID control method based on parallel dominant motion evaluation | |
| CN117873238A (en) | Plant growth control method and system, computer equipment and storage medium | |
| CN112615364A (en) | Novel wide-area intelligent cooperative control method for power grid stability control device | |
| Horie et al. | Multi-objective safe reinforcement learning: the relationship between multi-objective reinforcement learning and safe reinforcement learning | |
| CN114355848B (en) | Tension detection and intelligent control system | |
| CN117469774B (en) | Air conditioning system regulation and control method and device, electronic equipment and storage medium | |
| CN111625030A (en) | Greenhouse environment control method, device, equipment, system and storage medium | |
| US20220027708A1 (en) | Arithmetic apparatus, action determination method, and non-transitory computer readable medium storing control program | |
| CN113110061B (en) | Intelligent irrigation fuzzy control method and system based on improved particle swarm optimization | |
| Avila-Miranda et al. | An optimal and intelligent control strategy to ventilate a greenhouse | |
| CN114397809A (en) | Intelligent control system for material weighing big data detection and packaging | |
| CN114298130A (en) | Short-term wind power prediction method and device | |
| CN112906856A (en) | Livestock and poultry house ambient temperature detection system based on cloud platform | |
| CN115185184B (en) | Multi-parameter identification method for electrohydraulic position servo system | |
| CN118315000B (en) | Prediction method for material corrosion fatigue crack growth rate | |
| CN118501648B (en) | Semiconductor test control method and system based on big data | |
| CN113505914B (en) | Injection molding machine hydraulic system energy consumption prediction method, system and equipment for optimizing SVM | |
| CN117356411B (en) | Agricultural irrigation system based on soil humidity |
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 | ||
| GR01 | Patent grant |