A method and system for vapor heat release control of modular molten salt energy storage

Abstract

The application discloses a kind of modular molten salt energy storage steam heat release control method and system, it is related to molten salt energy storage technical field, including establishing the BP neural network model of modular molten salt energy storage heat release control system, training BP neural network model;Definition is normalized after output error, according to output error analysis BP neural network model;Variable rate adjustment technology is used to process the BP neural network model after analysis, establish the mathematical model of actual object opposite to it, complete the steam heat release control of modular molten salt energy storage.The application effectively improves the steam heat release control precision and response speed of modular molten salt energy storage system, ensures the stability and reliability of control system, realizes fine control to molten salt pump frequency, optimizes the heat exchange process of molten salt and water vapor, significantly improves the overall efficiency of energy storage system, reduces energy loss.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of molten salt energy storage, in particular to a steam heat release control method and system for modular molten salt energy storage. BACKGROUND

[0002] Currently, molten salt has become an ideal thermal energy storage medium due to its high specific heat capacity, high thermal stability and good thermal conductivity at high temperatures. The emergence of molten salt energy storage technology has solved the contradiction between intermittent energy supply and continuous demand. Traditional solar thermal power generation systems cannot collect enough solar radiation at night or on cloudy days. Molten salt energy storage can store excess heat during the day and release it when needed, thereby ensuring continuous power supply. Coupling molten salt energy storage systems with thermal power generating units to achieve thermal-electric decoupling and auxiliary peak shaving and frequency modulation is the current development trend.

[0003] Currently, common molten salt energy storage systems include double-tank and single-tank systems. These two forms of systems are usually large in size, high in investment cost and difficult to adjust, which limits their widespread application. Modular molten salt energy storage technology is an advanced thermal energy storage solution that combines the efficiency of molten salt energy storage with the flexibility of modular design, providing a new approach to the integration of renewable energy and the optimization of traditional thermal power generation. The modular design further enhances the scalability, economy and adaptability of the system.

[0004] Modular molten salt energy storage systems solve the above problems by dividing large energy storage units into multiple smaller and independently operable modules. Each module can be independently manufactured, tested, transported and installed, greatly reducing the complexity and cost of on-site construction. In addition, modular design allows the system to be flexibly expanded according to actual needs, whether it is initial construction or later expansion, it can quickly adapt to different load requirements. SUMMARY

[0005] In view of the problems existing in the prior art, the present application is proposed.

[0006] Therefore, the purpose of the present application is to provide a steam heat release control method and system for modular molten salt energy storage. To solve the problem of insufficient control accuracy of existing molten salt energy storage systems, the present application uses BP neural network combined with variable rate regulation technology.

[0007] To solve the above technical problems, the present application provides the following technical solutions:

[0008] In a first aspect, the embodiments of the present application provide a steam heat release control method for modular molten salt energy storage, which comprises: establishing a BP neural network model of a modular molten salt energy storage heat release control system, training the BP neural network model, and obtaining node parameters of the model; performing normalization processing on the node parameters, defining an output error after the normalization processing, and analyzing the BP neural network model according to the output error; processing the analyzed BP neural network model by using a variable rate adjustment technology, establishing a mathematical model of an actual object opposite to the processed BP neural network model, and completing steam heat release control of the modular molten salt energy storage.

[0009] As a preferred scheme of the steam heat release control method for modular molten salt energy storage, the establishment of the BP neural network model comprises extracting T y1 , T y2 , T y3 , and T y4 as medium temperatures of the molten salt according to N input energy storage heat release parameters.

[0010] The input comprises inputting the N energy storage heat release parameters through an input layer and outputting through a hidden layer, the hidden layer comprises obtaining a node number of the hidden layer corresponding to a node number of the input layer, the node number of the input layer comprises x1-x7, wherein x1→fy, x2→Lso, x3→Tsin, x4→T y1 , x5→T y2 , x6→T y3 , and x7→T y4 , at this time, x4→T y1 represents the hidden layer, the node number of the hidden layer is L=h, and the output formula of the jth neuron of the hidden layer is:

[0011]

[0012] wherein i=1, 2, 3...n, j=1, 2, 3...L, wh ij represents a weight value from the input layer to the hidden layer, a j represents a node threshold value of the hidden layer, h j represents the jth neuron of the hidden layer.

[0013] As a preferred scheme of the steam heat release control method for modular molten salt energy storage, the output through the hidden layer comprises introducing a node activation function to pre-process the jth neuron of the hidden layer, accelerating the learning speed of the BP neural network model, and calculating the jth neuron of the output layer, and the specific calculation formula is:

[0014]

[0015] wherein, y j represents the jth neuron of the output layer, wy j represents the weight value from the output layer to the hidden layer, b1 represents the node threshold value of the output layer, h j represents the jth neuron of the hidden layer.

[0016] A preliminary model of the neural network is established for the jth neuron of the hidden layer and the jth neuron of the output layer, and all node parameters of the input and output energy heat release are obtained according to the steam heat release under different conditions.

[0017] As a preferred scheme of the modular molten salt energy storage steam heat release control method, the training of the BP neural network model comprises setting the weight value from the input layer to the hidden layer and the weight value from the output layer to the hidden layer.

[0018] When the weight value from the input layer to the hidden layer is greater than the node threshold value of the hidden layer in the jth neuron, the BP neural network model enters the saturation zone, at this time, the BP neural network model converges, and the weight value from the input layer to the hidden layer and the weight value from the output layer to the hidden layer are transported to the saturation zone for setting, and the weight value from the input layer to the hidden layer and the weight value from the output layer to the hidden layer are randomly taken as non-zero values.

[0019] As a preferred scheme of the modular molten salt energy storage steam heat release control method, the normalization processing of the node parameters comprises uniform processing of all node parameters by using an excitation function, and the specific normalization calculation formula of the excitation function is:

[0020]

[0021] wherein, x represents the normalized value, x r represents the physical quantity in operation, x max represents the maximum value in the excitation function, x min represents the minimum value in the excitation function.

[0022] The specific formula for defining the output error after normalization processing is:

[0023]

[0024] wherein, represents the output value of the BP neural network model, y1 represents the actual value, and E represents the output error after normalization processing.

[0025] As a preferred scheme of the modular molten salt energy storage steam heat release control method, the analysis of the BP neural network model according to the output error comprises setting a minimum error ∈.

[0026] When E<∈, then the BP neural network model is established.

[0027] When E≥∈, then the weights of the jth neuron of the hidden layer and the jth neuron of the output layer are increased by the gradient descent method, and the specific increase calculation formula is:

[0028]

[0029]

[0030]

[0031]

[0032] Wherein, μ represents the learning rate, represents the operator, and y'1 represents the actual value first-order derivative.

[0033] As a preferred scheme of the steam heat release control method of the modular molten salt energy storage, wherein: the processing of the analyzed BP neural network model comprises iterative learning rate μ by using a variable rate adjustment technique, and a BP neural network model of the molten salt pump frequency fy is output based on the iterative time value of the adjusted learning rate μ, and the specific calculation formula of the variable rate adjustment technique is:

[0034]

[0035] Wherein, μ max represents the maximum learning rate, μ min represents the minimum learning rate, t max represents the maximum iteration number, t represents the current iteration number, and μ(t) represents the variable rate.

[0036] The steam heat release control comprises substituting the number of nodes of the input layer into the BP neural network model of the output molten salt pump frequency fy, transmitting the output molten salt pump frequency fy to the molten salt pump frequency converter, adjusting the heat exchange rate of the molten salt and water vapor with the change of the molten salt flow condition, and completing the steam heat release control.

[0037] In the second aspect, the embodiment of the present application provides a steam heat release control system of modular molten salt energy storage, which comprises: a training module, which establishes a BP neural network model of the modular molten salt energy storage heat release control system, and trains the BP neural network model; an analysis module, which defines the normalized output error, and analyzes the BP neural network model according to the output error; and a control module, which processes the analyzed BP neural network model by using a variable rate adjustment technique, establishes a mathematical model of the actual object opposite to it, and completes the steam heat release control of the modular molten salt energy storage.

[0038] In a third aspect, an embodiment of the present application provides a computer device, comprising a memory and a processor, and the memory stores a computer program, wherein the processor executes the computer program to implement any step of the modular molten salt energy storage steam heat release control method.

[0039] In a fourth aspect, an embodiment of the present application provides a computer readable storage medium, which stores a computer program, wherein the computer program is executed by a processor to implement any step of the modular molten salt energy storage steam heat release control method.

[0040] The present application has the following beneficial effects: the present application effectively improves the steam heat release control precision and response speed of the modular molten salt energy storage system by introducing the BP neural network and the variable rate adjustment technology, ensures the stability and reliability of the control system through the normalization processing of the node parameters and the accurate output error analysis, makes the neural network model quickly converge through the adjustment of the learning rate and the introduction of the excitation function, thereby realizing the fine control of the molten salt pump frequency, optimizing the heat exchange process of the molten salt and water vapor, significantly improving the overall efficiency of the energy storage system, and reducing the energy loss. BRIEF DESCRIPTION OF DRAWINGS

[0041] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the following will briefly introduce the drawings needed to be used in the embodiment description. Obviously, the drawings in the following description are only some embodiments of the present application, and for those skilled in the art, other drawings can also be obtained from these drawings without creative labor. Among them:

[0042] Figure 1 A specific flowchart of a modular molten salt energy storage steam heat release control method and system provided for an embodiment of the present application.

[0043] Figure 2 A modular molten salt energy storage system structure process flowchart of a modular molten salt energy storage steam heat release control method and system provided for an embodiment of the present application.

[0044] Figure 3 A BP neural network structure diagram of a modular molten salt energy storage steam heat release control method and system provided for an embodiment of the present application. DETAILED DESCRIPTION

[0045] In order to make the above objectives, characteristics and advantages of the present application more obvious and comprehensible, the specific embodiments of the present application will be described in detail below with reference to the drawings. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments of the present application, all the other embodiments obtained by those skilled in the art without creative work should fall within the protection scope of the present application.

[0046] In the following description, numerous specific details are set forth in order to provide a thorough understanding of the present application. The present application, however, can be practiced in a variety of ways beyond the specific details set forth herein without departing from the scope of the present application, and it is understood that the application is not limited in this respect.

[0047] Secondly, the "one embodiment" or "embodiment" referred to herein means that the specific features, structures or characteristics can be included in at least one implementation of the present application. The "in one embodiment" appearing in different places in the specification does not mean the same embodiment, nor is it an embodiment that is independent of or selected from other embodiments.

[0048] The present application is described in detail in conjunction with the schematic drawings. In the detailed description of the embodiments of the present application, the sectional view of the device structure is partially enlarged without the general proportion for the convenience of description, and the schematic drawings are only examples, which should not limit the scope of protection of the present application herein. In addition, the three-dimensional spatial dimensions of length, width and depth should be included in actual manufacturing.

[0049] Meanwhile, in the description of the present application, it should be noted that the terms "upper, lower, inner and outer" and the like indicate the orientation or positional relationship based on the orientation or positional relationship shown in the drawings, and are only for the convenience of describing the present application and simplifying the description, and do not indicate or imply that the indicated device or element must have a particular orientation, be constructed and operated in a particular orientation, and therefore cannot be understood as a limitation on the present application. In addition, the terms "first, second or third" are only for the purpose of description, and cannot be understood as indicating or implying relative importance.

[0050] In the present application, unless otherwise explicitly specified and limited, the terms "mounting, connecting, connection" should be understood broadly, for example: it can be fixed connection, detachable connection or integral connection; it can also be mechanical connection, electrical connection or direct connection, it can also be indirectly connected through an intermediate medium, or it can be the communication inside two elements. For those skilled in the art, the specific meaning of the above terms in the present application can be understood according to the specific circumstances.

[0051] Example 1

[0052] Reference Figures 1-3For the first embodiment of the present application, the embodiment provides a modular molten salt energy storage steam heat release control method, comprising:

[0053] S1: establishing a BP neural network model of the modular molten salt energy storage heat release control system, training the BP neural network model, and obtaining node parameters of the model.

[0054] Wherein, the BP neural network model comprises extracting T y1 , T y2 , T y3 and T y4 as the medium temperature of the molten salt according to the input N energy storage heat release parameters.

[0055] The input comprises inputting the N energy storage heat release parameters through the input layer and outputting through the hidden layer, the hidden layer comprises obtaining the number of nodes of the hidden layer corresponding to the number of nodes of the input layer, and the number of nodes of the input layer comprises x1-x7, wherein x1->fy, x2->Lso, x3->Tsin, x4->T y1 , x5->T y2 , x6->T y3 , x7->T y4 , at this time, x4->T y1 represents the hidden layer, the number of nodes of the hidden layer is L=h, and the output formula of the jth neuron of the hidden layer is:

[0056]

[0057] Wherein, i=1, 2, 3...n, j=1, 2, 3...L, wh ij represents the weight value from the input layer to the hidden layer, a j represents the node threshold value of the hidden layer, h j represents the jth neuron of the hidden layer.

[0058] Further, in the present application, a BP neural network model of the modular molten salt energy storage heat release control system is established to accurately control the steam heat release process. The model is constructed based on the input of N energy storage heat release parameters, which mainly include molten salt temperature, pressure, flow rate and other key indicators. For example, assuming that the input layer contains 12 nodes corresponding to 12 key energy storage parameters, after input in the input layer, the parameters are calculated and processed in the hidden layer, and the number of nodes in the hidden layer is determined by matching the number of nodes in the input layer. Assuming that the hidden layer contains 20 nodes, each hidden layer node is processed by an excitation function after the weight and threshold value of the input layer node, and the output result is further used to extract the medium temperature of the molten salt to reflect the current thermal state of the system. Through such a neural network model, the dynamic behavior of the molten salt energy storage system can be accurately simulated, thereby laying a foundation for subsequent accurate control. This method processes a large amount of input data to generate accurate output node parameters, providing efficient and real-time control capability for the system, effectively improving the overall efficiency and response speed of energy storage heat release.

[0059] S1.1: output through the hidden layer includes introducing a node activation function to pre-process the jth neuron of the hidden layer, accelerating the learning speed of the BP neural network model, and calculating the jth neuron of the output layer, the specific calculation formula is:

[0060]

[0061] wherein y j represents the jth neuron of the output layer, wy j represents the weight from the output layer to the hidden layer, b1 represents the node threshold value of the output layer, h j represents the jth neuron of the hidden layer.

[0062] A neural network preliminary model is established for the jth neuron of the hidden layer and the jth neuron of the output layer, and all node parameters of the energy storage heat release are obtained according to the steam heat release under different conditions.

[0063] Further, in the present application, in order to improve the learning efficiency of the BP neural network model, when output is performed through the hidden layer, a node activation function is used to preprocess each neuron of the hidden layer, specifically, assuming that the hidden layer contains 20 neurons, each neuron will be processed by the activation function, and the activation function plays a role in accelerating the convergence of the neural network. For example, when processing input data containing 1000 data points, the activation function can quickly filter out parameters closely related to steam heat release, and pass these processed data to the output layer, which may contain 5 neurons. The connection weights between each neuron of the output layer and the 20 neurons of the hidden layer are used to calculate the final output of the energy release heat parameter. In different working conditions, such as high load or low load, the neurons of the output layer will generate different control signals to accurately adjust the steam heat release of the system. Through this preliminary neural network model, the system can analyze and obtain various input and output energy release heat parameters in real time, providing reliable data support for subsequent control decisions, thereby significantly improving the accuracy and efficiency of control.

[0064] S1.2: training the BP neural network model includes setting the input layer to hidden layer weight and the output layer to hidden layer weight;

[0065] When the input layer to hidden layer weight is greater than the hidden layer node threshold in the jth neuron, then the BP neural network model enters the saturation zone, at this time, the BP neural network model converges, and the input layer to hidden layer weight and the output layer to hidden layer weight are transported to the saturation zone for setting. The input layer to hidden layer weight and the output layer to hidden layer weight are randomly assigned non-zero values.

[0066] In the training process of the present application, in order to ensure the effectiveness of the BP neural network model, the weights between the input layer and the hidden layer, and between the hidden layer and the output layer need to be accurately set. Assuming that the input layer has 12 nodes, the hidden layer has 20 nodes, and the output layer has 5 nodes, the connection weights between these nodes will be assigned non-zero values according to the random value mode in the initial stage. For example, the input layer to hidden layer weight may be set to a random value between 0.01 and 0.1. As the training progresses, when these weights gradually increase and exceed the node threshold of a neuron in the hidden layer, the network enters the so-called saturation zone. In this region, the BP neural network model begins to converge, that is, gradually reduces the error and optimizes the output effect of the network. The system will fix these weights in the saturation zone and further adjust to ensure that the model can still effectively learn and adapt to new data input in the saturated state. In this way, the model can avoid the dilemma of overtraining or undertraining, thereby ensuring the stability and accuracy of the system under different operating conditions and effectively improving the precision of steam heat release control. The neural network weight setting data is shown in Table 1.

[0067] Table 1 neural network weight setting data table

[0068] Node level Number of nodes Weight range Input layer 12 - Hidden layer 20 0.01 to 0.1 Output layer 5 - Input layer to hidden layer 12×20 0.01 to 0.1 Hidden layer to output layer 20×5 -

[0069] The table shows the number of nodes and weight settings in each layer of the BP neural network model. The input layer contains 12 nodes, the hidden layer has 20 nodes, and the output layer has 5 nodes. The table lists the weight range from the input layer to the hidden layer as 0.01 to 0.1, but the specific weight range from the hidden layer to the output layer is not explicitly stated. Overall, the table summarizes the initial settings of the model weights and their configuration between different layers.

[0070] S2: Normalizing the node parameters, defining the normalized output error, and analyzing the BP neural network model according to the output error.

[0071] Among them, the normalization processing of node parameters includes uniform processing of all node parameters by using an excitation function, and the specific normalization calculation formula of the excitation function is:

[0072]

[0073] Where x represents the normalized value, x r represents the physical quantity of operation, x max represents the maximum value in the excitation function, x min represents the minimum value in the excitation function.

[0074] The specific formula for defining the normalized output error is:

[0075]

[0076] Where, represents the output value of the BP neural network model, y1 represents the actual value, and E represents the normalized output error.

[0077] Further, in the present application, in order to improve the accuracy and stability of the BP neural network model, the normalization processing of node parameters is a key step. Specifically, when processing the steam heat release control of the molten salt energy storage system, multiple input parameters such as temperature, pressure, and flow rate need to be uniformly processed. These parameters may differ greatly in physical magnitude. In order to avoid the influence of these differences on the learning effect of the neural network, an excitation function is used to normalize these parameters. Assuming that the temperature range is 200℃ to 500℃ and the pressure range is 2MPa to 10MPa, the normalization processing converts these parameters to a unified scale, making them fluctuate within the same range (such as 0 to 1).

[0078] After completing the normalization, the system will further define the output error as an indicator to measure the performance of the neural network model, by calculating the difference between the normalized output value and the actual value, to determine the accuracy of the model, for example, in a steam heat release operation, if the normalized output temperature is 0.8, and the actual value corresponds to the normalized value of 0.75, then the output error will reflect this gap, according to these error data, the system can analyze and adjust the BP neural network model, to ensure that it can provide accurate control under different working conditions, this process not only improves the robustness of the model, but also enhances the adaptability of the system to complex dynamic environment, which helps to realize more efficient energy storage heat release management.

[0079] S2.1: Analyzing the BP neural network model according to the output error includes setting a minimum error

[0080] When E<∈, the BP neural network model is established;

[0081] When E≥∈, the weights of the jth neuron of the hidden layer and the jth neuron of the output layer are adjusted by the gradient descent method, and the specific adjustment formula is:

[0082]

[0083]

[0084]

[0085]

[0086] Wherein, μ represents the learning rate, represents the operator, y'1 represents the actual value first-order derivative

[0087] In the present application, the analysis of the output error is an important step for optimizing the BP neural network model, first, by setting a minimum error value, the system judges whether the training of the model has reached the ideal state, if the output error is less than this minimum error value, such as setting 0.001, it means that the model is accurate enough, at this time, the training of the model can be ended, if the output error is greater than the minimum error value, it means that the model still has deviation, which needs to be further optimized, at this time, the system will use the gradient descent method to adjust the weights in the neural network.

[0088] In actual operation, assuming that the hidden layer contains 20 neurons and the output layer contains 5 neurons, the system will fine-tune the weight of each neuron to reduce the error. For example, if it is found that the weight of a certain hidden layer neuron is too high, resulting in a large output error, the gradient descent method will adjust the weight value slightly to reduce the overall error. Assuming that the learning rate is set to 0.01, which means that the adjustment is relatively small to ensure that the model gradually converges to the optimal state. After multiple iterations, the system can gradually reduce the error to the set range, and finally complete the establishment of the BP neural network model. This fine error analysis and weight adjustment method ensures the accuracy and reliability of the model, which can provide efficient control in the complex energy storage and heat release process.

[0089] S3: The analyzed BP neural network model is processed by using a variable rate adjustment technique, and based on the processed BP neural network model, an actual object mathematical model opposite to it is established to complete the modular molten salt energy storage steam heat release control.

[0090] Among them, the processing of the analyzed BP neural network model includes iterating the learning rate μ by using a variable rate adjustment technique, and outputting the BP neural network model of the molten salt pump frequency fy based on the iteration value of the adjusted learning rate μ, and the specific calculation formula of the variable rate adjustment technique is:

[0091]

[0092] Among them, μ max represents the maximum learning rate, μ min represents the minimum learning rate, t max represents the maximum number of iterations, t represents the current iteration number, and μ(t) represents the variable rate.

[0093] The steam heat release control includes substituting the number of nodes in the input layer into the BP neural network model of the output molten salt pump frequency fy, transmitting the output molten salt pump frequency fy to the molten salt pump frequency converter, adjusting the heat exchange rate of molten salt and water vapor with the change of molten salt flow, and completing the steam heat release control.

[0094] Preferably, in the present application, the variable rate adjustment technique is used to optimize the learning rate of the BP neural network model. Initially, the system uses a larger learning rate, such as 0.1, to quickly converge the model; as the iteration proceeds, the learning rate is gradually reduced, possibly to 0.01 or lower, to fine-tune. After multiple iterations of optimization, the model calculates the appropriate molten salt pump frequency for the current operating condition, such as 1500 RPM, and transmits this frequency to the molten salt pump frequency converter, which controls the molten salt flow rate according to the adjusted frequency, thereby accurately adjusting the heat exchange rate of the molten salt and water vapor, achieving efficient control of the steam heat release process. Through this method, the system can maintain high response speed and control accuracy in actual operation.

[0095] In a preferred embodiment, a modular molten salt energy storage steam heat release control system, the system comprises a training module, which establishes a BP neural network model of the modular molten salt energy storage heat release control system, trains the BP neural network model; an analysis module, which defines the normalized output error, analyzes the BP neural network model according to the output error; a control module, which uses a variable rate adjustment technique to process the analyzed BP neural network model, establishes a mathematical model of the actual object opposite to it, and completes the modular molten salt energy storage steam heat release control.

[0096] The above-mentioned unit modules can be embedded in or independent of the processor in the computer device in hardware form, or stored in the memory in the computer device in software form, so as to be called and executed by the processor to perform the corresponding operations of the above-mentioned modules.

[0097] The computer device can be a terminal, and the computer device includes a processor, a memory, a communication interface, a display screen, and an input device connected through a system bus. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operating system and the computer program in the non-volatile storage medium to run. The communication interface of the computer device is used to communicate with external terminals in a wired or wireless manner. The wireless manner can be achieved through WIFI, operator network, NFC (Near Field Communication), or other technologies. The display screen of the computer device can be a liquid crystal display screen or an electronic ink display screen. The input device of the computer device can be a touch layer overlaid on the display screen, or a key, trackball, or touchpad arranged on the shell of the computer device. It can also be an external keyboard, touchpad, or mouse, etc.

[0098] To sum up, the application effectively improves the steam heat release control precision and response speed of the modular molten salt energy storage system by introducing the BP neural network and the variable rate adjustment technology, ensures the stability and reliability of the control system through the normalization processing of the node parameters and the accurate output error analysis, makes the neural network model quickly converge through adjusting the learning rate and introducing the excitation function, thereby realizing the fine control of the molten salt pump frequency, optimizing the heat exchange process of the molten salt and water vapor, significantly improving the overall efficiency of the energy storage system, and reducing the energy loss.

[0099] Embodiment 2

[0100] With reference to Figures 1-3 For the second embodiment of the application, the embodiment provides a steam heat release control method for modular molten salt energy storage. In order to verify the beneficial effects of the application, a simulation experiment is carried out for scientific demonstration.

[0101] During the experiment, a BP neural network model of the modular molten salt energy storage heat release control system was first established. In the early stage of the experiment, 15 nodes were set in the input layer, 25 nodes were included in the hidden layer, and 6 nodes were set in the output layer. In order to evaluate the performance of the model, the weight values from the input layer to the hidden layer were initialized and set between 0.02 and 0.08, and the weight values from the hidden layer to the output layer were set between 0.05 and 0.15. In the training stage, 1000 groups of experimental data were used, each group of data including molten salt temperature, flow rate and other parameters. The model training used a variable rate adjustment technology, the initial learning rate was 0.05, and gradually decreased to 0.005, and the training was iterated 5000 times, and the weight values were gradually adjusted to optimize the model performance.

[0102] In the test stage, new molten salt temperature data such as 300℃ and 320℃, and flow rate data such as 0.8L / s and 1.2L / s were input, and the model output molten salt pump frequency was 1800RPM and 1900RPM respectively. The frequency value was transmitted to the molten salt pump frequency converter for adjusting the heat exchange rate of molten salt and water vapor. Finally, the experimental results show that the system can accurately adjust the steam heat release process, improve the control precision, and reduce the error range to within 5%, verifying the effectiveness of the model in practical application. The experimental data is shown in Table 2 as follows:

[0103] Table 2 Summary of experimental data

[0104] Experiment content Data Number of input layer nodes 15 Number of hidden layer nodes 25 Number of output layer nodes 6 Input layer to hidden layer weight range 0.02 to 0.08 Hidden layer to output layer weight range 0.05 to 0.15 Number of experimental data sets 1000 Initial learning rate 0.05 Final learning rate 0.005 Number of training iterations 5000 Temperature data 300℃,320℃ Flow rate data 0.8 L / s, 1.2 L / s Output pump frequency 1800 RPM, 1900 RPM Control accuracy error range 5％

[0105] The table summarizes the key data in the experimental process, including the node number of the BP neural network model, the weight range and the specific parameters of training and testing, the table shows the node number and weight range of the input layer, hidden layer and output layer, the data group number, learning rate and iteration number in the training process, the temperature and flow rate data used in the test stage, and the pump frequency and control accuracy error range output by the model, the comparison of the technical scheme of the present application with the prior art is shown in the following table 3.

[0106] Table 3 Comparison table of the technical scheme of the present application with the prior art

[0107]

[0108] The table compares the main differences and advantages of the technical scheme of the present application with the prior art, the present application adopts the BP neural network model, which can handle more complex nonlinear problems compared with the traditional linear model, provides higher accuracy, the weight initialization is more accurate, the learning rate is self-adaptive adjustment, makes the model training more stable and efficient, in terms of data processing, through normalization improves the data consistency and model convergence, the control accuracy is significantly improved, the error range is reduced to 5%, and more data and training iteration number are used to enhance the generalization ability of the model, finally, the rapid adjustment ability of the output feedback improves the timeliness of the control reaction, in general, the present application is superior to the prior art in all aspects, provides a more accurate and stable control scheme.

[0109] It should be noted that the above examples are only used to illustrate the technical solutions of the present application and not to limit, although the present application has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present application can be modified or replaced equivalently without departing from the spirit and scope of the technical solutions of the present application, which should be covered in the scope of the claims of the present application.

Claims

1. A method of steam heat release control for modular molten salt energy storage, characterized by: Comprising, The BP neural network model of the modular molten salt energy storage heat release control system is established, the BP neural network model is trained, and node parameters of the model are acquired; The node parameters are normalized, an output error after the normalization is defined, and the BP neural network model is analyzed according to the output error; The analyzed BP neural network model is processed by using a variable rate regulation technology, an actual object mathematical model opposite to the processed BP neural network model is established based on the processed BP neural network model, and the steam heat release control of the modular molten salt energy storage is completed; The BP neural network model is established, which includes extracting temperature of the medium as molten salt; The input includes inputting the N energy storage heat release parameters through an input layer and outputting through a hidden layer, the hidden layer includes acquiring the number of nodes of the hidden layer corresponding to the number of nodes of the input layer, and the number of nodes of the input layer includes wherein, , , , , , , at this time, represents a hidden layer, the number of nodes of the hidden layer is , and the output formula of the i-th neuron of the hidden layer is: ​ wherein, , , denotes the input layer to hidden layer weights, denotes the hidden layer node threshold values, denotes the hidden layer's th neuron; The output through the hidden layer includes preprocessing the first neuron of the hidden layer by introducing a node activation function, accelerating the learning speed of the BP neural network model, and calculating the first neuron of the output layer, and the specific calculation formula is: wherein, represents the i-th neuron of the output layer, represents the weight from the output layer to the hidden layer, represents the node threshold of the output layer, represents the i-th neuron of the hidden layer, represents the i-th neuron of the hidden layer,​ The preliminary model of neural network is established for the first neurons of the hidden layer and the first neurons of the output layer, and all the node parameters of the input and output energy release are obtained according to the steam heat release under different conditions.

2. The modular molten salt energy storage vapor heat release control method of claim 1, wherein: The training of the BP neural network model comprises setting weight values from an input layer to a hidden layer and weight values from an output layer to the hidden layer; When the weight value from the input layer to the hidden layer is greater than the threshold value of the hidden layer node in the first neuron, the saturation zone of the BP neural network model is entered, at this time, the BP neural network model converges, and the weight value from the input layer to the hidden layer and the weight value from the output layer to the hidden layer are transported to the saturation zone for setting. The weight value from the input layer to the hidden layer and the weight value from the output layer to the hidden layer are randomly taken as non-zero values.

3. The modular molten salt energy storage vapor heat release control method of claim 2, wherein: The normalization processing of the node parameters comprises uniform processing of all node parameters by using an excitation function, and a specific normalization calculation formula of the excitation function is: wherein, denotes the normalized value, denotes a physical quantity of the operation, denotes the maximum value in the excitation function, denotes the minimum value in the excitation function; A specific formula of the output error after the normalization is defined as: wherein, represents an output value of the BP neural network model, represents an actual value, represents a normalized output error.

4. The modular molten salt energy storage vapor heat release control method of claim 3, wherein: The BP neural network model according to the output error analysis comprises setting a minimum error ; When the BP neural network model is established; When the first neuron of the output layer are adjusted by the gradient descent method, and the specific adjustment formula is as follows: the first neuron of the output layer are adjusted by the gradient descent method, and the specific adjustment formula is as follows:​ wherein, denotes the learning rate, denotes the operator, denotes the actual value first derivative.

5. The modular molten salt energy storage vapor heat release control method of claim 4, wherein: The processing of the analyzed BP neural network model includes using a variable rate adjustment technique to adjust the learning rate based on the value of the adjusted learning rate at the iteration, outputting the frequency of the molten salt pump The specific calculation formula of the variable rate adjustment technique is: wherein, denotes a maximum learning rate, denotes a minimum learning rate, denotes a maximum number of iterations, denotes a current number of iterations, denotes a variable rate; The steam heat release control comprises substituting the number of nodes of the input layer into a BP neural network model of the output molten salt pump frequency , and transmitting the output molten salt pump frequency to a molten salt pump frequency converter, so as to adjust the heat exchange rate of the molten salt and the steam with the change of the molten salt flow, and complete the steam heat release control.

6. A modular molten salt energy storage vapor heat release control system based on the modular molten salt energy storage vapor heat release control method of any of claims 1-5, characterized in that: Comprising, The training module establishes the BP neural network model of the modular molten salt energy storage heat release control system, and trains the BP neural network model; The analysis module defines the output error after the normalization, and analyzes the BP neural network model according to the output error; The control module processes the analyzed BP neural network model by using the variable rate regulation technology, establishes the actual object mathematical model opposite to the processed BP neural network model, and completes the steam heat release control of the modular molten salt energy storage. 7.A computer device, comprising a memory and a processor, wherein the memory stores a computer program, and the computer device is characterized in that: The processor executes the computer program to realize the steps of the steam heat release control method of the modular molten salt energy storage according to any one of claims 1-5.

8. A computer readable storage medium having stored thereon a computer program, characterized in that: The computer program is executed by the processor to realize the steps of the steam heat release control method of the modular molten salt energy storage according to any one of claims 1-5.

Images