Method for identifying primary frequency modulation capability of photothermal generator set based on QPSO-LSTM

Abstract

The application provides a kind of QPSO-LSTM-based photovoltaic thermal power unit primary frequency modulation capability identification method, comprising: determining the input variable and output variable of frequency modulation capability identification model;Based on QPSO-LSTM, an initial primary frequency modulation capability identification model is constructed;The number of hidden layer nodes, learning rate and network training times are optimized by QPSO algorithm to obtain the globally optimal model super parameter;The globally optimal model super parameter is input into the LSTM in the initial primary frequency modulation capability identification model, and the LSTM is trained based on the training set until the primary frequency modulation capability identification model is obtained.The scheme of the application realizes the accurate identification of the primary frequency modulation capability of photovoltaic thermal power unit by establishing a primary frequency modulation capability identification model, can timely guide the power plant staff to control the operation state of photovoltaic thermal power unit, avoid large range fluctuation, improve the stability and reliability of power grid.

Description

Technical Field

[0001] This invention relates to the field of power grid dispatching technology, and in particular to a method for identifying the primary frequency regulation capability of a solar thermal power generator based on QPSO-LSTM. Background Technology

[0002] Frequency stability is an important indicator of power quality in a power grid, and it is crucial for the safe and stable operation of the grid. Frequency regulation is an important ancillary service for grid-connected generating units, and it is classified into primary frequency regulation, secondary frequency regulation, and tertiary frequency regulation based on its regulatory function and time effect.

[0003] Primary frequency regulation refers to the automatic adjustment of the turbine control valve opening by the unit's speed control system to regulate the power output of the unit by changing the main steam flow when load balance changes cause the grid frequency to deviate from the rated value during grid-connected unit operation. This adjustment is achieved by modifying the main steam flow rate to adapt to changes in external load. Primary frequency regulation relies on the control system itself to adjust power, offering a fast response time. However, it is a differential regulation method and cannot maintain a constant grid frequency; it can only mitigate frequency variations.

[0004] In practical applications, improving the identification of the primary frequency regulation capability of solar thermal power (CSP) units helps assist power grid dispatching and ensures the safe and stable operation of the grid. Traditional methods for assessing primary frequency regulation capability can only calculate the frequency regulation capability index of CSP units and cannot observe the dynamic changes of the units. Therefore, existing technologies cannot accurately identify the primary frequency regulation capability of CSP units, resulting in poor grid stability and reliability. Summary of the Invention

[0005] This invention provides a method for identifying the primary frequency regulation capability of solar thermal power generators based on QPSO-LSTM, which solves the shortcomings of existing technologies that cannot accurately identify the primary frequency regulation capability of solar thermal power generators, resulting in poor grid stability and reliability. This method enables accurate identification of the primary frequency regulation capability of solar thermal power generators, allowing power plant staff to control the operating status of solar thermal power generators in a timely manner, avoiding large-scale fluctuations, and improving the stability and reliability of the power grid.

[0006] This invention provides a method for identifying the primary frequency regulation capability of a solar thermal power generator based on QPSO-LSTM, comprising: By analyzing the mechanism of solar thermal power generation units, the change in unit output is used as the output variable of the primary frequency regulation capability identification model, and multiple operating state variables that are strongly correlated with the change in unit output are determined as the input variables of the primary frequency regulation capability identification model. Based on QPSO-LSTM, an initial primary frequency modulation capability identification model is constructed. The number of hidden layer nodes, learning rate, and number of network training iterations are optimized using the QPSO algorithm to obtain the globally optimal model hyperparameters. The globally optimal model hyperparameters are input into the LSTM in the initial primary frequency regulation capability identification model, and the LSTM is trained based on the training set until the primary frequency regulation capability identification model is obtained; wherein, the training set includes multiple training data, each training data includes historical input variables and the historical unit output change corresponding to each historical input variable.

[0007] According to the present invention, a method for identifying the primary frequency regulation capability of a solar thermal power generation unit based on QPSO-LSTM is provided, wherein determining multiple operating state variables strongly correlated with the output change of the unit as input variables of the primary frequency regulation capability identification model includes: Based on historical operating data under primary frequency regulation conditions, by analyzing the mechanism of the solar thermal power generator set, multiple operating state variables that can fully reflect the actual operating state of the solar thermal power generator set are determined. Calculate the Pearson correlation coefficient between each operating state variable and the change in unit output, and determine the operating state variables with the Pearson correlation coefficient greater than the first threshold as operating state variables that are strongly correlated with the change in unit output. The input variables of the primary frequency regulation capability identification model are composed of multiple operating state variables that are strongly correlated with the changes in unit output.

[0008] According to the present invention, a method for identifying the primary frequency regulation capability of a solar thermal power generator based on QPSO-LSTM is provided. The method involves optimizing the number of hidden layer nodes, learning rate, and network training iterations using the QPSO algorithm to obtain globally optimal model hyperparameters, including: Initialize particle swarm optimization parameters; wherein, the particle swarm optimization parameters include population size, number of iterations, spatial dimension and particle position, and the particle position includes the number of hidden layer nodes, learning rate and number of network training iterations; The hyperparameters of the LSTM obtained from the particle position are used to train the current primary frequency modulation capability identification model and calculate the particle fitness value. Based on the fitness value, determine the optimal solution for each individual particle and the current global optimal solution for the entire particle swarm, and update the particle positions. If the change in the fitness value meets the requirements or the current iteration count reaches the maximum iteration count, then the optimal particle position is output as the globally optimal model hyperparameter; otherwise, the process returns to the step of training the current one-time frequency modulation capability identification model by using the hyperparameters of the LSTM obtained from the particle position and calculating the particle fitness value.

[0009] According to the present invention, a method for identifying the primary frequency regulation capability of a solar thermal power generation unit based on QPSO-LSTM is provided. After training the LSTM based on the training set until a primary frequency regulation capability identification model is obtained, the method further includes: Extract the actual input variables from the actual operating state variables of the solar thermal power generator set; The actual input variables are input into the trained primary frequency regulation capability identification model to obtain the predicted change in unit output output of the trained primary frequency regulation capability identification model.

[0010] According to the present invention, a method for identifying the primary frequency regulation capability of a solar thermal power generation unit based on QPSO-LSTM is provided. The method involves constructing an initial primary frequency regulation capability identification model based on QPSO-LSTM, comprising: LSTM consists of an input layer, a data preprocessing layer, an LSTM network layer, and a prediction output layer. An optimization layer is constructed based on the QPSO algorithm to optimize the number of hidden layer nodes, learning rate, and number of network training iterations in the LSTM model.

[0011] According to the present invention, a method for identifying the primary frequency regulation capability of a solar thermal power generator based on QPSO-LSTM is provided, the method further comprising: The identification effectiveness of primary frequency modulation capability is measured by root mean square error and / or mean absolute error.

[0012] The present invention also provides a primary frequency regulation capability identification device for a solar thermal power generator based on QPSO-LSTM, comprising: The determination module is used to analyze the mechanism of the solar thermal power generator set, take the change in unit output as the output variable of the primary frequency regulation capability identification model, and determine multiple operating state variables that are strongly correlated with the change in unit output as the input variables of the primary frequency regulation capability identification model. The building module is used to construct an initial primary frequency modulation capability identification model based on QPSO-LSTM. The optimization module is used to optimize the number of hidden layer nodes, learning rate, and network training times using the QPSO algorithm to obtain the globally optimal model hyperparameters. The training module is used to input the globally optimal model hyperparameters into the LSTM in the initial primary frequency regulation capability identification model, and train the LSTM based on the training set until the primary frequency regulation capability identification model is obtained; wherein, the training set includes multiple training data, each training data includes historical input variables and the historical unit output change corresponding to each historical input variable.

[0013] The present invention also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the primary frequency regulation capability identification method for a solar thermal generator based on QPSO-LSTM as described above.

[0014] The present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the primary frequency regulation capability identification method for solar thermal generator sets based on QPSO-LSTM as described above.

[0015] The present invention also provides a computer program product, including a computer program that, when executed by a processor, implements the primary frequency regulation capability identification method for solar thermal generator sets based on QPSO-LSTM as described above.

[0016] The present invention provides a method for identifying the primary frequency regulation capability of a solar thermal power (CSP) generator based on QPSO-LSTM. By analyzing the mechanism of the CSP generator, the method uses the change in generator output and several operating state variables strongly correlated with this change as input and output variables of the primary frequency regulation capability identification model. The change in generator output characterizes the dynamic adjustment capability of the CSP generator during primary frequency regulation. This invention comprehensively considers the operating state of the CSP generator to achieve accurate modeling of its primary frequency regulation capability. Furthermore, an initial primary frequency regulation capability identification model is constructed based on QPSO-LSTM; the network structure is optimized using the QPSO algorithm to obtain the final primary frequency regulation capability identification model. Therefore, the present invention, by establishing a primary frequency regulation capability identification model, achieves accurate identification of the primary frequency regulation capability of the CSP generator, enabling timely guidance for power plant personnel to control the operating state of the CSP generator, avoiding large-scale fluctuations, and improving the stability and reliability of the power grid. Attached Figure Description

[0017] To more clearly illustrate the technical solutions in this invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0018] Figure 1 This is one of the flowcharts illustrating the method for identifying the primary frequency regulation capability of a solar thermal power generator based on QPSO-LSTM provided by this invention.

[0019] Figure 2 This is a schematic diagram of the static characteristic curve of the primary frequency regulation of the solar thermal unit provided by the present invention.

[0020] Figure 3 This is a schematic diagram of the frequency modulation capability identification model provided by the present invention.

[0021] Figure 4 This is the second flowchart illustrating the method for identifying the primary frequency regulation capability of a solar thermal power generator based on QPSO-LSTM provided by this invention.

[0022] Figure 5 This is a schematic diagram of the structure of the QPSO-LSTM-based solar thermal generator set primary frequency regulation capability identification device provided by the present invention.

[0023] Figure 6 This is a schematic diagram of the structure of the electronic device provided by the present invention. Detailed Implementation

[0024] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of this invention. All other embodiments obtained by those skilled in the art based on the embodiments of this invention without creative effort are within the scope of protection of this invention.

[0025] Frequency stability is an important indicator of power quality in a power grid, and it is crucial for the safe and stable operation of the grid. Frequency regulation is an important ancillary service for grid-connected generating units, and it is classified into primary frequency regulation, secondary frequency regulation, and tertiary frequency regulation based on its regulatory function and time effect.

[0026] Primary frequency regulation refers to the automatic adjustment of the turbine control valve opening by the unit's speed control system to regulate the power grid frequency when load balance changes cause deviations from the rated value. This adjustment, achieved by changing the main steam flow, adjusts the unit's output to adapt to changes in external load. Primary frequency regulation relies on the control system itself to regulate power, offering a fast response time. However, it is a differential regulation method and cannot maintain a constant grid frequency; it can only mitigate frequency fluctuations. Secondary frequency regulation involves adjusting the load through manual or automatic control methods to restore the grid frequency. Common methods include load adjustments via central dispatch and automatic control systems in the units. Secondary frequency regulation is error-free regulation. Tertiary frequency regulation focuses on the economical allocation of load, rationally utilizing electrical energy while ensuring grid frequency stability and safety. It is designed for loads that change slowly but fluctuate significantly.

[0027] In practical applications, improving the identification of the primary frequency regulation capability of solar thermal power (CSP) units helps assist power grid dispatching and ensures the safe and stable operation of the grid. Traditional methods for assessing primary frequency regulation capability can only calculate the frequency regulation capability index of CSP units and cannot observe the dynamic changes of the units. Therefore, existing technologies cannot accurately identify the primary frequency regulation capability of CSP units, resulting in poor grid stability and reliability.

[0028] To address the aforementioned technical problems, this invention provides a method for identifying the primary frequency regulation capability of a solar thermal power generator based on QPSO-LSTM.

[0029] The technical solution of this application and how it solves the above-mentioned technical problems are described in detail below with specific embodiments. These specific embodiments can be combined with each other, and the same or similar concepts or processes may not be described again in some embodiments. The following is a combination of... Figures 1-4 This invention describes a method for identifying the primary frequency regulation capability of a solar thermal power generator based on QPSO-LSTM.

[0030] In practical applications, the execution entity of this QPSO-LSTM-based solar thermal power generator primary frequency regulation capability identification method can be a QPSO-LSTM-based solar thermal power generator primary frequency regulation capability identification device. There are various ways to implement this device. For example, it can be implemented through a computer program, such as application software; or, for example, a chip; it can also be implemented as a medium storing the relevant computer program, such as a USB flash drive or cloud storage; or it can be implemented through a physical device that integrates or installs the relevant computer program, such as a server.

[0031] Figure 1 This is one of the flowcharts illustrating the method for identifying the primary frequency regulation capability of a solar thermal power generator based on QPSO-LSTM provided by this invention. Figure 1 As shown, the method includes the following steps 101 to 105.

[0032] Step 101: By analyzing the mechanism of the solar thermal power generator set, the change in the unit's output is used as the output variable of the primary frequency regulation capability identification model.

[0033] Step 102: Determine multiple operating state variables that are strongly correlated with the changes in unit output as input variables for the primary frequency regulation capability identification model.

[0034] The change in generator output refers to the difference in output power between the start and end of a certain time period. In practice, the change in generator output may be caused by various factors, including changes in load demand, fluctuations in system frequency, and adjustments made to maintain the stability and security of the power system. The change in generator output can be used to describe the generator's responsiveness to changes in grid demand, or the change in generator performance under specific conditions. In this embodiment, the change in generator output characterizes the primary frequency regulation capability of the solar thermal power generator; therefore, the change in generator output is used as the output variable of the primary frequency regulation capability identification model.

[0035] Figure 2 This is a schematic diagram of the static characteristic curve of the primary frequency regulation of the solar thermal power unit provided by the present invention, as shown in the figure. Figure 2 As shown, taking a 50MW solar thermal power generator with a rated power of Pe as an example, the dead zone of the unit is usually set to ±2r / min. When the unit speed is within the dead zone, no primary frequency regulation is performed, and the change in unit output ΔPw is zero. When the unit speed deviation |Δn|>2r / min, the unit begins primary frequency regulation, requiring rapid adjustment of the frequency change within a short period of time. This is reflected in the primary frequency regulation static characteristic curve as a rapid increase or decrease in the change in unit output ΔPw, with a maximum limit of ±6%Pe.

[0036] Specifically, primary frequency regulation involves two directions. When fluctuations in external load cause the grid frequency to rise, the speed control system issues valve commands to close the regulating valves, resulting in a decrease in main steam flow, an increase in main steam pressure and temperature, and a decrease in actual power generation. Conversely, when fluctuations in external load cause the grid frequency to fall, the speed control system issues valve commands to open the regulating valves, resulting in an increase in main steam flow, a decrease in main steam pressure and temperature, and an increase in actual power generation. Primary frequency regulation mainly adjusts the turbine's regulating valves, utilizing the heat storage capacity of the steam-water working fluid and metal. This is a short-term load adjustment; to achieve long-term adjustment, it is also necessary to adjust the boiler-side feedwater and molten salt flow.

[0037] During primary frequency regulation, changes in parameters and external disturbances can affect the output of concentrated solar power (CSP) generators. When changes in external load cause a decrease in grid frequency, the main steam flow is increased by widening the valve opening to improve the CSP generator output and compensate for the power loss. Simultaneously, the main steam pressure and temperature decrease. Therefore, changes in operation-related parameters can affect the primary frequency regulation capability of CSP generators.

[0038] To establish a primary frequency regulation capability identification model, appropriate feature variables need to be selected as input variables for the model. The primary frequency regulation operation of a concentrated solar power (CSP) unit involves numerous parameters with strong coupling; changes in these parameters affect the unit's output. By analyzing the CSP unit's mechanism and determining the model's input variables based on the closeness of the relationship between each parameter and the model's output vector (the change in unit output), the model's input variables can be identified.

[0039] In practical applications, the degree of correlation between each parameter and the change in unit output can be determined by calculating the correlation coefficient between each parameter and the change in unit output. In practice, the correlation coefficient is an indicator characteristic quantity used to describe the direction and degree of correlation between variables; the larger the correlation coefficient, the stronger the correlation between the variables. In this embodiment, no specific limitation is made to the correlation coefficient; for example, the correlation coefficient can be the Pearson correlation coefficient, partial correlation coefficient, distance correlation coefficient, etc. Further, based on the correlation coefficient, multiple operating state variables that are strongly correlated with the change in unit output are selected as input variables for the primary frequency regulation capability identification model.

[0040] Specifically, in one possible implementation, step 102 includes: Based on historical operating data under primary frequency regulation conditions, by analyzing the mechanism of the solar thermal power generator set, multiple operating state variables that can fully reflect the actual operating state of the solar thermal power generator set are determined. Calculate the Pearson correlation coefficient between each operating state variable and the change in unit output, and determine the operating state variables with a Pearson correlation coefficient greater than the first threshold as operating state variables that are strongly correlated with the change in unit output. The input variables of the primary frequency regulation capability identification model are composed of multiple operating state variables that are strongly correlated with the changes in unit output.

[0041] As an example, historical operating data of a 50MW concentrated solar power (CSP) generator unit under primary frequency regulation conditions were obtained. By analyzing the mechanism of the CSP generator unit, several operating state variables that can fully reflect the actual operating state of the CSP generator unit were identified. These operating state variables include: unit load, main steam flow rate, speed, main steam pressure, main steam temperature, feedwater flow rate, total valve opening, frequency, reheat steam pressure, reheat steam temperature, regulating stage pressure, and high-pressure exhaust temperature.

[0042] Existing methods for assessing primary frequency regulation capability can only calculate the frequency regulation capability index of solar thermal power generators (CSP) units, but cannot observe the dynamic changes of the units. In this embodiment, multiple operating state variables that are strongly correlated with the changes in unit output are selected as input variables for the primary frequency regulation capability identification model. Based on this, a primary frequency regulation capability identification model is constructed to achieve the identification of the primary frequency regulation capability of CSP units, thereby improving the accuracy of the identification.

[0043] Furthermore, the Pearson correlation coefficient method was used to analyze and calculate the correlation between each operating state variable and the change in unit output. Specifically, the Pearson correlation coefficient between each operating state variable and the change in unit output was calculated. The formula for calculating the Pearson correlation coefficient K is as follows: In the above formula, n Indicates the number of data points ；X and Y Two sets of data, representing the degree of correlation to be calculated; and They represent X and Y The sample mean difference.

[0044] In practice, the Pearson correlation coefficient ranges from -1 to 1. When K < 0, it indicates that the variables change in the same direction. The absolute value of K reflects the degree of correlation between the variables. Generally, |K| > 0.8 is considered a strong correlation.

[0045] In this embodiment, multiple operating state variables strongly correlated with changes in unit output are used as input variables for the primary frequency regulation capability identification model. Specifically, based on correlation calculations, the multiple operating state variables strongly correlated with changes in unit output include: the actual power of the solar thermal power generator unit, load command, main steam temperature, main steam pressure, main steam flow rate, speed, and total valve opening. Further, these multiple operating state variables strongly correlated with changes in unit output are used as input variables for the primary frequency regulation capability identification model.

[0046] Step 103: Based on QPSO-LSTM, construct the initial primary frequency modulation capability identification model.

[0047] QPSO-LSTM (QPSO algorithm-LSTM) is a prediction model that combines Quantum Particle Swarm Optimization (QPSO) and Long Short-Term Memory (LSTM). In the QPSO-LSTM model, QPSO is used to optimize the hyperparameters of the LSTM, which may include the learning rate, the number of hidden layer nodes, and the number of training epochs. Optimization with QPSO can improve the prediction accuracy and convergence speed of the LSTM model. QPSO is a quantum computing-based optimization algorithm that quickly finds the optimal solution globally through information sharing and cooperation among particles. LSTM networks are a special type of recurrent neural network capable of learning long-term dependencies in sequential data. With appropriate parameter settings, it can effectively process time series data and predict future values. The core idea of ​​LSTM is to control the flow and storage of information through three gates (input gate, forget gate, and output gate), thereby solving the gradient vanishing problem in traditional RNNs.

[0048] Specifically, in one possible implementation, step 103 includes: LSTM consists of an input layer, a data preprocessing layer, an LSTM network layer, and a prediction output layer. An optimization layer is constructed based on the QPSO algorithm to optimize the number of hidden layer nodes, learning rate, and number of network training iterations in the LSTM model.

[0049] Figure 3 This is a schematic diagram of the primary frequency modulation capability identification model provided by the present invention, as shown below. Figure 3 As shown, constructing the initial primary frequency modulation (PMMC) capability identification model consists of two parts: building an LSTM model and optimizing the particle swarm optimization (PSO) parameters. The constructed LSTM model contains a two-hidden-layer network, and the selection of LSTM model parameters has a significant impact on accuracy. Therefore, the QPSO algorithm is proposed to optimize the relevant parameters of the LSTM, thereby establishing the initial PMMC capability identification model. It should be noted that the initial PMMC capability identification model refers to the untrained model; after training, the PMMC capability identification model can be obtained.

[0050] like Figure 3As shown, the Input Layer receives input variables (multiple operating state variables strongly correlated with changes in unit output) and transmits them to the Preprocessing Layer. The Preprocessing Layer normalizes the input variables before feeding them into the LSTM Layers. Each LSTM layer consists of multiple stacked LSTM units to increase the model's depth and learning capacity. Each LSTM unit updates its internal state at each time step of the sequence. The LSTM layers are responsible for learning the temporal dynamics and long-term dependencies of the input data. The LSTM layers control the flow of information through their unique gating mechanisms (forget gate, input gate, output gate) to capture key patterns in the time series data. The Output Layer denormalizes the information output by the LSTM layers and outputs the changes in unit output. During LSTM model training, the optimization layer optimizes the number of hidden nodes, learning rate, and number of training iterations.

[0051] Step 104: Optimize the number of hidden layer nodes, learning rate, and number of network training iterations using the QPSO algorithm to obtain the globally optimal model hyperparameters.

[0052] In practical applications, the selection of hyperparameters in a Long Short-Term Memory (LSTM) neural network directly affects its prediction accuracy. Therefore, the QPSO algorithm is used to find the optimal network parameters. Specifically, during the training of the LSM network model, the number of hidden layer neurons, the learning rate, and the number of training rounds are used as optimization variables for particles. By continuously updating the particle's velocity and position, the fitness value of the objective function is calculated and compared, and finally, the globally optimal model hyperparameters are found. The globally optimal model hyperparameters are then returned to the LSM network model structure for training.

[0053] As an example, in one possible implementation, Figure 4 This is the second flowchart illustrating the method for identifying the primary frequency regulation capability of a solar thermal power generator based on QPSO-LSTM provided by this invention. Figure 4 As shown, step 104 above includes steps 401 to 406.

[0054] Step 401: Initialize particle swarm parameters.

[0055] Among them, the particle swarm related parameters include population size, number of iterations, spatial dimension and particle position. Particle position includes the number of hidden layer nodes, learning rate and number of network training iterations.

[0056] Step 402: Train the current primary frequency modulation capability identification model based on the hyperparameters of the LSTM obtained from the particle position, and calculate the fitness value of the particle.

[0057] Specifically, the mean squared error of the training set is used as the particle fitness function. f Its expression is: In the above formula, N Indicates the number of samples in the training set; and These represent the predicted output value and the true value of the training set samples, respectively.

[0058] Step 403: Based on the fitness value, determine the current optimal solution for each individual particle and the current global optimal solution for the entire particle swarm, and update the position of the particles.

[0059] Step 404: Determine whether the termination condition is met.

[0060] The termination conditions mentioned above include: the change in fitness value meets the requirements or the current iteration number reaches the maximum iteration number.

[0061] Step 405: If the termination condition is met, output the optimal particle position as the globally optimal model hyperparameter.

[0062] Step 406: If the termination condition is not met, return to step 402.

[0063] In this embodiment, the number of hidden layer nodes, learning rate, and network training iterations are optimized using the QPSO algorithm to obtain the globally optimal model hyperparameters. This significantly improves the performance of the LSTM model, including prediction accuracy, convergence speed, and generalization ability. Furthermore, the QPSO algorithm automates the hyperparameter search process, reducing the need for manual debugging and saving time and resources.

[0064] Step 105: Input the globally optimal model hyperparameters into the LSTM in the initial primary frequency modulation capability identification model, and train the LSTM based on the training set until the primary frequency modulation capability identification model is obtained.

[0065] The training set includes multiple training data sets, each containing historical input variables and the corresponding historical changes in unit output for each historical input variable.

[0066] In practical applications, training, validation, and test sets can be set. Specifically, during training, the training set data is used to train the LSTM model, and the validation set is used to adjust the model's hyperparameters and perform model selection. After training is complete, a separate test set is used to evaluate the model's performance.

[0067] Optionally, in one possible implementation, the above method further includes: The identification effectiveness of primary frequency modulation capability is measured by root mean square error and / or mean absolute error.

[0068] Specifically, the root mean square error (RMSE) measures the square root of the mean of the squares of the differences between the predicted and actual values; a smaller RMSE indicates higher prediction accuracy. The mean absolute error (MAE) measures the average of the absolute values ​​of the differences between the predicted and actual values; similarly, a smaller MAE indicates higher prediction accuracy.

[0069] The formula for calculating the root mean square error (RMSE) is as follows: In the above formula, N Indicates the number of samples in the test set; and These represent the predicted output value and the true value of the test training set samples, respectively.

[0070] The formula for calculating the mean absolute error (MAE) is as follows: In the above formula, N Indicates the number of samples in the test set; and These represent the predicted output value and the true value of the test set sample, respectively.

[0071] In this embodiment, root mean square error and mean absolute error are introduced to measure the identification effect of the model, so as to achieve accurate evaluation of the model identification accuracy and improve the reliability and accuracy of the identification of the primary frequency regulation capability of the solar thermal power generation unit.

[0072] Based on the above explanation, the globally optimal model hyperparameters are input into the LSTM in the initial primary frequency regulation capability identification model, and the LSTM is trained using the training set to obtain the primary frequency regulation capability identification model. On this basis, the primary frequency regulation capability identification of a solar thermal power generation unit can be achieved using the trained primary frequency regulation capability identification model.

[0073] Specifically, in one possible implementation, after step 105, the method further includes: Extract the actual input variables from the actual operating state variables of the solar thermal power generator set; The actual input variables are input into the trained primary frequency regulation capability identification model to obtain the predicted unit output change output by the trained primary frequency regulation capability identification model.

[0074] Specifically, multiple operating state variables strongly correlated with changes in unit output are extracted from the actual operating state variables of the solar thermal power generator set to form actual input variables. For example, the actual input variables include: the actual power of the solar thermal power generator set, load command, main steam temperature, main steam pressure, main steam flow rate, speed, and total valve opening.

[0075] In this embodiment, actual input variables are extracted from the actual operating state variables of the solar thermal power generator unit and input into the trained primary frequency regulation capability identification model. The predicted output change of the unit can be obtained from the trained primary frequency regulation capability identification model, which can accurately identify the primary frequency regulation capability of the solar thermal power generator unit. This can guide power plant staff to control the operating status of the solar thermal power generator unit in a timely manner, avoid large-scale fluctuations, and improve the stability and reliability of the power grid.

[0076] The QPSO-LSTM-based method for identifying the primary frequency regulation capability of solar thermal power (CSP) generators provided in this embodiment analyzes the mechanism of CSP generators and uses the change in generator output and several operating state variables strongly correlated with the change in generator output as the input and output variables of the primary frequency regulation capability identification model. The change in generator output characterizes the dynamic adjustment capability of the CSP generator during primary frequency regulation. This scheme comprehensively considers the operating state of the CSP generator to achieve accurate modeling of its primary frequency regulation capability. Furthermore, an initial primary frequency regulation capability identification model is constructed based on QPSO-LSTM; the network structure is optimized using the QPSO algorithm to obtain the final primary frequency regulation capability identification model. Therefore, the scheme in this embodiment achieves accurate identification of the primary frequency regulation capability of CSP generators by establishing a primary frequency regulation capability identification model, which can promptly guide power plant personnel to control the operating state of CSP generators, avoid large-scale fluctuations, and improve the stability and reliability of the power grid.

[0077] The following describes the primary frequency regulation capability identification device for solar thermal power generators based on QPSO-LSTM provided by the present invention. The primary frequency regulation capability identification device for solar thermal power generators based on QPSO-LSTM described below can be referred to in correspondence with the primary frequency regulation capability identification method for solar thermal power generators based on QPSO-LSTM described above.

[0078] Figure 5 This is a schematic diagram of the primary frequency regulation capability identification device for solar thermal power generator sets based on QPSO-LSTM provided by the present invention, as shown below. Figure 5As shown, the primary frequency regulation capability identification device for solar thermal power generation units based on QPSO-LSTM includes: a determination module 51, a construction module 52, an optimization module 53, and a training module 54.

[0079] The aforementioned determining module 51 is used to analyze the mechanism of the solar thermal power generator unit, take the change in unit output as the output variable of the primary frequency regulation capability identification model, and determine multiple operating state variables that are strongly correlated with the change in unit output as the input variables of the primary frequency regulation capability identification model.

[0080] The change in generator output refers to the difference in output power between the start and end of a certain time period. In practice, the change in generator output may be caused by various factors, including changes in load demand, fluctuations in system frequency, and adjustments made to maintain the stability and security of the power system. The change in generator output can be used to describe the generator's responsiveness to changes in grid demand, or the change in generator performance under specific conditions. In this embodiment, the change in generator output characterizes the primary frequency regulation capability of the solar thermal power generator; therefore, the change in generator output is used as the output variable of the primary frequency regulation capability identification model.

[0081] Specifically, primary frequency regulation involves two directions. When fluctuations in external load cause the grid frequency to rise, the speed control system issues valve commands to close the regulating valves, resulting in a decrease in main steam flow, an increase in main steam pressure and temperature, and a decrease in actual power generation. Conversely, when fluctuations in external load cause the grid frequency to fall, the speed control system issues valve commands to open the regulating valves, resulting in an increase in main steam flow, a decrease in main steam pressure and temperature, and an increase in actual power generation. Primary frequency regulation mainly adjusts the turbine's regulating valves, utilizing the heat storage capacity of the steam-water working fluid and metal. This is a short-term load adjustment; to achieve long-term adjustment, it is also necessary to adjust the boiler-side feedwater and molten salt flow.

[0082] During primary frequency regulation, changes in parameters and external disturbances can affect the output of concentrated solar power (CSP) generators. When changes in external load cause a decrease in grid frequency, the main steam flow is increased by widening the valve opening to improve the CSP generator output and compensate for the power loss. Simultaneously, the main steam pressure and temperature decrease. Therefore, changes in operation-related parameters can affect the primary frequency regulation capability of CSP generators.

[0083] To establish a primary frequency regulation capability identification model, appropriate feature variables need to be selected as input variables for the model. The primary frequency regulation operation of a concentrated solar power (CSP) unit involves numerous parameters with strong coupling; changes in these parameters affect the unit's output. By analyzing the CSP unit's mechanism and determining the model's input variables based on the closeness of the relationship between each parameter and the model's output vector (the change in unit output), the model's input variables can be identified.

[0084] In practical applications, the degree of correlation between each parameter and the change in unit output can be determined by calculating the correlation coefficient between each parameter and the change in unit output. In practice, the correlation coefficient is an indicator characteristic quantity used to describe the direction and degree of correlation between variables; the larger the correlation coefficient, the stronger the correlation between the variables. In this embodiment, no specific limitation is made to the correlation coefficient; for example, the correlation coefficient can be the Pearson correlation coefficient, partial correlation coefficient, distance correlation coefficient, etc. Further, based on the correlation coefficient, multiple operating state variables that are strongly correlated with the change in unit output are selected as input variables for the primary frequency regulation capability identification model.

[0085] Specifically, in one possible implementation, when the determining module 51 is used to determine multiple operating state variables that are strongly correlated with the change in unit output as input variables for the primary frequency regulation capability identification model, it is specifically used for: Based on historical operating data under primary frequency regulation conditions, by analyzing the mechanism of the solar thermal power generator set, multiple operating state variables that can fully reflect the actual operating state of the solar thermal power generator set are determined. Calculate the Pearson correlation coefficient between each operating state variable and the change in unit output, and determine the operating state variables with a Pearson correlation coefficient greater than the first threshold as operating state variables that are strongly correlated with the change in unit output. The input variables of the primary frequency regulation capability identification model are composed of multiple operating state variables that are strongly correlated with the changes in unit output.

[0086] As an example, historical operating data of a 50MW concentrated solar power (CSP) generator unit under primary frequency regulation conditions were obtained. By analyzing the mechanism of the CSP generator unit, several operating state variables that can fully reflect the actual operating state of the CSP generator unit were identified. These operating state variables include: unit load, main steam flow rate, speed, main steam pressure, main steam temperature, feedwater flow rate, total valve opening, frequency, reheat steam pressure, reheat steam temperature, regulating stage pressure, and high-pressure exhaust temperature.

[0087] Existing methods for assessing primary frequency regulation capability can only calculate the frequency regulation capability index of solar thermal power generators (CSP) units, but cannot observe the dynamic changes of the units. In this embodiment, multiple operating state variables that are strongly correlated with the changes in unit output are selected as input variables for the primary frequency regulation capability identification model. Based on this, a primary frequency regulation capability identification model is constructed to achieve the identification of the primary frequency regulation capability of CSP units, thereby improving the accuracy of the identification.

[0088] Furthermore, the Pearson correlation coefficient method was used to analyze and calculate the correlation between each operating state variable and the change in unit output. Specifically, the Pearson correlation coefficient between each operating state variable and the change in unit output was calculated. The formula for calculating the Pearson correlation coefficient K is as follows: In the above formula, n Indicates the number of data points ；X and Y Two sets of data, representing the degree of correlation to be calculated; and They represent X and Y The sample mean difference.

[0089] In practice, the Pearson correlation coefficient ranges from -1 to 1. When K < 0, it indicates that the variables change in the same direction. The absolute value of K reflects the degree of correlation between the variables. Generally, |K| > 0.8 is considered a strong correlation.

[0090] In this embodiment, multiple operating state variables strongly correlated with changes in unit output are used as input variables for the primary frequency regulation capability identification model. Specifically, based on correlation calculations, the multiple operating state variables strongly correlated with changes in unit output include: the actual power of the solar thermal power generator unit, load command, main steam temperature, main steam pressure, main steam flow rate, speed, and total valve opening. Further, these multiple operating state variables strongly correlated with changes in unit output are used as input variables for the primary frequency regulation capability identification model.

[0091] The aforementioned module 52 is used to construct an initial primary frequency modulation capability identification model based on QPSO-LSTM.

[0092] QPSO-LSTM (QPSO algorithm-LSTM) is a prediction model that combines Quantum Particle Swarm Optimization (QPSO) and Long Short-Term Memory (LSTM). In the QPSO-LSTM model, QPSO is used to optimize the hyperparameters of the LSTM, which may include the learning rate, the number of hidden layer nodes, and the number of training epochs. Optimization with QPSO can improve the prediction accuracy and convergence speed of the LSTM model. QPSO is a quantum computing-based optimization algorithm that quickly finds the optimal solution globally through information sharing and cooperation among particles. LSTM networks are a special type of recurrent neural network capable of learning long-term dependencies in sequential data. With appropriate parameter settings, it can effectively process time series data and predict future values. The core idea of ​​LSTM is to control the flow and storage of information through three gates (input gate, forget gate, and output gate), thereby solving the gradient vanishing problem in traditional RNNs.

[0093] Specifically, in one possible implementation, the aforementioned building module 52 is specifically used for: LSTM consists of an input layer, a data preprocessing layer, an LSTM network layer, and a prediction output layer. An optimization layer is constructed based on the QPSO algorithm to optimize the number of hidden layer nodes, learning rate, and number of network training iterations in the LSTM model.

[0094] like Figure 3 As shown, constructing the initial primary frequency modulation (PMMC) capability identification model consists of two parts: building an LSTM model and optimizing the particle swarm optimization (PSO) parameters. The constructed LSTM model contains a two-hidden-layer network, and the selection of LSTM model parameters has a significant impact on accuracy. Therefore, the QPSO algorithm is proposed to optimize the relevant parameters of the LSTM, thereby establishing the initial PMMC capability identification model. It should be noted that the initial PMMC capability identification model refers to the untrained model; after training, the PMMC capability identification model can be obtained.

[0095] like Figure 3As shown, the Input Layer receives input variables (multiple operating state variables strongly correlated with changes in unit output) and transmits them to the Preprocessing Layer. The Preprocessing Layer normalizes the input variables before feeding them into the LSTM Layers. Each LSTM layer consists of multiple stacked LSTM units to increase the model's depth and learning capacity. Each LSTM unit updates its internal state at each time step of the sequence. The LSTM layers are responsible for learning the temporal dynamics and long-term dependencies of the input data. The LSTM layers control the flow of information through their unique gating mechanisms (forget gate, input gate, output gate) to capture key patterns in the time series data. The Output Layer denormalizes the information output by the LSTM layers and outputs the changes in unit output. During LSTM model training, the optimization layer optimizes the number of hidden nodes, learning rate, and number of training iterations.

[0096] The aforementioned optimization module 53 is used to optimize the number of hidden layer nodes, learning rate, and network training times using the QPSO algorithm to obtain the globally optimal model hyperparameters.

[0097] In practical applications, the selection of hyperparameters in a Long Short-Term Memory (LSTM) neural network directly affects its prediction accuracy. Therefore, the QPSO algorithm is used to find the optimal network parameters. Specifically, during the training of the LSM network model, the number of hidden layer neurons, the learning rate, and the number of training rounds are used as optimization variables for particles. By continuously updating the particle's velocity and position, the fitness value of the objective function is calculated and compared, and finally, the globally optimal model hyperparameters are found. The globally optimal model hyperparameters are then returned to the LSM network model structure for training.

[0098] As an example, the optimization module 53 mentioned above includes: The initialization unit is used to initialize particle swarm-related parameters.

[0099] Among them, the particle swarm related parameters include population size, number of iterations, spatial dimension and particle position. Particle position includes the number of hidden layer nodes, learning rate and number of network training iterations.

[0100] The computational unit is used to train the current primary frequency modulation capability identification model based on the hyperparameters of the LSTM obtained from the particle position, and to calculate the fitness value of the particle.

[0101] Specifically, the mean squared error of the training set is used as the particle fitness function. f Its expression is: In the above formula, N Indicates the number of samples in the training set; and These represent the predicted output value and the true value of the training set samples, respectively.

[0102] The update unit is used to determine the current optimal solution for each individual particle and the current global optimal solution for the entire particle swarm based on the fitness value, and to update the position of the particles.

[0103] The processing unit is used to output the optimal particle position as the globally optimal model hyperparameter if the change in fitness value meets the requirements or the current iteration number reaches the maximum iteration number; otherwise, it returns to the step of training the current one-time frequency modulation capability identification model by using the hyperparameters of the LSTM obtained from the particle position, and calculating the particle fitness value.

[0104] In this embodiment, the number of hidden layer nodes, learning rate, and network training iterations are optimized using the QPSO algorithm to obtain the globally optimal model hyperparameters. This significantly improves the performance of the LSTM model, including prediction accuracy, convergence speed, and generalization ability. Furthermore, the QPSO algorithm automates the hyperparameter search process, reducing the need for manual debugging and saving time and resources.

[0105] The aforementioned training module 54 is used to input the globally optimal model hyperparameters into the LSTM in the initial primary frequency modulation capability identification model, and to train the LSTM based on the training set until the primary frequency modulation capability identification model is obtained.

[0106] The training set includes multiple training data sets, each containing historical input variables and the corresponding historical changes in unit output for each historical input variable.

[0107] In practical applications, training, validation, and test sets can be set. Specifically, during training, the training set data is used to train the LSTM model, and the validation set is used to adjust the model's hyperparameters and perform model selection. After training is complete, a separate test set is used to evaluate the model's performance.

[0108] Optionally, in one possible implementation, the above-described apparatus further includes: The verification module is used to measure the identification effect of primary frequency modulation capability identification by root mean square error and / or mean absolute error.

[0109] Specifically, the root mean square error (RMSE) measures the square root of the mean of the squares of the differences between the predicted and actual values; a smaller RMSE indicates higher prediction accuracy. The mean absolute error (MAE) measures the average of the absolute values ​​of the differences between the predicted and actual values; similarly, a smaller MAE indicates higher prediction accuracy.

[0110] The formula for calculating the root mean square error (RMSE) is as follows: In the above formula, N Indicates the number of samples in the test set; and These represent the predicted output value and the true value of the test training set samples, respectively.

[0111] The formula for calculating the mean absolute error (MAE) is as follows: In the above formula, N Indicates the number of samples in the test set; and These represent the predicted output value and the true value of the test set sample, respectively.

[0112] In this embodiment, root mean square error and mean absolute error are introduced to measure the identification effect of the model, so as to achieve accurate evaluation of the model identification accuracy and improve the reliability and accuracy of the identification of the primary frequency regulation capability of the solar thermal power generation unit.

[0113] Based on the above explanation, the globally optimal model hyperparameters are input into the LSTM in the initial primary frequency regulation capability identification model, and the LSTM is trained using the training set to obtain the primary frequency regulation capability identification model. On this basis, the primary frequency regulation capability identification of a solar thermal power generation unit can be achieved using the trained primary frequency regulation capability identification model.

[0114] Specifically, in one possible implementation, the above-described apparatus further includes: an application module, configured to: Extract the actual input variables from the actual operating state variables of the solar thermal power generator set; The actual input variables are input into the trained primary frequency regulation capability identification model to obtain the predicted unit output change output by the trained primary frequency regulation capability identification model.

[0115] Specifically, multiple operating state variables strongly correlated with changes in unit output are extracted from the actual operating state variables of the solar thermal power generator set to form actual input variables. For example, the actual input variables include: the actual power of the solar thermal power generator set, load command, main steam temperature, main steam pressure, main steam flow rate, speed, and total valve opening.

[0116] In this embodiment, actual input variables are extracted from the actual operating state variables of the solar thermal power generator unit and input into the trained primary frequency regulation capability identification model. The predicted output change of the unit can be obtained from the trained primary frequency regulation capability identification model, which can accurately identify the primary frequency regulation capability of the solar thermal power generator unit. This can guide power plant staff to control the operating status of the solar thermal power generator unit in a timely manner, avoid large-scale fluctuations, and improve the stability and reliability of the power grid.

[0117] In the QPSO-LSTM-based primary frequency regulation capability identification device for solar thermal power (CSP) generators provided in this embodiment, the determination module analyzes the mechanism of the CSP generator and uses the change in generator output and several operating state variables strongly correlated with the change in generator output as the input and output variables of the primary frequency regulation capability identification model, respectively. The change in generator output characterizes the dynamic adjustment capability of the CSP generator during primary frequency regulation. This scheme comprehensively considers the operating state of the CSP generator to achieve accurate modeling of its primary frequency regulation capability. Furthermore, the construction module, based on QPSO-LSTM, constructs an initial primary frequency regulation capability identification model; the optimization module uses the QPSO algorithm to optimize the network structure, obtaining the final primary frequency regulation capability identification model. Therefore, this scheme, by establishing a primary frequency regulation capability identification model, achieves accurate identification of the primary frequency regulation capability of the CSP generator, enabling timely guidance for power plant personnel to control the operating state of the CSP generator, avoiding large-scale fluctuations, and improving the stability and reliability of the power grid.

[0118] Figure 6 This is a schematic diagram of the structure of the electronic device provided by the present invention, such as... Figure 6As shown, the electronic device may include: a processor 610, a communications interface 620, a memory 630, and a communications bus 640, wherein the processor 610, the communications interface 620, and the memory 630 communicate with each other through the communications bus 640. The processor 610 can call logic instructions in the memory 630 to execute a QPSO-LSTM-based method for identifying the primary frequency regulation capability of a solar thermal power generator. This method includes: analyzing the mechanism of the solar thermal power generator, using the change in generator output as the output variable of the primary frequency regulation capability identification model, and determining multiple operating state variables strongly correlated with the change in generator output as input variables of the primary frequency regulation capability identification model; constructing an initial primary frequency regulation capability identification model based on QPSO-LSTM; optimizing the number of hidden layer nodes, learning rate, and number of network training iterations using the QPSO algorithm to obtain globally optimal model hyperparameters; inputting the globally optimal model hyperparameters into the LSTM in the initial primary frequency regulation capability identification model, and training the LSTM based on the training set until the primary frequency regulation capability identification model is obtained; wherein, the training set includes multiple training data points, each training data point including historical input variables and the historical change in generator output corresponding to each historical input variable.

[0119] Furthermore, the logical instructions in the aforementioned memory 630 can be implemented as software functional units and, when sold or used as independent products, can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0120] On the other hand, the present invention also provides a computer program product, which includes a computer program that can be stored on a non-transitory computer-readable storage medium. When the computer program is executed by a processor, the computer can execute the QPSO-LSTM-based primary frequency regulation capability identification method for solar thermal power generators provided by the above methods. The method includes: analyzing the mechanism of the solar thermal power generator, using the change in unit output as the output variable of the primary frequency regulation capability identification model, and determining multiple operating state variables strongly correlated with the change in unit output as input variables of the primary frequency regulation capability identification model; constructing an initial primary frequency regulation capability identification model based on QPSO-LSTM; optimizing the number of hidden layer nodes, learning rate, and network training times using the QPSO algorithm to obtain the globally optimal model hyperparameters; inputting the globally optimal model hyperparameters into the LSTM in the initial primary frequency regulation capability identification model, and training the LSTM based on the training set until the primary frequency regulation capability identification model is obtained; wherein, the training set includes multiple training data, each training data including historical input variables and the historical unit output change corresponding to each historical input variable.

[0121] In another aspect, the present invention also provides a non-transitory computer-readable storage medium storing a computer program thereon. When executed by a processor, the computer program implements the QPSO-LSTM-based primary frequency regulation capability identification method for concentrated solar power (CSP) generators provided by the above methods. This method includes: analyzing the mechanism of the CSP generator, using the change in generator output as the output variable of the primary frequency regulation capability identification model, and determining multiple operating state variables strongly correlated with the change in generator output as input variables of the primary frequency regulation capability identification model; constructing an initial primary frequency regulation capability identification model based on QPSO-LSTM; optimizing the number of hidden layer nodes, learning rate, and network training iterations using the QPSO algorithm to obtain globally optimal model hyperparameters; inputting the globally optimal model hyperparameters into the LSTM in the initial primary frequency regulation capability identification model, and training the LSTM based on the training set until the primary frequency regulation capability identification model is obtained; wherein the training set includes multiple training data sets, each training data set including historical input variables and the historical generator output change corresponding to each historical input variable.

[0122] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Those skilled in the art can understand and implement this without any creative effort.

[0123] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods described in the various embodiments or some parts of the embodiments.

[0124] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for identifying the primary frequency regulation capability of a solar thermal power generator based on QPSO-LSTM, characterized in that, include: By analyzing the mechanism of solar thermal power generation units, the change in unit output is used as the output variable of the primary frequency regulation capability identification model, and multiple operating state variables that are strongly correlated with the change in unit output are determined as the input variables of the primary frequency regulation capability identification model. Based on QPSO-LSTM, an initial primary frequency modulation capability identification model is constructed. The number of hidden layer nodes, learning rate, and number of network training iterations are optimized using the QPSO algorithm to obtain the globally optimal model hyperparameters. The globally optimal model hyperparameters are input into the LSTM in the initial primary frequency regulation capability identification model, and the LSTM is trained based on the training set until the primary frequency regulation capability identification model is obtained; wherein, the training set includes multiple training data, each training data includes historical input variables and the historical unit output change corresponding to each historical input variable.

2. The method for identifying the primary frequency regulation capability of a solar thermal power generator based on QPSO-LSTM according to claim 1, characterized in that, The determination of multiple operating state variables that are strongly correlated with the change in unit output as input variables for the primary frequency regulation capability identification model includes: Based on historical operating data under primary frequency regulation conditions, by analyzing the mechanism of the solar thermal power generator set, multiple operating state variables that can fully reflect the actual operating state of the solar thermal power generator set are determined. Calculate the Pearson correlation coefficient between each operating state variable and the change in unit output, and determine the operating state variables with the Pearson correlation coefficient greater than the first threshold as operating state variables that are strongly correlated with the change in unit output. The input variables of the primary frequency regulation capability identification model are composed of multiple operating state variables that are strongly correlated with the changes in unit output.

3. The method for identifying the primary frequency regulation capability of a solar thermal power generator based on QPSO-LSTM according to claim 1, characterized in that, The process of optimizing the number of hidden layer nodes, learning rate, and network training iterations using the QPSO algorithm to obtain globally optimal model hyperparameters includes: Initialize particle swarm optimization parameters; wherein, the particle swarm optimization parameters include population size, number of iterations, spatial dimension and particle position, and the particle position includes the number of hidden layer nodes, learning rate and number of network training iterations; The hyperparameters of the LSTM obtained from the particle position are used to train the current primary frequency modulation capability identification model and calculate the particle fitness value. Based on the fitness value, determine the optimal solution for each individual particle and the current global optimal solution for the entire particle swarm, and update the particle positions. If the change in the fitness value meets the requirements or the current iteration count reaches the maximum iteration count, then the optimal particle position is output as the globally optimal model hyperparameter; otherwise, the process returns to the step of training the current one-time frequency modulation capability identification model by using the hyperparameters of the LSTM obtained from the particle position and calculating the particle fitness value.

4. The method for identifying the primary frequency regulation capability of a solar thermal power generator based on QPSO-LSTM according to claim 1, characterized in that, After training the LSTM based on the training set until a frequency modulation capability identification model is obtained, the method further includes: Extract the actual input variables from the actual operating state variables of the solar thermal power generator set; The actual input variables are input into the trained primary frequency regulation capability identification model to obtain the predicted change in unit output output of the trained primary frequency regulation capability identification model.

5. The method for identifying the primary frequency regulation capability of a solar thermal power generator based on QPSO-LSTM according to claim 1, characterized in that, The initial primary frequency modulation capability identification model based on QPSO-LSTM includes: LSTM consists of an input layer, a data preprocessing layer, an LSTM network layer, and a prediction output layer. An optimization layer is constructed based on the QPSO algorithm to optimize the number of hidden layer nodes, learning rate, and number of network training iterations in the LSTM model.

6. The method for identifying the primary frequency regulation capability of a solar thermal power generator based on QPSO-LSTM according to any one of claims 1-5, characterized in that, The method further includes: The identification effectiveness of primary frequency modulation capability is measured by root mean square error and / or mean absolute error.

7. A device for identifying the primary frequency regulation capability of a solar thermal power generator based on QPSO-LSTM, characterized in that, include: The determination module is used to analyze the mechanism of the solar thermal power generator set, take the change in unit output as the output variable of the primary frequency regulation capability identification model, and determine multiple operating state variables that are strongly correlated with the change in unit output as the input variables of the primary frequency regulation capability identification model. The building module is used to construct an initial primary frequency modulation capability identification model based on QPSO-LSTM. The optimization module is used to optimize the number of hidden layer nodes, learning rate, and network training times using the QPSO algorithm to obtain the globally optimal model hyperparameters. The training module is used to input the globally optimal model hyperparameters into the LSTM in the initial primary frequency regulation capability identification model, and train the LSTM based on the training set until the primary frequency regulation capability identification model is obtained; wherein, the training set includes multiple training data, each training data includes historical input variables and the historical unit output change corresponding to each historical input variable.

8. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the method for identifying the primary frequency regulation capability of a solar thermal power generator based on QPSO-LSTM as described in any one of claims 1 to 6.

9. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the method for identifying the primary frequency regulation capability of a solar thermal power generator based on QPSO-LSTM as described in any one of claims 1 to 6.

10. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by the processor, it implements the method for identifying the primary frequency regulation capability of a solar thermal power generator based on QPSO-LSTM as described in any one of claims 1 to 6.

Images