A molten salt energy storage agc adaptive control method and system

Abstract

The application provides a molten salt energy storage AGC adaptive control method and system, belongs to the automatic control and energy storage frequency modulation technical field, and can significantly alleviate or solve the problems that the PID parameters of the existing molten salt energy storage participating in AGC frequency modulation depend on manual setting, are difficult to adapt to frequency modulation rules and working condition changes, and are prone to control impact. The method obtains an AGC instruction value, an actual power value of a generator set and an actual output value of the molten salt energy storage, determines a target output value, an output control error, a normalized error and a frequency modulation performance weight group, and outputs candidate PID parameters through a pre-trained neural network. After being verified and corrected through an output response model, operation constraints and stability rules, the target PID parameters are generated, and then a control amount is output through a smooth transition and a preset PID algorithm. The method can improve the frequency modulation response speed, tracking accuracy and operation stability, and reduce the parameter switching impact.

Description

Technical Field

[0001] This invention belongs to the field of automatic control and energy storage frequency regulation technology, specifically relating to an adaptive control method and system for molten salt energy storage AGC. Background Technology

[0002] With the continuous increase in the installed capacity of new energy sources, the power system is placing higher demands on the peak-shaving, frequency regulation, and rapid power response capabilities of thermal power units. Automatic generation control (AGC) is an important means for the grid dispatching side to issue power regulation commands to the generation side based on system frequency deviation, tie-line power deviation, and regional control needs. For generator units participating in AGC frequency regulation, their comprehensive regulation performance is usually related to factors such as regulation rate, regulation accuracy, and response time. The regulation process not only affects the unit's performance evaluation results but also directly relates to the safety and stability of grid operation.

[0003] In the scenarios of flexible retrofitting of thermal power units and energy storage-assisted frequency regulation, molten salt energy storage systems, due to their large heat storage capacity, high operating temperature, and suitability for coupling with thermal systems, can participate in the unit's power response as an auxiliary regulation unit. In actual operation, the molten salt energy storage system typically determines the output target based on the difference between the AGC command issued by the dispatch center and the actual power of the generator unit. It then uses power regulators, heat exchange units, and related actuators to change the heat storage and release state, thereby assisting the unit's total output in tracking the AGC command. Because molten salt energy storage systems inherently possess thermal inertia, heat exchange lag, power regulator operating limits, molten salt temperature boundaries, and heat storage margin constraints, their control process differs from ordinary electrochemical energy storage or conventional motor servo systems. It requires ensuring response speed while avoiding overshoot, oscillation, and frequent actuator movements.

[0004] In existing molten salt energy storage-assisted AGC frequency regulation control, PID control remains a commonly used basic control method. Traditional PID parameters often rely on manual tuning or empirical parameters under fixed operating conditions. When AGC performance evaluation rules change, or when unit load, molten salt temperature, thermal storage margin, and power regulator response characteristics change, fixed PID parameters struggle to simultaneously ensure rapid response during large deviations and steady-state accuracy during small deviations. While some adaptive PID methods can adjust parameters based on error changes, they typically only consider the control error itself, failing to incorporate AGC comprehensive frequency regulation performance evaluation parameters, molten salt energy storage thermal state constraints, and power regulator execution constraints into the parameter adjustment process. This leads to a disconnect between control parameters and actual performance targets and equipment safety boundaries. Furthermore, directly using neural network-outputted PID parameters lacks stability verification, amplitude and speed limiting, and parameter smoothing mechanisms, which may cause abrupt changes in control inputs, closed-loop oscillations, or actuator shocks when AGC commands change abruptly or operating conditions change abruptly.

[0005] Therefore, there is an urgent need for an adaptive control method and system that can adapt to changes in AGC performance evaluation parameters while taking into account the operational constraints and closed-loop stability of molten salt energy storage systems. Summary of the Invention

[0006] The present invention aims to solve at least one of the technical problems existing in the prior art, and to provide a molten salt energy storage AGC adaptive control method and system.

[0007] This invention provides an adaptive control method for molten salt energy storage AGC, comprising the following steps: S1: Obtain the AGC command value, actual power value of the generator set, actual output value of the molten salt energy storage system and operating constraint parameters of the molten salt energy storage system issued by the dispatch center, and determine the target output value of the molten salt energy storage system based on the AGC command value and the actual power value of the generator set. S2: Based on the difference between the target output value and the actual output value of the molten salt energy storage system, determine the output control error for the current control cycle, and determine the error change rate based on the difference in output control errors between adjacent control cycles; furthermore, normalize the output control error based on the preset output benchmark value of the molten salt energy storage system to obtain a normalized output control error; and, obtain a pre-configured AGC frequency modulation performance rule parameter set, and determine a frequency modulation performance weighted set based on the AGC frequency modulation performance rule parameter set, the normalized output control error, and the error change rate. S3: Based on the frequency regulation performance weighted reassembly, the normalized output control error, the error change rate, and the operating status parameters of the molten salt energy storage system, construct a neural network input feature set, and input the neural network input feature set into a pre-trained neural network parameter model to output candidate PID parameters; S4: Obtain the pre-established molten salt energy storage system output response model, pre-configured operating constraint parameters, and preset PID stability verification rules. Based on the molten salt energy storage system output response model, the operating constraint parameters, and the preset PID stability verification rules, perform safety verification on the candidate PID parameters. If the candidate PID parameters do not meet the safety verification conditions, limit or scale the candidate PID parameters according to the preset control safety boundary to obtain the target PID parameters. S5: Using the PID parameters used in the previous control cycle as the transition start parameters, generate transition PID parameters from the transition start parameters to the target PID parameters according to the preset smooth transition rules. S6: Based on the transition PID parameters, calculate the control quantity for adjusting the output of the molten salt energy storage system according to the preset PID control algorithm, perform amplitude limiting, speed limiting and integral anti-saturation processing on the control quantity, and output the processed control quantity to the output adjustment actuator of the molten salt energy storage system.

[0008] Further, in step S1, the operating constraint parameters include at least one of the following: rated charge / discharge power, upper limit of output, lower limit of output, ramp rate limit, molten salt temperature, thermal storage margin, heat exchanger outlet temperature, and output regulation actuator action limit of the molten salt energy storage system. The target output value is determined based on the difference between the AGC command value and the actual power value of the generator set, and is subject to amplitude limiting based on the upper limit of output, the lower limit of output, and the climbing rate limit.

[0009] Specifically, in step S2, the AGC frequency regulation performance rule parameter set is determined by the currently executed power dispatch assessment rules or power market operation rules, and includes regulation rate evaluation coefficient, regulation accuracy evaluation coefficient, and response time evaluation coefficient. The frequency modulation performance weighting includes a regulation rate weight, a regulation accuracy weight, and a response time weight. The regulation rate weight, the regulation accuracy weight, and the response time weight are obtained by normalizing the regulation rate evaluation coefficient, the regulation accuracy evaluation coefficient, and the response time evaluation coefficient.

[0010] Specifically, in step S2, determining the frequency modulation performance weighting set based on the AGC frequency modulation performance rule parameter set, the normalized output control error, and the error change rate includes: Obtain a preset error partitioning rule, wherein the preset error partitioning rule includes a first error threshold and a second error threshold, and the first error threshold is greater than the second error threshold; When the normalized output control error is greater than the first error threshold, the adjustment rate weight is increased and the adjustment accuracy weight is decreased. When the normalized output control error is less than or equal to the second error threshold, the adjustment accuracy weight is increased and the adjustment rate weight is decreased. When the normalized output control error is greater than the second error threshold and less than or equal to the first error threshold, the adjustment rate weight and adjustment accuracy weight are continuously interpolated according to the error change rate; and The adjusted frequency modulation performance weighted recombination is subjected to first-order inertial filtering, and the filtered frequency modulation performance weighted recombination is input into the neural network parameter model.

[0011] Preferably, in step S3, the pre-trained neural network parameter model is obtained in the following manner: Data such as molten salt energy storage system status, AGC command data, generator power data, output regulation actuator response data, and PID parameter data under multiple sample operating conditions are collected or simulated. A comprehensive performance evaluation function is constructed using settling time, steady-state error, overshoot, rate of change of control quantity, and output regulation actuator motion constraints; and The PID parameters that enable the comprehensive performance evaluation function to meet the preset optimization conditions are used as training labels to train the neural network, thereby obtaining the pre-trained neural network parameter model.

[0012] Specifically, in step S4, the output response model is a first-order inertial plus pure time delay model, a second-order inertial model, or a state-space model identified based on the historical input and output data of the molten salt energy storage system. The safety verification includes at least one of the following: closed-loop pole location verification, phase margin verification, gain margin verification, integral gain upper limit verification, output regulating actuator control quantity limit verification, and molten salt temperature boundary verification.

[0013] Further, in step S4, when the candidate PID parameters do not meet the safety verification conditions, the candidate PID parameters are limited or scaled according to a preset control safety boundary, including: The allowable range of PID parameters is determined based on the output response model and the preset stability margin. The allowable range of PID parameters includes the allowable range of proportional gain, integral gain, and derivative gain. When any gain in the candidate PID parameters exceeds the corresponding allowable gain range, the gain exceeding the allowable gain range is corrected to the corresponding allowable gain range; and When the rate of change of the control quantity calculated based on the candidate PID parameters exceeds the action limit of the output regulating actuator, the candidate PID parameters are proportionally scaled and corrected.

[0014] Furthermore, in step S6, the preset PID control algorithm adopts the derivative-first PID control algorithm, which satisfies the following relationship:

[0015] in, To control the quantity, For proportional gain, For integral gain, For differential gain, To control output error, For integration variables, The filtered output value is the actual output value of the molten salt energy storage system after filtering.

[0016] Another aspect of the present invention provides a molten salt energy storage AGC adaptive control system, comprising: The measurement unit is used to acquire AGC command values, actual power values ​​of generator sets, actual output values ​​of molten salt energy storage systems, and operating status parameters of molten salt energy storage systems. The target output calculation unit is used to determine the target output value of the molten salt energy storage system based on the AGC command value and the actual power value of the generator set. The dynamic weight calculation unit is used to determine the output control error and error change rate based on the target output value and the actual output value of the molten salt energy storage system, normalize the output control error based on the preset output benchmark value of the molten salt energy storage system to obtain the normalized output control error, and determine the frequency regulation performance weight set based on the pre-configured AGC frequency regulation performance rule parameter set, the normalized output control error and the error change rate. The neural network parameter unit is used to construct a neural network input feature set based on the frequency modulation performance weighted reassembly, the normalized output control error, the error change rate, and the operating state parameters of the molten salt energy storage system, and input the neural network input feature set into the pre-trained neural network parameter model to output candidate PID parameters. The parameter safety verification unit is used to perform safety verification on the candidate PID parameters based on the pre-established molten salt energy storage system output response model, pre-configured operating constraint parameters, and preset PID stability verification rules. When the candidate PID parameters do not meet the safety verification conditions, the unit limits or scales the candidate PID parameters according to the preset control safety boundary to obtain the target PID parameters. A smooth transition PID control unit is used to generate transition PID parameters from the transition start parameters to the target PID parameters according to a preset smooth transition rule, using the PID parameters adopted in the previous control cycle as the transition start parameters. Based on the transition PID parameters, the control quantity is calculated, and the control quantity is subjected to amplitude limiting, speed limiting, and integral anti-saturation processing. An output regulating actuator is used to regulate the output of the molten salt energy storage system according to the control quantity.

[0017] Specifically, the system also includes an online correction unit and a security protection unit; The online correction unit is used to record the status data, PID parameter data and regulation performance data during the AGC regulation process, and to update the model parameters of the neural network parameter unit when the regulation performance data meets the preset update conditions. The safety protection unit is used to switch the PID parameters to preset safety parameters and limit the output change rate of the output regulating actuator when the molten salt temperature exceeds the limit, the heat storage margin is insufficient, the output regulating actuator responds abnormally, the control quantity exceeds the limit, or the closed-loop stability margin is insufficient.

[0018] The beneficial effects of this invention are as follows: This invention forms a closed-loop control based on the AGC command value, the actual power value of the generator set, and the actual output value of the molten salt energy storage system. It also adaptively adjusts the frequency regulation performance weighting based on the AGC frequency regulation performance rule parameter set, output control error, and error change rate, enabling the PID parameters to automatically optimize according to changes in frequency regulation assessment rules and operating conditions. Simultaneously, candidate PID parameters are generated through a neural network parameter model and safely corrected by combining the output response model, operating constraint parameters, and stability verification rules. This improves the response speed, regulation accuracy, and operational stability of the molten salt energy storage system when participating in AGC frequency regulation. Furthermore, through smooth transition, amplitude and speed limiting, integral anti-saturation, and derivative-first control, it reduces parameter switching shocks and sudden changes in control quantities, lowers the impact risk on the actuator, and improves the long-term reliability of the system. Attached Figure Description

[0019] Figure 1 This is a flowchart illustrating the steps of an adaptive control method for molten salt energy storage AGC according to a specific embodiment of the present invention. Figure 2 This is a connection block diagram of a molten salt energy storage AGC adaptive control system according to a specific embodiment of the present invention. Detailed Implementation

[0020] To enable those skilled in the art to better understand the technical solution of the present invention, the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0021] like Figure 1 As shown in the figure, a specific embodiment of the present invention provides an adaptive control method for molten salt energy storage AGC, which includes the following steps: S1: Obtain the AGC command value, actual power value of the generator set, actual output value of the molten salt energy storage system and operating constraint parameters of the molten salt energy storage system issued by the dispatch center, and determine the target output value of the molten salt energy storage system based on the AGC command value and the actual power value of the generator set. S2: Based on the difference between the target output value and the actual output value of the molten salt energy storage system, determine the output control error for the current control cycle, and determine the error change rate based on the difference in output control errors between adjacent control cycles; furthermore, normalize the output control error based on the preset output benchmark value of the molten salt energy storage system to obtain a normalized output control error; and acquire a pre-configured AGC frequency modulation performance rule parameter set, and determine a frequency modulation performance weighted set based on the AGC frequency modulation performance rule parameter set, the normalized output control error, and the error change rate; specifically, if the output control error for the current control cycle is... The output control error of the previous control cycle was The sampling period is Then the error change rate of the current control cycle It can be determined according to the following relationship:

[0022] in, .

[0023] Furthermore, to avoid abrupt changes in the frequency modulation performance weighting, a first-order inertial filter can be applied to the adjusted frequency modulation performance weighting. Let the adjusted frequency modulation performance weighting for the current control cycle be... The frequency modulation performance weights after filtering in the previous control cycle are recombined as follows: Then the frequency modulation performance weighted after filtering in the current control cycle is recombined. The following relationship must be satisfied:

[0024] in, These are the filter coefficients, and ; S3: Based on the frequency modulation performance weighted reassembly, the normalized output control error, the error change rate, and the operating state parameters of the molten salt energy storage system, a neural network input feature set is constructed. This neural network input feature set is then input into a pre-trained neural network parameter model, outputting candidate PID parameters, including proportional gain, integral gain, and derivative gain. As an example, the comprehensive performance evaluation function can be constructed based on settling time, steady-state error, overshoot, control quantity change rate, and response delay. A smaller value indicates better comprehensive control performance corresponding to the PID parameter. The preset optimization condition can be that the comprehensive performance evaluation function reaches its minimum value, or that settling time, steady-state error, overshoot, and control quantity change rate all meet the corresponding threshold requirements. Specifically, the normalized parameter values ​​output by the neural network parameter model can be converted into proportional gain, integral gain, and derivative gain through linear mapping. The mapping range of proportional gain, integral gain, and derivative gain can be predetermined based on historical tuning parameters, simulation optimization results, or equipment safety boundaries. S4: Obtain the pre-established molten salt energy storage system output response model, pre-configured operating constraint parameters, and preset PID stability verification rules. Based on the molten salt energy storage system output response model, the operating constraint parameters, and the preset PID stability verification rules, perform safety verification on the candidate PID parameters. If the candidate PID parameters do not meet the safety verification conditions, limit or scale the candidate PID parameters according to the preset control safety boundary to obtain the target PID parameters. In this embodiment, the response time weight is used as part of the neural network input feature set to participate in the generation of candidate PID parameters. When the response time weight is large, the neural network parameter model tends to output candidate PID parameters that can shorten the response delay and adjustment time, while still limiting its closed-loop stability margin by the preset PID stability verification rules. Specifically, the preset control safety boundary may include the allowable range of PID parameters, the allowable range of control quantity, the allowable range of control quantity change rate, the allowable range of integral cumulative quantity, and the allowable range of molten salt temperature. Among them, the allowable range of PID parameters can be determined by simulation scanning, historical tuning data statistics, or field test tuning based on the output response model and the preset stability margin. S5: Using the PID parameters adopted in the previous control cycle as the transition start parameters, generate transition PID parameters from the transition start parameters to the target PID parameters according to a preset smooth transition rule; specifically, the preset smooth transition rule can employ linear interpolation, a smoothstep function, or a first-order inertial transition method. Taking linear interpolation as an example, the PID parameters are represented as a parameter vector:

[0025] in, For proportional gain, For integral gain, This is the differential gain. If the PID parameters used in the previous control cycle were... The target PID parameters are Then the first Transition PID parameters for each transition control cycle The following relationship must be satisfied:

[0026] in, It is the transition coefficient, and .when When the transition PID parameter is equal to the PID parameter used in the previous control cycle; when At that time, the transition PID parameter is equal to the target PID parameter; S6: Based on the transition PID parameters, calculate the control quantity for adjusting the output of the molten salt energy storage system according to the preset PID control algorithm, perform amplitude limiting, speed limiting and integral anti-saturation processing on the control quantity, and output the processed control quantity to the output adjustment actuator of the molten salt energy storage system.

[0027] In one embodiment, the molten salt energy storage system and the generator set jointly participate in AGC frequency regulation. The AGC command value issued by the dispatch center serves as the combined output target for both the generator set and the energy storage side. The actual power value of the generator set is used to characterize the output currently undertaken by the generator set, and the target output value of the molten salt energy storage system is used to characterize the power deviation that the molten salt energy storage system needs to compensate for or absorb. Therefore, this embodiment enables the molten salt energy storage system to quickly compensate for the deviation between the actual power of the generator set and the AGC command, avoiding response lag or insufficient regulation accuracy caused by relying solely on the generator set's own adjustment. The target output value has a directional attribute: a positive value indicates that the molten salt energy storage system adds equivalent output power to the generator set or the grid side, while a negative value indicates that the molten salt energy storage system absorbs or transfers power from the generator set. Of course, the opposite positive and negative directions can also be set according to actual engineering control conventions, as long as they remain consistent in the control system.

[0028] Preferably, the actual output value of the molten salt energy storage system is obtained by a power measurement device, a thermal power conversion unit, or a feedback signal from the control system located on the output side of the molten salt energy storage system; the operating constraint parameters are jointly determined by the rated parameters of the molten salt energy storage system equipment, protection settings, real-time operating status, and the action capability of the output adjustment actuator.

[0029] Furthermore, the output regulation actuator includes at least one of the following: power regulation device, heat exchange regulation device, valve regulation device, frequency converter drive device, and power conversion device, capable of changing the charging power, releasing power, or external output power of the molten salt energy storage system. The control quantity output by the adaptive PID controller acts on the output regulation actuator, causing the actual output value of the molten salt energy storage system to converge towards the target output value.

[0030] Based on the above basic implementation method, in step S1, the operating constraint parameters include at least one of the following: rated charging and discharging power of the molten salt energy storage system, upper limit of output, lower limit of output, ramp rate limit, molten salt temperature, thermal storage margin, heat exchanger outlet temperature, and output regulation actuator action limit. The target output value is determined based on the difference between the AGC command value and the actual power value of the generator set, and is subject to amplitude limiting based on the upper limit of output, the lower limit of output, and the climbing rate limit.

[0031] Furthermore, the target output value can be determined as follows: the difference between the AGC command value and the actual power value of the generator set is used as the initial target output value of the molten salt energy storage system. The initial target output value is then constrained according to the upper limit, lower limit, and ramp rate limit of the molten salt energy storage system to obtain the final target output value. This avoids the target output value from exceeding the current executable range of the molten salt energy storage system.

[0032] Furthermore, when the initial target output value exceeds the upper limit, the target output value is limited to the upper limit; when the initial target output value is lower than the lower limit, the target output value is limited to the lower limit; when the target output change in adjacent control cycles exceeds the ramp rate limit, the target output change is limited according to the ramp rate limit. The molten salt temperature, thermal reserve, and heat exchanger outlet temperature are used to determine whether the molten salt energy storage system currently has the ability to continue charging or releasing heat, and are used to correct the upper limit, lower limit, or ramp rate limit.

[0033] In one specific implementation, in step S2, the AGC frequency regulation performance rule parameter set is determined by the currently executed power dispatch assessment rules or power market operation rules, and includes regulation rate evaluation coefficient, regulation accuracy evaluation coefficient, and response time evaluation coefficient. The frequency regulation performance weighting includes regulation rate weight, regulation accuracy weight, and response time weight. The regulation rate weight, regulation accuracy weight, and response time weight are obtained by normalizing the regulation rate evaluation coefficient, regulation accuracy evaluation coefficient, and response time evaluation coefficient, and are corrected according to the relationship between the normalized output control error and the preset error threshold. This allows the molten salt energy storage system to increase the regulation rate weight during the large error stage, increase the regulation accuracy weight during the small error stage, and retain the constraint on the response time factor based on the response time evaluation coefficient.

[0034] In this embodiment, the AGC frequency regulation performance rule parameter set can be pre-written into the control system by operators according to the currently executed power dispatch assessment rules or power market operation rules, or it can be issued to the adaptive PID controller by the host computer, dispatch interface, or plant-level monitoring system. The regulation rate evaluation coefficient, regulation accuracy evaluation coefficient, and response time evaluation coefficient are used to characterize the degree of attention paid to regulation speed, steady-state tracking accuracy, and response timeliness in the current AGC frequency regulation assessment rules, respectively.

[0035] Furthermore, when the power dispatch assessment rules or power market operation rules are adjusted, only the AGC frequency regulation performance rule parameter set needs to be updated. The adaptive PID controller can then redetermine the frequency regulation performance weight set based on the updated rule parameter set and output candidate PID parameters that match the current rules through the neural network parameter model, thereby reducing the workload of manual PID tuning due to changes in assessment rules.

[0036] In another specific embodiment, in step S2, determining the frequency modulation performance weighting set based on the AGC frequency modulation performance rule parameter set, the normalized output control error, and the error change rate includes: A preset error partitioning rule is obtained, which includes a first error threshold and a second error threshold, wherein the first error threshold is greater than the second error threshold. When the normalized output control error is greater than the first error threshold, the adjustment rate weight is increased and the adjustment accuracy weight is decreased. When the normalized output control error is less than or equal to the second error threshold, the adjustment accuracy weight is increased and the adjustment rate weight is decreased. When the normalized output control error is greater than the second error threshold and less than or equal to the first error threshold, the adjustment rate weight and the adjustment accuracy weight are continuously interpolated according to the error change rate. A first-order inertial filtering process is performed on the adjusted frequency modulation performance weight reassembly, and the filtered frequency modulation performance weight reassembly is input into the neural network parameter model.

[0037] In this embodiment, the preset error partitioning rule is used to divide the AGC tracking process of the molten salt energy storage system into a large error adjustment stage, a medium error adjustment stage, and a small error adjustment stage. The large error adjustment stage aims to quickly reduce the output deviation, the small error adjustment stage aims to reduce steady-state error and reduce frequent adjustments, and the medium error adjustment stage balances the adjustment rate and adjustment accuracy.

[0038] Specifically, the normalized output control error can be obtained by comparing the output control error of the current control cycle with a preset output benchmark value. The preset output benchmark value includes at least one of the following: the rated output value of the molten salt energy storage system, the maximum allowable output adjustment range, the allowable adjustment range of the AGC command, or the absolute value of the target output value. The first error threshold and the second error threshold can be preset based on the rated output of the molten salt energy storage system, historical AGC adjustment data, or simulation test results. As an example, the first error threshold can be between 0.2 and 0.3, and the second error threshold can be between 0.05 and 0.1.

[0039] Furthermore, when the normalized output control error is between the second error threshold and the first error threshold, the convergence trend of the current output deviation can be determined based on the error change rate. When the error change rate indicates that the output deviation convergence speed is slow, the adjustment rate weight is increased; when the error change rate indicates that the output deviation has converged rapidly, the adjustment accuracy weight is increased. By performing first-order inertial filtering on the adjusted frequency modulation performance weight reorganization, frequent jumps in PID parameters caused by abrupt weight changes can be avoided.

[0040] In another specific embodiment, in step S3, the pre-trained neural network parameter model is obtained as follows: Multiple sample operating conditions of the molten salt energy storage system's state data, AGC command data, generator power data, output regulation actuator response data, and PID parameter data are collected or simulated; a comprehensive performance evaluation function is constructed using adjustment time, steady-state error, overshoot, control quantity change rate, and output regulation actuator action constraints; and PID parameters that enable the comprehensive performance evaluation function to satisfy preset optimization conditions are used as training labels to train the neural network, resulting in the pre-trained neural network parameter model. When constructing the comprehensive performance evaluation function, the response time weight can be used to adjust the evaluation weight of the corresponding item for response delay time. When the response time weight increases, the weight of the corresponding item for response delay time in the comprehensive performance evaluation function increases, making the training labels more inclined to select PID parameters with shorter response delay times; when the response time weight is small, the comprehensive performance evaluation function focuses more on adjustment accuracy, steady-state error, or control quantity stability. Therefore, the response time weights not only participate in the construction of the neural network input feature set, but also in the training label selection process, which can avoid the response time weights being used only as formal parameters without actually affecting the generation of PID parameters.

[0041] Furthermore, the comprehensive performance evaluation function is used to evaluate the overall control effect of a set of PID parameters under sample operating conditions. As an example, the comprehensive performance evaluation function satisfies the following relationship:

[0042] in, This is the value of the comprehensive performance evaluation function; To adjust the time; This is the steady-state error; This is the overshoot. The rate of change of the control quantity is used as an evaluation value; For response delay time; To constrain penalty items; , , , and These are the preset normalized benchmark values ​​for the corresponding indicators; to These are the preset evaluation weight coefficients.

[0043] The constraint penalty item This is used to characterize whether the control process corresponding to the candidate PID parameters violates the output regulating actuator action constraints, molten salt temperature boundaries, heat storage margin boundaries, or closed-loop stability constraints; when the above constraints are not violated... Take zero or a small value when the above constraints are violated. Take a penalty value greater than zero.

[0044] During the training sample generation process, PID parameters can be searched or optimized through simulation for each set of sample conditions, which will improve the comprehensive performance evaluation function. The PID parameters that achieve the minimum value and satisfy the preset PID stability verification rules are used as the training labels for the sample operating condition; alternatively, PID parameters whose settling time, steady-state error, overshoot, control variable rate of change, and response delay all meet the corresponding threshold requirements can be used as training labels. This enables the neural network parameter model to learn the mapping relationship between PID parameters under different frequency modulation performance weight reconfigurations, error states, and molten salt energy storage operating states.

[0045] Furthermore, the operating status parameters of the molten salt energy storage system include at least one of the following: molten salt temperature, thermal reserve, actual output value, actual output change rate, heat exchanger outlet temperature, opening degree of the output regulation actuator, action limit of the output regulation actuator, and current operating condition type. The current operating condition type includes at least one of the following: start-up regulation condition, normal AGC tracking condition, rapid load change condition, external disturbance condition, and limited output condition.

[0046] Specifically, the neural network input feature set may include adjustment rate weights, adjustment accuracy weights, response time weights, normalized output control error, error change rate, normalized molten salt temperature value, normalized thermal storage margin value, normalized actual output value, and operating condition type code. The pre-trained neural network parameter model can employ a multilayer feedforward neural network, a radial basis function neural network, or other models capable of establishing a nonlinear mapping relationship between input features and PID parameters. The proportional gain, integral gain, and derivative gain output from the neural network can first undergo parameter range mapping before being used as candidate PID parameters in subsequent safety verification steps.

[0047] In another specific embodiment, in step S4, the output response model is a first-order inertial plus pure time-delay model, a second-order inertial model, or a state-space model identified based on the historical input and output data of the molten salt energy storage system; the safety verification includes at least one of closed-loop pole position verification, phase margin verification, gain margin verification, integral gain upper limit verification, output regulation actuator control quantity limit verification, and molten salt temperature boundary verification.

[0048] Furthermore, the historical input-output data includes at least one of the following: historical control quantity, actual output value of the molten salt energy storage system, rate of change of control quantity, output response delay, molten salt temperature, and thermal margin. By identifying the historical input-output data, an output response model can be obtained to characterize the dynamic relationship between the control quantity and the actual output value. The output response model is used to predict the closed-loop response characteristics of the molten salt energy storage system under the action of candidate PID parameters.

[0049] Specifically, when the output response model adopts a first-order inertial plus pure time-delay model, the control quantity can be used as the model input, and the actual output value of the molten salt energy storage system can be used as the model output, and the gain, time constant, and pure time-delay time can be identified. When the output response model adopts a second-order inertial model or a state-space model, the corresponding model parameters can be identified based on historical input and output data. The preset PID stability verification rules may include at least one of the following: the real part of the closed-loop pole is less than zero, the phase margin is not less than a preset phase margin threshold, the gain margin is not less than a preset gain margin threshold, the integral gain does not exceed a preset integral gain upper limit, the control quantity does not exceed the allowable control range of the output regulating actuator, and the molten salt temperature does not exceed a preset temperature boundary.

[0050] In another specific embodiment, in step S4, when the candidate PID parameter does not meet the safety verification conditions, the candidate PID parameter is subjected to limiting or scaling correction according to a preset control safety boundary, including: The allowable range of PID parameters is determined based on the output response model and the preset stability margin. The allowable range of PID parameters includes the allowable range of proportional gain, integral gain, and derivative gain. When any gain in the candidate PID parameters exceeds the corresponding allowable gain range, the gain exceeding the corresponding allowable gain range is corrected to the corresponding allowable gain range. And when the rate of change of the control quantity calculated based on the candidate PID parameters exceeds the action limit of the output regulating actuator, the candidate PID parameters are proportionally scaled and corrected.

[0051] Furthermore, the preset control safety boundary is determined based on the rated parameters of the molten salt energy storage system equipment, the action constraints of the output regulating actuator, the molten salt temperature protection setpoint, the thermal storage margin protection setpoint, and the closed-loop stability margin requirements. The preset control safety boundary is used at least to constrain the PID parameter value range, control variable amplitude, control variable rate of change, and integral cumulative quantity. Furthermore, the preset control safety boundary is used to limit the safe operating range of candidate PID parameters and their corresponding control processes. The preset control safety boundary includes at least one of the following: PID parameter safety boundary, control quantity safety boundary, control quantity change rate safety boundary, integral cumulative quantity safety boundary, and thermal state safety boundary.

[0052] Specifically, the PID parameter safety boundaries include the allowable ranges of proportional gain, integral gain, and derivative gain; the control quantity safety boundaries include the upper and lower limits of the control quantity that the output regulating actuator is allowed to receive; the control quantity change rate safety boundaries include the maximum allowable change value of the control quantity within adjacent control cycles; the integral cumulative quantity safety boundaries include the upper and lower limits of the allowable integral cumulative quantity of the output control error; and the thermal state safety boundaries include the allowable ranges of molten salt temperature, thermal storage margin, and heat exchanger outlet temperature.

[0053] In one specific implementation, if the candidate PID parameter is The safety boundary for PID parameters is:

[0054] If any gain in the candidate PID parameters exceeds the corresponding allowable range, the gain is corrected to the allowable range. As an example, the corrected proportional gain can be determined according to the following relationship:

[0055] Integral gain and derivative gain can be limited using the same method.

[0056] Furthermore, if the control quantity calculated based on the candidate PID parameters is The control quantity of the previous control cycle is Then the safety boundary of the control quantity and the safety boundary of the rate of change of the control quantity can be expressed as:

[0057] in, and These are the lower and upper limits of the permissible control quantity for the output regulating actuator, respectively. This represents the maximum allowable variation of the control quantity in adjacent control cycles. When the control quantity or its rate of change exceeds this range, the candidate PID parameters are subjected to amplitude limiting correction or proportional scaling correction so that the control quantity corresponding to the corrected target PID parameters meets the preset control safety boundary.

[0058] Furthermore, the thermal state safety boundary is used to prevent the molten salt energy storage system from continuing to perform excessive heat charging or releasing regulation when the molten salt temperature exceeds the limit, the heat storage margin is insufficient, or the heat exchange capacity is limited. When the molten salt temperature, heat storage margin, or heat exchanger outlet temperature reaches the corresponding protection threshold, the proportional gain and integral gain in the target PID parameters are reduced, or the target PID parameters are switched to preset safety parameters to limit the output variation of the molten salt energy storage system.

[0059] Furthermore, when any gain in the candidate PID parameters exceeds the corresponding allowable gain range, the gain is corrected to the boundary value of the allowable gain range. When a candidate PID parameter is within the allowable gain range but the rate of change of the control quantity calculated based on that candidate PID parameter exceeds the action limit of the output regulating actuator, the proportional gain, integral gain, and derivative gain are scaled by the same scaling factor so that the rate of change of the control quantity corresponding to the corrected candidate PID parameter meets the action limit of the output regulating actuator. This reduces the risk of actuator shock and closed-loop oscillation while maintaining the relative proportional relationship of the PID parameters.

[0060] In one specific implementation, the preset smooth transition rule is used to limit the rate of change of the PID parameters when switching from the PID parameters used in the previous control cycle to the target PID parameters. The PID parameters can be represented as parameter vectors. ,in, For proportional gain, For integral gain, This is the differential gain. If the PID parameters used in the previous control cycle were... The target PID parameters are Then the first Transition PID parameters for each transition control cycle It can be determined according to the following relationship:

[0061] in, It is the transition coefficient, and As an example, the transition coefficient can be determined using the smoothstep function:

[0062] in, , For the preset number of transition cycles, and when season .

[0063] Furthermore, the changes in PID parameters between adjacent control cycles can also be limited, so that... The variation does not exceed the preset parameter variation limit. This avoids sudden changes in the control quantity caused by abrupt changes in PID parameters, reducing the impact on the output regulating actuator.

[0064] In one specific implementation, in step S6, the preset PID control algorithm adopts the derivative-first PID control algorithm, which satisfies the following relationship:

[0065] in, To control the quantity, For proportional gain, For integral gain, For differential gain, To control output error, For integration variables, The filtered output value is the actual output value of the molten salt energy storage system after filtering. When calculating the control quantity, the control quantity is subjected to amplitude and speed limiting processing, and when the control quantity reaches the preset amplitude limit boundary, the integral accumulation is subjected to anti-saturation correction.

[0066] In this embodiment, the differential term acts on the filtered output value of the molten salt energy storage system after filtering, rather than directly on the output control error. Therefore, when the AGC command value changes abruptly, the proportional and integral terms can still participate in the adjustment based on the output control error, while the differential term will not experience excessive differential shocks due to sudden changes in the setpoint, thereby improving the smoothness of the output adjustment actuator.

[0067] Specifically, the filtering process can employ first-order low-pass filtering, moving average filtering, or amplitude limiting filtering. The amplitude limiting process ensures that the control quantity does not exceed the allowable control range of the output regulating actuator; the speed limiting process ensures that the change in control quantity between adjacent control cycles does not exceed the operating rate limit of the output regulating actuator; and the integral anti-saturation process maintains, limits, or backcalculates the integral accumulation of the output control error when the control quantity reaches the preset amplitude limit boundary, preventing the continuous accumulation of the integral term from causing the control quantity to saturate for a long time.

[0068] In one specific embodiment, a molten salt energy storage AGC adaptive control system is provided, comprising: The measurement unit is used to acquire AGC command values, actual power values ​​of generator sets, actual output values ​​of molten salt energy storage systems, and operating status parameters of molten salt energy storage systems. The target output calculation unit is used to determine the target output value of the molten salt energy storage system based on the AGC command value and the actual power value of the generator set. The dynamic weight calculation unit is used to determine the output control error and error change rate based on the target output value and the actual output value of the molten salt energy storage system, normalize the output control error based on the preset output benchmark value of the molten salt energy storage system to obtain the normalized output control error, and determine the frequency regulation performance weight set based on the pre-configured AGC frequency regulation performance rule parameter set, the normalized output control error and the error change rate. The neural network parameter unit is used to construct a neural network input feature set based on the frequency modulation performance weighted reassembly, the normalized output control error, the error change rate, and the operating state parameters of the molten salt energy storage system, and input the neural network input feature set into the pre-trained neural network parameter model to output candidate PID parameters. The parameter safety verification unit is used to perform safety verification on the candidate PID parameters based on the pre-established molten salt energy storage system output response model, pre-configured operating constraint parameters, and preset PID stability verification rules. When the candidate PID parameters do not meet the safety verification conditions, the unit limits or scales the candidate PID parameters according to the preset control safety boundary to obtain the target PID parameters. A smooth transition PID control unit is used to generate transition PID parameters from the transition start parameters to the target PID parameters according to a preset smooth transition rule, using the PID parameters adopted in the previous control cycle as the transition start parameters. Based on the transition PID parameters, the control quantity is calculated, and the control quantity is subjected to amplitude limiting, speed limiting, and integral anti-saturation processing. An output regulating actuator is used to regulate the output of the molten salt energy storage system according to the control quantity.

[0069] In this embodiment, the measurement unit, target output calculation unit, dynamic weight calculation unit, neural network parameter unit, parameter safety verification unit, and smooth transition PID control unit can be integrated into the controller of the molten salt energy storage system, the plant-level monitoring system, or a separate AGC auxiliary control device. The measurement unit is communicatively connected to the dispatch interface, the generator power measurement device, and the molten salt energy storage system output measurement device, and the smooth transition PID control unit is communicatively connected to the output regulation actuator.

[0070] Specifically, the output of the target output calculation unit is connected to the input of the dynamic weight calculation unit, the output of the dynamic weight calculation unit is connected to the input of the neural network parameter unit, the output of the neural network parameter unit is connected to the input of the parameter safety verification unit, the output of the parameter safety verification unit is connected to the input of the smooth transition PID control unit, and the output of the smooth transition PID control unit is connected to the output adjustment actuator. After the output adjustment actuator adjusts the output of the molten salt energy storage system, the measurement unit again collects the actual output value of the molten salt energy storage system, thereby forming a closed-loop control.

[0071] In one specific implementation, the system further includes an online correction unit and a security protection unit; The online correction unit is used to record the status data, PID parameter data, and regulation performance data during the AGC regulation process, and to update the model parameters of the neural network parameter unit when the regulation performance data meets the preset update conditions; the safety protection unit is used to switch the PID parameters to preset safety parameters and limit the output change rate of the output regulation actuator when the molten salt temperature exceeds the limit, the heat storage margin is insufficient, the output regulation actuator responds abnormally, the control quantity exceeds the limit, or the closed-loop stability margin is insufficient.

[0072] In this embodiment, the online correction unit is used to continuously record the neural network input feature set, candidate PID parameters, target PID parameters, control quantity, actual output value, settling time, steady-state error, and overshoot during the actual operation of the molten salt energy storage system, and to form an operating sample based on the recording results. The operating sample can be stored in an experience sample library for periodic updating or verification of the neural network parameter model.

[0073] Specifically, when the adjustment time, steady-state error, overshoot, and rate of change of control quantity corresponding to a certain operating sample all meet the preset performance requirements, the operating sample is added to the training sample set as a valid sample; when the adjustment performance corresponding to a certain operating sample is lower than the preset performance requirements, a local parameter search or rollback to the preset safety parameters can be triggered. When the safety protection unit detects that the molten salt temperature exceeds the limit, the thermal storage margin is insufficient, the output regulation actuator responds abnormally, the control quantity exceeds the limit, or the closed-loop stability margin is insufficient, the safety protection logic is executed first to keep the molten salt energy storage system within the allowable operating range.

[0074] In one specific implementation, taking a 300MW thermal power unit equipped with a molten salt energy storage system participating in AGC frequency regulation as an example, the AGC command value issued by the dispatch center is 240MW, and the actual power of the generator unit is 228MW. Therefore, the initial target output value of the molten salt energy storage system is 12MW. If the current upper limit of the molten salt energy storage system's output is 20MW, the lower limit is -15MW, and the ramp rate limit is 2MW / s, then the initial target output value does not exceed the constraints of the upper limit, lower limit, and ramp rate limit. Therefore, 12MW can be used as the target output value for the current control cycle.

[0075] In this embodiment, if the actual output of the molten salt energy storage system in the current control cycle is 8MW, then the output control error for the current control cycle is 4MW. If the output control error in the previous control cycle was 5MW, and the sampling period is 1s, then the error change rate is -1MW / s. If the preset output benchmark value is 20MW, then the normalized output control error is 0.2. After determining the frequency regulation performance weight set according to the currently executed AGC frequency regulation performance rule parameter set, the frequency regulation performance weight set, normalized output control error, error change rate, molten salt temperature, thermal storage margin, and actual output value are input into the neural network parameter model, and candidate PID parameters are output.

[0076] Specifically, the parameter safety verification unit performs closed-loop stability verification on the candidate PID parameters based on the pre-identified first-order inertial plus pure time delay model. If the candidate PID parameter meets the stability verification conditions, it is used as the target PID parameter. If the rate of change of the control quantity corresponding to the candidate PID parameter exceeds the action limit of the output regulating actuator, the candidate PID parameter is proportionally scaled and corrected according to the preset control safety boundary to obtain the target PID parameter. Subsequently, the smooth transition PID control unit uses the PID parameters adopted in the previous control cycle as the transition start parameters to generate transition PID parameters, and calculates the processed control quantity based on the transition PID parameters, so that the actual output value of the molten salt energy storage system gradually approaches the target output value.

[0077] In summary, this embodiment has at least the following technical effects: The target output value of the molten salt energy storage system is determined by the AGC command value and the actual power value of the generator set. The closed-loop control is formed by combining the actual output value of the molten salt energy storage system. This enables the molten salt energy storage system to quickly compensate for the AGC adjustment deviation of the generator set and improve the ability of the generator set and the energy storage system to track the AGC command. The frequency regulation performance weight set is determined by pre-configured AGC frequency regulation performance rule parameter set, and the frequency regulation performance weight set is corrected according to the normalized output control error and error change rate, so that the control parameters can be automatically adjusted as the dispatch assessment rules, power market operation rules and actual error status change, reducing the workload of manually retuning PID parameters. By using a pre-trained neural network parameter model to establish a nonlinear mapping relationship between frequency modulation performance weight reassembly, normalized output control error, error change rate, molten salt energy storage system operating state parameters and candidate PID parameters, the real-time performance and adaptability of PID parameter generation under different operating conditions can be improved. After the neural network outputs candidate PID parameters, it further combines the power response model of the molten salt energy storage system, operating constraint parameters, preset PID stability verification rules and preset control safety boundaries to perform safety verification and correction, which can reduce the risk of closed-loop oscillation, control quantity exceeding limits or actuator impact during adaptive parameter adjustment. By generating transition PID parameters through smooth transition rules and performing amplitude limiting, speed limiting, and integral anti-saturation processing on the control quantity, the sudden changes in control quantity caused by PID parameter switching can be reduced, thereby improving the smoothness of the output regulating actuator. By employing a derivative-first PID control algorithm, the derivative term acts on the filtered output value of the molten salt energy storage system after filtering, which reduces the derivative impact when the AGC command changes abruptly and improves the response stability of the molten salt energy storage system when participating in AGC frequency regulation.

[0078] It is understood that the above embodiments are merely exemplary implementations used to illustrate the principles of the present invention, and the present invention is not limited thereto. For those skilled in the art, various modifications and improvements can be made without departing from the spirit and essence of the present invention, and these modifications and improvements are also considered to be within the scope of protection of the present invention.

Claims

1. A molten salt energy storage AGC adaptive control method, characterized in that, The steps include the following: S1: Obtain the AGC command value, actual power value of the generator set, actual output value of the molten salt energy storage system and operating constraint parameters of the molten salt energy storage system issued by the dispatch center, and determine the target output value of the molten salt energy storage system based on the AGC command value and the actual power value of the generator set. S2: Determine the output control error of the current control cycle based on the difference between the target output value and the actual output value of the molten salt energy storage system, and determine the error change rate based on the difference in output control errors between adjacent control cycles. Furthermore, the output control error is normalized according to the preset output benchmark value of the molten salt energy storage system to obtain the normalized output control error; and a pre-configured AGC frequency modulation performance rule parameter set is obtained, and a frequency modulation performance weight set is determined according to the AGC frequency modulation performance rule parameter set, the normalized output control error and the error change rate. S3: Based on the frequency regulation performance weighted reassembly, the normalized output control error, the error change rate, and the operating status parameters of the molten salt energy storage system, construct a neural network input feature set, and input the neural network input feature set into a pre-trained neural network parameter model to output candidate PID parameters; S4: Obtain the pre-established molten salt energy storage system output response model, pre-configured operating constraint parameters, and preset PID stability verification rules. Based on the molten salt energy storage system output response model, the operating constraint parameters, and the preset PID stability verification rules, perform safety verification on the candidate PID parameters. If the candidate PID parameters do not meet the safety verification conditions, limit or scale the candidate PID parameters according to the preset control safety boundary to obtain the target PID parameters. S5: Using the PID parameters used in the previous control cycle as the transition start parameters, generate transition PID parameters from the transition start parameters to the target PID parameters according to the preset smooth transition rules. S6: Based on the transition PID parameters, calculate the control quantity for adjusting the output of the molten salt energy storage system according to the preset PID control algorithm, perform amplitude limiting, speed limiting and integral anti-saturation processing on the control quantity, and output the processed control quantity to the output adjustment actuator of the molten salt energy storage system.

2. The molten salt energy storage AGC adaptive control method according to claim 1, characterized in that, In step S1, the operating constraint parameters include at least one of the following: rated charge / discharge power, upper limit of output, lower limit of output, ramp rate limit, molten salt temperature, thermal storage margin, heat exchanger outlet temperature, and output regulation actuator action limit of the molten salt energy storage system. The target output value is determined based on the difference between the AGC command value and the actual power value of the generator set, and is subject to amplitude limiting based on the upper limit of output, the lower limit of output, and the climbing rate limit.

3. The molten salt energy storage AGC adaptive control method according to claim 1, characterized in that, In step S2, the AGC frequency regulation performance rule parameter set is determined by the currently executed power dispatch assessment rules or power market operation rules, and includes regulation rate evaluation coefficient, regulation accuracy evaluation coefficient, and response time evaluation coefficient. The frequency modulation performance weighting includes a regulation rate weight, a regulation accuracy weight, and a response time weight. The regulation rate weight, the regulation accuracy weight, and the response time weight are obtained by normalizing the regulation rate evaluation coefficient, the regulation accuracy evaluation coefficient, and the response time evaluation coefficient.

4. The molten salt energy storage AGC adaptive control method according to claim 3, characterized in that, In step S2, the frequency modulation performance weighting set is determined based on the AGC frequency modulation performance rule parameter set, the normalized output control error, and the error change rate, including: Obtain a preset error partitioning rule, wherein the preset error partitioning rule includes a first error threshold and a second error threshold, and the first error threshold is greater than the second error threshold; When the normalized output control error is greater than the first error threshold, the adjustment rate weight is increased and the adjustment accuracy weight is decreased. When the normalized output control error is less than or equal to the second error threshold, the adjustment accuracy weight is increased and the adjustment rate weight is decreased. When the normalized output control error is greater than the second error threshold and less than or equal to the first error threshold, the adjustment rate weight and adjustment accuracy weight are continuously interpolated according to the error change rate; and The adjusted frequency modulation performance weighted recombination is subjected to first-order inertial filtering, and the filtered frequency modulation performance weighted recombination is input into the neural network parameter model.

5. The molten salt energy storage AGC adaptive control method according to claim 1, characterized in that, In step S3, the pre-trained neural network parameter model is obtained in the following manner: Data such as molten salt energy storage system status, AGC command data, generator power data, output regulation actuator response data, and PID parameter data are collected or simulated under multiple sample operating conditions. A comprehensive performance evaluation function is constructed using settling time, steady-state error, overshoot, rate of change of control quantity, and action constraints of the output regulating actuator; as well as The PID parameters that enable the comprehensive performance evaluation function to meet the preset optimization conditions are used as training labels to train the neural network, thereby obtaining the pre-trained neural network parameter model.

6. The molten salt energy storage AGC adaptive control method according to claim 1, characterized in that, In step S4, the output response model is a first-order inertial plus pure time delay model, a second-order inertial model, or a state-space model identified based on the historical input and output data of the molten salt energy storage system. The safety verification includes at least one of the following: closed-loop pole location verification, phase margin verification, gain margin verification, integral gain upper limit verification, output regulating actuator control quantity limit verification, and molten salt temperature boundary verification.

7. The molten salt energy storage AGC adaptive control method according to claim 1, characterized in that, In step S4, when the candidate PID parameter does not meet the safety verification conditions, the candidate PID parameter is subjected to limiting or scaling correction according to a preset control safety boundary, including: The allowable range of PID parameters is determined based on the output response model and the preset stability margin. The allowable range of PID parameters includes the allowable range of proportional gain, integral gain, and derivative gain. When any gain in the candidate PID parameters exceeds the corresponding allowable gain range, the gain exceeding the allowable gain range is corrected to the corresponding allowable gain range; and When the rate of change of the control quantity calculated based on the candidate PID parameters exceeds the action limit of the output regulating actuator, the candidate PID parameters are proportionally scaled and corrected.

8. The molten salt energy storage AGC adaptive control method according to any one of claims 1 to 7, characterized in that, In step S6, the preset PID control algorithm adopts the derivative-first PID control algorithm, which satisfies the following relationship: in, To control the quantity, For proportional gain, For integral gain, For differential gain, To control output error, For integration variables, The filtered output value is the actual output value of the molten salt energy storage system after filtering.

9. A molten salt energy storage AGC adaptive control system, characterized in that, include: The measurement unit is used to acquire AGC command values, actual power values ​​of generator sets, actual output values ​​of molten salt energy storage systems, and operating status parameters of molten salt energy storage systems. The target output calculation unit is used to determine the target output value of the molten salt energy storage system based on the AGC command value and the actual power value of the generator set. The dynamic weight calculation unit is used to determine the output control error and error change rate based on the target output value and the actual output value of the molten salt energy storage system, normalize the output control error based on the preset output benchmark value of the molten salt energy storage system to obtain the normalized output control error, and determine the frequency regulation performance weight set based on the pre-configured AGC frequency regulation performance rule parameter set, the normalized output control error and the error change rate. The neural network parameter unit is used to construct a neural network input feature set based on the frequency modulation performance weighted reassembly, the normalized output control error, the error change rate, and the operating state parameters of the molten salt energy storage system, and input the neural network input feature set into the pre-trained neural network parameter model to output candidate PID parameters. The parameter safety verification unit is used to perform safety verification on the candidate PID parameters based on the pre-established molten salt energy storage system output response model, pre-configured operating constraint parameters, and preset PID stability verification rules. When the candidate PID parameters do not meet the safety verification conditions, the unit limits or scales the candidate PID parameters according to the preset control safety boundary to obtain the target PID parameters. A smooth transition PID control unit is used to take the PID parameters used in the previous control cycle as the transition start parameters, generate transition PID parameters from the transition start parameters to the target PID parameters according to the preset smooth transition rules, calculate the control quantity based on the transition PID parameters, and perform amplitude limiting, speed limiting and integral anti-saturation processing on the control quantity. as well as An output regulating actuator is used to regulate the output of the molten salt energy storage system according to the control quantity.

10. The molten salt energy storage AGC adaptive control system according to claim 9, characterized in that, The system also includes an online correction unit and a security protection unit; The online correction unit is used to record the status data, PID parameter data and regulation performance data during the AGC regulation process, and to update the model parameters of the neural network parameter unit when the regulation performance data meets the preset update conditions. The safety protection unit is used to switch the PID parameters to preset safety parameters and limit the output change rate of the output regulating actuator when the molten salt temperature exceeds the limit, the heat storage margin is insufficient, the output regulating actuator responds abnormally, the control quantity exceeds the limit, or the closed-loop stability margin is insufficient.

Images