Method, system, device and storage medium for estimating primary frequency regulation margin of thermal power unit

Abstract

The present application discloses a method, system, device and storage medium for estimating the primary frequency regulation margin of a thermal power unit, which belongs to the field of thermal control technology of thermal power plants. The method comprises obtaining operation data of the primary frequency regulation of the thermal power unit, determining a time domain data group of the thermal power unit based on the operation data, and obtaining initial parameter information based on the time domain data group; optimizing the initial parameter information to obtain key parameter information; building an initial mechanism model combining a DEH system and a steam turbine model, and a deep learning model embedded in the initial mechanism model, inputting key parameters and valve opening information into the deep learning model to obtain an optimized control signal, optimizing the initial mechanism model based on the optimized control signal to obtain a primary frequency regulation margin estimation model; inputting a frequency difference and a power given value into the primary frequency regulation margin estimation model, and determining the margin data achieved by the primary frequency regulation of the thermal power unit. The method can effectively capture nonlinear and time-varying characteristics, and the primary frequency regulation margin prediction is more accurate.

Description

Technical Field

[0001] The present application relates to the technical field of thermal power plant automatic control, and in particular to a method, system, device and storage medium for estimating primary frequency regulation margin of a thermal power unit. Background Art

[0002] In the modern new power system, with the massive access to new energy sources such as large-scale wind power and photovoltaic power generation, the system's reliable power supply capacity has been significantly weakened. In order to absorb new energy, thermal power units must perform deep peak regulation on a regular basis. A large number of test results under deep peak regulation conditions have shown that the deterioration of the safety and grid-related performance of thermal power units has become a common phenomenon, and their power regulation and bottom-line capacity has also been weakened.

[0003] At present, in order to ensure that the units are capable of participating in primary frequency regulation, most thermal power plants do not fully open the regulating steam valves of the units. They retain a certain regulating margin through throttling to ensure that when the frequency regulation command is received, the regulating steam valves can be quickly opened to increase the unit load. Nowadays, the primary frequency regulation mechanism model method is often used to evaluate the primary frequency regulation capability of the unit. The conventional primary frequency regulation mechanism model consists of a digital electro-hydraulic control system (Digital Electro-Hydraulic, DEH) and a steam turbine prime mover model, which uses DEH and steam turbine prime mover models to respond to the primary frequency regulation of the power grid. The primary frequency regulation test command issued by the dispatching center under simulated frequency disturbances, through the constructed primary frequency regulation mechanism model, has an intuitive display of the primary frequency regulation performance of the unit over a period of time from static and dynamic perspectives.

[0004] However, the volume time constant of the front steam chamber of the high-pressure cylinder and the volume time constant of the reheater in the turbine model will change with the unit load. At the same time, when the unit is running with variable load, the load change will cause the efficiency of the high-pressure cylinder and the medium and low-pressure cylinders to change, and the power distribution coefficient and power overregulation coefficient will also change with the load change during the variable load process. This leads to a significant difference between the turbine work model under deep peak load regulation conditions and the rated condition.

[0005] At the same time, the traditional method for estimating the primary frequency regulation margin of thermal power units requires the mechanism model to continuously adjust parameters to adapt to different operating conditions under deep peak regulation conditions, which makes it difficult to capture nonlinear and time-varying characteristics, and the prediction accuracy of the model is affected. Summary of the invention

[0006] The present application aims to solve the problems in the prior art of estimating the primary frequency regulation margin of thermal power units, such as constantly adjusting parameters to adapt to different operating conditions, difficulty in capturing nonlinear and time-varying characteristics, which affects the prediction accuracy. Therefore, a method, system, device and storage medium for estimating the primary frequency regulation margin of thermal power units are proposed.

[0007] In order to achieve the above objectives, in a first aspect, the present application provides a method for estimating a primary frequency regulation margin of a thermal power unit, comprising:

[0008] Acquire the operation data of the primary frequency modulation of the thermal power unit, and determine the time domain data group of the thermal power unit based on the operation data, wherein the time domain data group is used to characterize the operation parameters of the thermal power unit at different historical time points;

[0009] Building a steam turbine model, and inputting the time domain data group into the steam turbine model to obtain initial parameter information;

[0010] Optimizing the initial parameter information based on the Hippo optimization algorithm to obtain key parameter information, wherein the key parameter information is used to characterize multiple volume time constants of the steam turbine;

[0011] Building an initial mechanism model combining the DEH system and the steam turbine model, and a deep learning model embedded in the initial mechanism model, inputting the key parameter information and valve opening information into the deep learning model to obtain an optimized control signal, and optimizing the initial mechanism model based on the optimized control signal to obtain a primary frequency regulation margin estimation model;

[0012] The frequency difference signal and the power given value are input into the primary frequency regulation margin estimation model to determine the margin data achieved by the primary frequency regulation of the thermal power unit.

[0013] In some possible embodiments, the obtaining of the operation data of the primary frequency modulation of the thermal power unit, and determining the time domain data group of the thermal power unit based on the operation data, wherein the time domain data group is used to characterize the operation parameters of the thermal power unit at different historical time points, includes:

[0014] Collecting the operating data of the thermal power unit under different rated loads and frequency difference step signal conditions;

[0015] Based on the running data, the time point index of the time series array is screened out, and the valid data segment is intercepted and the abnormal batch data containing abnormal values ​​is eliminated;

[0016] An interpolation function is created, and based on the interpolation function, the time points of each historical data are aligned to the time points of the parameters to obtain a time domain data group including the operating time, main steam pressure, comprehensive valve position and power.

[0017] In some possible embodiments, the step of building a steam turbine model and inputting the time domain data set into the steam turbine model to obtain initial parameter information includes:

[0018] The steam turbine model building includes the high-pressure steam chamber volume link, the reheater volume link, the low-pressure connecting pipe steam chamber volume link and the high-pressure cylinder power natural overmodulation coefficient link, and

[0019] The time series data based on the main steam pressure and the comprehensive valve position is used as the input of the steam turbine model, and the actual load is used as the output of the steam turbine model.

[0020] In some possible embodiments, the optimization of the initial parameter information based on the Hippo optimization algorithm to obtain key parameter information, the key parameter information is used to characterize multiple volume time constants of the steam turbine, including:

[0021] Initial parameter information is obtained according to the steam turbine model identification, and each sub-data in the initial parameter information is iteratively optimized based on the Hippo optimization algorithm to obtain key parameter information, wherein the key parameter information at least includes the high-pressure steam chamber volume time constant, the reheater volume time constant, the low-pressure connecting pipe steam chamber volume time constant and the high-pressure cylinder power natural overmodulation coefficient.

[0022] In some possible embodiments, the initial parameter information is obtained according to the turbine model identification, and each sub-data in the initial parameter information is iteratively optimized based on the Hippo optimization algorithm to obtain key parameter information, including:

[0023] Set the initial value range of each sub-data in the initial parameter information, where:

[0024] The initialization expression is:

[0025] In the formula, is the position of the corresponding sub-data, and are the lower and upper limits of the initial values ​​of the corresponding sub-data, is a random number between 0 and 1;

[0026] Iterate each sub-data after initialization, and determine whether the number of iterations reaches a preset iteration threshold or whether the fitness value is less than a preset fitness threshold;

[0027] If so, output the corresponding high-pressure steam chamber volume time constant, reheater volume time constant, low-pressure connecting pipe steam chamber volume time constant and high-pressure cylinder power natural overmodulation coefficient value.

[0028] In some possible embodiments, the initial mechanism model combining the DEH system and the steam turbine model is constructed, and a deep learning model is embedded in the initial mechanism model, the key parameter information and valve opening information are input into the deep learning model to obtain an optimized control signal, and the initial mechanism model is optimized based on the optimized control signal to obtain a primary frequency regulation margin estimation model, including:

[0029] Build a target optimization model, in which the power regulator parameters, electro-hydraulic converter time constant, and oil motor time constant are used as model inputs, the optimized control signal is used as model output, and the high-pressure steam chamber volume time constant, reheater volume time constant, low-pressure connecting pipe steam chamber volume time constant, and high-pressure cylinder power natural overshoot coefficient are used as model feedback;

[0030] Extracting features from the input and output data, constructing a time series array from the extracted features using a sliding time window method, and dividing the reconstructed time series array into a training set and a validation set;

[0031] Training the target optimization model based on the training set, and verifying the trained target optimization model based on the verification set;

[0032] Determine the difference between the predicted results of training and verification outputs and the actual observed values ​​based on the preset cross entropy loss function;

[0033] The network weights of the target optimization model are adaptively adjusted based on the difference value to obtain a deep learning model.

[0034] In some possible embodiments, the step of inputting a frequency difference signal and a power given value into the primary frequency regulation margin estimation model to determine the margin data achieved by the primary frequency regulation of the thermal power unit includes:

[0035] Inputting a frequency difference signal and a power given value into the primary frequency modulation margin estimation model to obtain a primary frequency modulation response waveform;

[0036] Based on the primary frequency regulation response waveform, the primary frequency regulation margin of the thermal power unit under different frequency deviations is obtained, wherein:

[0037] The primary frequency modulation margin includes additional frequency modulation capability, margin response time, and margin recovery time.

[0038] Compared with the prior art, the technical solution provided by the embodiment of the present application includes at least the following beneficial effects or advantages:

[0039] 1) In the primary frequency regulation margin estimation method provided in the present application, a pre-created steam turbine prime mover model is used, and based on running the model and using the Hippo optimization algorithm, the optimal high-pressure steam chamber volume time constant, reheater volume time constant, low-pressure connecting pipe steam chamber volume time constant, and high-pressure cylinder power natural overmodulation coefficient are identified, thereby optimizing the unit's operating performance and frequency regulation capability. A deep learning model is constructed, and the deep learning model is obtained by training and verifying the historical data of the primary frequency regulation of the thermal power unit. The deep learning model is embedded in the initial mechanism model combined with the DEH system and the steam turbine model. The optimized control signal is obtained by inputting the key parameter information and the valve opening information into the deep learning model. The initial mechanism model is optimized based on the optimized control signal to obtain a primary frequency regulation margin estimation model. By inputting the frequency difference signal and the power given value into the estimation model, the accurate performance data of the primary frequency regulation under different frequency difference signals is obtained. Under the deep peak regulation condition, the mechanism model does not need to continuously adjust the parameters to adapt to different operating states, and the nonlinear and time-varying characteristics are well captured, and the primary frequency regulation margin prediction accuracy is higher.

[0040] In a second aspect, the present application provides a system for estimating the primary frequency regulation margin of a thermal power unit, comprising:

[0041] an acquisition module, configured to acquire operation data of a primary frequency modulation of a thermal power unit, and determine a time domain data group of the thermal power unit based on the operation data, wherein the time domain data group is used to characterize operation parameters of the thermal power unit at different historical time points;

[0042] A model building module is configured to build a steam turbine model, an initial mechanism model combining a DEH system and the steam turbine model, and a deep learning model embedded in the initial mechanism model;

[0043] A first determination module is configured to input the time domain data group into the steam turbine model to obtain initial parameter information;

[0044] A data optimization module is configured to optimize the initial parameter information based on a Hippo optimization algorithm to determine key parameter information, wherein the key parameter information is used to characterize a plurality of volume time constants of the steam turbine;

[0045] A second determination module is configured to input the key parameter information and valve opening information into the deep learning model to obtain an optimized control signal, and optimize the initial mechanism model based on the optimized control signal to determine a primary frequency modulation margin estimation model;

[0046] The third determination module is configured to input a frequency difference signal and a power given value into the primary frequency regulation margin estimation model to determine the margin data achieved by the primary frequency regulation of the thermal power unit.

[0047] In a third aspect, the present application further provides an electronic device, including:

[0048] at least one processor; and

[0049] a memory communicatively connected to the at least one processor; wherein,

[0050] The memory stores instructions that can be executed by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute the steps of the method for estimating the primary frequency regulation margin of a thermal power unit provided in the first aspect above.

[0051] In a fourth aspect, the present application further provides a computer-readable storage medium having computer program instructions stored thereon, and when the computer program instructions are executed by a processor, the steps of the method for estimating the primary frequency regulation margin of a thermal power unit provided in the first aspect above are implemented.

[0052] It can be understood that the beneficial effects of the technical solutions provided in the second, third and fourth aspects can be found in the relevant description of the first aspect, and will not be repeated here.

[0053] Additional aspects and advantages of the present application will be given in part in the description below, and in part will become apparent from the description below, or will be learned through the practice of the present application. BRIEF DESCRIPTION OF THE DRAWINGS

[0054] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0055] Figure 1 is a flow chart of a method for estimating a primary frequency regulation margin of a thermal power unit according to an embodiment of the present application;

[0056] Figure 2 is a control principle flow chart of a method for estimating a primary frequency regulation margin of a thermal power unit according to an embodiment of the present application;

[0057] Figure 3 is a control principle flow chart of a steam turbine model according to an embodiment of the present application;

[0058] Figure 4 It is a control principle flow chart of a primary frequency regulation model diagram of a thermal power unit based on deep learning according to an embodiment of the present application;

[0059] Figure 5 is a block diagram of a system for estimating primary frequency regulation margin of a thermal power unit according to an embodiment of the present application;

[0060] Figure 6 is a structural block diagram of an electronic device according to an embodiment of the present application; DETAILED DESCRIPTION

[0061] The embodiments of the present application are described in detail below. The embodiments described with reference to the accompanying drawings are exemplary. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0062] It should be noted that, unless otherwise defined, all technical and scientific terms used herein have the same meaning as those commonly understood by those skilled in the art to which this application belongs. The terms used herein in the specification of this application are only for the purpose of describing specific implementations and are not intended to limit this application. The term "and / or" used herein includes any and all combinations of one or more of the related listed items.

[0063] See also Figures 1 to 4 This embodiment provides a method for estimating a primary frequency regulation margin of a thermal power unit, comprising:

[0064] Step S100: acquiring the operation data of the primary frequency modulation of the thermal power unit, and determining the time domain data group of the thermal power unit based on the operation data, wherein the time domain data group is used to characterize the operation parameters of the thermal power unit at different historical time points;

[0065] In this step, it should be noted that for the operation data of the primary frequency regulation of thermal power units, the operation information of thermal power units from the power grid dispatching center can be systematically collected and analyzed. This information should at least include but is not limited to key parameters such as main steam pressure, valve opening, and actual load. In order to ensure the accuracy and reliability of subsequent data analysis, the collected raw operation data can be subjected to strict data preprocessing steps. This process involves the use of advanced data cleaning techniques to eliminate noise, identify and correct outliers or missing values, and use standardization or normalization methods to unify data of different dimensions for subsequent advanced data analysis and modeling.

[0066] Optionally, for the time domain data group, the operating data of the frequency modulation under the typical working conditions of the power plant units in the power grid dispatching center can be collected, combined with the analysis and summary of the thermal characteristics data of the thermal power units, and in the data preprocessing step, the operating data of the frequency difference step signal test of +4% and -4% under the steady-state working conditions of 60%, 75% and 90% rated load of the thermal power units can be collected, and the time point index of the time series array can be filtered out, the valid data segment can be intercepted, and the abnormal batch data containing abnormal values ​​can be eliminated. Create an interpolation function to align the time points of each historical data to the time point of the parameter, and obtain a time domain data group consisting of operating time, main steam pressure, comprehensive valve position and power.

[0067] It should also be noted that the time domain data group can also be obtained based on the operating data of the test under other rated load steady-state conditions of the thermal power unit and other frequency difference step signal conditions. The specific selection can be made according to actual needs. The acquired time domain data group can be used for data input and output of the model in the subsequent steps of this embodiment, as well as the construction of a deep learning model.

[0068] Step S200: constructing a steam turbine model, and inputting the time domain data group into the steam turbine model to obtain initial parameter information;

[0069] In this step, see Figure 3 , Figure 3 A control principle flow chart of the steam turbine model provided in an embodiment of the present application is shown. The construction of the steam turbine model can be created based on existing modeling software, such as MATLAB / Simulink, and the specific selection can be made according to actual needs. The model is based on the timing data of the main steam pressure and the integrated valve position as the input of the steam turbine model, and based on the actual load (power) as the output of the steam turbine model. The construction of the steam turbine model includes the high-pressure steam chamber volume link, the reheater volume link, the low-pressure connecting pipe steam chamber volume link and the high-pressure cylinder power natural overmodulation coefficient link.

[0070] It should be noted that 1) for the high-pressure steam chamber, the volume time constant represents the time scale from the high-pressure steam entering the steam chamber to reaching a steady-state response, which is used to describe the dynamic characteristics of the steam pressure change inside the high-pressure cylinder; 2) for the reheater, the volume time constant reflects the speed of steam pressure change in the reheater, which is crucial for evaluating the impact of the reheater on the dynamic response of the system; 3) for the low-pressure connecting pipe steam chamber, the volume time constant measures the rate of pressure change of the low-pressure steam flowing from the high-pressure cylinder to the low-pressure cylinder, which is extremely important for understanding the dynamic behavior of energy transfer in the entire turbine system; 4) for the high-pressure cylinder power natural overshoot coefficient, it defines the transient response characteristics of the high-pressure cylinder output power relative to the input signal, especially the overshoot phenomenon that may occur during the regulation process.

[0071] It should also be noted that by building a steam turbine model, the steam turbine prime mover model takes the main steam pressure and the integrated valve position as input and the unit load as output. The main links in the model are three first-order inertia links, which represent the front steam chamber of the high-pressure cylinder, the volume effect of the reheater, and the low-pressure connecting pipe steam chamber. The time domain data group is input into the steam turbine model to obtain the initial parameter information, which contains multiple important dynamic characteristic parameters, including the high-pressure steam chamber volume time constant, the reheater volume time constant, the low-pressure connecting pipe steam chamber volume time constant, and the high-pressure cylinder power natural overshoot coefficient.

[0072] Step S300: optimizing the initial parameter information based on the Hippo optimization algorithm to obtain key parameter information, wherein the key parameter information is used to characterize multiple volume time constants of the steam turbine;

[0073] In this step, since the initial parameter information determined in step S200 has certain noise and is not the optimal parameter for the operating performance and frequency regulation capability of the unit, the initial parameter information is optimized by the Hippo optimization algorithm to obtain key parameter information, and the initial value range of each sub-data in the initial parameter information is set, where:

[0074] The initialization expression is:

[0075] In the formula, is the position of the corresponding sub-data, and are the lower and upper limits of the initial values ​​of the corresponding sub-data, is a random number between 0 and 1;

[0076] Then iterate each sub-data after initialization, and determine whether the number of iterations reaches the preset iteration threshold or whether the fitness value is less than the preset fitness threshold. If yes, output the corresponding high-pressure steam chamber volume time constant, reheater volume time constant, low-pressure connecting pipe steam chamber volume time constant and high-pressure cylinder power natural overmodulation coefficient value; if not, repeat the steps of the Hippo optimization algorithm.

[0077] In one example, the initial parameter information is optimized based on the Hippo optimization algorithm, specifically including:

[0078] ① Initialize the population:

[0079] Set each parameter ( : High pressure steam chamber volume time constant, : Reheater volume time constant, : Low-pressure connecting pipe steam chamber volume time constant, : The initial value range of the natural over-modulation coefficient of the high-pressure cylinder power. Use the formula:

[0080]

[0081] in:

[0082] express location;

[0083] and Respectively represent the lower and upper limits of the initial values ​​of the parameters;

[0084] is a random number between 0 and 1;

[0085] ② Location update:

[0086] In each iteration, the dominant hippo is identified , i.e., the hippopotamus with the optimal objective function value;

[0087] Update the position of each hippopotamus to tend towards the position of the dominant hippopotamus ;

[0088] Update the position of the male hippopotamus using the following formula;

[0089]

[0090] in:

[0091] is a random number that controls the step size;

[0092] If the number of iterations is greater than 6, update the positions of female and immature hippos:

[0093]

[0094] in:

[0095] It is also a random number that controls the step size;

[0096] ③Hippopotamus defends against predators (exploration phase):

[0097] Simulate the hippopotamus's strategy of defending against predators by quickly turning towards them and making warning calls to drive off predators;

[0098] Update the hippo's position to defend against predators using the following formula:

[0099]

[0100] in:

[0101] is a random number that controls the step size;

[0102] is the position of the predator, which can be obtained by random generation;

[0103] if , the hippopotamus quickly turns toward the predator and changes position to defend itself;

[0104] ④Hippo escapes from predator (exploitation stage):

[0105] When hippos sense danger, they will try to flee from the predator and seek the nearest safe location;

[0106] Update the hippo's position to escape the predator using the following formula:

[0107]

[0108] in:

[0109] is a random number that controls the step size;

[0110] It is a randomly generated safe location;

[0111] If the new location If the objective function value of is better than the current position, the new position is accepted; otherwise, the original position remains unchanged;

[0112] ⑤Repeat steps ② to ④:

[0113] Repeat the above steps until the termination condition is met, the maximum number of iterations is reached, or the fitness value no longer improves significantly;

[0114] ⑥Output optimal solution:

[0115] Output the hippopotamus position with the best fitness value As the final solution.

[0116] Similarly, the parameter , , The final solution is obtained in the same way as above.

[0117] Through these steps, the key parameter information of the thermal power unit can be accurately identified, thereby optimizing the unit's operating performance and frequency regulation capabilities.

[0118] Step S400: constructing an initial mechanism model combining the DEH system and the steam turbine model, and a deep learning model embedded in the initial mechanism model, inputting the key parameter information and valve opening information into the deep learning model to obtain an optimized control signal, and optimizing the initial mechanism model based on the optimized control signal to obtain a primary frequency regulation margin estimation model;

[0119] In this step, for the DEH system, the Digital Electro-Hydraulic Control System (DEH) is an advanced technology used to control and adjust mechanical equipment, especially in large-scale power equipment (such as steam turbines, generators, etc.). An initial mechanism model combining the DEH system and the steam turbine model is built to simulate the control and dynamic characteristics of thermal power units. The DEH system realizes electro-hydraulic signal conversion and adjusts the valve opening to control the steam flow and pressure.

[0120] Optionally, for the construction of a deep learning model, a target optimization model is constructed, wherein a target optimization model is constructed, wherein power regulator parameters, electro-hydraulic converter time constants, and oil motor time constants are used as model inputs, an optimized control signal is used as model output, and high-pressure steam chamber volume time constants, reheater volume time constants, low-pressure connecting pipe steam chamber volume time constants, and high-pressure cylinder power natural overmodulation coefficients are used as model feedback; features are extracted from input and output data, and a sliding time window method is used to construct a time series array from the extracted features, and the reconstructed time series array is divided into a training set and a validation set; the target optimization model is trained based on the training set, and the trained target optimization model is verified based on the validation set; the difference between the predicted results of the training and verification outputs and the actual observed values ​​is determined based on a preset cross entropy loss function, and the network weights of the target optimization model are adjusted based on the difference values ​​to obtain a deep learning model.

[0121] In one example, a combination of Kolmogorov–Arnold Networks (KAN) and BiLSTM can be used for the deep learning model, by obtaining historical operation data and real-time monitoring data of the unit, including steam pressure, temperature, valve opening, speed, and electro-hydraulic regulation signals, etc. These data are denoised and normalized, and key features are extracted through principal component analysis (PCA) to reduce the dimension and redundancy of the data, thereby improving the efficiency and accuracy of the model. Denoising and normalizing the data are existing technical means and will not be elaborated here.

[0122] Optionally, the preprocessed data is input into a Kolmogorov-Arnold network (KAN). The KAN network uses its embedded nonlinear mapping function to transform the input linear signal into a high-dimensional nonlinear feature representation. Specifically, KAN constructs several nonlinear basis functions to perform layer-by-layer feature mapping and combination of the input signal. In this way, KAN enhances the ability to capture complex nonlinear relationships in the input signal and generates a new feature space, so that the subsequent BiLSTM network can better understand the dynamic characteristics of the thermal power unit.

[0123] The high-dimensional features output from the KAN network are input into the Bidirectional Long Short-Term Memory Network (BiLSTM). The BiLSTM consists of two parts, the forward LSTM and the backward LSTM, which process the forward and backward information of the sequence respectively. At each time step, the forward LSTM starts from the first time step of the input sequence and processes the input features of each time step sequentially. The forget gate of the forward LSTM determines how much of the cell state information from the previous time step to forget based on the input features of the current time step and the previous hidden state. The input gate generates an input gate value through the current input features and the previous hidden state, and updates the cell state in combination with the forget gate information. The output gate controls the part passed from the updated cell state to the hidden state to generate the forward output signal. The backward LSTM starts from the last time step of the input sequence and processes the input features of each time step in reverse order. The forget gate of the backward LSTM also determines how much of the cell state information from the next time step to forget based on the input features of the current time step and the next hidden state. The input gate combines the current input features and the next hidden state to generate an input gate value, and combines it with the forget gate information to update the cell state. The output gate controls what is passed from the updated cell state to the hidden state to generate the backward output signal.

[0124] The output signals generated by the forward LSTM and the backward LSTM are merged. Specifically, at each time step, the forward output signal and the backward output signal are concatenated to form a combined vector containing bidirectional information. The merged bidirectional output signal is input to the fully connected layer and further processed by linear transformation and activation function. The purpose of this step is to compress the high-dimensional features into a more manageable output form and generate a specific control signal output in the last fully connected layer. The loss function is defined to measure the difference between the model output and the target value. The network parameters of BiLSTM and KAN are optimized by the back-propagation algorithm, so that the model can gradually reduce the estimation error.

[0125] In this way, the closed-loop control mechanism enables the thermal power unit to respond to load changes and system disturbances in real time. By continuously monitoring the system's output signal, it ensures real-time adjustment of the control signal, thereby generating the optimal primary frequency regulation margin estimation model.

[0126] Step S500: inputting a frequency difference signal and a power given value into the primary frequency regulation margin estimation model to determine the margin data achieved by the primary frequency regulation of the thermal power unit.

[0127] In this step, the frequency difference signal and the power given value are input into the primary frequency regulation margin estimation model. Based on the frequency difference signal and the power given value, the opening of the control integrated valve position can be obtained, thereby obtaining the primary frequency regulation response waveform; based on the primary frequency regulation response waveform, the primary frequency regulation margin of the thermal power unit under different frequency deviations is obtained, wherein the primary frequency regulation margin includes additional frequency regulation capacity, margin response time and margin recovery time.

[0128] Specifically, for the operation process of the primary frequency regulation margin estimation model, the frequency difference signal is converted into a regulation signal, input into the KAN-BiLSTM deep learning model, and fed back to the electro-hydraulic regulation system and turbine prime mover model of the thermal power unit. According to the real-time monitoring data and the generated optimization control signal, the opening of the electro-hydraulic valve is dynamically adjusted to control the steam flow and pressure. By monitoring and recording key parameters in the frequency regulation process, such as valve opening, steam flow, pressure change, unit load and frequency deviation, the margin data achieved by the thermal power unit in this primary frequency regulation process are calculated. These margin data include actual frequency regulation margin, frequency regulation response time and steady-state error, etc., which are used to evaluate the performance of the thermal power unit in responding to frequency changes.

[0129] It should be explained that the primary frequency regulation margin refers to the additional frequency regulation capacity that a thermal power unit can provide during the primary frequency regulation process to cope with unexpected frequency changes. The primary frequency regulation margin includes but is not limited to:

[0130] Additional frequency regulation capacity, that is, the frequency regulation capacity that thermal power units can provide after reaching their maximum frequency regulation capacity;

[0131] Margin response time, i.e. the response time of thermal power units from maximum frequency regulation capability to additional frequency regulation capability;

[0132] Margin recovery time refers to the time interval for a thermal power unit to recover from the additional frequency regulation capability to the steady-state operation state.

[0133] In the steps of the primary frequency regulation margin estimation method provided in the above embodiment, a pre-created steam turbine prime mover model is used, and based on running the model and using the Hippo optimization algorithm, the optimal high-pressure steam chamber volume time constant, reheater volume time constant, low-pressure connecting pipe steam chamber volume time constant, and high-pressure cylinder power natural overmodulation coefficient are identified, thereby optimizing the unit's operating performance and frequency regulation capability. A deep learning model is constructed, and the deep learning model is obtained by training and verifying the historical data of the primary frequency regulation of the thermal power unit. The deep learning model is embedded in the initial mechanism model combined with the DEH system and the steam turbine model. The optimized control signal is obtained by inputting the key parameter information and the valve opening information into the deep learning model. The initial mechanism model is optimized based on the optimized control signal to obtain a primary frequency regulation margin estimation model. The frequency difference signal and the power given value are input into the estimation model to obtain the performance data of the precise primary frequency regulation under different frequency difference signals. Under the deep peak regulation condition, the mechanism model does not need to continuously adjust parameters to adapt to different operating states, and can well capture the nonlinear and time-varying characteristics, so the prediction accuracy of the primary frequency regulation margin is higher.

[0134] In one example, continue to see Figure 4 , provides a practical application of the above-mentioned method for estimating the primary frequency regulation margin of thermal power units:

[0135] First, collect and analyze the operating information of thermal power units.

[0136] The primary frequency regulation information of the power plant includes operating time, main steam pressure, comprehensive valve position and unit load. The collected data is preprocessed, and the time point index of the time series array is filtered out using the where function of NumPy, the valid data segment is intercepted, and the abnormal batch data containing abnormal values ​​is eliminated; the interp1d function of SciPy is used to create an interpolation function to align the time points of each historical data to the time point of the parameter, and obtain a time domain data group consisting of operating time, main steam pressure, comprehensive valve position and power.

[0137] Secondly, build a steam turbine prime mover model.

[0138] The steam turbine prime mover model takes the main steam pressure and the integrated valve position as input and the unit load as output. The main links in the model are three first-order inertia links, which represent the volume effect of the high-pressure cylinder front steam chamber, the reheater, and the low-pressure connecting pipe steam chamber. FLP, FIP, and FHP in the model represent the low-pressure cylinder power proportional coefficient, the medium-pressure cylinder power proportional coefficient, and the high-pressure cylinder power proportional coefficient, respectively. These three proportional coefficients are set to fixed values ​​by factory parameters. Ty and Tc in the model represent the electro-hydraulic converter time constant and the oil motor time constant, respectively. The constants are set to fixed values ​​by factory parameters. The simulation model is run to identify the high-pressure steam chamber volume time constant, the reheater volume time constant, the low-pressure connecting pipe steam chamber volume time constant, and the high-pressure cylinder power natural overshoot coefficient, and the optimal parameters are shown in Table 1.

[0139] Table 1: Steam turbine parameter list

[0140]

[0141] Secondly, a thermal power unit mechanism model and a KAN-BiLSTM neural network deep learning model were established.

[0142] The mechanism model of the thermal power unit takes the power given P and the frequency difference as input, and obtains the output unit power through the power regulator PID, electro-hydraulic converter, oil motor, high-pressure cylinder front steam chamber, reheater volume effect, and low-pressure connecting pipe steam chamber. In general, the power regulator parameters, electro-hydraulic converter time constant, and oil motor time constant are considered to be fixed values, and are usually set according to the factory parameters of the unit equipment. The high-pressure steam chamber volume time constant, reheater volume time constant, low-pressure connecting pipe steam chamber volume time constant, and high-pressure cylinder power natural overshoot coefficient are parameters obtained by identifying the prime mover model. The KAN-BiLSTM neural network is constructed. The KAN-BiLSTM neural network is composed of a combination of Kolmogorov–Arnold Networks (KAN) and a bidirectional long short-term memory network (BiLSTM). The power regulator parameters, electro-hydraulic converter time constant, and oil motor time constant are used as inputs, and the high-pressure steam chamber volume time constant, reheater volume time constant, low-pressure connecting pipe steam chamber volume time constant, and high-pressure cylinder power natural overmodulation coefficient are used as feedback. The input and output are extracted for features, and the sliding time window method is used to construct a time series array from the extracted features. The reconstructed time series array is loaded into the KAN-BiLSTM model for model training. The time series array is divided into a training set and a validation set. Usually 80% of the data is used as the training set, and 20% of the data is used as the validation set. During the training process, the cross entropy loss function is used to measure the reconstruction error between the model prediction result and the true label. The trained KAN-BiLSTM model is combined with the mechanism model of the thermal power unit to obtain the optimal primary frequency modulation model.

[0143] Finally, the optimized single frequency modulation model is run using the signal with a specific frequency deviation as the input variable.

[0144] Specifically, the frequency deviation is set to +4% and -4%, and these deviation signals are respectively input into the aforementioned optimized primary frequency regulation margin estimation model. During the operation of the model, the changes in the output power of the generator set caused by these frequency deviation signals are recorded, and the waveforms of these power changes are monitored and visualized in real time using a high-precision oscilloscope. Through detailed analysis and evaluation of the obtained waveform data, the trend of the power output of the generator set changing with the frequency deviation signal is observed. In addition, with the help of these visualization results, we can also deeply explore the impact mechanism of frequency deviation on the power output of the generator set.

[0145] In summary, based on a series of experimental data and observation results, the primary frequency regulation capability and dynamic performance of the generator set are further evaluated. In particular, the ability of the generator set to maintain stable operation and meet the grid frequency control requirements under different frequency deviation conditions is determined.

[0146] See also Figure 5 , Figure 5 The block diagram of the primary frequency regulation margin estimation system 200 for a thermal power plant provided in this embodiment is shown. The primary frequency regulation margin estimation system 200 for a thermal power plant includes:

[0147] An acquisition module 210 is configured to acquire operation data of a primary frequency modulation of a thermal power unit, and determine a time domain data group of the thermal power unit based on the operation data, wherein the time domain data group is used to characterize operation parameters of the thermal power unit at different historical time points;

[0148] The model building module 220 is configured to build a steam turbine model, an initial mechanism model combining the DEH system and the steam turbine model, and a deep learning model embedded in the initial mechanism model.

[0149] The first determination module 230 is configured to input the time domain data group into the steam turbine model to obtain initial parameter information,

[0150] A data optimization module 240 is configured to optimize the initial parameter information based on a Hippo optimization algorithm to determine key parameter information, wherein the key parameter information is used to characterize a plurality of volume time constants of the steam turbine;

[0151] The second determination module 250 is configured to input the key parameter information and the valve opening information into the deep learning model to obtain an optimized control signal, and optimize the initial mechanism model based on the optimized control signal to determine a primary frequency modulation margin estimation model;

[0152] The third determination module 260 is configured to input the frequency difference signal and the power given value into the primary frequency regulation margin estimation model to determine the margin data achieved by the primary frequency regulation of the thermal power unit.

[0153] It can be understood that when the thermal power unit primary frequency regulation margin estimation system 200 in this embodiment is implemented, each module runs the above Figures 1 to 4 The steps of the method for estimating the primary frequency regulation margin of a thermal power unit in the corresponding embodiment, and the technical effects that can be achieved can be referred to the above Figures 1 to 4 The technical effects achieved in the corresponding embodiments are not described in detail here.

[0154] See also Figure 6 , Figure 6 5 is a structural block diagram of an electronic device provided in an embodiment of the present application. The server 500 of the electronic device includes: a processor 501, a memory 502, and a computer program 503 stored in the memory 502 and executable on the processor 501, such as a program of a method for estimating the primary frequency regulation margin of a thermal power plant. When the processor 501 executes the computer program 503, the steps of the method for estimating the primary frequency regulation margin of a thermal power plant in the above-mentioned embodiments are implemented, such as Figure 1 Alternatively, the processor 501 executes the computer program 503 to implement the above Figure 5 The functions of each module in the corresponding embodiment are, for example, Figure 5 For details on the functions of the modules (eg, acquisition module 210), please refer to Figure 5 The relevant descriptions in the corresponding embodiments are not repeated here.

[0155] Exemplarily, the computer program 503 may be divided into one or more units, one or more units are stored in the memory 502, and are executed by the processor 501 to complete the technical solution provided in the above embodiment. One or more units may be a series of computer program instruction segments that can complete specific functions, and the instruction segments are used to describe the execution process of the computer program 503 in the server 500.

[0156] The electronic device may include, but is not limited to, a processor 501 and a memory 502. Those skilled in the art will appreciate that Figure 6 It is only an example of the server 500 in the electronic device and does not constitute a limitation of the server 500. It may include more or fewer components than shown in the figure, or a combination of certain components, or different components. For example, the turntable terminal device may also include an input and output terminal device, a network access terminal device, a bus, etc.

[0157] The processor 501 may be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor may be a microprocessor or any conventional processor, etc.

[0158] The memory 502 may be an internal storage unit of the server 500, such as a hard disk or memory of the server 500. The memory 502 may also be an external storage terminal device of the server 500, such as a plug-in hard disk, a smart media card (SMC), a secure digital (SD) card, a flash card, etc. equipped on the server 500. Further, the memory 502 may also include both an internal storage unit of the server 500 and an external storage terminal device. The memory 502 is used to store computer programs and other programs and data required by the turntable terminal device. The memory 502 may also be used to temporarily store data that has been output or is to be output.

[0159] In some embodiments, a computer-readable storage medium is further provided, wherein the computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps of the method for estimating the primary frequency regulation margin of a thermal power unit as described in the above embodiment are implemented.

[0160] In addition, each functional unit in each embodiment of the present application may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.

[0161] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Among them, the computer-readable storage medium can be non-volatile or volatile. Based on this understanding, the present application implements all or part of the processes in the above-mentioned embodiment method, and can also be completed by instructing the relevant hardware through a computer program. The computer program can be stored in a computer-readable storage medium, and the computer program can implement the steps of the above-mentioned various method embodiments when executed by the processor. Among them, the computer program includes computer program code, and the computer program code can be in source code form, object code form, executable file or some intermediate form. The computer-readable storage medium may include: any entity or device capable of carrying computer program code, recording medium, U disk, mobile hard disk, disk, optical disk, computer memory, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), electrical carrier signal, telecommunication signal and software distribution medium, etc.

[0162] The terms "first", "second", "third", etc. in the specification and claims of the present application and the drawings are used to distinguish different objects rather than to describe a specific order. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions. For example, a series of steps or units are included, or optionally, steps or units not listed are included, or optionally, other steps or units inherent to these processes, methods, products or devices are included.

[0163] Only the part relevant to the present application is shown in the accompanying drawings, but not all of the content. Before discussing the exemplary embodiments in more detail, it should be mentioned that some exemplary embodiments are described as processing or methods depicted as flow charts. Although the flow chart describes each operation (or step) as a sequential process, many of the operations therein can be implemented in parallel, concurrently or simultaneously. In addition, the order of each operation can be rearranged. When its operation is completed, the process can be terminated, but it can also have additional steps not included in the accompanying drawings. The process can correspond to a method, function, procedure, subroutine, subprogram, etc.

[0164] The terms "component", "module", "system", "unit", etc. used in this specification are used to represent computer-related entities, hardware, firmware, a combination of hardware and software, software, or software in execution. For example, a unit can be, but is not limited to, a process running on a processor, a processor, an object, an executable file, an execution thread, a program, and / or distributed between two or more computers. In addition, these units can be executed from various computer-readable media having various data structures stored thereon. Units can communicate through local and / or remote processes, for example, based on signals having one or more data packets (e.g., data from a second unit interacting with another unit in a local system, a distributed system, and / or a network. For example, the Internet interacts with other systems via signals).

[0165] In the description of this specification, the description with reference to the terms "one embodiment", "some embodiments", "illustrative embodiments", "examples", "specific examples", or "some examples" means that the specific features, structures, materials, or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present application. In this specification, the schematic representation of the above terms does not necessarily refer to the same embodiment or example.

[0166] Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments. Mentioning "embodiment" in this article means that the specific features, structures or characteristics described in conjunction with the embodiment may be included in at least one embodiment of the present embodiment application. The appearance of this phrase in various positions in the specification is not necessarily the same embodiment, nor is it an independent or alternative embodiment that is mutually exclusive with other embodiments. It can be explicitly and implicitly understood by those skilled in the art that the embodiments described herein can be combined with other embodiments. Based on the embodiments in this application, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of this application.

[0167] Although the embodiments of the present application have been shown and described, those skilled in the art will appreciate that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present application, and that the scope of the present application is defined by the claims and their equivalents.

[0168] Those skilled in the art will readily appreciate other embodiments of the present application after considering the specification and practicing the disclosure disclosed herein. The present application is intended to cover any modification, use or adaptation of the present application, which follows the general principles of the present application and includes common knowledge or customary techniques in the art that are not disclosed in the present application. The specification and examples are intended to be exemplary only, and the true scope and spirit of the present application are indicated by the following claims.

Claims

1. A method for estimating primary frequency regulation margin of a thermal power unit, characterized in that: include: Acquire the operation data of the primary frequency modulation of the thermal power unit, and determine the time domain data group of the thermal power unit based on the operation data, wherein the time domain data group is used to characterize the operation parameters of the thermal power unit at different historical time points; Building a steam turbine model, inputting the time domain data group into the steam turbine model to obtain initial parameter information, wherein the steam turbine model building includes a high-pressure steam chamber volume link, a reheater volume link, a low-pressure connecting pipe steam chamber volume link and a high-pressure cylinder power natural overshoot coefficient link, and time series data based on main steam pressure and comprehensive valve position as input of the steam turbine model, and actual load as output of the steam turbine model; Optimizing the initial parameter information based on the Hippo optimization algorithm to obtain key parameter information, wherein the key parameter information is used to characterize multiple volume time constants of the steam turbine; An initial mechanism model combining the DEH system and the steam turbine model is constructed, and a deep learning model embedded in the initial mechanism model is constructed, and the key parameter information and valve opening information are input into the deep learning model to obtain an optimized control signal, and the initial mechanism model is optimized based on the optimized control signal to obtain a primary frequency regulation margin estimation model, which includes constructing a target optimization model, using power regulator parameters, electro-hydraulic converter time constants, and oil motor time constants as model inputs, optimizing control signals as model outputs, and using high-pressure steam chamber volume time constants, reheater volume time constants, low-pressure connecting pipe steam chamber volume time constants, and high-pressure cylinder power natural overmodulation coefficients as model feedbacks. Extract features from the input and output data, use a sliding time window method to construct a time series array from the extracted features, and divide the reconstructed time series array into a training set and a validation set. The target optimization model is trained based on the training set, and the trained target optimization model is verified based on the verification set. Based on the preset cross entropy loss function, the difference between the predicted results of training and verification output and the actual observed values ​​is determined. Adaptively adjusting the network weights of the target optimization model based on the difference value to obtain a deep learning model; The frequency difference signal and the power given value are input into the primary frequency regulation margin estimation model to determine the margin data achieved by the primary frequency regulation of the thermal power unit.

2. A method for estimating primary frequency regulation margin of a thermal power unit according to claim 1, characterized in that: The obtaining of the operation data of the primary frequency modulation of the thermal power unit, and determining the time domain data group of the thermal power unit based on the operation data, wherein the time domain data group is used to characterize the operation parameters of the thermal power unit at different historical time points, includes: Collecting the operating data of the thermal power unit under different rated loads and frequency difference step signal conditions; Based on the running data, the time point index of the time series array is screened out, and the valid data segment is intercepted and the abnormal batch data containing abnormal values ​​is eliminated; An interpolation function is created, and based on the interpolation function, the time points of each historical data are aligned to the time points of the parameters to obtain a time domain data group including the operating time, main steam pressure, comprehensive valve position and power.

3. A method for estimating primary frequency regulation margin of a thermal power unit according to claim 1, characterized in that: The key parameter information is obtained by optimizing the initial parameter information based on the Hippo optimization algorithm, and the key parameter information is used to characterize multiple volume time constants of the steam turbine, including: Initial parameter information is obtained according to the steam turbine model identification, and each sub-data in the initial parameter information is iteratively optimized based on the Hippo optimization algorithm to obtain key parameter information, wherein the key parameter information at least includes the high-pressure steam chamber volume time constant, the reheater volume time constant, the low-pressure connecting pipe steam chamber volume time constant and the high-pressure cylinder power natural overmodulation coefficient.

4. A method for estimating primary frequency regulation margin of a thermal power unit according to claim 3, characterized in that: The initial parameter information is obtained according to the steam turbine model identification, and each sub-data in the initial parameter information is iteratively optimized based on the Hippo optimization algorithm to obtain key parameter information, including: Set the initial value range of each sub-data in the initial parameter information, where: The initialization expression is: In the formula, is the position of the corresponding sub-data, and are the lower and upper limits of the initial values ​​of the corresponding sub-data, is a random number between 0 and 1; Iterate each sub-data after initialization, and determine whether the number of iterations reaches a preset iteration threshold or whether the fitness value is less than a preset fitness threshold; If so, output the corresponding high-pressure steam chamber volume time constant, reheater volume time constant, low-pressure connecting pipe steam chamber volume time constant and high-pressure cylinder power natural overmodulation coefficient value.

5. A method for estimating primary frequency regulation margin of a thermal power unit according to claim 1, characterized in that: The step of inputting a frequency difference signal and a power given value into the primary frequency regulation margin estimation model to determine the margin data achieved by the primary frequency regulation of the thermal power unit includes: Inputting a frequency difference signal and a power given value into the primary frequency modulation margin estimation model to obtain a primary frequency modulation response waveform; Based on the primary frequency regulation response waveform, the primary frequency regulation margin of the thermal power unit under different frequency deviations is obtained, wherein: The primary frequency modulation margin includes additional frequency modulation capability, margin response time, and margin recovery time.

6. A method system for estimating primary frequency regulation margin of a thermal power unit, characterized in that: include: an acquisition module, configured to acquire operation data of a primary frequency modulation of a thermal power unit, and determine a time domain data group of the thermal power unit based on the operation data, wherein the time domain data group is used to characterize operation parameters of the thermal power unit at different historical time points; The model building module is configured to build a steam turbine model, an initial mechanism model combining a DEH system and the steam turbine model, and a deep learning model embedded in the initial mechanism model, wherein the steam turbine model building includes a high-pressure steam chamber volume link, a reheater volume link, a low-pressure connecting pipe steam chamber volume link, and a high-pressure cylinder power natural overshoot coefficient link, as well as time series data based on main steam pressure and comprehensive valve position as input of the steam turbine model, and actual load as output of the steam turbine model; wherein, it includes building a target optimization model, with power regulator parameters, electro-hydraulic converter time constant, and oil motor time constant as model input, optimized control signal as model output, and high-pressure steam chamber volume time constant, reheater volume time constant, low-pressure connecting pipe steam chamber volume time constant, and high-pressure cylinder power natural overshoot coefficient as model feedback, Extract features from the input and output data, use a sliding time window method to construct a time series array from the extracted features, and divide the reconstructed time series array into a training set and a validation set. The target optimization model is trained based on the training set, and the trained target optimization model is verified based on the verification set. Based on the preset cross entropy loss function, the difference between the predicted results of training and verification output and the actual observed values ​​is determined. Adaptively adjusting the network weights of the target optimization model based on the difference value to obtain a deep learning model; A first determination module is configured to input the time domain data group into the steam turbine model to obtain initial parameter information; A data optimization module is configured to optimize the initial parameter information based on a Hippo optimization algorithm to determine key parameter information, wherein the key parameter information is used to characterize a plurality of volume time constants of the steam turbine; A second determination module is configured to input the key parameter information and valve opening information into the deep learning model to obtain an optimized control signal, and optimize the initial mechanism model based on the optimized control signal to determine a primary frequency modulation margin estimation model; The third determination module is configured to input a frequency difference signal and a power given value into the primary frequency regulation margin estimation model to determine the margin data achieved by the primary frequency regulation of the thermal power unit.

7. An electronic device, characterized in that: include: at least one processor; as well as a memory communicatively connected to the at least one processor; wherein, The memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute the steps of a method for estimating the primary frequency regulation margin of a thermal power unit according to any one of claims 1-6.

8. A computer-readable storage medium having computer program instructions stored thereon, characterized in that: When the computer program instructions are executed by a processor, the steps of a method for estimating primary frequency regulation margin of a thermal power unit according to any one of claims 1 to 6 are implemented.

Images