Hydraulic power plant AGC frequency modulation strategy optimization method and device based on big data analysis

Abstract

The invention provides a hydraulic power plant AGC frequency modulation strategy optimization method and device based on big data analysis. The method comprises the steps that current unit operation parameters of a unit of a target hydraulic power plant, current power system operation parameters of a power system controlled by the target hydraulic power plant and current external environment parameters of the target hydraulic power plant are collected; big data analysis is carried out on the current power system operation parameters and the current external environment parameters, and frequency modulation key features related to AGC frequency modulation are extracted; inputting the frequency modulation key features into a pre-trained frequency modulation strategy prediction model to obtain first power generation prediction power and second power generation prediction power output by the frequency modulation strategy prediction model; determining target power generation power based on the first power generation predicted power and the second power generation predicted power; target unit operation parameters are determined based on the target generation power and the current unit operation parameters, and units of the target hydraulic power plant are controlled to operate according to the target unit operation parameters. The reliability and the stability of the power system are improved.

Description

Technical Field

[0001] This invention relates to the field of computer technology, and in particular to a method and apparatus for optimizing the frequency regulation strategy of hydropower plants based on big data analysis. Background Technology

[0002] With the continuous expansion of the power system and the increasing proportion of renewable energy integration, the frequency stability of the power system faces more severe challenges. As an important frequency-regulating power source in the power system, hydropower plants play a crucial role in maintaining the frequency stability of the power system through their Automatic Generation Control (AGC) frequency regulation performance.

[0003] Traditional AGC frequency regulation in hydropower plants often relies on simple control strategies and empirical models, which are insufficient to adapt to the complex and ever-changing operating conditions of power systems and external environmental changes. For example, rapid fluctuations in power system load, the intermittency and uncertainty of renewable energy generation, and differences in external environments under different seasons and climates can all significantly impact the frequency regulation performance of hydropower plants. Therefore, traditional methods may encounter problems such as untimely frequency regulation and insufficient frequency regulation accuracy when dealing with complex situations, leading to power system frequency fluctuations exceeding the allowable range and affecting the safe, stable operation and reliability of the power system. Summary of the Invention

[0004] This invention provides a method and apparatus for optimizing the frequency regulation strategy of hydropower plants based on big data analysis, in order to improve the reliability and stability of power systems.

[0005] In a first aspect, the present invention provides a method for optimizing the frequency regulation strategy of AGC in hydropower plants based on big data analysis, including: The current operating parameters of the generating units of the target hydropower plant, the current operating parameters of the power system controlled by the target hydropower plant, and the current external environmental parameters of the target hydropower plant are collected. Big data analysis is performed on the current power system operating parameters and the current external environment parameters to extract key frequency regulation features related to AGC frequency regulation; The frequency regulation key features are input into a pre-trained frequency regulation strategy prediction model to obtain the first and second power generation predictions output by the frequency regulation strategy prediction model. The target power generation is determined based on the first power generation forecast and the second power generation forecast. The target unit operating parameters are determined based on the target power generation and the current unit operating parameters, and the units of the target hydropower plant are controlled to operate at the target unit operating parameters. The frequency regulation strategy prediction model includes a first prediction model and a second prediction model. The first power generation prediction is based on the output of the first prediction model, and the second power generation prediction is based on the output of the second prediction model. The first prediction model is trained on a preset deep learning network based on the frequency regulation features of the target sample and its corresponding power generation label results. The second prediction model is trained on a preset multilayer perceptron network based on the frequency regulation features of the target sample and its corresponding power generation label results.

[0006] Secondly, the present invention also provides a hydropower plant AGC frequency regulation strategy optimization device based on big data analysis, applied to the hydropower plant AGC frequency regulation strategy optimization method based on big data analysis as described in the first aspect; the hydropower plant AGC frequency regulation strategy optimization device based on big data analysis includes: The data acquisition module is used to collect the current unit operating parameters of the target hydropower plant's generating units, the current power system operating parameters of the power system controlled by the target hydropower plant, and the current external environmental parameters of the target hydropower plant. The data analysis module is used to perform big data analysis on the current power system operating parameters and the current external environment parameters, and extract key frequency regulation features related to AGC frequency regulation. The model prediction module is used to input the key frequency regulation features into a pre-trained frequency regulation strategy prediction model to obtain the first and second power generation predictions output by the frequency regulation strategy prediction model. A power determination module is used to determine a target power generation based on the first power generation prediction power and the second power generation prediction power. The frequency regulation module is used to determine the target unit operating parameters based on the target power generation and the current unit operating parameters, and to control the units of the target hydropower plant to operate at the target unit operating parameters; The frequency regulation strategy prediction model includes a first prediction model and a second prediction model. The first power generation prediction is based on the output of the first prediction model, and the second power generation prediction is based on the output of the second prediction model. The first prediction model is trained on a preset deep learning network based on the frequency regulation features of the target sample and its corresponding power generation label results. The second prediction model is trained on a preset multilayer perceptron network based on the frequency regulation features of the target sample and its corresponding power generation label results.

[0007] Thirdly, the present invention also provides an electronic device, comprising: a memory for storing computer software programs; and a processor for reading and executing the computer software programs, thereby implementing the hydropower plant AGC frequency regulation strategy optimization method based on big data analysis as described above.

[0008] Fourthly, the present invention also provides a non-transitory computer-readable storage medium storing a computer software program, which, when executed by a processor, implements the above-described method for optimizing the frequency regulation strategy of a hydropower plant based on big data analysis.

[0009] Fifthly, the present invention also provides a computer program product, including a computer program that, when executed by a processor, implements the hydropower plant AGC frequency regulation strategy optimization method based on big data analysis as described above.

[0010] The hydropower plant AGC frequency regulation strategy optimization method based on big data analysis provided in this invention can deeply and comprehensively mine key frequency regulation features related to AGC frequency regulation by performing big data analysis on power system operating parameters and external environmental parameters. These key frequency regulation features are then input into a first prediction model trained based on a deep learning network and a second prediction model trained based on a multilayer perceptron network, respectively, to obtain a first predicted power generation and a second predicted power generation. Using two different models for prediction allows for a comprehensive consideration of the advantages and characteristics of each model, improving the accuracy and reliability of power generation prediction. Based on the two predicted power outputs, a target power generation is determined. Further adjustments are made to the current unit operating parameters based on the target power generation to obtain the target unit operating parameters, and the unit is controlled to operate accordingly. This allows for dynamic adjustment of unit power generation based on the real-time operating status of the power system and changes in the external environment, effectively reducing power system frequency fluctuations and improving the reliability and stability of the power system. Attached Figure Description

[0011] Figure 1 This is a flowchart illustrating the method for optimizing the frequency regulation strategy of a hydropower plant based on big data analysis, as provided in an embodiment of the present invention. Figure 2 This is a schematic diagram of the structure of the hydropower plant AGC frequency regulation strategy optimization device based on big data analysis provided in an embodiment of the present invention; Figure 3 An embodiment diagram of the electronic device provided in this invention; Figure 4 An embodiment diagram of a computer-readable storage medium provided in accordance with the present invention. Detailed Implementation

[0012] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0013] In the description of this invention, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of indicated technical features. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of the stated features. In the description of this invention, "a plurality of" means two or more, unless otherwise explicitly specified.

[0014] In the description of this invention, the term "for example" is used to mean "used as an example, illustration, or description." Any embodiment described as "for example" in this invention is not necessarily to be construed as being more preferred or advantageous than other embodiments. The following description is provided to enable any person skilled in the art to make and use the invention. Details are set forth in the following description for purposes of explanation. It should be understood that those skilled in the art will recognize that the invention can be made without using these specific details. In other instances, well-known structures and processes will not be described in detail to avoid obscuring the description of the invention with unnecessary detail. Therefore, the invention is not intended to be limited to the embodiments shown, but is consistent with the broadest scope of the principles and features disclosed herein.

[0015] Optional, see below Figure 1 , Figure 1 This is a flowchart illustrating the hydropower plant AGC frequency regulation strategy optimization method based on big data analysis provided by the present invention. In this embodiment of the invention, the executing entity of the hydropower plant AGC frequency regulation strategy optimization method based on big data analysis is a frequency regulation optimization device. Therefore, the hydropower plant AGC frequency regulation strategy optimization method based on big data analysis includes: Step 10: Collect the current operating parameters of the generating units of the target hydropower plant, the current operating parameters of the power system controlled by the target hydropower plant, and the current external environmental parameters of the target hydropower plant.

[0016] Optionally, the unit operating parameters in the embodiments of the present invention may include turbine speed, guide vane opening, current output power of the unit, temperature and pressure of each component, etc.; power system operating parameters may include overall grid frequency, real-time power transmitted by each tie line, voltage amplitude and phase angle of each node, etc.; external environmental parameters may include current temperature, precipitation (affecting water flow), air humidity, etc.

[0017] Therefore, the frequency regulation optimization device collects the current unit operating parameters of any target hydropower plant's generating units. ,in, This indicates the dimension of the current unit operating parameters. Indicates the first Individual unit operating parameters, such as This represents the current output power of the generator unit. The turbine speed is measured, and simultaneously, the frequency regulation optimization device collects the current power system operating parameters of the power system controlled by the target hydropower plant. ,in, The dimension representing the current operating parameters of the power system. Indicates the first Power system operating parameters, such as For tie line power, This is the grid frequency. Simultaneously, the frequency regulation optimization device collects the current external environmental parameters of the target hydropower plant. ,in, This indicates the dimension of the current external environment parameters. Indicates the first External environmental parameters, such as The incoming water flow rate Temperature.

[0018] Specifically, unit operating parameters can be obtained through sensors installed at various key locations within the unit, such as power sensors at the generator output and speed sensors at the turbine main shaft. Power system operating parameters are collected and aggregated by the power system monitoring system; for example, smart meters in the power grid and monitoring devices in substations upload relevant data to the dispatch center in real time. Frequency regulation optimization devices can access the dispatch center. External environmental parameters, such as inflow water flow, can be obtained using flow meters installed upstream of the river and at the intake. Meteorological data (temperature, precipitation, etc.) can be obtained through data sharing from meteorological stations or through small meteorological monitoring equipment installed by the hydropower plant itself.

[0019] It should be noted that the unit operating parameters in the embodiments of the present invention refer to the unit's power generation capacity.

[0020] Step 20: Perform big data analysis on the current power system operating parameters and current external environment parameters to extract key frequency regulation features related to AGC frequency regulation.

[0021] Optionally, the frequency regulation optimization device analyzes the characteristics of each parameter in the current power system operating parameters and the characteristics of each parameter in the current external environment parameters through big data analysis, and the correlation between them and the AGC frequency regulation target. In this embodiment of the invention, the Pearson correlation coefficient can be used to measure the correlation, and the key frequency regulation characteristics related to AGC frequency regulation in the current power system operating parameters and the current external environment parameters can be obtained.

[0022] Optionally, the key frequency regulation features in the embodiments of the present invention include grid frequency deviation features, tie line power deviation features, inflow water flow features, and power load variation features.

[0023] 1. Extraction of power grid frequency deviation characteristics: Obtain time-series data of the power grid frequency from the current power system operating parameters and record the frequency values. The time interval is The time series length is .

[0024] In one embodiment, the rated frequency of the power grid is Calculate the frequency deviation sequence ,in, .

[0025] For short-term fluctuation feature extraction: a sliding window technique is used, with the window size set to [value missing]. Calculate the standard deviation of the frequency deviation within each window to obtain the standard deviation sequence. : ; The standard deviation series reflects the degree of fluctuation in the power grid frequency in the short term.

[0026] Calculate the maximum rate of change of frequency deviation within each window. : ; Among them, the maximum rate of change of frequency deviation reflects how fast the frequency changes in a short period of time.

[0027] For long-term deviation feature extraction: calculate the average frequency deviation over a time period (e.g., one day). : ; The average frequency deviation reflects the overall offset of the power grid frequency.

[0028] Calculate the root mean square value of the frequency deviation : ; The root mean square value of the frequency deviation is used to measure the overall magnitude of the frequency deviation.

[0029] 2. Extraction of tie line power deviation features: Time series data of tie line power collection , length is To filter the data and remove high-frequency noise interference, a low-pass filter, such as a Butterworth low-pass filter, can be used.

[0030] In one embodiment, the planned power of the tie line is Then the power deviation sequence of the tie line is .

[0031] Fluctuation Feature Extraction: Calculating the Autocorrelation Function of Power Deviation : ; By analyzing the autocorrelation function The attenuation characteristics yielded the periodic fluctuations in power deviation.

[0032] The power deviation sequence is decomposed into subsequences of different frequencies using Discrete Wavelet Transform (DWT). For example, decomposition using the Daubechies wavelet basis function yields approximation coefficients and detail coefficients. The approximation coefficients reflect the low-frequency components (long-term trend) of the power deviation, while the detail coefficients reflect the high-frequency components (short-term fluctuations). Analyzing the energy distribution of the detail coefficients allows determination of the main fluctuation frequency range of the power deviation.

[0033] Deviation amplitude feature extraction: Calculate the maximum absolute value of the power deviation. : ; Among them, the maximum value of the absolute value of the power deviation. The maximum magnitude of the deviation.

[0034] Calculate the average value of the absolute value of the power deviation : ; Among them, the average value of the absolute value of the power deviation Used to measure the overall level of deviation.

[0035] 3. Extraction of inflow flow characteristics Time series data of water inflow obtained from external environmental parameters. , length is Check the completeness and accuracy of the data. For missing data points, the average of adjacent data or estimations based on relevant factors such as season and precipitation can be used to supplement them.

[0036] Trend Feature Extraction: Calculating the First-Order Difference Sequence of Inflow Water Flow Analyze the sign and magnitude of the first-order difference sequence to determine whether the inflow rate is increasing, decreasing, or stable. Employ polynomial fitting methods, such as quadratic polynomial fitting. The long-term trend of water flow rate is further determined based on the sign and magnitude of the fitting coefficients.

[0037] Periodic feature extraction: Fast Fourier Transform (FFT) is performed on the inflow flow data to obtain the spectrum. (in one embodiment) (Even numbers). Find the peak value in the spectrum; the corresponding frequency is the main periodic component of the inflow. Use wavelet analysis to perform time-frequency analysis on the inflow, obtaining a time-frequency graph. Observe the energy distribution at different times and frequencies from the time-frequency graph to determine the multi-period characteristics of the inflow and the time range of each period.

[0038] Flow amplitude feature extraction: Calculate the maximum inflow flow rate and minimum value This gives the range of flow rate variation. The average inflow flow rate is then calculated. and standard deviation It is used to describe the overall level and degree of fluctuation of traffic.

[0039] 4. Extraction of power load variation characteristics Time series data of power load are obtained from power system operating parameters. , length is .

[0040] Rate of change feature extraction: Calculating the first-order difference sequence of power load Calculate the average of the first-order difference sequence. and standard deviation It is used to describe the average rate and fluctuation of load changes.

[0041] Calculate the percentile of the load change rate, such as the 25th percentile. 50th percentile (median) and the 75th percentile Percentiles can be used to obtain the distribution of load change rates, for example, The difference can reflect the degree of dispersion of the load change rate.

[0042] Periodicity and trend feature extraction: Perform spectral analysis on power load data, such as FFT or wavelet analysis, similar to water flow feature extraction, to determine the main periodic components and long-term trends of load changes.

[0043] Seasonal decomposition methods, such as seasonally based decomposition using moving averages (STL), are employed to decompose load data into seasonal components, trend components, and residual components. The periodicity and amplitude variations of the seasonal components, as well as the slope and direction of the trend components, are analyzed to obtain the seasonal patterns and long-term trends of load changes.

[0044] Step 30: Input the key frequency regulation features into the pre-trained frequency regulation strategy prediction model to obtain the first and second power generation predictions output by the frequency regulation strategy prediction model.

[0045] Optionally, the frequency regulation optimization device of this embodiment of the invention has a pre-trained frequency regulation strategy prediction model embedded in it. The frequency regulation strategy prediction model is obtained by fusing a first prediction model and a second prediction model. The first prediction model is obtained by training a preset deep learning network based on the frequency regulation features of the target sample and its corresponding power generation label results. The second prediction model is obtained by training a preset multilayer perceptron network based on the frequency regulation features of the target sample and its corresponding power generation label results. The specific training process of the frequency regulation strategy prediction model is described in steps 401 to 403.

[0046] In this embodiment of the invention, the preset deep learning network includes multiple convolutional layers, pooling layers, and fully connected layers, and the preset multilayer perceptron network includes an input layer, multiple hidden layers, and an output layer. Therefore, the frequency modulation optimization device inputs key frequency modulation features into a pre-trained frequency modulation strategy prediction model. The first prediction model performs calculations on the key frequency modulation features through multiple convolutional layers, pooling layers, and fully connected layers, outputting a first predicted power generation. In this embodiment of the invention, the combined operations of each layer are simplified as functions. Therefore, the first predicted power generation It can be represented as: ; in, This indicates the key characteristics of frequency modulation. This represents the output weight coefficients of the first prediction model. This represents the bias term of the first prediction model.

[0047] Furthermore, the second prediction model performs calculations on the key features of frequency regulation through an input layer, multiple hidden layers, and an output layer, outputting the second predicted power generation. In this embodiment of the invention, the combined operations of each layer are simplified as functions. Therefore, the second power generation forecast It can be represented as: ; in, This represents the weighting coefficients of the second prediction model. This represents the bias term of the second prediction model.

[0048] Step 40: Determine the target power generation based on the first power generation forecast and the second power generation forecast.

[0049] Furthermore, the frequency regulation optimization device determines the target power generation based on the first and second predicted power generation, as detailed below: The frequency regulation optimization device calculates a first weighting coefficient for the first power generation prediction and a second weighting coefficient for the second power generation prediction based on the first power generation prediction and the second power generation prediction. The first weighting coefficient can be expressed as first power generation prediction / (first power generation prediction + second power generation prediction), and the second weighting coefficient can be expressed as second power generation prediction / (first power generation prediction + second power generation prediction).

[0050] Furthermore, the frequency regulation optimization device performs a weighted calculation based on the first predicted power generation and its corresponding first weight coefficient, and the second predicted power generation and its corresponding second weight coefficient to obtain the target power generation. Therefore, the target power generation = first predicted power generation * first weight coefficient + second predicted power generation * second weight coefficient.

[0051] Step 50: Determine the target unit operating parameters based on the target power generation and the current unit operating parameters, and control the units of the target hydropower plant to operate at the target unit operating parameters.

[0052] Furthermore, the frequency regulation optimization device determines the target unit operating parameters based on the target power generation and the current unit operating parameters. Specifically, the frequency regulation optimization device calculates the absolute value of the power generation difference between the target power generation and the current unit operating parameters, and determines whether the absolute value of the power generation difference is greater than or equal to a preset power difference value, which is set according to actual conditions, such as 0.01 or 0.05. If the absolute value of the power generation difference is less than or equal to the preset power difference value, the frequency regulation optimization device determines the target power generation or the current unit operating parameters as the target unit operating parameters.

[0053] If the absolute value of the power generation difference is determined to be greater than the preset power difference value, the frequency regulation optimization device adjusts the current unit operating parameters based on the absolute value of the power generation difference to obtain the target unit operating parameters. Specifically, if the target power generation is greater than the current unit operating parameters, the calculation formula for the target unit operating parameters is: ; in, Indicates the operating parameters of the target unit. This indicates the current operating parameters of the unit. This represents the difference between the target power generation and the current operating parameters of the generating unit. This represents the base of the exponential function.

[0054] If the target power generation is less than the current unit operating parameters, the formula for calculating the target unit operating parameters is: .

[0055] Furthermore, the frequency regulation optimization device controls the units of the target hydropower plant to operate according to the target unit operating parameters.

[0056] This invention, through big data analysis of power system operating parameters and external environmental parameters, can deeply and comprehensively uncover key frequency regulation features related to AGC (Automatic Generation Control) frequency regulation. These key features are then input into a first prediction model trained on a deep learning network and a second prediction model trained on a multilayer perceptron network, respectively, to obtain a first predicted power generation and a second predicted power generation. Using two different models for prediction allows for a comprehensive consideration of their respective advantages and characteristics, improving the accuracy and reliability of power generation prediction. Based on these two predicted power outputs, a target power generation is determined. Further adjustments are made to the current unit operating parameters based on the target power generation to obtain the target unit operating parameters, and the units are controlled to operate accordingly. This allows for dynamic adjustment of unit power generation based on the real-time operating status of the power system and changes in the external environment, effectively reducing power system frequency fluctuations and improving the reliability and stability of the power system.

[0057] In one embodiment, before training the frequency regulation strategy prediction model, it is necessary to screen the frequency regulation features of the original samples. The frequency regulation strategy prediction model is then trained using the screened target sample frequency regulation features and their corresponding power generation label results, as detailed in steps a to d: Step a: Obtain the frequency modulation features of multiple raw samples and their corresponding power generation label results; Step b: Input each original sample frequency regulation feature into the data optimization model to obtain the power generation prediction result of each original sample frequency regulation feature output by the data optimization model.

[0058] Specifically, the frequency regulation optimization device acquires multiple original sample frequency regulation features and their corresponding power generation label results. The original sample frequency regulation features are also obtained through steps 10 and 20.

[0059] Furthermore, the frequency regulation optimization device inputs the frequency regulation features of each original sample into the data optimization model, obtaining the power generation prediction result of each original sample frequency regulation feature output by the data optimization model. The data optimization model is built based on a machine learning or deep learning framework, and its structure may include multiple hidden layers. Therefore, the data is processed according to the model through nonlinear transformations of each hidden layer (e.g., using activation functions such as ReLU, Sigmoid, etc.), and finally, the output layer outputs the predicted power generation value corresponding to each original sample frequency regulation feature.

[0060] Step c: Based on the loss function of the data optimization model, the loss function value of each original sample frequency regulation feature is determined according to the power generation label result and power generation prediction result of each original sample frequency regulation feature. Step d: Iterate through the loss function value of each original sample frequency modulation feature, and determine the original sample frequency modulation features whose loss function value is less than or equal to the first preset difference threshold as the target sample frequency modulation features, and obtain the target sample frequency modulation features and their corresponding power generation label results.

[0061] Furthermore, the frequency regulation optimization device inputs the power generation label results and power generation prediction results of each original sample frequency regulation feature into the loss function of the data optimization model to calculate the loss value, thereby obtaining the loss function value of each original sample frequency regulation feature. The loss function value of each original sample frequency modulation feature. The calculation formula is as follows: ; in, Indicates the preset weight, and ; Indicates the first The original sample frequency modulation features and their corresponding power generation labels Indicates the first Predicted power generation values ​​based on the frequency modulation characteristics of each original sample. Indicates the preset coefficient. This represents the maximum value function.

[0062] Furthermore, the frequency regulation optimization device iterates through the loss function value of each original sample frequency regulation feature, and determines the original sample frequency regulation feature whose loss function value is less than or equal to the first preset difference threshold as the target sample frequency regulation feature, thereby obtaining the target sample frequency regulation feature and its corresponding power generation label result. The first preset difference threshold is set according to actual conditions, such as the first preset difference threshold being 0.02, 0.05, etc.

[0063] In this embodiment of the invention, before training the frequency regulation strategy prediction model, the frequency regulation features of the original samples are screened. The frequency regulation features of the screened target samples and their corresponding power generation label results are used to train a frequency regulation strategy prediction model with better performance, so that the prediction results of the frequency regulation strategy prediction model are more accurate.

[0064] In one embodiment, steps 401 to 403 are described as follows: Step 401: Input the frequency modulation features of the target sample into a preset deep learning network to obtain the first power prediction value output by the preset deep learning network; based on the first loss function in the preset deep learning network, obtain the first loss function value according to the first power prediction value and the power generation label result corresponding to the frequency modulation features of the target sample.

[0065] Optionally, the frequency modulation optimization device inputs the frequency modulation features of the target sample into a preset deep learning network. The preset deep learning network performs calculations on the frequency modulation features of the target sample through multiple convolutional layers, pooling layers, and fully connected layers, and outputs the first power prediction value of the frequency modulation features of the target sample.

[0066] Furthermore, the frequency regulation optimization device inputs the power generation label result of the frequency regulation feature of the target sample and the first power prediction value into the first loss function of the preset deep learning network to calculate the loss value, thereby obtaining the first loss function value of the frequency regulation feature of each target sample. Wherein, the first loss function value of the frequency modulation feature of each target sample The calculation formula is as follows: ; in, Indicates the first The first power prediction value of the frequency modulation characteristics of each target sample Indicates the first The power generation labeling results of the frequency modulation characteristics of each target sample.

[0067] Step 402: Input the frequency modulation features of the target sample into a preset multilayer perceptron network to obtain the second power prediction value output by the preset multilayer perceptron network; based on the second loss function in the preset multilayer perceptron network, obtain the second loss function value according to the second power prediction value and the power generation label result corresponding to the frequency modulation features of the target sample.

[0068] Furthermore, the frequency modulation optimization device inputs the frequency modulation features of the target sample into a preset multilayer perceptron network. The preset multilayer perceptron network performs calculations on the frequency modulation features of the target sample through an input layer, multiple hidden layers, and an output layer, and outputs a second power prediction value of the frequency modulation features of the target sample.

[0069] Furthermore, the frequency regulation optimization device inputs the power generation label result and the second power prediction value corresponding to the frequency regulation feature of the target sample into the second loss function of the preset multilayer perceptron network to calculate the loss value, thereby obtaining the second loss function value of the frequency regulation feature of each target sample. The second loss function value of the frequency modulation feature of each target sample. The calculation formula is as follows: ; in, Indicates the first The second power prediction value of the frequency modulation characteristics of the target sample.

[0070] Step 403: Based on the first loss function value, the second loss function value, and the preset constraints, the model is trained to obtain the frequency modulation strategy prediction model.

[0071] Furthermore, the embodiments of the present invention set preset constraints for model training, wherein the preset constraints are: the loss function value is less than a preset loss threshold, and the difference between the loss function values ​​output by different models is less than a second preset difference threshold. The preset loss threshold and the second preset difference threshold are set according to actual conditions.

[0072] Therefore, the frequency modulation optimization device trains the model based on the first loss function value, the second loss function value, and preset constraints to obtain the frequency modulation strategy prediction model, as detailed below: For the case where the value of the first loss function is greater than or equal to the preset loss threshold, and the value of the second loss function is greater than or equal to the preset loss threshold: If it is determined that the value of the first loss function is greater than or equal to a preset loss threshold, and the value of the second loss function is greater than or equal to a preset loss threshold, the frequency modulation optimization device adjusts the parameters of the first objective optimization function of the preset deep learning network, and adjusts the parameters of the second objective optimization function of the preset multilayer perceptron network, wherein the first objective optimization function is: ; in, Denotes the first objective function. Indicates the number of samples. Indicates the output dimension. Represents an exponential function. This represents the neuron connection weight matrix; Indicates the first Frequency modulation characteristics of each target sample; Indicates the first Transpose of the frequency modulation features of each target sample This represents the bias vector. The first term in the linear transformation result represents the... The absolute value of each element.

[0073] In deep learning networks, the neuron connection weight matrix is ​​one such component. Dimensions Indicates the strength of connections between different neurons. The corresponding number of neurons in the input layer or the number of neurons in the previous layer. This corresponds to the number of neurons in the next layer. In a fully connected layer, the neuron connection weight matrix... Each element It was determined that the first layer from the previous layer... The number of neurons is transferred to the current layer. Connection weights of each neuron. Bias vector. The dimension is Bias vector Each element in This represents the bias term for each neuron in the corresponding layer. The dimension of the target sample frequency modulation feature is... .

[0074] Optionally, the second objective function is: ; in, Indicates the number of network layers. Indicates the first The weight matrix of the layer network, Indicates the first The weight matrix of the layer network, Indicates the first Bias terms of the layer network; This indicates that the multilayer perceptron network starts from the first layer and proceeds sequentially through... The final linear transformation result is obtained by performing multiple linear transformations on the weight matrix and bias vector of the layer network. The innermost layer is the linear transformation result of the first layer. In the final linear transformation result, the first... The absolute value of each element.

[0075] Optionally, the process of adjusting the parameters of the first objective optimization function of the preset deep learning network is as follows: Step 1 Initialization: First, initialize the neuron connection weight matrix. and bias vector Perform random initialization (for example, determine appropriate mean and standard deviation to initialize weights based on factors such as the number of input and output features, and the bias is usually initialized to 0 or a small value).

[0076] Step 2: Calculate the objective optimization function value: [The value of the current function is missing from the original text.] , Frequency modulation features of target samples Substitute into the first objective function The calculations are performed to obtain the corresponding function values. As mentioned earlier, the first objective optimization function... The calculation is based on the frequency modulation features of each target sample. and each output dimension Therefore, calculations should be performed first. Then, sum the squares of all the results.

[0077] Step 3: Parameter Update: Update using gradient-based optimization algorithms (such as stochastic gradient descent, Adam, etc.). and Calculate the first objective function. about and The gradient (obtained through differentiation methods such as the chain rule) is updated in the opposite direction of the gradient, based on parameters such as the learning rate determined by the optimization algorithm. and ,For example The update formula is (in, It's the learning rate. It concerns the first objective function. (gradient) The same applies to updates.

[0078] Iterative loop: Repeat steps 2 and 3, continuously adjusting. and The value of makes the first objective function... The value changes in the desired direction (usually decreasing, because the function is designed to reach a certain reasonable optimization state) until the preset stopping condition is met.

[0079] Second objective optimization function The parameter adjustment process is the same.

[0080] Furthermore, the frequency modulation optimization device predicts the frequency modulation features of the target sample using a pre-set deep learning network with adjusted parameters, and calculates the corresponding loss function value based on the loss function and the prediction result. Simultaneously, the frequency modulation optimization device predicts the frequency modulation features of the target sample using a pre-set multilayer perceptron network with adjusted parameters, and calculates the corresponding loss function value based on the loss function and the prediction result.

[0081] Furthermore, if it is determined that the loss function value of the preset deep learning network after parameter adjustment is less than the preset loss threshold, and the loss function value of the preset multilayer perceptron network after parameter adjustment is less than the preset loss threshold, and the difference between the loss function value of the preset deep learning network after parameter adjustment and the loss function value of the preset multilayer perceptron network after parameter adjustment is less than the second preset difference threshold, then it is determined that the model training is complete, and the trained first prediction model and second prediction model are obtained. The first prediction model and the second prediction model are then fused to obtain the frequency modulation strategy prediction model.

[0082] Furthermore, if it is determined that the loss function value of the parameter-adjusted preset deep learning network is greater than or equal to a preset loss threshold, or / and the loss function value of the parameter-adjusted preset multilayer perceptron network is greater than or equal to a preset loss threshold, or / and the difference between the loss function value of the parameter-adjusted preset deep learning network and the loss function value of the parameter-adjusted preset multilayer perceptron network is greater than or equal to a second preset difference threshold, then the process is repeated until the loss function value of the parameter-adjusted preset deep learning network is less than the preset loss threshold, and the loss function value of the parameter-adjusted preset multilayer perceptron network is less than the preset loss threshold, and the difference between the loss function value of the parameter-adjusted preset deep learning network and the loss function value of the parameter-adjusted preset multilayer perceptron network is less than the second preset difference threshold. Then, a first prediction model and a second prediction model are obtained, and the first prediction model and the second prediction model are fused to obtain a frequency modulation strategy prediction model.

[0083] For cases where the first loss function value is greater than or equal to a preset loss threshold and the second loss function value is less than a preset loss threshold, or where the first loss function value is less than a preset loss threshold and the second loss function value is greater than or equal to a preset loss threshold, this embodiment of the invention uses the case where the first loss function value is greater than or equal to a preset loss threshold and the second loss function value is less than a preset loss threshold as an example: If it is determined that the value of the first loss function is greater than or equal to the preset loss threshold, and the value of the second loss function is less than the preset loss threshold, then it is only necessary to adjust the parameters of the first objective optimization function of the preset deep learning network. Therefore, the frequency modulation optimization device adjusts the parameters of the first objective optimization function of the preset deep learning network.

[0084] Furthermore, the frequency modulation optimization device predicts the frequency modulation features of the target sample through a preset deep learning network with adjusted parameters, and calculates the corresponding loss function value based on the prediction results and the loss function.

[0085] Furthermore, if it is determined that the loss function value of the preset deep learning network after parameter adjustment is less than the preset loss threshold, and the difference between the loss function value of the preset deep learning network after parameter adjustment and the second loss function value is less than the second preset difference threshold, then it is determined that the model training is complete, and the trained first prediction model and second prediction model are obtained. The first prediction model and the second prediction model are then fused to obtain the frequency modulation strategy prediction model.

[0086] Furthermore, if it is determined that the loss function value of the preset deep learning network after parameter adjustment is greater than or equal to the preset loss threshold, or / and the difference between the loss function value of the preset deep learning network after parameter adjustment and the second loss function value is greater than or equal to the second preset difference threshold, then the process is repeated until the loss function value output by the preset deep learning network after parameter adjustment is less than the preset loss threshold, and the difference between the loss function value of the preset deep learning network after parameter adjustment and the second loss function value is less than the second preset difference threshold. Then, the first prediction model and the second prediction model are obtained, and the first prediction model and the second prediction model are fused to obtain the frequency modulation strategy prediction model.

[0087] For the case where the value of the first loss function is less than a preset loss threshold, and the value of the second loss function is less than a preset loss threshold: If it is determined that the first loss function value is less than a preset loss threshold, and the second loss function value is less than a preset loss threshold, the frequency modulation optimization device calculates the loss function difference between the first loss function value and the second loss function value. At this time, if it is determined that the loss function difference is less than the second preset difference threshold, it means that the model training is complete, and the trained first prediction model and second prediction model are obtained. The first prediction model and the second prediction model are then fused to obtain the frequency modulation strategy prediction model.

[0088] Furthermore, if the difference in the loss functions is determined to be greater than or equal to a second preset difference threshold, the frequency modulation optimization device determines the magnitude relationship between the first loss function value and the second loss function value. This magnitude relationship can be either the first loss function value being greater than the second loss function value, or the first loss function value being less than the second loss function value. For the magnitude relationship where the first loss function value is greater than the second loss function value, the parameters of the first objective optimization function of the preset deep learning network need to be adjusted. For the magnitude relationship where the first loss function value is less than the second loss function value, the parameters of the second objective optimization function of the preset multilayer perceptron network need to be adjusted. This embodiment of the invention uses the case where the magnitude relationship is the first loss function value being greater than the second loss function value as an example for illustration: Therefore, if the value of the first loss function is greater than the value of the second loss function, the frequency modulation optimization device adjusts the parameters of the first objective optimization function of the preset deep learning network.

[0089] Furthermore, the frequency modulation optimization device predicts the frequency modulation features of the target sample through a preset deep learning network with adjusted parameters, and calculates the corresponding loss function value based on the loss function and the prediction result.

[0090] Furthermore, if the difference between the loss function value of the preset deep learning network after parameter adjustment and the second loss function value is less than the second preset difference threshold, then the model training is determined to be complete, and the trained first prediction model and second prediction model are obtained. The first prediction model and the second prediction model are then fused to obtain the frequency modulation strategy prediction model.

[0091] Furthermore, if the difference between the loss function value of the preset deep learning network after parameter adjustment and the second loss function value is greater than or equal to the second preset difference threshold, the process is repeated until the difference between the loss function value of the preset deep learning network after parameter adjustment and the second loss function value is less than the second preset difference threshold. Then, the first prediction model and the second prediction model are obtained, and the first prediction model and the second prediction model are fused to obtain the frequency modulation strategy prediction model.

[0092] The embodiments of this invention continuously optimize the deep learning network and the multilayer perceptron network to construct the optimal frequency regulation strategy prediction model. Therefore, the frequency regulation strategy prediction model can accurately predict the first and second power generation predictions, improving the accuracy and reliability of power generation prediction. This further enables accurate target power generation control of unit operation, effectively reducing power system frequency fluctuations and improving the reliability and stability of the power system.

[0093] Furthermore, the hydropower plant AGC frequency regulation strategy optimization device based on big data analysis provided by the present invention will be described below. The hydropower plant AGC frequency regulation strategy optimization device based on big data analysis described below corresponds to the hydropower plant AGC frequency regulation strategy optimization method based on big data analysis described above.

[0094] Optional, refer to Figure 2 , Figure 2 This is a schematic diagram of the structure of the hydropower plant AGC frequency regulation strategy optimization device based on big data analysis provided by the present invention. The hydropower plant AGC frequency regulation strategy optimization device based on big data analysis includes: The data acquisition module 210 is used to collect the current unit operating parameters of the target hydropower plant's generating units, the current power system operating parameters of the power system controlled by the target hydropower plant, and the current external environmental parameters of the target hydropower plant. Data analysis module 220 is used to perform big data analysis on current power system operating parameters and current external environment parameters, and extract key frequency regulation features related to AGC frequency regulation; The model prediction module 230 is used to input key frequency regulation features into the frequency regulation strategy prediction model to obtain the first and second power generation prediction outputs of the frequency regulation strategy prediction model. The power determination module 240 is used to determine the target power generation based on the first power generation prediction power and the second power generation prediction power. The frequency regulation module 250 is used to determine the target unit operating parameters based on the target power generation and the current unit operating parameters, and to control the units of the target hydropower plant to operate at the target unit operating parameters.

[0095] This invention, through big data analysis of power system operating parameters and external environmental parameters, can deeply and comprehensively uncover key frequency regulation features related to AGC (Automatic Generation Control) frequency regulation. These key features are then input into a first prediction model trained on a deep learning network and a second prediction model trained on a multilayer perceptron network, respectively, to obtain a first predicted power generation and a second predicted power generation. Using two different models for prediction allows for a comprehensive consideration of their respective advantages and characteristics, improving the accuracy and reliability of power generation prediction. Based on these two predicted power outputs, a target power generation is determined. Further adjustments are made to the current unit operating parameters based on the target power generation to obtain the target unit operating parameters, and the units are controlled to operate accordingly. This allows for dynamic adjustment of unit power generation based on the real-time operating status of the power system and changes in the external environment, effectively reducing power system frequency fluctuations and improving the reliability and stability of the power system.

[0096] Please see Figure 3 , Figure 3 An embodiment diagram of an electronic device provided in accordance with the present invention. For example... Figure 3 As shown, this embodiment of the invention provides an electronic device 300, including a memory 310, a processor 320, and a computer program 311 stored in the memory 310 and executable on the processor 320. When the processor 320 executes the computer program 311, it performs the following steps: Collect the current operating parameters of the generating units of the target hydropower plant, the current operating parameters of the power system controlled by the target hydropower plant, and the current external environmental parameters of the target hydropower plant; Big data analysis is conducted on current power system operating parameters and current external environmental parameters to extract key frequency regulation features related to AGC frequency regulation; The key features of frequency regulation are input into the pre-trained frequency regulation strategy prediction model to obtain the first and second power generation prediction outputs of the frequency regulation strategy prediction model. The target power generation is determined based on the first and second power generation forecasts. The target unit operating parameters are determined based on the target power generation and the current unit operating parameters, and the units of the target hydropower plant are controlled to operate at the target unit operating parameters.

[0097] Please see Figure 4 , Figure 4An embodiment diagram of a computer-readable storage medium provided in accordance with an embodiment of the present invention is shown. Figure 4 As shown, this embodiment provides a computer-readable storage medium 400 on which a computer program 311 is stored. When the computer program 311 is executed by a processor, it performs the following steps: Collect the current operating parameters of the generating units of the target hydropower plant, the current operating parameters of the power system controlled by the target hydropower plant, and the current external environmental parameters of the target hydropower plant; Big data analysis is conducted on current power system operating parameters and current external environmental parameters to extract key frequency regulation features related to AGC frequency regulation; The key features of frequency regulation are input into the pre-trained frequency regulation strategy prediction model to obtain the first and second power generation prediction outputs of the frequency regulation strategy prediction model. The target power generation is determined based on the first and second power generation forecasts. The target unit operating parameters are determined based on the target power generation and the current unit operating parameters, and the units of the target hydropower plant are controlled to operate at the target unit operating parameters.

[0098] On the other hand, the present invention also provides a computer program product, which includes a computer program that can be stored on a non-transitory computer-readable storage medium. When the computer program is executed by a processor, the computer is able to execute the hydropower plant AGC frequency regulation strategy optimization method based on big data analysis provided by the above methods. The method includes: Collect the current operating parameters of the generating units of the target hydropower plant, the current operating parameters of the power system controlled by the target hydropower plant, and the current external environmental parameters of the target hydropower plant; Big data analysis is conducted on current power system operating parameters and current external environmental parameters to extract key frequency regulation features related to AGC frequency regulation; The key features of frequency regulation are input into the pre-trained frequency regulation strategy prediction model to obtain the first and second power generation prediction outputs of the frequency regulation strategy prediction model. The target power generation is determined based on the first and second power generation forecasts. The target unit operating parameters are determined based on the target power generation and the current unit operating parameters, and the units of the target hydropower plant are controlled to operate at the target unit operating parameters.

[0099] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Those skilled in the art can understand and implement this without any creative effort.

[0100] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods described in the various embodiments or some parts of the embodiments.

[0101] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for optimizing AGC frequency regulation strategy in hydropower plants based on big data analysis, characterized in that, include: The current operating parameters of the generating units of the target hydropower plant, the current operating parameters of the power system controlled by the target hydropower plant, and the current external environmental parameters of the target hydropower plant are collected. Big data analysis is performed on the current power system operating parameters and the current external environment parameters to extract key frequency regulation features related to AGC frequency regulation; The frequency regulation key features are input into a pre-trained frequency regulation strategy prediction model to obtain the first and second power generation predictions output by the frequency regulation strategy prediction model. The target power generation is determined based on the first power generation forecast and the second power generation forecast. The target unit operating parameters are determined based on the target power generation and the current unit operating parameters, and the units of the target hydropower plant are controlled to operate at the target unit operating parameters. The frequency regulation strategy prediction model includes a first prediction model and a second prediction model. The first power generation prediction is based on the output of the first prediction model, and the second power generation prediction is based on the output of the second prediction model. The first prediction model is trained on a preset deep learning network based on the frequency regulation features of the target sample and its corresponding power generation label results. The second prediction model is trained on a preset multilayer perceptron network based on the frequency regulation features of the target sample and its corresponding power generation label results.

2. The method for optimizing the frequency regulation strategy of hydropower plants based on big data analysis according to claim 1, characterized in that, The process of determining the target unit operating parameters based on the target power generation and the current unit operating parameters includes: Determine the absolute value of the difference between the target power generation and the power generation of the current unit operating parameters; If the absolute value of the power generation difference is greater than the preset power difference, the current unit operating parameters are adjusted based on the absolute value of the power generation difference to obtain the target unit operating parameters; If the absolute value of the power generation difference is less than or equal to the preset power difference, then the target power generation or the current unit operating parameters are determined as the target unit operating parameters.

3. The method for optimizing the frequency regulation strategy of hydropower plants based on big data analysis according to claim 1, characterized in that, The process of obtaining the frequency modulation features of the target sample and its corresponding power generation label results includes the following steps: Obtain the frequency modulation features of multiple raw samples and their corresponding power generation labels; Each original sample frequency regulation feature is input into the data optimization model to obtain the power generation prediction result of each original sample frequency regulation feature output by the data optimization model; The loss function based on the data optimization model determines the value of the loss function for each original sample frequency regulation feature based on the power generation label result and power generation prediction result of each original sample frequency regulation feature. The loss function value of each original sample frequency modulation feature is traversed, and the original sample frequency modulation features whose loss function value is less than or equal to the first preset difference threshold are determined as target sample frequency modulation features, thus obtaining the target sample frequency modulation features and their corresponding power generation label results.

4. The method for optimizing the frequency regulation strategy of hydropower plants based on big data analysis according to claim 1, characterized in that, The training process of the frequency modulation strategy prediction model includes: The target sample frequency modulation feature is input into the preset deep learning network to obtain the first power prediction value output by the preset deep learning network; based on the first loss function in the preset deep learning network, the first loss function value is obtained according to the first power prediction value and the power generation label result corresponding to the target sample frequency modulation feature; The target sample frequency modulation feature is input into the preset multilayer perceptron network to obtain the second power prediction value output by the preset multilayer perceptron network; based on the second loss function in the preset multilayer perceptron network, the second loss function value is obtained according to the second power prediction value and the power generation label result corresponding to the target sample frequency modulation feature; The frequency modulation strategy prediction model is obtained by training the model based on the first loss function value, the second loss function value and the preset constraints; the preset constraints are: the loss function value is less than the preset loss threshold, and the difference between the loss function values ​​output by different models is less than the second preset difference threshold.

5. The method for optimizing the frequency regulation strategy of a hydropower plant based on big data analysis according to claim 4, characterized in that, The step of training the model based on the first loss function value, the second loss function value, and preset constraints to obtain the frequency modulation strategy prediction model includes: If the value of the first loss function is greater than or equal to the preset loss threshold, and the value of the second loss function is greater than or equal to the preset loss threshold, then the parameters of the first objective optimization function of the preset deep learning network are adjusted, and the parameters of the second objective optimization function of the preset multilayer perceptron network are adjusted. The frequency modulation features of the target sample are predicted based on a preset deep learning network with adjusted parameters, and the loss function value is determined based on the prediction result; and the frequency modulation features of the target sample are predicted based on a preset multilayer perceptron network with adjusted parameters, and the loss function value is determined based on the prediction result. The process is repeated until the loss function values ​​output by the preset deep learning network after parameter adjustment and the preset multilayer perceptron network after parameter adjustment satisfy the preset constraints, thus obtaining the first prediction model and the second prediction model; the first prediction model and the second prediction model are then fused to obtain the frequency modulation strategy prediction model.

6. The method for optimizing the frequency regulation strategy of hydropower plants based on big data analysis according to claim 4, characterized in that, The step of training the model based on the first loss function value, the second loss function value, and preset constraints to obtain the frequency modulation strategy prediction model includes: If the value of the first loss function is greater than or equal to the preset loss threshold, and the value of the second loss function is less than the preset loss threshold, then the parameters of the first objective optimization function of the preset deep learning network are adjusted. The frequency modulation features of the target sample are predicted based on the preset deep learning network with adjusted parameters, and the loss function value is determined according to the prediction result. The first prediction model and the second prediction model are obtained after the loss function value output by the preset deep learning network with adjusted parameters and the second loss function value satisfy the preset constraint conditions. The first prediction model and the second prediction model are then fused to obtain the frequency modulation strategy prediction model.

7. The method for optimizing the frequency regulation strategy of hydropower plants based on big data analysis according to claim 4, characterized in that, The step of training the model based on the first loss function value, the second loss function value, and preset constraints to obtain the frequency modulation strategy prediction model includes: If the first loss function value is less than the preset loss threshold, and the second loss function value is less than the preset loss threshold, then the loss function difference between the first loss function value and the second loss function value is determined. If the difference in the loss function is greater than or equal to the second preset difference threshold, and the first loss function value is greater than the second loss function value, then the parameters of the first objective optimization function of the preset deep learning network are adjusted. The frequency modulation features of the target sample are predicted based on the preset deep learning network with adjusted parameters, and the loss function value is determined according to the prediction result. The first prediction model and the second prediction model are obtained after the loss function value output by the preset deep learning network with adjusted parameters and the second loss function value satisfy the preset constraint conditions. The first prediction model and the second prediction model are then fused to obtain the frequency modulation strategy prediction model.

8. The method for optimizing the frequency regulation strategy of a hydropower plant based on big data analysis according to any one of claims 1 to 7, characterized in that, Determining the target power generation based on the first predicted power generation and the second predicted power generation includes: Based on the first power generation forecast and the second power generation forecast, a first weighting coefficient for the first power generation forecast and a second weighting coefficient for the second power generation forecast are determined respectively. The target power generation is obtained by weighting the first predicted power generation and its corresponding first weighting coefficient, and the second predicted power generation and its corresponding second weighting coefficient.

9. A device for optimizing AGC frequency regulation strategy in hydropower plants based on big data analysis, characterized in that, The method for optimizing AGC frequency regulation strategy in hydropower plants based on big data analysis, as described in any one of claims 1 to 8; The hydropower plant AGC frequency regulation strategy optimization device based on big data analysis includes: The data acquisition module is used to collect the current unit operating parameters of the target hydropower plant's generating units, the current power system operating parameters of the power system controlled by the target hydropower plant, and the current external environmental parameters of the target hydropower plant. The data analysis module is used to perform big data analysis on the current power system operating parameters and the current external environment parameters, and extract key frequency regulation features related to AGC frequency regulation. The model prediction module is used to input the key frequency regulation features into a pre-trained frequency regulation strategy prediction model to obtain the first and second power generation predictions output by the frequency regulation strategy prediction model. A power determination module is used to determine a target power generation based on the first power generation prediction power and the second power generation prediction power. The frequency regulation module is used to determine the target unit operating parameters based on the target power generation and the current unit operating parameters, and to control the units of the target hydropower plant to operate at the target unit operating parameters; The frequency regulation strategy prediction model includes a first prediction model and a second prediction model. The first power generation prediction is based on the output of the first prediction model, and the second power generation prediction is based on the output of the second prediction model. The first prediction model is trained on a preset deep learning network based on the frequency regulation features of the target sample and its corresponding power generation label results. The second prediction model is trained on a preset multilayer perceptron network based on the frequency regulation features of the target sample and its corresponding power generation label results.

10. A non-transitory computer-readable storage medium, wherein a computer software program is stored therein, characterized in that, When the computer software program is executed by the processor, it implements the hydropower plant AGC frequency regulation strategy optimization method based on big data analysis as described in any one of claims 1 to 8.

Images