Hydroelectric generating set power active control method, system and equipment based on frequency prediction and medium

Abstract

The invention discloses a hydroelectric generating set power active control method, system and device based on frequency prediction and a medium, and belongs to the technical field of power control, and the method comprises the steps: collecting basic operation data of a hydroelectric generating set adjusting system in real time, and carrying out the preprocessing of the basic operation data; time sequence characteristics of the power grid frequency data are extracted, and a neural network is adopted to generate a power grid frequency prediction sequence in a future time window; the power grid frequency prediction sequence is converted into a dynamic power feed-forward compensation amount, feed-forward compensation amount optimization calculation is carried out, the feed-forward compensation amount is superposed to the output end of a feedback controller, and a composite power instruction input value hydroelectric generating set is formed; and updating a neural network input sequence in each control period by adopting a rolling time domain prediction strategy, dynamically correcting a future prediction result, and realizing controlled quantity updating based on actual frequency deviation. According to the method, the compensation weight is dynamically updated by using the online adaptive algorithm, the prediction error and the system performance target are considered, and the compensation amount distribution is optimized.

Description

Technical Field

[0001] This invention relates to the field of power control technology, specifically to a method, system, device, and medium for active power control of hydropower units based on frequency prediction. Background Technology

[0002] With the large-scale integration of renewable energy sources such as wind and solar power into the power grid, power system frequency regulation faces higher requirements for dynamic response. Hydropower units, as key resources for grid frequency support and inertial response, play a crucial role in grid regulation due to their rapid start-up, shutdown, and regulation capabilities. In scenarios with a high proportion of renewable energy connected to the grid, the rapid power response of hydropower units not only helps to smooth out frequency disturbances but also improves system stability and safety margins, which is of great significance for ensuring the safe operation of the power grid.

[0003] Currently, power regulation of hydropower units mainly relies on the unit's hydraulic governor and proportional-integral controller, using load distribution strategies and primary droop characteristics to achieve feedback regulation of frequency deviation. The unit speed control system receives real-time frequency deviation and issues increase / decrease commands to achieve closed-loop power control. However, due to the inherent time delay of the unit control loop and water flow characteristics, traditional feedback control modes often exhibit response lag during frequency abrupt changes or rapid fluctuations, making it difficult to provide sufficient feedforward support in a timely manner. Furthermore, existing methods generally neglect the prediction and utilization of future grid frequency trends, remaining limited to feedback control based on real-time frequency measurements. This results in the unit's inability to compensate in advance during the initial stages of frequency disturbances, restricting its rapid response capability and support effectiveness. While some research has attempted to apply wind and solar power forecasting tools to power system dispatching, research on accurate prediction of grid frequency disturbances and feedforward compensation strategies for hydropower units remains lacking. Therefore, there is an urgent need for a composite control method that combines short-term grid frequency prediction with the dynamic characteristics of the unit to overcome the phase lag limitations of traditional control structures and achieve active power support for hydropower units during the frequency disturbance propagation phase. Summary of the Invention

[0004] In view of the above-mentioned problems, the present invention is proposed.

[0005] Therefore, the technical problem addressed by this invention is how to solve the problem that existing methods generally neglect the prediction and utilization of future grid frequency trends, remaining limited to feedback control based on real-time frequency measurements. This results in generating units being unable to compensate in advance during the initial stages of frequency disturbances, thus restricting their rapid response capabilities and support effectiveness. Furthermore, although existing research has attempted to apply wind and solar power forecasting tools to power system dispatching, research on accurate prediction of grid frequency disturbances and feedforward compensation strategies for hydropower units remains lacking.

[0006] To address the aforementioned technical problems, this invention provides the following technical solution: a method for active power control of hydropower units based on frequency prediction, comprising: real-time acquisition of basic operating data of the hydropower unit regulation system, including historical frequency sequences of the power grid, unit output status, and power grid load fluctuation information, and preprocessing the basic operating data; extracting the temporal characteristics of the power grid frequency data, and using a neural network to generate a power grid frequency prediction sequence within a future time window; converting the power grid frequency prediction sequence into a dynamic power feedforward compensation quantity, and performing optimized calculation of the feedforward compensation quantity in conjunction with a preset power-frequency mapping relationship and unit regulation constraints, and superimposing the feedforward compensation quantity onto the output of the feedback controller to form a composite power command input value hydropower unit; and employing a rolling time-domain prediction strategy to update the neural network input sequence in each control cycle, dynamically correcting future prediction results, and realizing control quantity updates based on actual frequency deviations.

[0007] As a preferred embodiment of the active power control method for hydropower units based on frequency prediction described in this invention, the real-time acquisition of basic operating data of the hydropower unit regulation system includes: acquiring historical frequency data sequences of the power grid to form an original frequency sequence; for outliers in the original frequency sequence, using a sliding median filtering method to eliminate instantaneous disturbances, and normalizing the filtered power grid frequency values; the normalized power grid frequency sequence satisfies a standard normal distribution.

[0008] As a preferred embodiment of the active power control method for hydropower units based on frequency prediction described in this invention, the step of extracting the time-series features of the power grid frequency data and generating a power grid frequency prediction sequence within a future time window using a neural network includes: defining the sliding window length and the prediction time domain based on historical observation step sizes; and generating training sample pairs by segmenting the sequence through the sliding window.

[0009] in, The first of the historical frequency window data Group input samples; The first step of the future frequency prediction target Group output labels; This represents the number of frequency sample indices. ; This represents the total number of samples at the sampling time. The length of the sliding window; It is the set of real numbers; for A matrix of order real numbers; For the normalized first The group's power grid frequency, For the normalized first The group's power grid frequency, For the normalized first The group's power grid frequency, For the normalized first The group's power grid frequency.

[0010] As a preferred embodiment of the frequency prediction-based active power control method for hydropower units described in this invention, the step of converting the grid frequency prediction sequence into a dynamic power feedforward compensation quantity includes: dividing the grid frequency sample into a dataset, and setting the feedforward... The group of samples is the training set, and then... The sample group serves as the validation set; a neural network is used for frequency prediction, and the output power grid frequency prediction sequence is represented as follows:

[0011] in, For power grid frequency prediction sequences, for Predicted power grid frequency at time 10:00. for Predicted power grid frequency at time 10:00. To predict the time domain, The sampling time is given; based on the frequency prediction output sequence, the prediction frequency deviation is defined as:

[0012] in, To predict frequency deviation, To predict frequency, For actual frequency; for prediction time domain frequency deviation sequence within The time-varying gain coefficient is expressed as:

[0013] in, For time-varying gain coefficients, This is the speed governor control coefficient. For nonlinear gain coefficients, For activation function, To predict the step size, For the first Predicted frequency deviation at time, This is the frequency threshold.

[0014] This preferred scheme converts the frequency prediction result into a timing deviation and combines it with a time-varying nonlinear gain adjustment mechanism, enabling the power compensation response to have dynamic adjustment capability. The designed gain coefficient, combined with the activation function, can enhance the control effect when the frequency deviation exceeds the threshold, while maintaining stability when the deviation is small. This effectively enhances the adaptive adjustment capability of the hydropower unit to frequency anomalies and provides a faster response in the early stages of disturbance.

[0015] As a preferred embodiment of the frequency prediction-based active power control method for hydropower units described in this invention, the step of converting the grid frequency prediction sequence into a dynamic power feedforward compensation quantity further includes generating the feedforward compensation quantity by introducing a time decay factor, and dynamically adjusting the compensation weights at different time points within the prediction time domain through exponential weighting, expressed as:

[0016] in, The decay time constant, The time decay factor, Let be the inertial time constant of the water flow in the water diversion system; then the dynamic feedforward compensation is expressed as:

[0017] in, This is the feedforward compensation amount. This is the current predicted time.

[0018] This preferred scheme introduces a time decay factor using an exponential weighting method to differentiate the confidence level of deviations at different time points within the prediction time domain; it can significantly reduce the impact of medium- and long-term prediction errors on power regulation stability and strengthen the control weight of short-term prediction results.

[0019] As a preferred embodiment of the frequency prediction-based active power control method for hydropower units described in this invention, the step of optimizing the feedforward compensation by combining a preset power-frequency mapping relationship and unit regulation constraints includes embedding the generated feedforward compensation into a PID control architecture, and the composite power regulation command of the unit after superimposing the feedforward compensation is expressed as follows:

[0020] in, This is a composite command for unit power regulation. This is the unit's original power regulation command. To control deviation, The integral coefficient of the turbine governor controller. The differential coefficients of the turbine governor controller; The input to the hydroelectric generator's guide vane servo actuator, together with the field frequency feedback, forms a secondary closed loop for dynamic output correction. Based on the turbine characteristic curve, the nonlinear relationship between the guide vane opening and the combined power regulation command of the unit is fitted, and an inverse hyperbolic tangent function is introduced to describe the static characteristics, expressed as:

[0021] in, The slope of the turbine guide vane opening-power regulation curve. This is the power normalization coefficient. This is the opening offset coefficient; The target guide vane opening; The inverse hyperbolic tangent function is represented; the unit's guide vane servo system is driven by pulse width modulation signals issued by the local execution unit, and the duty cycle-opening conversion formula is expressed as:

[0022] in, For PWM duty cycle, This is the minimum opening of the unit's guide vanes. This represents the maximum opening of the unit's guide vanes. This represents the minimum effective range of the duty cycle. The maximum effective range of the duty cycle is defined as the maximum value. After mapping and conversion, a pulse width modulation signal is output to drive the guide vane servo motor, realizing closed-loop execution from power command to mechanical action.

[0023] As a preferred embodiment of the frequency prediction-based active power control method for hydropower units described in this invention, the method employs a rolling time-domain prediction strategy to update the neural network input sequence in each control cycle, dynamically correcting future prediction results to achieve control quantity updates based on actual frequency deviations. This includes real-time correction of the prediction expression based on the dynamic equation of the power grid frequency, as follows:

[0024] in, Let be the system's inertial constant. This is the load damping coefficient. This is the gain coefficient of the turbine output power in relation to the operating head and guide vane opening. For the guide vane opening of the unit, For the unit's working head, This represents the frequency deviation.

[0025] This invention provides an active power control system for hydropower units based on frequency prediction.

[0026] To address the aforementioned technical problems, this invention provides the following technical solution: a frequency-predictive-based active power control system for hydropower units, comprising: a power grid frequency acquisition and preprocessing module, a multi-scale time-series prediction module, a dynamic compensation quantity generation module, and a composite control execution module; the power grid frequency acquisition and preprocessing module is used to acquire basic operating data of the hydropower unit regulation system in real time, including historical power grid frequency sequences, unit output status, and power grid load fluctuation information, and preprocesses the basic operating data; the multi-scale time-series prediction module is used to extract the time-series features of the power grid frequency data and generate a power grid frequency prediction sequence within a future time window using a neural network; the dynamic compensation quantity generation module is used to convert the power grid frequency prediction sequence into a dynamic power feedforward compensation quantity, perform feedforward compensation quantity optimization calculation based on a preset power-frequency mapping relationship and unit regulation constraints, and superimpose the feedforward compensation quantity onto the output of the feedback controller to form a composite power command input value hydropower unit; the composite control execution module is used to adopt a rolling time-domain prediction strategy to update the neural network input sequence in each control cycle, dynamically correct future prediction results, and realize control quantity updates based on actual frequency deviations.

[0027] The present invention provides a computer device, including a memory and a processor, wherein the memory stores a computer program, characterized in that the processor executes the computer program to implement the steps of the above-described active power control method for hydropower units based on frequency prediction.

[0028] The present invention provides a computer-readable storage medium having a computer program stored thereon, characterized in that, when the computer program is executed by a processor, it implements the steps of the above-described active power control method for hydropower units based on frequency prediction.

[0029] The beneficial effects of this invention are as follows: This invention employs a multi-scale LSTM structure to model frequency characteristics at different time scales, achieving joint prediction of macro trends and micro fluctuations. Simultaneously, it utilizes an online adaptive algorithm to dynamically update compensation weights, balancing prediction error with system performance objectives and optimizing compensation allocation. By predicting short-term fluctuations in grid frequency in advance, frequency deviation information can be obtained before the actual occurrence of disturbances. By dynamically adjusting the prediction compensation weights, a feedforward compensation amount is generated and superimposed on the closed-loop control loop, enabling hydropower units to implement power increases or decreases in advance during the frequency disturbance propagation stage. This mechanism overcomes the phase lag limitation of traditional PI control, shortening the unit power response time under a load step of 0.1 pu by 11.2 seconds, effectively improving the response speed of hydropower units to grid frequency disturbances. Attached Figure Description

[0030] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0031] Figure 1 The above is a flowchart of an active power control method for hydropower units based on frequency prediction, provided as an embodiment of the present invention.

[0032] Figure 2 A comparison chart of LSTM prediction results and analysis data for an active power control method for hydropower units based on frequency prediction, provided in an embodiment of the present invention.

[0033] Figure 3 A comparison of power command signals for hydropower units using different control methods under load step disturbances, provided by an embodiment of the present invention, for an active power control method for hydropower units based on frequency prediction. Figure 1 .

[0034] Figure 4 A comparison of power command signals for hydropower units using different control methods under load step disturbances, provided by an embodiment of the present invention, for an active power control method for hydropower units based on frequency prediction. Figure 2 .

[0035] Figure 5 This is a comparison diagram of guide vane opening using different control methods in an active power control method for hydropower units based on frequency prediction, provided as an embodiment of the present invention.

[0036] Figure 6 This is a comparison chart of the output power of hydropower units using different control methods under load step disturbances, provided by an embodiment of the present invention, which is a method for active power control of hydropower units based on frequency prediction. Detailed Implementation

[0037] To make the above-mentioned objects, features, and advantages of the present invention more apparent and understandable, specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of them. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the protection scope of the present invention.

[0038] Example 1, referring to Figure 1 This is one embodiment of the present invention, which provides a method for active power control of hydropower units based on frequency prediction, comprising: S1. Real-time acquisition of basic operating data of the hydropower unit regulation system, including historical frequency sequences of the power grid, unit output status and power grid load fluctuation information, and preprocessing of the basic operating data.

[0039] S2. Extract the temporal features of the power grid frequency data and use a neural network to generate a power grid frequency prediction sequence within the future time window.

[0040] S3. Convert the grid frequency prediction sequence into a dynamic power feedforward compensation quantity. Combine the preset power-frequency mapping relationship and unit regulation constraints to optimize the feedforward compensation quantity calculation. Then, superimpose the feedforward compensation quantity onto the output of the feedback controller to form a composite power command input value for the hydropower unit.

[0041] S4. A rolling time-domain prediction strategy is adopted to update the neural network input sequence in each control cycle, dynamically correct future prediction results, and realize the update of control quantity based on actual frequency deviation.

[0042] Furthermore, in step S1, historical frequency data sequences of the power grid are collected, with a sampling period of T. s This forms the original frequency sequence:

[0043] in, This is the original frequency time series. For the first The measured frequency of the power grid at time 1 is the 1st The measured frequency of the power grid for each sampled sample, i.e. The measured frequency of the power grid at the first moment. The measured frequency of the power grid at the second moment. For the first The measured frequency of the power grid at each sampling time. The total number of samples at the sampling time. It is the set of real numbers.

[0044] For outliers in historical frequency sequences, a moving median filter is used to eliminate transient disturbances.

[0045] in, It is a median function. The filter window half-width, This is the filtered frequency value. For the first The measured frequency of the power grid at a given moment. For the first The measured frequency of the power grid at a given moment.

[0046] To avoid the impact of numerical differences on model training, the filtered power grid frequency values ​​are normalized:

[0047] in, This represents the mean of the filtered power grid frequency sequence; The standard deviation of the filtered power grid frequency sequence; This represents the normalized frequency value.

[0048] Normalized power grid frequency sequence It follows a standard normal distribution.

[0049] Furthermore, in step S2, the power grid frequency samples are constructed based on a sliding window. The normalized power grid frequency sequence is then... The dataset is converted into a supervised learning sample set to capture temporal dynamics. First, the sliding window length is defined based on the historical observation step size. With prediction time domain Considering the characteristics of hydropower units participating in power grid frequency regulation, real-time communication, and online calculation time, then ≤10.

[0050] Sequence segmentation using a sliding window Generate training sample pairs :

[0051] in, The first of the historical frequency window data Group input samples; The first step of the future frequency prediction target Group output labels; This represents the number of frequency sample indices. ; This represents the total number of samples at the sampling time. The length of the sliding window; It is the set of real numbers; for A matrix of order real numbers; For the normalized first The group's power grid frequency, For the normalized first The group's power grid frequency, For the normalized first The group's power grid frequency, For the normalized first The group's power grid frequency.

[0052] Furthermore, in step S3, the power grid frequency samples are divided into datasets, letting the previous... The group of samples is the training set, and then... The training and validation sets are configured using a stratified random partitioning strategy to ensure that the training and validation sets cover different frequency fluctuation patterns.

[0053] To achieve accurate ultra-short-term frequency prediction for the power grid, a deep network model with long-term memory capability is established as follows. First, the basic network topology is designed, determining the variable sequence dimension of the input layer, the gating mechanism of the hidden layers, and the data type of the output layer. The gating mechanism formula for the Long Short-Term Memory (LSTM) neural network is as follows:

[0054] in, for Input features at any time, i.e. ; for Always hide your status; for Unit state at any given time; The input weight matrix, The hidden state term weight matrix; For bias terms, For the dimension of the hidden layer, The dimension of the input feature. Let be the candidate cell state vector. Represents the Hadamard product. For the Sigmoid function:

[0055] For hyperbolic tangent function:

[0056] The output layer uses a certain dropout probability to randomly disable some neurons to prevent overfitting, resulting in the following output layer mapping formula:

[0057] in, To output the weight matrix; For output bias; This refers to the output layer unit state.

[0058] Configure and optimize network parameters to improve the accuracy and adaptability of frequency sequence prediction. The optimizer parameter update law is as follows:

[0059] in, For the first Network parameters for the next iteration; The learning rate; This represents the first-order moment estimate after bias correction. It is the first moment of the parameter gradient, used to estimate the gradient mean, accelerate convergence and reduce oscillations; This represents the second-order moment estimate after bias correction. , , representing the estimated decay rates of the first and second moments, respectively. The second moment of the parameter gradient is used to estimate the gradient variance and adaptively adjust the learning rate.

[0060] The training loss function is defined as follows:

[0061] in, The sample size for each training step; The model represents the input frequency sequence. The predicted output, For the input frequency sequence The actual output.

[0062] The power grid frequency prediction sequence output in this step This will serve as the reference input for power feedforward compensation, providing the data foundation for active control in subsequent control stages. For power grid frequency prediction sequences, for Predicted power grid frequency at time 10:00. for Predicted power grid frequency at time 10:00. To predict the time domain, The sampling point is the sampling time.

[0063] Based on the frequency prediction output sequence, the prediction frequency bias is defined as follows:

[0064] in, To predict frequency deviation, To predict frequency, This refers to the actual frequency.

[0065] Combining the predicted frequency deviation with the dynamic characteristics of the hydropower unit, feedforward compensation calculations for the hydropower unit's power response are performed. The hydropower unit's regulating system is affected by the water hammer effect of the water diversion system, mechanical and hydraulic actions, and the characteristics of flow torque inertia. Its power response and frequency disturbance process exhibit nonlinear phase lag characteristics. The dynamic equation reflecting the turbine's power output characteristics is used, and the predicted frequency deviation needs to be substituted into this equation. The specific formula is as follows:

[0066] in, This refers to the speed governor control coefficient; For frequency measurement delay time; The hydraulic response time constant of the speed governor; The inertial time constant of the water flow in the water diversion system; For the Laplace operator, For frequency deviation, This refers to the unit's power response.

[0067] For prediction time domain frequency deviation sequence within The time-varying gain coefficient is designed as follows:

[0068] in, For time-varying gain coefficients, This is the speed governor control coefficient. For nonlinear gain coefficients, For activation function, To predict the step size, For the first Predicted frequency deviation at time, This is a frequency threshold; when the prediction deviation exceeds the threshold, the coefficient... The increased nonlinearity enhances the compensation strength against large disturbances.

[0069] The generated feedforward compensation is introduced into the time decay factor, and the compensation weights at different time points in the prediction time domain are dynamically adjusted through exponential weighting to reduce the impact of long-term prediction uncertainty on control stability and prioritize responses to high-confidence short-term prediction biases.

[0070] in, The decay time constant, The time decay factor, The inertial time constant of the water flow in the water diversion system.

[0071] The dynamic feedforward compensation amount is then expressed as:

[0072] in, This is the feedforward compensation amount. This is the current predicted time.

[0073] The feedforward compensation amount generated in the previous step By embedding a traditional proportional-integral-derivative (PID) control architecture, a predictive-feedback collaborative mechanism is formed to overcome the response speed limitations of single-loop regulation. The composite power regulation command of the unit after superimposing feedforward compensation can be expressed as:

[0074] in, This is a composite command for unit power regulation. This is the unit's original power regulation command. To control deviation, The integral coefficient of the turbine governor controller. For the differential coefficient of the turbine governor controller.

[0075] feedforward term It provides an advanced regulation function to compensate for the inherent mechanical delay of the turbine regulation system.

[0076] Will The input hydropower unit guide vane servo actuator, together with the field frequency feedback, forms a secondary closed loop, dynamically correcting the output and enabling the unit power to quickly track the prediction deviation.

[0077] Based on the turbine characteristic curve, the nonlinear relationship between guide vane opening and composite power command is fitted, and an inverse hyperbolic tangent function is introduced to describe its static characteristics:

[0078] in, The slope of the turbine guide vane opening-power regulation curve. This is the power normalization coefficient. This is the opening offset coefficient. , , All are fitting coefficients for the characteristic curves of the water turbine; The target guide vane opening; This represents the inverse hyperbolic tangent function, used to fit nonlinear saturation characteristics.

[0079] The unit's guide vane servo system is driven by PWM signals issued by the local execution unit, and its duty cycle-opening conversion formula is as follows:

[0080] in, For PWM duty cycle, This is the minimum opening of the unit's guide vanes. This represents the maximum opening of the unit's guide vanes. This represents the minimum effective range of the duty cycle. This represents the maximum effective range of the duty cycle.

[0081] After completing the above mapping and conversion, the field execution unit outputs a pulse width modulation (PWM) signal to drive the guide vane servo motor, realizing closed-loop execution from power command to mechanical action.

[0082] The constraints of the unit's actuator mechanism are as follows:

[0083] To prevent sudden pressure changes in the water supply pipe.

[0084] in, The guide vane operation status under the unit power regulation command, corresponding to The guide vane's operational status.

[0085] Furthermore, in step S4, an execution-feedback closed-loop dynamic control framework is constructed to achieve coordinated adaptive regulation of the power grid frequency and the hydropower unit power. First, a power grid frequency window sliding mechanism is established: within each control cycle, the first segment of historical sequence data is deleted, and the latest measured data value is inserted. Generate the updated input sequence:

[0086] in, This is the updated input sequence.

[0087] A dynamic truncation strategy is employed. When a sudden change in system frequency is detected, the historical sequence is cleared, and the window is reconstructed centered on the current point to avoid outdated data interfering with predictions. Based on the updated... Generate the future The 10-step prediction sequence can be represented as: .

[0088] Real-time correction and prediction based on power grid frequency dynamic equations:

[0089] in, Let be the system's inertial constant. This is the load damping coefficient. This is the gain coefficient of the turbine output power in relation to the operating head and guide vane opening. The guide vane opening of the unit is used to dynamically adjust power regulation in real time according to the unit's status. This refers to the unit's operating head, measured in meters, and is generally considered a constant in dynamic equations. This represents the frequency deviation.

[0090] Example 2, refer to Figures 2-6This invention provides an active power control method for hydropower units based on frequency prediction, and scientific demonstration is carried out through experiments to verify the beneficial effects of the invention.

[0091] Step 1: Accurate extraction and prediction of power grid frequency dynamic characteristics are achieved by constructing an LSTM time series prediction model with a sliding window mechanism. Specific data preparation is as follows: In the actual system, a synchronous phasor measurement unit (PMU) is used to collect power grid frequency data in real time, with a sampling period set to ≤100ms to ensure the capture of high-frequency dynamic characteristics. To address noise interference, a sliding median filtering algorithm is used with a window width of 6 sampling points to effectively eliminate instantaneous impulse noise. Missing data is filled in within a finite time window using linear interpolation to avoid the impact of long-term data gaps on model training.

[0092] Typical power grid frequency disturbance time series data were selected as an example to verify the effectiveness of the prediction method. Data standardization was performed using the Z-score method, which normalized the given frequency data using the global mean and standard deviation to ensure consistent data distribution in the input model. In the sliding window mechanism, the input window length was set to 20 sampling points, and the output step size was 5 steps (predicting frequency changes in the next 500ms), which fully balanced the ability to capture short-term fluctuation characteristics and long-term trends.

[0093] The LSTM prediction model is designed with a single-input, single-output structure, with 500 hidden layers. Extensive experiments have verified that this size achieves the optimal balance between prediction accuracy and computational efficiency. During training, the Adam optimizer is used with an initial learning rate of 0.1, coupled with an early stopping strategy to prevent overfitting. Typical power fluctuation data from a renewable energy power plant, totaling 800 samples, are selected. The first 600 samples are used for training, and the validation set retains 200 samples to ensure the model's generalization ability. Comparison of LSTM prediction results with analyzed data is shown below. Figure 2 As shown in the figure, the overall prediction results of LSTM show a high degree of consistency with the actual analysis data trends.

[0094] Step 2: Dynamic Adjustment Coefficient Design and Power Feedforward Compensation Mapping. Specific data preparation is as follows: Based on the predicted frequency deviation, and considering the water hammer effect and mechanical inertia characteristics of the hydropower unit, a dynamic adjustment coefficient design and feedforward compensation mechanism are implemented. Taking a certain power station as an example, its water hammer time constant in the water intake pipe... =2.2s, governor time constant =0.3, based on this characteristic, the dynamic adjustment coefficient Designed as a time-varying gain, its basic adjustment gain is set as the original load frequency adjustment gain of the unit, corresponding to the amount of power that needs to be compensated for per Hz deviation when the unit is at full load.

[0095] After detecting power grid frequency fluctuations, The system then adaptively adjusts based on the cumulative magnitude of the prediction deviation, with the sensitivity coefficient set to 0.3 and the maximum frequency deviation threshold. Set the Hz to 0.02. When the prediction deviation exceeds the threshold, A nonlinear mapping is performed to enhance the compensation strength for large disturbances. Simultaneously, a time decay factor is introduced into the calculation of the feedforward compensation, using a decay time constant. =1.2s, with higher weight for recent forecasts and exponentially decreasing weight for long-term forecasts, effectively suppressing the interference of forecast uncertainty on control commands.

[0096] Step 3: Composite Power Command Generation and Closed-Loop Control. Specific data preparation is as follows: Based on the traditional PID controller, the feedforward compensation is weighted and superimposed to generate a composite power command. After determining the original PID control coefficients of the unit, the guide vane opening control adopts PWM modulation technology, and the carrier frequency is dynamically adjusted according to the frequency deviation. A comparison of guide vane opening PWM between conventional PID control and the frequency prediction-based active power control method is provided. Figure 3 and Figure 4 As shown, where Figure 3 This is the power command signal for a 0.1 load step disturbance under the original control structure. Figure 4 The figure shows the power command signal used in the active control method based on grid prediction proposed in this invention. As can be seen from the figure, the active power control method increases the duration of the power command signal, thereby enabling rapid tracking of the unit's power setpoint.

[0097] Step 4: Construct an execution-feedback closed-loop dynamic control framework to achieve coordinated adaptive regulation of grid frequency and hydropower unit power. Specific data preparation is as follows: Within each 100ms control cycle, the latest measured grid frequency is injected into the LSTM input sequence, replacing the oldest data, maintaining a window length of 60 points. The prediction deviation at the current moment is exponentially decayed to correct the prediction value for the next 10 steps. Furthermore, by combining the grid frequency dynamic equation with the power response characteristics of the turbine regulation system, dynamic compensation of the hydropower unit power regulation based on the actual frequency deviation is achieved. The output power response of the hydropower unit using the active power response control method is compared with the opening degree. Figure 5 and Figure 6 As shown in the figure, the active power control method can adjust the guide vane opening more quickly after a frequency disturbance, improving power response speed and shortening the duration of the transient process. Compared to the original control structure, the guide vane opening changes more rapidly under active control, and the power output decrease is smoother, effectively improving the dynamic performance of frequency regulation and system stability. Especially... Figure 6 In this study, the active control scheme reduced the response time by approximately 11.2 seconds, significantly improving the speed and accuracy of the control system.

[0098] Example 4 is an embodiment of the present invention. This embodiment provides an active power control system for hydropower units based on frequency prediction, including a power grid frequency acquisition and preprocessing module, a multi-scale time series prediction module, a dynamic compensation quantity generation module, and a composite control execution module.

[0099] The power grid frequency acquisition and preprocessing module is used to acquire basic operating data of the hydropower unit regulation system in real time, including historical frequency sequences of the power grid, unit output status and power grid load fluctuation information, and to preprocess the basic operating data.

[0100] The multi-scale time series prediction module is used to extract the time series features of power grid frequency data and uses a neural network to generate a power grid frequency prediction sequence within the future time window.

[0101] The dynamic compensation quantity generation module is used to convert the power grid frequency prediction sequence into a dynamic power feedforward compensation quantity. It combines the preset power-frequency mapping relationship and unit regulation constraints to perform feedforward compensation quantity optimization calculation, and then superimposes the feedforward compensation quantity onto the output of the feedback controller to form a composite power command input value for the hydropower unit.

[0102] The composite control execution module is used to adopt a rolling time-domain prediction strategy to update the neural network input sequence in each control cycle, dynamically correct future prediction results, and realize the control quantity update based on the actual frequency deviation.

[0103] This embodiment also provides an electronic device applicable to a frequency prediction-based active power control method for hydropower units, comprising: a memory and a processor; the memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions to implement the frequency prediction-based active power control method for hydropower units as proposed in the above embodiment.

[0104] This embodiment also provides a storage medium storing a computer program that, when executed by a processor, implements a frequency prediction-based active power control method for hydropower units as proposed in the above embodiment.

[0105] The storage medium proposed in this embodiment belongs to the same inventive concept as the method for active power control of hydropower units based on frequency prediction proposed in the above embodiments. Technical details not described in detail in this embodiment can be found in the above embodiments, and this embodiment has the same beneficial effects as the above embodiments.

[0106] Based on the above description of the implementation methods, those skilled in the art can clearly understand that the present invention can be implemented using software and necessary general-purpose hardware, and of course, it can also be implemented using hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of the present invention, 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 a computer floppy disk, read-only memory (ROM), random access memory (RAM), flash memory, hard disk, or optical disk, etc., including several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods of the various embodiments of the present invention.

[0107] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A method for active power control of hydropower units based on frequency prediction, characterized in that: include, Real-time acquisition of basic operating data of hydropower unit regulation system, including historical frequency sequence of power grid, unit output status and power grid load fluctuation information, and preprocessing of basic operating data; Extract the temporal features of power grid frequency data and use a neural network to generate a power grid frequency prediction sequence within a future time window; The power grid frequency prediction sequence is converted into a dynamic power feedforward compensation quantity. The feedforward compensation quantity is optimized by combining the preset power-frequency mapping relationship and the unit regulation constraints. The feedforward compensation quantity is then superimposed on the output of the feedback controller to form a composite power command input value for the hydropower unit. A rolling time-domain prediction strategy is adopted to update the neural network input sequence in each control cycle, dynamically correct future prediction results, and realize the update of control quantity based on actual frequency deviation.

2. The active power control method for hydropower units based on frequency prediction as described in claim 1, characterized in that: The real-time acquisition of basic operating data of the hydropower unit regulation system includes, Collect historical frequency data sequences of the power grid to form the original frequency sequence; For outliers in the original frequency sequence, the moving median filtering method is used to eliminate instantaneous disturbances, and the filtered power grid frequency values ​​are normalized. The normalized power grid frequency sequence follows a standard normal distribution.

3. The active power control method for hydropower units based on frequency prediction as described in claim 2, characterized in that: The extraction of time-series features from power grid frequency data, followed by the generation of a future time window power grid frequency prediction sequence using a neural network, includes... Define the sliding window length and the prediction time domain based on the historical observation step size; The sequence is segmented using a sliding window, and training sample pairs are generated as follows: in, The first of the historical frequency window data Group input samples; The first step of the future frequency prediction target Group output labels; This represents the number of frequency sample indices. ; This represents the total number of samples at the sampling time. The length of the sliding window; It is the set of real numbers; for A matrix of order real numbers; For the normalized first The group's power grid frequency, For the normalized first The group's power grid frequency, For the normalized first The group's power grid frequency, For the normalized first The group's power grid frequency.

4. The active power control method for hydropower units based on frequency prediction as described in claim 3, characterized in that: The process of converting the grid frequency prediction sequence into a dynamic power feedforward compensation quantity includes, Divide the power grid frequency samples into datasets, and let the first... The group of samples is the training set, and then... The group of samples serves as the validation set; Frequency prediction is performed using a neural network, and the output power grid frequency prediction sequence is represented as follows: in, For power grid frequency prediction sequences, for Predicted power grid frequency at time 10:

00. for Predicted power grid frequency at time 10:

00. To predict the time domain, Sampling time; Based on the frequency prediction output sequence, the prediction frequency deviation is defined as follows: in, To predict frequency deviation, To predict frequency, This is the actual frequency; For prediction time domain frequency deviation sequence within The design time-varying gain coefficient is expressed as follows: in, For time-varying gain coefficients, This is the speed governor control coefficient. For nonlinear gain coefficients, For activation function, To predict the step size, For the first Predicted frequency deviation at time, This is the frequency threshold.

5. The active power control method for hydropower units based on frequency prediction as described in claim 4, characterized in that: The process of converting the grid frequency prediction sequence into a dynamic power feedforward compensation quantity also includes, The feedforward compensation is generated by introducing a time decay factor. The compensation weights at different time points in the prediction time domain are dynamically adjusted through exponential weighting, as shown below. in, The decay time constant, The time decay factor, The inertial time constant of the water flow in the water diversion system; The dynamic feedforward compensation amount is then expressed as: in, This is the feedforward compensation amount. This is the current predicted time.

6. The active power control method for hydropower units based on frequency prediction as described in claim 5, characterized in that: The step of optimizing the feedforward compensation amount by combining the preset power-frequency mapping relationship and unit regulation constraints includes, The generated feedforward compensation is embedded into the PID control architecture, and the composite power regulation command of the unit after superimposing the feedforward compensation is expressed as follows: in, This is a composite command for unit power regulation. This is the unit's original power regulation command. To control deviation, The integral coefficient of the turbine governor controller. The differential coefficient of the turbine governor controller; Will The input hydro-generator guide vane servo actuator, together with the field frequency feedback, forms a secondary closed loop to dynamically correct the output. Based on the turbine characteristic curves, the nonlinear relationship between the guide vane opening and the combined power regulation command of the unit is fitted. An inverse hyperbolic tangent function is introduced to describe the static characteristics, expressed as: in, The slope of the turbine guide vane opening-power regulation curve. This is the power normalization coefficient. This is the opening offset coefficient; The target guide vane opening; Represents the inverse hyperbolic tangent function; The turbine guide vane servo system is driven by pulse width modulation signals issued by the local execution unit. The duty cycle-opening conversion formula is expressed as follows: in, For PWM duty cycle, This is the minimum opening of the unit's guide vanes. This represents the maximum opening of the unit's guide vanes. This represents the minimum effective range of the duty cycle. This represents the maximum effective range of the duty cycle. After mapping and conversion are completed, a pulse width modulation signal is output to drive the guide vane servo motor, realizing closed-loop execution from power command to mechanical action.

7. The active power control method for hydropower units based on frequency prediction as described in claim 6, characterized in that: The aforementioned rolling time-domain prediction strategy updates the neural network input sequence in each control cycle, dynamically correcting future prediction results to achieve control quantity updates based on actual frequency deviations. The real-time correction prediction based on the power grid frequency dynamic equation is expressed as follows: in, Let be the system's inertial constant. This is the load damping coefficient. This is the gain coefficient between the turbine output power and the operating head and guide vane opening. For the turbine guide vane opening, For the unit's working head, This represents the frequency deviation.

8. A frequency-predictive-based active power control system for hydropower units, employing the frequency-predictive-based active power control method for hydropower units as described in any one of claims 1 to 7, characterized in that, include: The module includes a power grid frequency acquisition and preprocessing module, a multi-scale time series prediction module, a dynamic compensation quantity generation module, and a composite control execution module. The power grid frequency acquisition and preprocessing module is used to acquire basic operating data of the hydropower unit regulation system in real time, including historical frequency sequences of the power grid, unit output status and power grid load fluctuation information, and to preprocess the basic operating data. The multi-scale time series prediction module is used to extract the time series features of power grid frequency data and use a neural network to generate a power grid frequency prediction sequence within a future time window. The dynamic compensation quantity generation module is used to convert the power grid frequency prediction sequence into a dynamic power feedforward compensation quantity, and to perform feedforward compensation quantity optimization calculation in combination with the preset power-frequency mapping relationship and unit regulation constraints. The feedforward compensation quantity is then superimposed on the output of the feedback controller to form a composite power command input value hydropower unit. The composite control execution module is used to employ a rolling time-domain prediction strategy to update the neural network input sequence in each control cycle, dynamically correct future prediction results, and realize control quantity updates based on actual frequency deviations.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the active power control method for hydropower units based on frequency prediction as described in any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the steps of the active power control method for hydropower units based on frequency prediction as described in any one of claims 1 to 7.

Images