A primary frequency modulation optimization control method and system based on adaptive gain adjustment

Abstract

This invention discloses a primary frequency regulation optimization control method and system based on adaptive gain adjustment, comprising: acquiring and preprocessing historical data of power grid operating state parameters and power grid frequency to construct a power grid frequency stability classification model; acquiring and normalizing power grid operating parameters in real time, and obtaining power grid operating state classification results based on the normalized data; making a preliminary judgment on stability based on the power grid operating parameters and the classification model; if unstable, calculating the power grid frequency stability index and determining whether it is below a threshold; constructing an adaptive gain adjustment model; if the power grid operating state classification results do not meet the requirements or are below the threshold, obtaining the optimal gain parameter for primary frequency regulation control based on the power grid frequency stability index and the adaptive gain adjustment model, and implementing optimized control based on the optimal gain parameter. This invention achieves precise adaptive control of the power grid frequency, providing a strong guarantee for the safe and stable operation of the power system.

Description

Technical Field

[0001] This invention belongs to the field of information technology, and in particular relates to a primary frequency modulation optimization control method and system based on adaptive gain adjustment. Background Technology

[0002] Primary frequency regulation control in power systems is a key technology for maintaining grid frequency stability. However, in actual operation, the grid's operating state is dynamic, and the factors affecting grid stability are complex. Traditional primary frequency regulation control uses fixed gain parameters, which are difficult to adapt to real-time changes in the grid's operating state, resulting in poor control performance and potentially causing drastic fluctuations in grid frequency. How to dynamically adjust the gain parameters of primary frequency regulation control based on the real-time operating state of the grid to achieve precise control of the grid frequency is a pressing technical challenge.

[0003] To address this issue, it is necessary to deeply analyze the inherent mechanisms of power grid operation and identify key variables affecting grid stability. Through extensive data collection and analysis, a power grid stability assessment model is established to calculate the grid stability index in real time. Based on this, an adaptive gain adjustment model is constructed, using the grid stability index as the optimization objective and the key variables as inputs, to optimize the gain parameters through intelligent algorithms. The optimized gain parameters are promptly fed back to the primary frequency regulation control system for precise adjustment of generator units. During grid operation, continuous signal acquisition and dynamic updates to the adaptive gain adjustment model are also required to adapt to changes in grid operating conditions. Solving this technical challenge requires a strong foundation in power system theory, extensive data analysis experience, and advanced intelligent optimization algorithms. Summary of the Invention

[0004] To address the aforementioned technical problems, this invention proposes a primary frequency modulation optimization control method and system based on adaptive gain adjustment, thereby resolving the issues present in the prior art.

[0005] To achieve the above objectives, the present invention provides a primary frequency modulation optimization control method based on adaptive gain adjustment, comprising:

[0006] Historical data on power grid operating status parameters and power grid frequency are acquired and preprocessed. A power grid frequency stability classification model is then constructed based on the preprocessed historical data.

[0007] Real-time acquisition and standardization of power grid operation parameters yields a standardized power grid operation parameter dataset; based on this dataset, power grid operation status classification results are obtained.

[0008] Based on the power grid operating parameters and the power grid frequency stability classification model, the stability is initially judged. If it is unstable, the power grid frequency stability index is calculated through the pre-trained stability index calculation model and it is determined whether it is below the threshold.

[0009] An adaptive gain adjustment model is constructed. If the power grid operation state classification result does not meet the requirements or the power grid frequency stability index is lower than the threshold, the power grid frequency stability index is used as the input of the adaptive gain adjustment model to obtain the optimal gain parameter of the primary frequency regulation control. Based on the optimal gain parameter of the primary frequency regulation control, optimized control is achieved.

[0010] Optionally, the process of constructing a power grid frequency stability classification model includes:

[0011] Historical data of power grid operating status parameters are cleaned and normalized to obtain standardized power grid operating status parameter data. A mapping relationship between power grid operating status parameters and power grid frequency is constructed. A classification model is built based on the support vector machine algorithm. The classification model is trained based on the standardized power grid operating status parameter data, historical power grid frequency data and the mapping relationship to obtain a power grid frequency stability classification model.

[0012] Optionally, the real-time acquired power grid operating parameters are denoised and standardized to obtain a standardized power grid operating parameter dataset; features are extracted from the standardized power grid operating parameter dataset to obtain a power grid operating state feature vector; and a power grid operating state classification result is obtained based on the power grid operating state feature vector.

[0013] Optionally, the process of calculating the power grid frequency stability index includes:

[0014] Sensors and measuring devices are deployed at key nodes of the power grid to collect power grid operating parameters in real time; the collected power grid operating parameter data is preprocessed; a mathematical model describing the stability of the power grid is constructed, the preprocessed power grid operating parameters are input into the mathematical model, and the power grid frequency stability index is calculated by solving the model equations.

[0015] Optionally, a stability index threshold can be set based on expert experience to determine whether the power grid frequency stability index exceeds the stability index threshold. If it is lower than the stability index threshold, an early warning can be issued. The optimal gain parameters for primary frequency regulation control can be obtained based on the power grid frequency stability index and the adaptive gain adjustment model.

[0016] Optionally, the process of obtaining the optimal gain parameters for primary frequency modulation control also includes:

[0017] Data on the operation status of the power grid is acquired, and multiple power grid frequency stability indices are calculated as input variables for the fuzzy controller. Based on the numerical range of the power grid frequency stability indices, a fuzzy control rule base is established, forming multiple fuzzy control subsystems. For each fuzzy control subsystem, fuzzy inference calculations are performed to obtain the fuzzy gain parameters for primary frequency regulation control. The fuzzy gain parameters of multiple fuzzy control subsystems are weighted and fused, and the optimal gain parameters for primary frequency regulation control are obtained through defuzzification processing.

[0018] Optionally, before inputting the power grid frequency stability index into the adaptive gain adjustment model, the safety margin of the power grid is calculated. Based on the calculation results of the power grid frequency stability index and the safety margin, and combined with the optimization control strategy in the expert knowledge base, a real-time power grid optimization control strategy is obtained.

[0019] Optionally, the process of achieving optimized control based on the optimal gain parameters of primary frequency regulation control includes: integrating the optimal gain parameters of primary frequency regulation control with grid frequency deviation and active power, and generating primary frequency regulation control commands in conjunction with real-time grid optimization control strategies; determining whether the primary frequency regulation control commands meet preset rationality conditions; if they do, sending the commands to the target generator set through the communication network; if they do not meet the conditions, recalculating the optimal gain parameters of primary frequency regulation control.

[0020] Optionally, power grid operating parameters and generator set status information are acquired as input data for the adaptive gain adjustment model, and the model is trained and optimized in real time using signal dynamic update technology. Specifically, the acquired power grid operating parameters and generator set status information are subjected to noise removal and feature extraction using data preprocessing technology to obtain high-quality input data. The preprocessed data is then input into the adaptive gain adjustment model, and the model is continuously optimized using machine learning algorithms.

[0021] This invention provides a primary frequency modulation optimization control system based on adaptive gain adjustment, comprising:

[0022] The power grid frequency stability mathematical modeling module is used to acquire and preprocess historical data of power grid operating status parameters and power grid frequency, and to build a power grid frequency stability classification model based on the preprocessed historical data.

[0023] The online monitoring module for power grid operating parameters is used to acquire and standardize power grid operating parameters in real time, thereby obtaining a standardized dataset of power grid operating parameters; and to obtain power grid operating status classification results based on the standardized dataset of power grid operating parameters.

[0024] The power grid stability judgment module makes a preliminary judgment on stability based on the power grid operating parameters and the power grid frequency stability classification model. If it is unstable, it calculates the power grid frequency stability index through a pre-trained stability index calculation model and determines whether it is below the threshold.

[0025] An adaptive gain adjustment module is used to construct an adaptive gain adjustment model. If the power grid operation status classification result does not meet the requirements or the power grid frequency stability index is lower than the threshold, the power grid frequency stability index is used as the input of the adaptive gain adjustment model to obtain the optimal gain parameter for primary frequency regulation control.

[0026] The primary frequency regulation control command generation module is used to integrate the optimal gain parameters of primary frequency regulation control with the grid frequency deviation and active power, and generate primary frequency regulation control commands in combination with real-time grid optimization control strategies; it determines whether the primary frequency regulation control command meets the preset rationality conditions. If it does, the command is sent to the target generator unit through the communication network; if it does not, the optimal gain parameters of primary frequency regulation control are recalculated.

[0027] The adaptive gain adjustment model optimization module is used to acquire power grid operating parameters and generator set status information as input data for the adaptive gain adjustment model. The model is trained and optimized in real time through signal dynamic update technology. Specifically, the acquired power grid operating parameters and generator set status information are subjected to noise removal and feature extraction using data preprocessing techniques to obtain high-quality input data. The preprocessed data is then input into the adaptive gain adjustment model, and the model is continuously optimized through machine learning algorithms.

[0028] Compared with the prior art, the present invention has the following advantages and technical effects:

[0029] This invention establishes a mathematical model of power grid operation status and frequency stability. Power grid operating parameters are acquired through online monitoring, and a power grid stability index is calculated. When the stability index falls below a preset threshold, the power grid is deemed unstable, requiring primary frequency regulation control. For different stability indices, an adaptive gain adjustment model is established using a fuzzy control algorithm. The stability index is used as input, and the optimal gain parameter is obtained through fuzzy inference. Based on this parameter, a primary frequency regulation control command is generated and issued to each generator unit to adjust the active power output. During the control process, operating parameters and generator unit status information are continuously collected. The model is trained and optimized online using dynamic signal update technology, continuously improving accuracy and robustness. This invention achieves precise adaptive control of the power grid frequency, effectively improving the stability and reliability of the power grid and providing strong support for the safe and stable operation of the power system. Attached Figure Description

[0030] The accompanying drawings, which form part of this application, are used to provide a further understanding of this application. The illustrative embodiments and descriptions of this application are used to explain this application and do not constitute an undue limitation of this application. In the drawings:

[0031] Figure 1This is a flowchart of a method according to an embodiment of the present invention. Detailed Implementation

[0032] It should be noted that, unless otherwise specified, the embodiments and features described in this application can be combined with each other. This application will now be described in detail with reference to the accompanying drawings and embodiments.

[0033] It should be noted that the steps shown in the flowchart in the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions, and although a logical order is shown in the flowchart, in some cases the steps shown or described may be executed in a different order than that shown here.

[0034] Example 1

[0035] like Figure 1 As shown, this embodiment provides a primary frequency modulation optimization control method based on adaptive gain adjustment, including:

[0036] To achieve the above objectives, the present invention provides a primary frequency modulation optimization control method based on adaptive gain adjustment, comprising:

[0037] Historical data on power grid operating status parameters and power grid frequency are acquired and preprocessed. A power grid frequency stability classification model is then constructed based on the preprocessed historical data.

[0038] In some specific implementations, the process of constructing a power grid frequency stability classification model includes:

[0039] Historical data of power grid operating status parameters are cleaned and normalized to obtain standardized power grid operating status parameter data. A mapping relationship between power grid operating status parameters and power grid frequency is constructed. A classification model is built based on the support vector machine algorithm. The classification model is trained based on the standardized power grid operating status parameter data, historical power grid frequency data and the mapping relationship to obtain a power grid frequency stability classification model.

[0040] Specifically, based on power system theory, the power grid operating state parameters and power grid frequency stability parameters are obtained, and a mapping relationship between the power grid operating state and power grid frequency stability is established. For the obtained power grid operating state parameters, data preprocessing techniques are used to clean and normalize the parameter data to obtain standardized power grid operating state parameters. Based on the standardized power grid operating state parameters, the support vector machine algorithm in machine learning is used to train and establish a classification model of power grid operating state and power grid frequency stability. During the training process of the classification model, the hyperparameters such as the kernel function and penalty coefficient of the support vector machine algorithm are optimized through grid search and cross-validation methods to improve the classification performance of the model.

[0041] For example, the relationship between power grid operating status and frequency stability is one of the core issues in power system operation. First, it's necessary to obtain power grid operating status parameters, such as generator power, load demand, and line power flow, as well as frequency stability parameters, such as system inertia and frequency regulation capability. These parameters reflect the real-time operating status of the power grid. Preprocessing the acquired raw data is essential. For example, generator power data can be normalized to unify the output of units with different capacities into the 0-1 range, facilitating subsequent modeling and analysis. Outliers also need to be processed, such as removing obviously erroneous data caused by communication failures. Support Vector Machines (SVMs) are excellent classification algorithms suitable for determining power grid frequency stability. During training, a grid search method can be used to optimize the kernel function parameters. For example, for a Gaussian kernel, different γ values ​​(such as 0.1, 1, 10, etc.) can be tried to select the parameter with the best classification effect. Simultaneously, the penalty coefficient C should be optimized to balance model complexity and classification accuracy.

[0042] Real-time acquisition and standardization of power grid operation parameters yields a standardized power grid operation parameter dataset; based on this dataset, power grid operation status classification results are obtained.

[0043] In some specific implementations, the real-time acquired power grid operating parameters are denoised and standardized to obtain a standardized power grid operating parameter dataset; features are extracted from the standardized power grid operating parameter dataset to obtain a power grid operating state feature vector; and a power grid operating state classification result is obtained based on the power grid operating state feature vector.

[0044] Specifically, based on pre-established threshold ranges for power grid operating parameters, it is determined whether the real-time collected power grid operating parameters exceed these ranges. If they do, an early warning signal is triggered, indicating an abnormal power grid operating status. Power grid operating parameters, such as voltage, current, active power, and reactive power, are collected in real-time using online monitoring technology and transmitted to a data processing center. At the data processing center, data cleaning and preprocessing techniques are used to denoise and standardize the collected power grid operating parameter data, resulting in a standardized power grid operating parameter dataset. Feature parameters are extracted from this standardized dataset to construct a power grid operating status feature vector, which serves as input data for judging the power grid operating status. A support vector machine algorithm is then used to classify the power grid operating status feature vector, obtaining the classification result to determine whether the current power grid operating status is normal.

[0045] For example, the threshold range of power grid operating parameters is a crucial foundation for ensuring the safe and stable operation of the power grid. For instance, for a 220kV transmission line, the normal operating voltage range is typically 198kV to 242kV. If the voltage value exceeds this range during real-time monitoring, the system will trigger an early warning signal. This early warning mechanism can promptly detect potential power grid operational anomalies, providing important information for maintenance personnel to take appropriate measures. Online monitoring technology is key to acquiring real-time power grid operating parameters. Taking substations as an example, sensors installed on key equipment such as main transformers and circuit breakers can collect parameters such as voltage, current, active power, and reactive power in real time. This data is transmitted to a data processing center via a communication network, providing raw data support for subsequent analysis. Data cleaning and preprocessing are crucial steps in ensuring data quality. For example, the collected voltage data may contain outliers due to equipment failure or communication interference. Methods such as median filtering can effectively remove this noise and improve data reliability. Simultaneously, normalizing parameters of different dimensions can make subsequent analysis more accurate. Feature extraction is a key step in constructing a feature vector of the power grid operating state. For example, indicators such as voltage deviation rate and power factor can be calculated as feature parameters. These feature parameters can better reflect the power grid's operating status and provide effective input for subsequent classification. Support vector machine algorithms have shown good performance in classifying power grid operating status. By training on historical data, a classification model capable of distinguishing between normal and abnormal operating states can be established. For example, when the input feature vector shows that the voltage deviation rate exceeds 5% and the power factor is below 0.85, the model may classify it as an abnormal state.

[0046] Based on the power grid operating parameters and the power grid frequency stability classification model, the stability is initially judged. If it is unstable, the power grid frequency stability index is calculated through the pre-trained stability index calculation model and it is determined whether it is below the threshold.

[0047] In some specific implementations, the process of calculating the power grid frequency stability index includes:

[0048] Sensors and measuring devices are deployed at key nodes of the power grid to collect power grid operating parameters in real time; the collected power grid operating parameter data is preprocessed; a mathematical model describing the stability of the power grid is constructed, the preprocessed power grid operating parameters are input into the mathematical model, and the power grid frequency stability index is calculated by solving the model equations.

[0049] Specifically, sensors and measuring devices are deployed at key nodes of the power grid to collect real-time power grid operating parameters, including voltage, current, active power, and reactive power. The collected power grid operating parameter data is transmitted to a data center, where it undergoes preprocessing to remove outliers and noise, resulting in standardized parameter data. Based on the physical model and operating characteristics of the power system, a mathematical model describing power grid stability is established, which comprehensively considers both static and dynamic stability. The real-time collected power grid parameters are input into the established mathematical model, and by solving the model equations, a stability index reflecting the current level of power grid stability is calculated. An adaptive gain adjustment model is constructed. If the power grid operating state classification result does not meet the requirements or the power grid frequency stability index is below a threshold, the power grid frequency stability index is used as the input to the adaptive gain adjustment model to obtain the optimal gain parameter for primary frequency regulation control. Optimal control is then achieved based on the optimal gain parameter for primary frequency regulation control.

[0050] In some specific implementations, a stability index threshold is set based on expert experience to determine whether the power grid frequency stability index exceeds the stability index threshold. If it is lower than the stability index threshold, an early warning is issued. The optimal gain parameters for primary frequency regulation control are obtained based on the power grid frequency stability index and the adaptive gain adjustment model.

[0051] Specifically, it determines whether the calculated stability index exceeds the preset safety threshold. If it does, an early warning is issued indicating that there is a risk of instability in the power grid operation, and measures need to be taken.

[0052] In some specific implementations, the process of obtaining the optimal gain parameter for primary frequency modulation control further includes:

[0053] Data on the operation status of the power grid is acquired, and multiple power grid frequency stability indices are calculated as input variables for the fuzzy controller. Based on the numerical range of the power grid frequency stability indices, a fuzzy control rule base is established, forming multiple fuzzy control subsystems. For each fuzzy control subsystem, fuzzy inference calculations are performed to obtain the fuzzy gain parameters for primary frequency regulation control. The fuzzy gain parameters of multiple fuzzy control subsystems are weighted and fused, and the optimal gain parameters for primary frequency regulation control are obtained through defuzzification processing.

[0054] Specifically, the process involves acquiring power grid operation status data, calculating multiple power grid stability indices as input variables for the fuzzy controller, establishing a fuzzy control rule base based on the numerical range of the power grid stability indices, forming multiple fuzzy control subsystems, performing fuzzy inference calculations for each fuzzy control subsystem to obtain the fuzzy gain parameters for primary frequency regulation control, and weighted fusing the fuzzy gain parameters of multiple fuzzy control subsystems. Through defuzzification processing, the optimal gain parameters for primary frequency regulation control are determined.

[0055] For example, firstly, real-time operating status data of the power grid is acquired, such as generator output power, line power flow, and node voltage. This data forms the basis for assessing grid stability. Then, based on this data, multiple grid stability indices are calculated, such as the frequency stability index, voltage stability index, and small-disturbance stability index. Assume that at a certain moment, the calculated frequency stability index is 0.95, the voltage stability index is 0.98, and the small-disturbance stability index is 0.92. These indices collectively reflect the current stable state of the power grid. These three stability indices are used as input variables for a fuzzy controller. A fuzzy control rule base is established based on the numerical range of each grid stability index and actual operating experience. For example, the range of the frequency stability index is divided into three fuzzy sets: low, medium, and high. When the frequency stability index is low, it means that the grid frequency stability is poor, requiring a larger frequency regulation gain; when the frequency stability index is high, a smaller frequency regulation gain can be used. Corresponding fuzzy control rules are formulated for different combinations of stability indices. For example, when the frequency stability index is "low" and the voltage stability index is "medium," the output fuzzy gain parameter is "large"; when the frequency stability index is "high" and the voltage stability index is "low," the output fuzzy gain parameter is "medium." These rules constitute the fuzzy control subsystem. For each fuzzy control subsystem, fuzzy inference calculations are performed based on the current grid stability index. For example, if the current frequency stability index is 0.95, belonging to the "high" fuzzy set, and the voltage stability index is 0.98, belonging to the "medium" fuzzy set, the fuzzy gain parameter for primary frequency regulation control can be inferred to be "medium" according to the corresponding fuzzy rules. The fuzzy gain parameters of multiple fuzzy control subsystems are weighted and fused to obtain a comprehensive fuzzy gain parameter. For example, if the fuzzy gain parameters output by the three subsystems are "large," "medium," and "small," respectively, a weighted average is performed based on the weights of each subsystem, ultimately resulting in a comprehensive fuzzy gain parameter that is "medium-large." Then, through defuzzification processing, the fuzzy gain parameter is converted into a precise value. For example, the "medium to large" fuzzy gain parameter can be converted into a specific value, such as 1.2, which will be used as the actual gain parameter for primary frequency modulation control.

[0056] Furthermore, if the grid stability index changes, the fuzzy controller is triggered to adaptively adjust the gain parameters and update the fuzzy rule base; the optimized gain parameters are applied to the primary frequency regulation control algorithm to adjust the active power output of the generator set in real time and maintain grid frequency stability; the trend of grid stability index changes is continuously monitored to evaluate the effect of primary frequency regulation control, and the above steps are repeated when necessary to achieve adaptive optimization of the gain parameters.

[0057] Specifically, if the grid stability index changes, for example, the frequency stability index drops from 0.95 to 0.85, the fuzzy controller is triggered to adaptively adjust the gain parameters. Based on the new stability index, fuzzy inference and weighted fusion are performed again to obtain new fuzzy gain parameters. Simultaneously, the fuzzy rule base is updated based on historical operating data and expert experience to adapt to different grid operating states. The optimized gain parameters are applied to the primary frequency regulation control algorithm to adjust the active power output of the generator sets in real time. For example, when the grid frequency drops, based on the gain parameter of 1.2, the generator sets are controlled to increase active power output to raise the grid frequency. In this way, a rapid response to grid frequency changes can be achieved, maintaining the frequency within the allowable range. Using fuzzy control, control parameters can be adaptively adjusted according to the grid stability state, improving the accuracy and robustness of control. During the primary frequency regulation control process, the changing trend of the grid stability index is continuously monitored to evaluate the effectiveness of the primary frequency regulation control. For example, it is observed whether the frequency stability index has returned to the normal range and whether the voltage stability index remains stable. If the control effect is unsatisfactory, for example, if the frequency stability index remains low, the above steps need to be repeated, the gain parameters readjusted, and the fuzzy rule base optimized until a satisfactory control effect is achieved. This feedback mechanism enables adaptive optimization of the gain parameters, improving the performance of primary frequency regulation control. Employing fuzzy control and adaptive optimization techniques can enhance the stability and reliability of the power grid, ensuring the safe operation of the power system.

[0058] In some specific implementations, before inputting the power grid frequency stability index into the adaptive gain adjustment model, the power grid safety margin is calculated. Based on the calculation results of the power grid frequency stability index and safety margin, and combined with the optimization control strategy in the expert knowledge base, a real-time power grid optimization control strategy is obtained.

[0059] In some specific implementations, the process of achieving optimized control based on the optimal gain parameter of primary frequency regulation control includes: integrating the optimal gain parameter of primary frequency regulation control with the grid frequency deviation and active power, and generating a primary frequency regulation control command in conjunction with the real-time grid optimization control strategy; determining whether the primary frequency regulation control command meets the preset rationality conditions; if it does, sending the command to the target generator unit through the communication network; if it does not, recalculating the optimal gain parameter of primary frequency regulation control.

[0060] Specifically, based on real-time grid operating parameters, an adaptive gain adjustment model is used to calculate the optimal gain parameters, which serve as the basis for generating frequency regulation control commands. The gain parameters output by the adaptive gain adjustment model are fused with parameters such as grid frequency deviation and active power, and a primary frequency regulation control command is generated using a machine learning algorithm. The generated primary frequency regulation control command is then evaluated to determine if it meets preset rationality conditions. If it does, the command is sent to the target generator unit via the communication network; otherwise, the process returns to step 1 to recalculate the gain parameters. In the communication network, security mechanisms such as encrypted transmission and authentication are employed to ensure the integrity and reliability of the primary frequency regulation control command, preventing tampering or impersonation. Upon receiving the primary frequency regulation control command, the generator unit extracts the active power adjustment value from the command and determines the final active power output adjustment value based on the unit's operating status and regulation capacity. The generator unit's control system adjusts parameters such as feedwater and turbine start-up / shutdown based on the determined active power output adjustment value to achieve precise control of active power output and feeds the adjustment results back to the dispatch center. The system acquires real-time grid frequency data to determine whether the frequency regulation control effect has achieved the expected target. If it has not, the system feeds back parameters such as frequency deviation to the adaptive gain adjustment model for the next round of iterative optimization. If it has achieved the target, the system maintains the current frequency regulation state and continuously monitors changes in grid frequency.

[0061] For example, the adaptive gain adjustment model is a core component of a primary frequency regulation control system, optimizing control gain through real-time analysis of grid operating parameters. For instance, when the grid frequency deviation is large, the model may output larger proportional and integral gains to accelerate the adjustment speed; conversely, when the frequency is close to the rated value, the gain may be reduced to avoid over-adjustment. Machine learning algorithms play a crucial role in generating frequency regulation control commands. Taking Support Vector Machines (SVMs) as an example, frequency deviations and active power changes in historical data can be used as features, with corresponding frequency regulation commands as labels for training. The trained model can quickly generate appropriate frequency regulation commands based on the current grid state. Determining the reasonableness of commands is a key aspect of ensuring system security. Multiple judgment conditions can be set, such as command amplitude not exceeding 10% of the unit's rated power and adjustment rate not exceeding the unit's ramp rate. If the conditions are not met, the system will recalculate the gain and generate new commands until the requirements are met. Communication security is paramount for the power system. AES encryption can be used to encrypt frequency regulation commands, and digital signature technology can be used to ensure command integrity. Simultaneously, a VPN channel can be established between the dispatch center and power plants to further enhance communication security. After receiving commands, generator sets need to process them according to their own characteristics. For example, for thermal power units, it may be necessary to adjust both the feedwater flow and fuel supply simultaneously; while for hydropower units, the main adjustment is the guide vane opening. Different types of units have different regulation characteristics, thus requiring targeted control strategies. Precise control of active power is the foundation for frequency regulation. Taking thermal power units as an example, generator output can be controlled by adjusting parameters such as boiler feedwater flow and turbine regulating valve opening. The control system needs to consider factors such as the unit's thermodynamic and mechanical characteristics, employing algorithms such as PID control to achieve fast and precise power regulation. Frequency regulation effect evaluation is a continuous process. A standard of within ±0.02Hz can be set as the acceptable standard. If this standard is not met, the system will feed back the deviation between the actual frequency and the target frequency, as well as the overshoot during the regulation process, to the adaptive model for the next round of optimization. This closed-loop feedback mechanism can continuously improve the system's frequency regulation performance. Through the coordinated operation of the above-mentioned links, the primary frequency regulation control system can effectively maintain grid frequency stability. This not only improves power quality but also provides an important guarantee for the safe and stable operation of the power grid. Meanwhile, the application of advanced technologies such as adaptive regulation and machine learning has enabled the system to have a stronger level of adaptability and intelligence, and to better cope with the complex and ever-changing power grid operating environment.

[0062] In some specific implementations, grid operating parameters and generator status information are acquired and used as input data for an adaptive gain adjustment model. The model is then trained and optimized in real time using dynamic signal update technology. Specifically, the acquired grid operating parameters and generator status information are preprocessed using data preprocessing techniques to remove noise and extract features, resulting in high-quality input data. The preprocessed data is then input into the adaptive gain adjustment model, and the model is continuously optimized using machine learning algorithms.

[0063] Specifically, grid operating parameters and generator status information are acquired and used as input data for an adaptive gain adjustment model. The model is then trained and optimized in real time using dynamic signal update technology. For the acquired grid operating parameters and generator status information, data preprocessing techniques are employed for noise removal and feature extraction to obtain high-quality input data. This preprocessed data is then input into the adaptive gain adjustment model, which is trained using machine learning algorithms such as support vector machines or neural networks to continuously optimize model parameters. During model training, cross-validation and early stopping mechanisms are introduced to avoid overfitting and improve the model's generalization ability and robustness. Based on the optimized adaptive gain adjustment model obtained from the training, newly acquired grid operating parameters and generator status information are predicted and analyzed to obtain the optimal gain adjustment strategy. This optimal gain adjustment strategy is applied to frequency regulation control, adaptively adjusting the generator output power to achieve stable grid frequency control. Continuous monitoring of grid operating status and generator performance is maintained. If anomalies occur or prediction accuracy decreases, retraining and optimization of the adaptive gain adjustment model are triggered to ensure the reliability and effectiveness of frequency regulation control.

[0064] For example, the adaptive gain adjustment model is a core technology of power grid frequency regulation control. Its input data includes power grid operating parameters and generator status information. For instance, power grid operating parameters may include system frequency, active power balance, and voltage level, while generator status information may include generator output, speed, and excitation current. This data often contains noise and redundancy, thus requiring preprocessing. Data preprocessing techniques can employ filtering algorithms to remove high-frequency noise, such as using low-pass filters or median filters. Feature extraction can utilize dimensionality reduction methods such as Principal Component Analysis (PCA) to extract the most representative features. For example, for a single generator unit, the original data may contain dozens of parameters; PCA can reduce this to a few key features, significantly reducing subsequent computation. Machine learning algorithms play a crucial role in the adaptive gain adjustment model. Support Vector Machines (SVMs) are suitable for small-sample learning and can effectively handle sudden events in the power grid. Neural networks, on the other hand, excel at processing large amounts of historical data and can capture complex nonlinear relationships. For example, a multilayer perceptron network can be constructed, with the input layer corresponding to preprocessed features, the hidden layer using the ReLU activation function, and the output layer providing the optimal gain parameters. To prevent overfitting, cross-validation is a commonly used technique. K-fold cross-validation can be employed, dividing the dataset into K parts, using K-1 parts for training and 1 part for validation each time, repeating this process K times. An early stopping mechanism stops training when the validation set performance no longer improves. Combining these two methods effectively enhances the model's generalization ability. The application of an optimal gain adjustment strategy is crucial. For example, when a deviation from the rated system frequency is detected, the model might suggest increasing the system gain for fast-responding units and appropriately decreasing the gain for slower-responding units. This differentiated strategy fully utilizes the characteristics of various units, achieving more precise frequency control. Continuous monitoring and model retraining mechanisms ensure the long-term reliability of the system. For example, a sliding time window can be set to periodically evaluate the model's prediction accuracy over a recent period. If the accuracy falls below a preset threshold, such as 95%, the model retraining process is triggered. This adapts to seasonal changes in grid load characteristics or dynamic changes in the system caused by new power generation equipment. Through this adaptive gain adjustment technique, the grid frequency regulation control system can better cope with challenges such as load fluctuations and new energy integration, improving the stability and reliability of the grid. At the same time, more precise control can reduce wear and tear on generator sets, extend equipment life, and achieve both economic and technological benefits.

[0065] Example 2

[0066] This embodiment provides a primary frequency modulation optimization control system based on adaptive gain adjustment, including:

[0067] The power grid frequency stability mathematical modeling module is used to acquire and preprocess historical data of power grid operating status parameters and power grid frequency, and to build a power grid frequency stability classification model based on the preprocessed historical data.

[0068] The online monitoring module for power grid operating parameters is used to acquire and standardize power grid operating parameters in real time, thereby obtaining a standardized dataset of power grid operating parameters; and to obtain power grid operating status classification results based on the standardized dataset of power grid operating parameters.

[0069] The power grid stability judgment module makes a preliminary judgment on stability based on the power grid operating parameters and the power grid frequency stability classification model. If it is unstable, it calculates the power grid frequency stability index through a pre-trained stability index calculation model and determines whether it is below the threshold.

[0070] An adaptive gain adjustment module is used to construct an adaptive gain adjustment model. If the power grid operation status classification result does not meet the requirements or the power grid frequency stability index is lower than the threshold, the power grid frequency stability index is used as the input of the adaptive gain adjustment model to obtain the optimal gain parameter for primary frequency regulation control.

[0071] The primary frequency regulation control command generation module is used to integrate the optimal gain parameters of primary frequency regulation control with the grid frequency deviation and active power, and generate primary frequency regulation control commands in combination with real-time grid optimization control strategies; it determines whether the primary frequency regulation control command meets the preset rationality conditions. If it does, the command is sent to the target generator unit through the communication network; if it does not, the optimal gain parameters of primary frequency regulation control are recalculated.

[0072] The adaptive gain adjustment model optimization module is used to acquire power grid operating parameters and generator set status information as input data for the adaptive gain adjustment model. The model is trained and optimized in real time through signal dynamic update technology. Specifically, the acquired power grid operating parameters and generator set status information are subjected to noise removal and feature extraction using data preprocessing techniques to obtain high-quality input data. The preprocessed data is then input into the adaptive gain adjustment model, and the model is continuously optimized through machine learning algorithms.

[0073] The above are merely preferred embodiments of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A method for optimizing control of frequency modulation based on adaptive gain adjustment, characterized in that, The method comprises the following steps: obtaining historical data of power grid operation state parameters and power grid frequency and preprocessing the historical data, and constructing a power grid stability classification model based on the preprocessed historical data; The process of constructing the power grid stability classification model comprises: performing data cleaning and normalization processing on the historical data of the power grid operation state parameters to obtain standardized power grid operation state parameter data; constructing a mapping relationship between the power grid operation state parameters and the power grid frequency, constructing a classification model based on a support vector machine algorithm, training the classification model based on the standardized power grid operation state parameter data, the historical data of the power grid frequency and the mapping relationship, and obtaining the power grid stability classification model; obtaining power grid operation state parameters in real time and normalizing the power grid operation state parameters to obtain a power grid operation state parameter standard data set; obtaining a power grid operation state classification result based on the power grid operation state parameter standard data set; preliminarily judging the stability based on the power grid operation state parameters and the power grid stability classification model, and if the stability is unstable, calculating a power grid stability index by using a pre-trained stability index calculation model and judging whether the power grid stability index is lower than a stability index threshold value; constructing an adaptive gain adjustment model, if the power grid operation state classification result does not meet the requirements or the power grid stability index is lower than the stability index threshold value, inputting the power grid stability index into the adaptive gain adjustment model to obtain an optimal gain parameter of primary frequency modulation control, and realizing optimized control based on the optimal gain parameter of primary frequency modulation control; setting a stability index threshold value based on expert experience, judging whether the power grid stability index exceeds the stability index threshold value, if the power grid stability index is lower than the stability index threshold value, issuing a warning, and obtaining the optimal gain parameter of primary frequency modulation control based on the power grid stability index and the adaptive gain adjustment model; The process of obtaining the optimal gain parameter of primary frequency modulation control further comprises: obtaining power grid operation state parameter data, calculating a plurality of power grid stability indexes as input variables of a fuzzy controller, the plurality of power grid stability indexes comprising a frequency stability index, a voltage stability index and a small disturbance stability index, establishing a fuzzy control rule base according to the numerical range of the power grid stability indexes to form a plurality of fuzzy control subsystems, performing fuzzy reasoning calculation for each fuzzy control subsystem to obtain a fuzzy gain parameter of primary frequency modulation control, and performing weighted fusion on the fuzzy gain parameters of the plurality of fuzzy control subsystems to obtain the optimal gain parameter of primary frequency modulation control through inverse fuzzy processing; before inputting the power grid stability index into the adaptive gain adjustment model, calculating a safety margin of the power grid, and obtaining a real-time power grid optimized control strategy based on the calculation results of the power grid stability index and the safety margin and in combination with an optimized control strategy in an expert knowledge base; The process of realizing optimized control based on the optimal gain parameter of primary frequency modulation control comprises: fusing the optimal gain parameter of primary frequency modulation control with a power grid frequency deviation and active power, generating a primary frequency modulation control instruction in combination with the real-time power grid optimized control strategy, judging whether the primary frequency modulation control instruction meets a preset rationality condition, if yes, issuing the instruction to a target generator set through a communication network, and if no, recalculating the optimal gain parameter of primary frequency modulation control. Further comprising: obtaining power grid operation state parameters and generator set state information as input data of the adaptive gain adjustment model, and performing real-time training and optimization on the model through signal dynamic updating technology; wherein, the obtained power grid operation state parameters and generator set state information are subjected to noise removal and feature extraction through data preprocessing technology to obtain high-quality input data; the preprocessed data is input into the adaptive gain adjustment model, and the model is continuously optimized through a machine learning algorithm.

2. The primary frequency regulation optimization control method based on adaptive gain adjustment according to claim 1, characterized in that, the real-time obtained power grid operation state parameters are subjected to denoising and standardization processing to obtain a power grid operation state parameter standard data set; the power grid operation state parameter standard data set is subjected to feature extraction to obtain a power grid operation state feature vector, and a power grid operation state classification result is obtained based on the power grid operation state feature vector.

3. The primary frequency regulation optimization control method based on adaptive gain adjustment according to claim 1, characterized in that, the process of calculating the power grid stability index comprises: sensors are arranged at key nodes of the power grid to collect power grid operation state parameters in real time; the collected power grid operation state parameter data are subjected to data preprocessing; a mathematical model describing the stability of the power grid is constructed, the preprocessed power grid operation state parameters are input into the mathematical model, and the power grid stability index is calculated by solving the model equation.

4. A system for implementing the method of any one of claims 1 to 3, based on adaptive gain adjustment, characterized in that, comprise: a power grid frequency stability mathematical modeling module, configured to obtain and preprocess historical data of power grid operation state parameters and power grid frequency, and construct a power grid stability classification model based on the preprocessed historical data; a power grid operation parameter online monitoring module, configured to obtain power grid operation state parameters in real time and normalize them to obtain a power grid operation state parameter standard data set; and obtain a power grid operation state classification result based on the power grid operation state parameter standard data set; a power grid stability state judgment module, configured to preliminarily judge the stability based on the power grid operation state parameters and the power grid stability classification model, and if the power grid is unstable, calculate the power grid stability index through a pre-trained stability index calculation model and judge whether the power grid stability index is lower than a stability index threshold value; an adaptive gain adjustment module, configured to construct an adaptive gain adjustment model, and if the power grid operation state classification result does not meet the requirements or the power grid stability index is lower than the stability index threshold value, input the power grid stability index into the adaptive gain adjustment model as input data to obtain optimal gain parameters of primary frequency regulation control; a primary frequency regulation control instruction generation module, configured to fuse the optimal gain parameters of primary frequency regulation control with power grid frequency deviation and active power, combine real-time power grid optimization control strategies to generate primary frequency regulation control instructions, judge whether the primary frequency regulation control instructions meet preset rationality conditions, and if yes, issue the instructions to target generator sets through a communication network; and if not, recalculate the optimal gain parameters of primary frequency regulation control. The adaptive gain adjustment model optimization module is configured to acquire power grid operation state parameters and generator set state information as input data of the adaptive gain adjustment model, and to perform real-time training and optimization on the model through signal dynamic updating technology; wherein, the acquired power grid operation state parameters and generator set state information are subjected to noise removal and feature extraction through data preprocessing technology to obtain high-quality input data; The preprocessed data is input into the adaptive gain adjustment model, and the model is continuously optimized through a machine learning algorithm.

Images