Hierarchical safety early warning method based on smart power grid analog simulation

Abstract

The invention relates to the technical field of smart power grids, and discloses a hierarchical safety early warning method based on smart power grid analog simulation, and the method comprises the following steps: S1, collecting the operation data of a smart power grid, including the real-time monitoring data of voltage, current, power factor and temperature, and processing the collected operation data; s2, optimizing estimation of a power grid state by using an extended Kalman filtering algorithm; s3, establishing a nonlinear dynamic model of the smart power grid, and predicting the health state of the power grid according to the dynamic behavior of the power grid; and S4, constructing a Bayesian network model, evaluating the risk of the power grid through probabilistic reasoning, and judging the health condition of the power grid. According to the invention, through combination of intelligent power grid analog simulation, deep reinforcement learning and adaptive optimal control, intelligence, adaptive adjustment and real-time response capability of the power grid early warning system are realized, and stability and emergency response efficiency of the power grid in a complex environment are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of smart grids, and in particular to a hierarchical safety early warning method based on smart grid simulation. Background Art

[0003] The existing power grid management system usually adopts an early warning mechanism based on manually set thresholds, which can detect partial faults and anomalies of the power grid at an early stage and issue alarms. These systems can quickly identify problems and trigger warnings when obvious faults occur in the power grid through real-time monitoring of key parameters, and then take predetermined emergency measures. Traditional monitoring systems also have a relatively mature hardware architecture and low deployment costs, which are suitable for relatively simple power grid operation environments.

[0004] However, the power grid monitoring and early warning systems of the existing technology have some shortcomings; the existing systems mostly rely on manually set thresholds and lack intelligent dynamic adjustment and real-time evaluation capabilities. Since the threshold settings are static, they cannot adapt to changes in the power grid in real time, especially when dealing with complex and sudden power grid risks, which can easily lead to delayed responses or inaccurate early warnings; in addition, the existing systems are not flexible enough in resource scheduling during emergency responses, and often can only rely on preset rules and lack the ability to flexibly adjust according to the actual situation of the power grid, which leads to the inability to respond effectively in a timely manner in some complex situations. Summary of the invention

[0005] In view of the deficiencies in the prior art, the present invention provides a hierarchical safety early warning method based on smart grid simulation, which solves the problem that the existing power grid early warning system lacks intelligence, adaptive adjustment and real-time response capabilities.

[0006] To achieve the above objectives, the present invention is implemented through the following technical solutions: a hierarchical safety early warning method based on smart grid simulation, the method comprising the following steps: S1. Collect the operation data of the smart grid, including real-time monitoring data of voltage, current, power factor and temperature, and process the collected operation data; S2, using the extended Kalman filter algorithm to optimize the estimation of the power grid state; S3. Establishing a nonlinear dynamic model of the smart grid and predicting the health status of the grid according to the dynamic behavior of the grid; S4. Construct a Bayesian network model to evaluate the risk of the power grid through probabilistic reasoning and determine the health status of the power grid; S5. Based on the Nash equilibrium model in game theory, the behaviors of multiple decision-making entities in the power grid are collaboratively optimized to achieve global stability of the power grid; S6. Use the maximum entropy principle in information theory to optimize the setting of warning levels and information flow transmission strategies, adjust the information allocation and transmission methods of each warning level, and ensure effective information flow; S7, based on adaptive optimal control and deep reinforcement learning, adjust the operation strategy of the power grid in real time and optimize the early warning response; S8. Automatically trigger graded safety warning responses based on the health status of the power grid and risk assessment results.

[0007] Preferably, the method of optimizing the estimation of the power grid state by using the extended Kalman filter algorithm includes: Optimize and estimate the nonlinear state of the power grid based on the extended Kalman filter (EKF) algorithm; Provide real-time updates on the health of the grid, use grid measurement data and dynamic models to correct grid status predictions, and reduce the impact of noise on system estimates; By calculating the Kalman gain to adjust the error estimate of the system, the accuracy of power grid state prediction can be further improved.

[0008] Preferably, the establishment of a nonlinear dynamic model of the smart grid includes: Using the operation data, the state vector of the power grid is constructed, and the dynamic relationship between the various states of the power grid is described by nonlinear functions; Introducing disturbance terms to simulate the impact of external factors on the power grid system, so that the power grid model can better adapt to different operating environments; The stability of the power grid model is analyzed using Lyapunov stability theory to ensure that the model remains stable under various disturbances.

[0009] Preferably, the constructing of the Bayesian network model includes: By establishing the conditional dependency relationship between various parameters of the power grid, a Bayesian network of the power grid is constructed; According to the historical operation data of the power grid, the conditional probability distribution of each node is calculated, and its risk probability is inferred according to the current power grid status; Based on the inference results, the risk level of the power grid under different operating conditions is evaluated, and decision support for power grid operation is provided through Bayesian reasoning.

[0010] Preferably, the behaviors of optimizing multiple decision-making entities of the power grid based on the Nash equilibrium model in game theory include: Set up multiple control entities of the power grid, including load dispatch, generator dispatch and backup power supply, as participants in the game; Using Nash equilibrium solutions, the decision-making of each subject is optimized through game theory to achieve the global optimal grid operation state; Each entity adjusts its decision-making behavior according to the health status of the power grid and the results of risk assessment to ensure the stable operation of the power grid.

[0011] Preferably, the optimization of the setting of warning levels and information flow transmission strategies using the maximum entropy principle in information theory includes: Based on the maximum entropy principle in information theory, the system can quickly transmit risk information by optimizing the information distribution of each warning level in the power grid; Set up corresponding information transmission mechanisms for different warning levels to ensure the effectiveness of information flow; Under different risk levels, the information distribution strategy is dynamically adjusted according to the status of the power grid and real-time data to reduce redundant information.

[0012] Preferably, the real-time adjustment of the operation strategy of the power grid based on adaptive optimal control and deep reinforcement learning includes: Introducing the deep reinforcement learning (DQN) algorithm, using real-time data from the power grid to train the intelligent decision-making system and adaptively adjust the control strategy of the power grid; Combined with optimal control theory, the emergency response mechanism of the power grid is optimized by learning the operation mode and risk status of the power grid in real time; Through reinforcement learning training, the system can continuously optimize the timing of warnings and response intensity, reducing false alarms and missed alarms.

[0013] Preferably, the automatic triggering of a hierarchical safety warning response based on the health status of the power grid and the risk assessment results includes: According to the evaluation results of the health status of the power grid, the corresponding warning level is automatically selected, and the warning response is triggered through the intelligent decision-making system; In high-risk situations, the system can automatically adjust the grid's operating strategy, including enabling backup power sources and adjusting load distribution; In low-risk situations, the system is able to choose non-intervention responses to keep the grid stable.

[0014] Preferably, the method predicts the risk of the power grid based on the power grid simulation model further comprises: Using dynamic power grid simulation models, combined with historical power grid data and real-time operation data, potential risks of the power grid can be predicted from multiple angles; Through big data analysis technology, we simulate and predict the risk scenarios of different failure modes and equipment failures, and identify in advance the security threats that the power grid may face.

[0015] Preferably, the feedback mechanism of the power grid health status and risk assessment results includes: Periodically update the health status and risk assessment results of the power grid through real-time monitoring and data feedback systems; The system continuously adjusts early warning strategies and response behaviors based on a feedback mechanism to ensure that the power grid can flexibly respond to emergencies in different operating environments.

[0016] The present invention provides a hierarchical safety early warning method based on smart grid simulation, which has the following beneficial effects: 1. The present invention uses an automated hierarchical safety warning response mechanism to automatically adjust the warning level in real time according to the health status and risk assessment of the power grid. Compared with traditional manual threshold setting, this intelligent system greatly improves the accuracy and timeliness of warnings and reduces false alarms and delays that may be caused by human operations.

[0017] 2. By combining adaptive optimal control with deep reinforcement learning, the power grid can dynamically adjust its dispatch strategy in the face of uncertain load fluctuations and equipment failures. This makes the power grid more flexible and can automatically adapt to real-time changes, avoiding the rigidity and inefficiency of traditional fixed strategies.

[0018] 3. Combining the maximum entropy principle in information theory, the present invention optimizes the information flow transmission and the setting of warning levels. By reducing the transmission of redundant information, the efficiency of information transmission is improved, ensuring that key warning information can reach decision makers quickly, and avoiding the slow response caused by information overload in traditional systems.

[0019] 4. The present invention optimizes the behavior of each decision-making subject of the power grid through the Nash equilibrium model in game theory, avoiding the local optimal problem in traditional management methods. The system can coordinate the decisions of all parties to ensure the global stability of the power grid and maximize the use of resources in a complex environment. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] Figure 1 The figure is a flow chart of the method of the present invention. DETAILED DESCRIPTION

[0021] The following will be combined with the drawings in the specification of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0022] Please see attached Figure 1 The embodiment of the present invention provides a hierarchical safety early warning method based on smart grid simulation, the method comprising the following steps: S1. Collect the operation data of the smart grid, including real-time monitoring data of voltage, current, power factor and temperature, and process the collected operation data; By collecting and processing the operation data of the smart grid, the accuracy of subsequent dynamic modeling, state estimation, risk assessment and other steps is ensured. The collection and processing of data directly determines the reliability of the early warning of the entire system. In this step, the system comprehensively collects and processes the real-time operation data of the power grid to ensure the quality and consistency of the data. High-quality data is the key to power grid state prediction and safety early warning. Therefore, data collection and preprocessing of the power grid are very important.

[0023] In this embodiment, the operation data collection of the smart grid mainly involves real-time monitoring data such as voltage, current, power factor, temperature, etc. of the power grid. The system monitors multiple key nodes of the power grid in real time through a sensor network and collects these operation data. In addition, the system also collects historical fault records, equipment status, maintenance logs and other data of the power grid to provide long-term data support for the operation of the power grid. Through the various types of data collected, the system can more comprehensively evaluate the health of the power grid.

[0024] In this embodiment, data collection is usually achieved through sensors installed at each node of the power grid. These sensors monitor and record various operating parameters of the power grid in real time. Specifically, the voltage sensor measures the voltage of each node in the power grid, the current sensor measures the current, the power factor sensor measures the relationship between the power grid load and the power supply, and the temperature sensor monitors the operating temperature of the power grid equipment (such as transformers, switchgear, etc.). These sensors can continuously and stably collect data and transmit it to the central processing unit of the system through the communication network.

[0025] Specifically, the acquisition of parameters such as voltage, current, power factor and temperature can be basically calculated using the following formula: ; in: Indicates the power grid node at time Power (in watts); For the grid node at time The voltage value (unit: volt); For the grid node at time The current value (unit: ampere); is the phase angle between voltage and current; is the power factor, which indicates the phase difference between voltage and current.

[0026] This formula shows how to calculate the power of the grid, which is obtained by multiplying the voltage, current and power factor. The system calculates the power of each node in real time as an important indicator of the health of the grid.

[0027] In the processing stage after data collection, the system pre-processes the data through a variety of methods. First, the system removes noise. Due to factors such as sensor errors and electromagnetic interference, the collected signals may contain noise, which will affect the subsequent analysis results. Therefore, it is necessary to use filtering algorithms (such as Kalman filtering, mean filtering, etc.) to remove noise and ensure the accuracy of the data.

[0028] In one possible implementation, the system optimizes and corrects the collected data through the Kalman filter algorithm. Kalman filtering is a recursive estimation algorithm that can provide optimal state estimation in a noisy environment. By combining the system model and measurement data, Kalman filtering can effectively filter out external noise and improve data quality.

[0029] For power grid data processing, the core of the Kalman filter is to make the best estimate of the system state by updating the equation. The system updates the power grid state through the following formula: ; in: For the power grid at all times The updated estimated state of Indicates time Estimated vector of grid status (such as voltage, current, power, etc.); is the Kalman gain, and the calculation formula is: ; in: is the actual measurement data; is the predicted value of the grid state (calculated by the system model); It's time The covariance matrix of the state estimate; is the covariance of the measurement noise; is the measurement matrix; represents the transpose of the measurement matrix HHH.

[0030] Through this formula, the system can update the state estimate of the power grid based on new measurement data and reduce the impact of noise on the system estimate.

[0031] Next is outlier detection. Grid equipment may fail or sensors may have problems, resulting in abnormal data. The system needs to detect and exclude these abnormal data to prevent them from affecting subsequent decisions. Outlier detection can be done by setting thresholds. For example, if a data value deviates too much from the normal range, it will be considered an outlier and removed.

[0032] In this case, the system also needs to normalize the data. Because the magnitudes of different types of data such as voltage and current vary greatly, if they are analyzed directly, some parameters may have too much influence on the final results. By normalizing the data, parameters of different magnitudes can be converted into a unified standard range to ensure that each data item has an equal impact on the analysis results.

[0033] For example, suppose the voltage value collected by a sensor is , the current value is , then the following normalization formula can be used: ; in: and are the minimum and maximum values ​​of the voltage data respectively; and are the minimum and maximum values ​​of the current data respectively; are the normalized voltage and current data, and their value range is [0, 1].

[0034] Through normalization, the voltage and current data will be mapped to a uniform scale range, making subsequent model analysis more accurate.

[0035] As an option, in some embodiments, the system may also combine time series analysis methods, such as Fourier transform or wavelet transform, to perform frequency domain analysis on the current and voltage waveforms of the power grid to identify periodic changes in the power grid. In this way, the system can effectively identify vibrations of power grid equipment or fluctuations in current load, thereby providing more detailed monitoring of the health status of the power grid.

[0036] Through data cleaning, denoising, outlier detection and normalization processing, the system can avoid false alarms and missed alarms caused by poor data quality. Through strict data quality management, it provides solid data support for subsequent model predictions, risk assessments, etc.

[0037] S2, using the extended Kalman filter algorithm to optimize the estimation of the power grid state; The state estimation of the power grid is optimized by the extended Kalman filter (EKF) algorithm. The Kalman filter can provide the best estimation in a noisy data environment, thus providing more accurate information for the health status prediction of the power grid.

[0038] In this embodiment, the extended Kalman filter algorithm (EKF) is applied to the nonlinear state estimation of the power grid. Since the power grid system is a nonlinear dynamic system, the Kalman filter algorithm is more effective than the traditional linear Kalman filter in processing nonlinear systems. Through the extended Kalman filter, the system can eliminate the influence of noise on the state estimation while monitoring the power grid state in real time, thereby improving the accuracy of the state estimation.

[0039] In one possible implementation, the system first predicts the state of the power grid based on a nonlinear dynamic model of the power grid. The model describes the relationship between various state variables of the power grid, such as voltage, current, power factor, etc. The model can use the following state update equation: ; in: It represents the power grid system at time Status; It is the grid state vector, which includes state parameters such as voltage, current, and power; is the control input vector, which includes control parameters such as grid dispatch and load management; It is a nonlinear function of the power grid system, describing the dynamic relationship of the power grid state evolving over time; It is the random disturbance term of the system, simulating the impact of external factors on the power grid state, such as equipment failure, environmental changes, etc.

[0040] In this prediction model, the update of the grid state is determined by the dynamic characteristics of the grid and the control input, and the impact of external disturbances needs to be considered. Therefore, the system predicts the grid state through this dynamic model.

[0041] In the update step of the extended Kalman filter, the system uses real-time measurement data to correct the predicted grid state. Specifically, the system calculates the Kalman gain by comparing the difference between the predicted value and the actual measurement data, and uses it to adjust the estimate of the grid state.

[0042] The calculation of the Kalman gain determines how to combine the measured data with the predicted data. The larger the Kalman gain, the higher the system's trust in the current measured value.

[0043] By weighting the Kalman gain, the system can effectively fuse the predicted value and the actual measurement data to obtain a more accurate estimation of the power grid state.

[0044] Through this updating process, the system can adjust its estimate of the grid state in real time, gradually correct the prediction error, and improve the accuracy of the state estimation.

[0045] In the extended Kalman filter, the covariance matrix It reflects the uncertainty of the system's estimation of the power grid state. By updating the covariance matrix, the system can dynamically adjust the trust in predictions and measurements. The covariance matrix update formula is as follows: ; in: Indicates time The covariance matrix of the updated state estimate; It's time The covariance matrix of the state estimate; is the unit matrix, which is used to represent the unit transformation when the covariance matrix is ​​updated; is the Kalman gain; is the measurement matrix.

[0046] By updating the covariance matrix, the system can reflect the reliability of the current estimation results and further improve the estimation accuracy of the power grid status.

[0047] Through the extended Kalman filter algorithm, the system can handle the nonlinear changes of the power grid state in real time and optimize the estimation of the power grid state in the presence of noise. By combining the measurement data and the prediction model, the extended Kalman filter greatly improves the accuracy of the power grid state estimation. Especially when the power grid faces external disturbances or measurement errors, the extended Kalman filter can effectively reduce uncertainty and improve the stability and reliability of the system.

[0048] S3. Establish a nonlinear dynamic model of the smart grid and predict the health status of the grid based on the dynamic behavior of the grid; In step S3, the nonlinear dynamic model of the power grid is an accurate description of the complex behavior of the power grid. The state of the power grid usually changes over time, including various key parameters such as voltage, current, and power. The relationship between these states is nonlinear and cannot be simply described by linear equations. Therefore, nonlinear dynamic equations must be used to accurately simulate the changes in the power grid.

[0049] In general, the nonlinear dynamic equations of the power grid can be constructed by considering the interactions among various state parameters of the power grid, control inputs (such as load scheduling, backup power control), and external disturbances (such as equipment failures, environmental factors, etc.).

[0050] The nonlinear dynamic characteristics of the power grid are manifested in many aspects, such as the relationship between voltage and current, the dependence between power and load, etc. In order to accurately describe these relationships, the system uses nonlinear equations to simulate the state evolution of the power grid. For example, the relationship between voltage, current and power can be expressed by a nonlinear function of the following form: ; in: Indicates the power grid system at time The nonlinear dynamic evolution of describes the change of the power grid state over time; For the power grid at all times The state vector of For the control input vector, it includes external factors or operation instructions that affect the grid state; Indicates the first where n is the nonlinear dynamic evolution function of the state variables in the power grid, i.e., the dimension of the power grid.

[0051] The dynamic behavior of the power grid is not only affected by the control input, but also by external disturbances. External disturbances include equipment failures, sudden load increases, weather changes, etc. These factors are usually unpredictable, so disturbance terms need to be added to the model. to simulate these effects.

[0052] In some embodiments, the disturbance term It can be modeled by a random process, such as the Gaussian noise model: ; in: Indicates at time The external disturbance term usually represents factors such as equipment failure, grid load change or environmental interference; represents a mean of 0 and a variance of Gaussian noise with variance Controls the intensity of the perturbation.

[0053] The introduction of this random disturbance term makes the model more realistic and can take into account various uncertainties that the power grid may encounter in actual operation.

[0054] In order to ensure the stability of the power grid state estimation, the system applies the Lyapunov stability theory to analyze the stability of the power grid model. The Lyapunov function is usually used to evaluate the stability of a system when it is disturbed. In the dynamic model of the power grid, the Lyapunov function can be defined as: ; in: is the Lyapunov function, which represents the energy of the grid state or the system stability; It is the state vector of the power grid, which contains the key state parameters of the power grid (such as voltage, current, etc.); is a positive definite matrix, usually defined according to the characteristics of the power grid. It is used to describe the stability of the system; is the grid state vector The transpose of ensures that the Lyapunov function is a scalar; This is a quadratic expression that represents the specific form of the Lyapunov function.

[0055] The core of Lyapunov stability analysis is to determine whether the system is stable by calculating the derivative of the Lyapunov function. If the derivative is less than zero, it means that the power grid system is stable under external disturbances; if it is greater than zero, the system may be unstable. Therefore, the derivative of the Lyapunov function provides important information about the stability of the system.

[0056] In order to further improve the prediction accuracy of the power grid status, the system can use machine learning methods to optimize nonlinear dynamic models. For example, by using technologies such as support vector machines (SVM) or deep neural networks (DNN), the model can be trained with historical data to automatically optimize the parameters of the power grid dynamic model, thereby enhancing the adaptability and accuracy of the power grid model.

[0057] Specifically, machine learning methods can help the system automatically identify the complex relationship between grid status and control inputs, reduce the complexity of manual modeling, and improve the predictive ability of the model.

[0058] By introducing nonlinear dynamic models, the system can more accurately simulate the actual behavior of the power grid, especially in the face of complex factors such as load fluctuations, equipment failures and environmental changes. Nonlinear models can provide more accurate predictions of power grid status. and Lyapunov stability analysis ensure the robustness and stability of the power grid model, and can maintain high prediction accuracy in an environment with strong uncertainty.

[0059] Specifically, by introducing external disturbance simulation, the prediction of the health status of the power grid is no longer limited to idealized scenarios, but can take into account the uncertainties in actual operation. This makes the health assessment of the power grid more comprehensive and accurate, and can effectively identify potential risks.

[0060] Through Lyapunov stability analysis and disturbance term simulation, the system ensures the stability and robustness of the power grid model. The introduction of machine learning methods to optimize nonlinear dynamic models further improves the accuracy of power grid state prediction. Ultimately, these technologies provide reliable data support for subsequent risk assessment and early warning, ensuring the safety and stability of power grid operation.

[0061] S4. Construct a Bayesian network model to evaluate the risk of the power grid through probabilistic reasoning and determine the health status of the power grid; S4 performs risk assessment by constructing a Bayesian network model between grid state variables. A Bayesian network is a graphical model based on probabilistic reasoning that is used to represent the conditional dependencies between random variables and calculate the risk of the grid under different operating conditions through probabilistic reasoning.

[0062] Bayesian networks use directed acyclic graphs (DAGs) to represent dependencies between random variables. In power grids, there may be complex interdependencies between state variables, such as the relationship between current, voltage, and power factor. Therefore, through Bayesian networks, the system can efficiently handle these dependencies and perform conditional probabilistic reasoning.

[0063] The core of the Bayesian network model lies in the calculation of joint probability distribution. The joint probability distribution of all state variables in the power grid can be expressed by the conditional probability distribution of each state variable. Specifically, the joint probability distribution of the power grid state variables is It is expressed by the following formula: ; in: Indicates the power grid status The joint probability distribution of reflects the overall risk of the power grid in different states. The joint probability distribution can help assess the overall risk of the power grid state; Represents a specific state variable in the power grid (such as voltage, current, power factor, etc.). These state variables determine the health of the power grid; Represents state variables The parent node set refers to the state variable Other related state variables. These parent nodes describe the conditional dependencies between various state variables of the power grid; Indicates that at a given parent node Under the condition that The conditional probability distribution of each state variable is estimated through the historical data and real-time data of the power grid, reflecting the risks under different states.

[0064] Each node in the Bayesian network represents a grid state variable. The conditional probability distribution of each state variable can be learned from historical data or estimated from the existing grid physical model. The conditional probability distribution may depend on the current and power factor ,Right now: ; in: Indicates that at a known current and power factor In the case of The conditional probability distribution of ; It is the voltage state variable in the power grid, reflecting the voltage health status of the power grid; and : The current and power factor state variables in the power grid affect the change of voltage respectively.

[0065] The Bayesian network model calculates the risk of the power grid through an inference algorithm. This process usually includes forward reasoning and backward reasoning. The goal of reasoning is to infer the risk of each state variable of the power grid based on known data, and then evaluate the overall risk of the power grid.

[0066] Specifically, if the system obtains real-time monitoring data (e.g., voltage, current, etc.), the Bayesian inference algorithm will update the probability distribution of the power grid state based on this data and calculate the risk probability of the power grid. , the system is able to assess the health of the grid and identify potential risks.

[0067] In order to classify the health status of the power grid, the system divides the risk level of the power grid into multiple levels. By calculating the risk probability of the power grid status, the system can automatically determine the current risk level of the power grid. For example, the risk level of the power grid can be divided into three levels: low risk, medium risk and high risk. The specific classification criteria are as follows: Low risk level: When the risk probability of the power grid is less than 10%, it means that the power grid is in normal operation.

[0068] Medium risk level: When the risk probability of the power grid is between 10% and 30%, it means that the power grid may have certain risks and needs to be closely monitored.

[0069] High risk level: When the risk probability of the power grid exceeds 30%, it means that the power grid is at high risk and immediate measures need to be taken.

[0070] To ensure the accuracy of grid risk assessment, the probability distribution in the Bayesian network is learned through historical data and real-time data. Specifically, the system can use methods such as the expectation maximization (EM) algorithm or variational inference to train the model to estimate the conditional probability between grid state variables.

[0071] During the reasoning process, the system calculates the overall risk of the power grid based on the existing power grid status (such as voltage, current, power factor, etc.) and the learned probability distribution.

[0072] By using the Bayesian network model for risk assessment, the system can effectively quantify the health of the power grid and assess the risk level of the power grid under different operating conditions. The Bayesian network can take into account the complex relationship between power grid state variables and update based on real-time data, thereby providing accurate data support for the early warning response of the power grid.

[0073] S5. Based on the Nash equilibrium model in game theory, the behaviors of multiple decision-making entities in the power grid are collaboratively optimized to achieve global stability of the power grid; S5 optimizes the behaviors of multiple decision-making entities (such as load scheduling, generator scheduling, etc.) in the power grid by using the Nash equilibrium model in game theory to achieve global stability of the power grid. The application of game theory models in the power grid aims to solve the problems of cooperation and conflict between multiple decision-making entities and optimize the strategies of each entity so that the power grid can maintain stable and efficient operation under different operating conditions.

[0074] Nash equilibrium in game theory refers to a system with multiple decision-makers. When all decision-makers choose their own strategies, no one decision-maker can improve its benefits by unilaterally changing its strategy. The behavior of each decision-maker will affect other decision-makers, so the system needs to achieve a global balance to ensure the optimal strategy of each decision-maker.

[0075] The behavior of each decision-making subject can be represented by a strategy set. Assume that there are decision-making entity, each decision-making entity Choose a strategy , and the strategy combination of all subjects is recorded as .

[0076] Each decision-making subject in the power grid has a payoff function (or utility function) that is used to evaluate the consequences of choosing a strategy given the strategies of other subjects. The payoff function of each decision-making subject can be expressed as: ; in: : Indicates the decision-making subject Given other decision-making subject strategies Next, select Strategy the benefits or utility of For decision-making body The strategy represents the different operations or behaviors that the decision-making subject can choose. In the power grid, the strategy of load dispatch may be the load distribution plan, and the strategy of generator dispatch may be the adjustment plan of power generation; is the strategy combination of all decision-making entities in the power grid, representing the set of behavioral decisions of all entities; For decision-making body The profit function of .

[0077] The basic definition of Nash equilibrium is: in a game, if the strategy of each decision-maker is optimal (that is, if the strategies of other decision-makers remain unchanged, changing their own strategies will not lead to higher returns), then this strategy combination is a Nash equilibrium. The specific mathematical expression is: ; in: It represents the optimal strategy of each decision-making subject under the Nash equilibrium state; Represents the strategy combination of all decision-making entities in the power grid; Represents decision-making subject The return under a certain strategy combination.

[0078] Specifically, when the strategies of all decision-making entities reach Nash equilibrium, the behaviors of the various decision-making entities in the power grid will not get better results due to the unilateral change of strategy by one entity, so the power grid is in a stable state.

[0079] The Nash equilibrium in game theory needs to be solved through optimization algorithms. Optimization algorithms (such as gradient descent, genetic algorithms, simulated annealing, etc.) can be used to solve the optimal strategy of each decision-making entity. Through continuous iteration, the strategy of the decision-making entity will gradually adjust until the Nash equilibrium is reached.

[0080] As an option, we can use global optimization methods such as simulated annealing to repeatedly try different strategy combinations and gradually approach the optimal solution, thereby obtaining the global optimal stable state of the power grid.

[0081] In a power grid, the behaviors of multiple decision-makers may affect each other. For example, the decision on load dispatch may affect the strategy of generator dispatch, and vice versa. Game theory models optimize the strategy of each decision-maker so that the power grid can remain stable when facing different operating states.

[0082] The goal of load dispatch may be to minimize load fluctuations and grid operation costs.

[0083] The goal of generator scheduling may be to schedule the operation of generators according to the grid load demand to ensure the stability and optimal distribution of power supply.

[0084] By solving the Nash equilibrium of the game model, the system can coordinate the behaviors of these different decision-makers and avoid local optimality that affects the overall stability of the power grid.

[0085] Through the Nash equilibrium model in game theory, the system can optimize the various dispatching decisions of the power grid while evaluating the risks of the power grid, ensuring that the power grid can respond quickly and restore stability when faced with emergencies. For example, the power grid may face load surges, generator failures, etc. Through the decision-making mechanism optimized by game theory, the power grid can intelligently dispatch generators and loads to minimize system risks and ensure stable power supply.

[0086] Specifically, through this game optimization mechanism, load scheduling and generator scheduling can optimize resource allocation and control strategies without sacrificing grid stability, thereby improving the operating efficiency and emergency response capabilities of the grid.

[0087] By optimizing the behaviors of multiple decision-makers in the power grid based on Nash equilibrium in game theory, the system can effectively avoid conflicts between local decisions and global goals, and ensure the global stability of the power grid. The Nash equilibrium model enables multiple decision-makers to coordinate and cooperate to jointly optimize the operation status of the power grid, thereby reducing the uncertainty and risks in the operation of the power grid and ensuring the safe, stable and efficient operation of the power grid.

[0088] Specifically, game-theoretic optimization mechanisms can: Coordinate the behaviors of different decision-makers (such as load scheduling, generator scheduling, etc.) to avoid overall instability caused by local optimal solutions; Improve the emergency response capability of the power grid to ensure that the power grid can quickly restore stability in the event of an emergency; Optimize resource allocation, reduce grid operating costs, and improve overall efficiency.

[0089] By building a game model and solving Nash equilibrium, the system can optimize decisions such as load scheduling and generator scheduling to ensure efficient operation of the power grid. This method provides a strong theoretical basis and technical support for the intelligent scheduling and safety management of the power grid.

[0090] S6. Use the maximum entropy principle in information theory to optimize the setting of warning levels and information flow transmission strategies, adjust the information allocation and transmission methods of each warning level, and ensure effective information flow; The maximum entropy principle is used to optimize the setting of the power grid warning level and the transmission strategy of information flow. This principle maximizes the effective transmission of information under different risk scenarios by maximizing the uncertainty of the system under known constraints, avoiding the transmission of redundant information, and optimizing the response capability of the power grid system.

[0091] The core idea of ​​the maximum entropy principle is to optimize information flow by selecting a probability distribution that maximizes information uncertainty. The application of this principle can optimize the distribution of warning levels and effectively transmit key information, reducing information overload and delay.

[0092] Information Entropy It measures the uncertainty of information, or the possible redundancy in information transmission. In the power grid early warning system, the size of entropy reflects the efficiency and accuracy of information transmission. The maximum entropy principle requires that under known constraints, the probability distribution of maximum entropy be selected so that the system can transmit as much information as possible while avoiding redundancy.

[0093] The information entropy calculation formula is as follows: ; in: It represents the entropy of the power grid warning information and measures the total uncertainty of the information. If the information entropy in the system is high, it means that the system contains a large amount of information, the uncertainty is high, and the efficiency and diversity of information transmission are also high; Indicates warning information item Different warning information items represent different power grid risk states. Indicates the probability of occurrence of these risk information; Representation information item The amount of information is inversely proportional to the probability of an event. The smaller the probability, the greater the amount of information. Through entropy calculation, the system can quantify the possibility of different power grid states and effectively allocate information at different warning levels.

[0094] By calculating the maximum entropy, the system can ensure that the most valuable information is transmitted in the early warning mechanism of the power grid and avoid the redundant transmission of useless information.

[0095] During the risk assessment process of the power grid, the system will dynamically set different warning levels according to the real-time health status of the power grid. Each warning level represents the response level of the power grid under a certain risk state and matches the health index and risk probability of the power grid.

[0096] The maximum entropy principle optimizes the setting of warning levels through the following formula: ; in: Indicates time The warning level of the power grid is dynamically set based on the risk assessment results and information entropy value to ensure that the warning response can be targeted at the actual risk status of the power grid; Indicates time The grid risk assessment results may include the health index of the grid, the probability of various risks, etc. The grid risk assessment results are the key basis for adjusting the warning level; Indicates time The information entropy of the grid is used to measure the redundancy and amount of information in the grid early warning information. A higher entropy value means that the uncertainty of the grid state is higher and more risk information needs to be transmitted in a timely manner.

[0097] By maximizing information entropy, the system can optimize information distribution and prioritize the delivery of important and urgent risk information to relevant decision-makers, ensuring that the power grid's early warning mechanism is both efficient and accurate.

[0098] In order to improve the transmission efficiency of power grid early warning information, the system adopts the maximum entropy principle to optimize the information flow transmission strategy. The optimization of information flow not only improves the speed at which decision-makers receive early warning information, but also ensures the accuracy of the information.

[0099] Information flow ( ) is a measure of the efficiency of information transmission in the power grid. The information flow calculation formula is as follows: ; in: Indicates time The information flow rate measures the efficiency of information transmission in the power grid. The larger the information flow rate, the more efficient the information transmission; Representation information item The probability distribution of , indicating the importance or urgency of the information item; Representation information item The transmission time or propagation speed of information. This parameter reflects the transmission delay of information and is usually affected by the power grid communication network and information channels.

[0100] The system dynamically optimizes the transmission path of information flow through the maximum entropy principle to ensure that when an abnormality occurs in the power grid, information can be quickly and accurately transmitted to relevant decision-makers, reducing information delays and redundancy and improving the emergency response capabilities of the power grid.

[0101] By introducing the maximum entropy principle, the system can effectively optimize the setting of warning levels and accurately adjust the information flow transmission strategy. Specifically, the maximum entropy optimization mechanism can bring the following advantages: By maximizing information entropy, the system ensures that critical information is delivered first, avoiding information redundancy and overload.

[0102] The system can dynamically adjust the warning level based on the health status and risk assessment of the power grid, ensuring that the power grid can make the most appropriate response when facing different risk levels.

[0103] By optimizing the transmission path of information flow, the system ensures that power grid managers can obtain the most critical information in the shortest time and make decisions quickly.

[0104] By calculating information entropy and dynamically adjusting the distribution and transmission of early warning information, the system can effectively improve the efficiency and accuracy of power grid emergency response. The application of the maximum entropy principle provides an efficient optimization solution for power grid risk management, which helps to reduce the transmission of redundant information when emergencies occur, ensure that the most critical risk information can be transmitted in a timely manner, and thus ensure the safe and stable operation of the power grid.

[0105] S7, based on adaptive optimal control and deep reinforcement learning, adjust the operation strategy of the power grid in real time and optimize the early warning response; By introducing adaptive optimal control and deep reinforcement learning (DRL), the goal is to enable the power grid to dynamically adjust its operation strategy based on real-time data. This strategy can effectively deal with uncertain load fluctuations, failures of power grid equipment and other emergencies, and optimize the performance and stability of the power grid. The relevant formulas and parameters are explained in detail below.

[0106] The optimization objective function used by adaptive optimal control in power grid operation is usually expressed as a cost function or a performance function. Its main purpose is to adjust the control strategy in real time through a feedback mechanism to ensure the optimal operation state of the power grid. The optimization objective can be expressed as the following integral formula: ; in: represents the optimization objective, i.e. the integral of the grid operation cost or system efficiency. Usually the goal is to minimize this value; For the power grid at all times The state vector contains key information of the power grid, such as voltage, current, load, etc. is the state penalty matrix, which reflects the impact of the grid state on the cost and is usually used to emphasize the importance of certain key states (such as voltage and current); To control the input vector, it includes the dispatching strategy of the power grid, such as generator dispatching, load dispatching, etc. The control input penalty matrix reflects the impact of the control strategy on the cost or efficiency of power grid operation and is often used to adjust the cost of control actions.

[0107] The optimization goal is to minimize the operating cost and maintain the stability and efficiency of the power grid. Specifically, adaptive optimal control adjusts the state vector and control input To optimize the operation status of the power grid.

[0108] Deep reinforcement learning combines the advantages of deep learning and reinforcement learning and can handle high-dimensional and complex environments. In power grid dispatching, deep reinforcement learning optimizes decision-making strategies through environmental feedback. The reinforcement learning model of the power grid is usually expressed as the following Q-learning model: ; in: Indicates in status Take action The optimal action value function of , that is, the long-term reward of the decision; Represents the state vector of the power grid, including various real-time data of the power grid, such as voltage, current, load, etc.; Indicates the action or control strategy taken under a given state, such as adjusting the load, starting the backup power supply, etc.; Indicates immediate rewards, usually related to how well the grid operates, cost savings, stability, etc.; is the discount factor, which controls the importance the system places on future rewards. The larger the discount factor, the stronger the system's focus on long-term goals; For current action After that, the next state reached by the power grid; In the new state Next, all possible actions.

[0109] Specifically, in the operation of the power grid, the deep reinforcement learning model continuously adjusts the dispatching strategy by combining exploration and utilization. To maximize long-term benefits and ensure the efficient and stable operation of the power grid.

[0110] The combination of adaptive optimal control and deep reinforcement learning enables the power grid to adapt to changing load demands and emergencies in real time. In general, the two methods each play to their strengths: adaptive control provides a framework for stability and efficiency, while deep reinforcement learning improves decision-making quality and adaptive capabilities by continuously learning the behavior of the power grid.

[0111] In some embodiments, the objective function combining adaptive optimal control and deep reinforcement learning may be as follows: ; in: is the joint optimization objective, which represents the goal of optimizing the power grid operation under the consideration of optimal control and reinforcement learning.

[0112] is a weighting factor used to balance the impact of adaptive optimal control and deep reinforcement learning. This factor controls the trade-off between the control strategy adjustment and deep reinforcement learning adjustment.

[0113] Specifically, the joint optimization method enables the system to improve the effectiveness of the control strategy through reinforcement learning while ensuring stability, so that the power grid can better cope with dynamic changes and emergencies.

[0114] The core of deep reinforcement learning lies in the balance between exploration and utilization. Exploration refers to trying new and unknown operations, while utilization refers to making decisions using known optimal strategies. In the power grid, deep reinforcement learning continuously explores new scheduling strategies, finds the best control strategy, and gradually improves it in actual operation. In order to enhance the learning efficiency and long-term stability of the system, the system updates the Q value function through the following formula: ; in: The learning rate indicates the speed at which new information updates the model, and usually controls the impact of new information in each learning process. For current action After that, the next state reached by the power grid; Indicates the next state All possible actions under It is an instant reward that reflects the impact of the current grid strategy on the operating status.

[0115] As an alternative, deep reinforcement learning can be approximated by neural networks. , that is, using convolutional neural networks or fully connected neural networks to process large-scale data and complex state space of the power grid.

[0116] Step S7 combines adaptive optimal control with deep reinforcement learning to provide the power grid with the ability to adjust in real time, enabling it to respond quickly to dynamic loads and emergencies. Adaptive control ensures the stability and efficiency of the power grid, while deep reinforcement learning enables the power grid to efficiently respond to complex changes by continuously optimizing the dispatch strategy.

[0117] S8. Automatically trigger hierarchical safety warning responses based on the health status of the power grid and risk assessment results; In step S8, the system automatically triggers different levels of early warning responses based on the health status assessment and risk assessment results of the power grid. The power grid health status assessment is derived from the Bayesian network model and other prediction models in the previous steps, while the risk assessment combines real-time data and historical data. In order to trigger different early warning responses based on the assessment results, the system needs to establish a grading mechanism. The specific formula is as follows: ; in: Indicates time The warning level of the power grid is dynamically set by the system according to the health status of the power grid and the results of risk assessment, and is usually divided into low, medium and high levels; Indicates time The grid risk assessment result is usually an assessment value that comprehensively considers the current health status of the grid, potential risks, and historical data. The result is calculated by combining real-time data and historical data; Indicates time The information entropy of the grid measures the uncertainty and redundancy of the information about the current health status of the grid. Higher entropy indicates greater uncertainty and may require a higher priority early warning response.

[0118] This formula helps the system dynamically adjust the warning level based on real-time assessment results. Higher risk assessment values ​​and higher uncertainty (i.e., higher entropy values) usually trigger higher warning levels, indicating the need for more urgent emergency response.

[0119] By automatically triggering early warning responses, the system can automatically initiate early warnings and take corresponding emergency measures when abnormalities or potential risks occur in the power grid. According to the health status and risk assessment of the power grid, the system automatically adjusts the response strategy to ensure that the power grid quickly recovers stability in emergencies. The automated early warning response can be expressed as the following optimization formula: ; in: Indicates at time The control input or dispatch decision determines the early warning response measures that the system needs to take. Control input includes operations such as adjusting load and enabling backup power supply; It represents the cost of the response measures, which is usually related to the actual cost of implementing the response strategy, equipment scheduling, etc. The choice of control strategy aims to minimize the cost; is a regulatory factor that controls the weight of the impact of risk assessment and health status on the response strategy. This factor balances the effectiveness of the response measures with the actual cost; Indicates that control input is taken After that, the risk assessment and health status of the power grid affect the stability and security of the power grid. This item reflects the mitigation effect of different early warning responses on the risk of the power grid.

[0120] Through this formula, the system can flexibly select the optimal emergency response strategy based on the health status of the power grid and the results of risk assessment, which not only reduces the response cost of the system but also ensures the safety and stability of the power grid.

[0121] During the operation of the power grid, the health status of the power grid and risk assessment are closely related, and the assessment of the health status is usually the basis of risk assessment. Through real-time monitoring and assessment of the power grid status and operation data, the system can promptly identify potential safety hazards and automatically trigger the response mechanism. The formula for combining the health status assessment of the power grid with the risk assessment is usually as follows: ; in: is the risk assessment result of the power grid, indicating the time The risk level of the power grid. The risk assessment of the power grid is determined by factors such as the health status of the power grid, load demand, and equipment status; Real-time monitoring data representing the state of the power grid, such as voltage, current, frequency, etc. These data reflect the current health of the power grid; Represents historical data of the power grid, often used for trend analysis and forecasting to help identify potential risks.

[0122] Specifically, the system uses real-time data and historical data Conduct a comprehensive analysis to obtain the risk assessment value of the power grid at the current moment This value is the basis for triggering different levels of early warning responses.

[0123] In order to further optimize the accuracy and timeliness of the power grid early warning response, the system continuously optimizes the response strategy through deep learning and reinforcement learning methods. The optimization of the early warning response is continuously updated through the feedback mechanism. The goal of the power grid is to minimize the response time while ensuring the stability of the power grid. The response strategy of the power grid is adjusted through the following optimization formula: ; in: represents the target of the optimized response strategy, which aims to minimize the cost and improve the efficiency of grid operation; Indicates that response measures are being taken Cost at the time of The weighting factor is used to adjust the cost of the control input and the impact of the grid health status to ensure that the grid operates in the optimal state.

[0124] Through this optimization formula, the system can adjust emergency response strategies in real time to ensure that the power grid can recover quickly and remain stable, while optimizing the cost and resource allocation of the response.

[0125] By dynamically adjusting the response strategy, the power grid can quickly take appropriate measures when faced with emergencies and complex environments, thereby ensuring the efficient and safe operation of the power grid. The automated early warning response mechanism not only improves the emergency response speed and accuracy of the power grid, but also optimizes the resource allocation and cost management of the power grid.

[0126] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A hierarchical safety early warning method based on smart grid simulation, characterized in that: The method comprises the following steps: S1. Collect the operation data of the smart grid, including real-time monitoring data of voltage, current, power factor and temperature, and process the collected operation data; S2, using the extended Kalman filter algorithm to optimize the estimation of the power grid state; S3. Establishing a nonlinear dynamic model of the smart grid and predicting the health status of the grid according to the dynamic behavior of the grid; S4. Construct a Bayesian network model to evaluate the risk of the power grid through probabilistic reasoning and determine the health status of the power grid; S5. Based on the Nash equilibrium model in game theory, the behaviors of multiple decision-making entities in the power grid are collaboratively optimized to ensure the global stability of the power grid; S6. Use the maximum entropy principle in information theory to optimize the setting of warning levels and information flow transmission strategies, and adjust the information distribution and transmission methods of each warning level; S7, based on adaptive optimal control and deep reinforcement learning, adjust the operation strategy of the power grid in real time and optimize the early warning response; S8. Automatically trigger graded safety warning responses based on the health status of the power grid and risk assessment results.

2. A hierarchical safety early warning method based on smart grid simulation according to claim 1, characterized in that: The method of optimizing the estimation of the power grid state by using the extended Kalman filter algorithm includes: Optimize and estimate the nonlinear state of the power grid based on the extended Kalman filter algorithm; Provide real-time updates on the health of the grid, use grid measurement data and dynamic models to correct grid status predictions, and reduce the impact of noise on system estimates; By calculating the Kalman gain to adjust the error estimate of the system, the accuracy of grid state prediction is improved.

3. The hierarchical safety early warning method based on smart grid simulation according to claim 1 is characterized in that: The nonlinear dynamic model of the smart grid is established including: Using the operation data, the state vector of the power grid is constructed, and the dynamic relationship between the various states of the power grid is described by nonlinear functions; Introducing disturbance terms to simulate the impact of external factors on the power grid system, so that the power grid model can better adapt to different operating environments; The stability of the power grid model is analyzed using Lyapunov stability theory to ensure that the model remains stable under various disturbances.

4. The hierarchical safety early warning method based on smart grid simulation according to claim 1 is characterized in that: The Bayesian network model construction includes: By establishing the conditional dependency relationship between various parameters of the power grid, a Bayesian network of the power grid is constructed; According to the historical operation data of the power grid, the conditional probability distribution of each node is calculated, and its risk probability is inferred according to the current power grid status; Based on the inference results, the risk level of the power grid under different operating conditions is evaluated, and decision support for power grid operation is provided through Bayesian reasoning.

5. The hierarchical safety early warning method based on smart grid simulation according to claim 1 is characterized in that: The behaviors of multiple decision-making entities in optimizing the power grid based on the Nash equilibrium model in game theory include: Set up multiple control entities of the power grid, including load dispatch, generator dispatch and backup power supply, as participants in the game; Using Nash equilibrium solutions, the decision-making of each subject is optimized through game theory to achieve the global optimal grid operation state; Each entity adjusts its decision-making behavior according to the health status of the power grid and the results of risk assessment to ensure the stable operation of the power grid.

6. The hierarchical safety early warning method based on smart grid simulation according to claim 1 is characterized in that: The optimization of the setting of the warning level and the information flow transmission strategy using the maximum entropy principle in information theory includes: Based on the maximum entropy principle in information theory, the system can quickly transmit risk information by optimizing the information distribution of each warning level in the power grid; Set up corresponding information transmission mechanisms for different warning levels to ensure the effectiveness of information flow; Under different risk levels, the information distribution strategy is dynamically adjusted according to the status of the power grid and real-time data to reduce redundant information.

7. The hierarchical safety early warning method based on smart grid simulation according to claim 1 is characterized in that: The operation strategy of adjusting the power grid in real time based on adaptive optimal control and deep reinforcement learning includes: Introducing deep reinforcement learning algorithms, using real-time data from the power grid to train intelligent decision-making systems and adaptively adjust the control strategy of the power grid; Combined with optimal control theory, the emergency response mechanism of the power grid is optimized by learning the operation mode and risk status of the power grid in real time; Through reinforcement learning training, the system can continuously optimize the timing of warnings and response intensity, reducing false alarms and missed alarms.

8. The hierarchical safety early warning method based on smart grid simulation according to claim 1 is characterized in that: The automatic triggering of graded safety warning responses based on the health status of the power grid and the risk assessment results includes: According to the evaluation results of the health status of the power grid, the corresponding warning level is automatically selected, and the warning response is triggered through the intelligent decision-making system; In high-risk situations, the system can automatically adjust the grid's operating strategy, including enabling backup power sources and adjusting load distribution; In low-risk situations, the system is able to choose non-intervention responses to keep the grid stable.

9. The hierarchical safety early warning method based on smart grid simulation according to claim 1 is characterized in that: The method predicts the risk of the power grid based on the power grid simulation model and further includes: Using dynamic power grid simulation models, combined with historical power grid data and real-time operation data, potential risks of the power grid can be predicted from multiple angles; Through big data analysis technology, we simulate and predict the risk scenarios of different failure modes and equipment failures, and identify in advance the security threats that the power grid may face.

10. The hierarchical safety early warning method based on smart grid simulation according to claim 1 is characterized in that: The feedback mechanism of the grid health status and risk assessment results includes: Periodically update the health status and risk assessment results of the power grid through real-time monitoring and data feedback systems; The system continuously adjusts early warning strategies and response behaviors based on a feedback mechanism to ensure that the power grid can flexibly respond to emergencies in different operating environments.

Images