Intelligent dispatching system of transformer area power supply grid

Abstract

The invention relates to the technical field of electric power design, in particular to an intelligent dispatching system of a transformer area power supply grid. According to the method, a self-adaptive sampling strategy of real-time load fluctuation and distributed energy output characteristics is adopted; when the system detects the rapid change or abnormal fluctuation of the load, the data acquisition frequency is automatically improved, so that the capture of key information is ensured; when the load is stable, the sampling frequency is reduced, and the collection of redundant data is reduced; adaptive sampling can dynamically adjust the acquisition frequency, it is guaranteed that important data are not lost at the critical moment, and meanwhile the system burden is reduced; a hierarchical data processing architecture is adopted, data processing is divided into an edge layer and a cloud layer, the edge layer is responsible for conducting primary processing on data collected in a transformer area, including denoising, feature extraction and the like, and the data quality is guaranteed; only the key data subjected to primary processing is uploaded to a cloud end for data analysis and decision optimization; therefore, the burden of data transmission and processing is reduced, and the real-time response capability of the scheduling system is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of electric power design, and in particular to an intelligent dispatching system for a power grid in an area. Background Art

[0002] Power dispatching is an effective management method used to ensure the safe and stable operation of the power grid, reliable external power supply, and orderly progress of various power production operations.

[0003] Chinese Patent Publication No.: CN112713603A discloses an intelligent power supply device on the low-voltage side of a power grid station, including a power generation system, a power line, a transformer and a plurality of distributed energy storage systems; by setting up a plurality of distributed energy storage systems, when the power supply pressure in the peak period is large in the load-saturated area, the power supply pressure of the power grid equipment can be shared, thereby avoiding the power grid equipment from being in a heavy load or overloaded operating state for a long time and extending its service life; further, the load level of the power grid equipment in the load valley period can be increased, thereby achieving peak shaving and valley filling, and improving the utilization rate of the power grid equipment; further, by installing a distributed energy storage system on the low-voltage side, making full use of the peak-to-valley difference of the power grid for charging and discharging and adjusting the power factor of the converter, the problem of low voltage in the distribution network in some remote mountainous areas can be effectively solved.

[0004] The current power supply dispatching system for substations includes the above-mentioned schemes, which usually use fixed sampling frequencies and preset monitoring points for data collection. However, the power load and distributed energy output in the substation often change rapidly due to weather changes, residents' electricity consumption behavior, and other unexpected factors. The fixed sampling frequency cannot capture these changes in real time, resulting in a lag in the dispatching system and the inability to adjust the power supply strategy in time. At the same time, during periods of relatively stable loads, fixed-frequency sampling often collects a large amount of redundant data, which will occupy a large amount of storage and transmission resources and increase the burden of data processing. On the other hand, when the load changes rapidly, fixed-frequency sampling may miss key information, resulting in insufficient data analysis and inability to support effective dispatching decisions. It cannot be adjusted according to the actual characteristics of different substations. This one-size-fits-all approach is difficult to meet the specific needs of different substations, resulting in insufficient accuracy and efficiency of dispatching decisions. Therefore, there is an urgent need for an intelligent dispatching system for the substation power supply grid to solve such problems. Summary of the invention

[0005] To this end, the present invention provides an intelligent dispatching system for a power supply grid in a substation area, so as to overcome the problems in the prior art of fixed sampling frequency, which is unable to capture dynamic changes in real time, resulting in a lag in the dispatching system and an inability to adjust the power supply strategy in time; when the load is relatively stable, a large amount of storage and transmission resources will be occupied, increasing the burden of data processing, and key information will be missed when the load changes rapidly, resulting in insufficient data analysis and an inability to support effective dispatching decisions; and the problem of being unable to make adjustments according to the actual characteristics of different substations.

[0006] To achieve the above object, the present invention provides an intelligent dispatching system for a power grid in an area, comprising:

[0007] The data acquisition module dynamically adjusts the data acquisition frequency according to the real-time power load volatility and the output characteristics of distributed energy. It adopts an adaptive sampling strategy to collect data, and the data collected by the adaptive sampling strategy is directly transmitted to the hierarchical data processing module for preliminary processing and analysis. This enables timely capture of key data when the power load changes or the output of distributed energy fluctuates. The sampling frequency is reduced when the load is stable, thereby reducing redundant data collection and reducing the storage and processing burden of the system.

[0008] The layered data processing module includes two layers, the edge layer and the cloud layer. The edge layer is responsible for the preliminary processing of the data collected in the substation area. Only the data that has been preliminarily processed is uploaded to the cloud layer for analysis and decision optimization. The processed data will be uploaded to the cloud layer, i.e. the cloud. The edge-processed data provides support for the optimization decision module based on the characteristics of the substation area.

[0009] The optimization decision module uses data from edge computing and cloud analysis to deeply explore and analyze the characteristics of different substations and generate personalized scheduling strategies. The scheduling strategies generated by the optimization decision module will be directly input into the dynamic scheduling response module to perform scheduling adjustments in real time.

[0010] The dynamic dispatch response module makes dynamic dispatch adjustments based on the dispatch strategy generated by the optimization decision module and combined with real-time monitoring data. It uses the adaptive dynamic programming ADP algorithm to adjust the dispatch strategy according to real-time conditions. It quickly responds to the fluctuations of distributed energy and load changes to ensure the stability of the power grid. The dynamic dispatch execution results are fed back to the distributed collaborative dispatch module for coordination and optimization at the global level.

[0011] The distributed collaborative dispatching module divides the substation power grid into multiple autonomous sub-regions. Each region makes independent localized dispatching decisions while coordinating at the global level. This ensures the overall stability and efficiency of the power grid and supports rapid response to local fluctuations.

[0012] Furthermore, the data acquisition module includes:

[0013] Real-time monitoring unit, which monitors real-time data of power load and distributed energy output to identify rapid load changes and abnormal fluctuations;

[0014] Adaptive adjustment unit dynamically adjusts the sampling frequency according to the monitoring results, accurately captures key data and reduces redundant data;

[0015] The hierarchical data processing modules include:

[0016] The edge layer is used to perform preliminary data processing on the data;

[0017] The cloud layer, where the preliminarily processed data is uploaded to, where it is analyzed and global decision-making is optimized;

[0018] The optimization decision module optimizes based on the characteristics of the substation, and the dispatch strategy dynamically adjusts the dispatch based on the load characteristics of each substation and the access status of distributed energy to achieve optimal resource allocation and fault prevention;

[0019] The optimization decision module includes:

[0020] Data mining unit, which uses machine learning algorithms to analyze historical data and real-time status of each substation to identify patterns and anomalies;

[0021] The personalized scheduling unit generates personalized scheduling strategies for each substation based on data mining results and dynamically adjusts resource allocation;

[0022] The dynamic scheduling response module includes:

[0023] Real-time scheduling unit, based on ADP algorithm, dynamically adjusts scheduling strategy;

[0024] Rapid response unit, which quickly adjusts dispatch when load changes or fluctuations in distributed energy output are detected to reduce voltage fluctuations and grid risks;

[0025] The distributed collaborative scheduling module includes:

[0026] Autonomous sub-region scheduling units are used to adapt to local needs, and each sub-region makes independent scheduling decisions;

[0027] The global collaborative optimization unit is used to coordinate the dispatching strategies of each sub-region to ensure the coordination and optimization of the global power grid, especially when distributed energy is widely connected;

[0028] Divide the power grid in the distributed collaborative dispatch module:

[0029] Divide the area power grid into multiple autonomous sub-areas based on geographical location, load characteristics, and distributed energy access;

[0030] Each sub-region has an independent balance of power supply and demand and makes independent dispatch decisions;

[0031] A local dispatch center is configured in each autonomous sub-region to monitor the power load, distributed energy output, and energy storage status in the sub-region in real time;

[0032] An autonomous dispatching system is deployed for each sub-region to make independent dispatching decisions based on real-time data in the region. The dispatching decisions include load management, energy storage dispatching, and distributed energy optimization.

[0033] Furthermore, in the data acquisition module, multi-factor sampling trigger conditions are defined, and the prediction error parameters are introduced when adjusting the acquisition frequency:

[0034] The power load change is defined as ΔL(t), and the rate of change of the power load is captured by the second-order derivative to capture the short-term change trend: Where PL(t) represents the power load at time t, Δt represents the time interval, ΔL(t) represents the rate of change of power load, which reflects the acceleration of load and captures the short-term fluctuation trend, d2PL(t) represents the second-order differential of function PL(t), that is, the second rate of change, and dt2 represents the square of time t. Indicates the acceleration of the power load PL(t) changing with time t; first-order derivative Represents the speed of change of the power load PL(t), that is, the speed of increase or decrease of the load; the second-order derivative The rate of change represents the speed of change;

[0035] The multi-factor sampling trigger condition also includes the distributed energy output volatility RE(t). The distributed energy output volatility takes into account the relative change and historical volatility of the output: Where PE(t) represents the output power of distributed energy at time t, represents the moving average of distributed energy output, σΕ represents the standard deviation of distributed energy output, and ε represents a small constant, which is 10-6;

[0036] The multi-factor sampling trigger condition also includes the prediction error EP(t), which is measured by weighted absolute error: Where PL,pred(t) represents the predicted value of the power load at time t, PL,actual(t) represents the actual value of the power load at time t, wi is the weight of different time windows, n is the number of time windows, and i is the specific time window.

[0037] Furthermore, in the data acquisition module, an adaptive sampling strategy for dynamically adjusting the acquisition frequency is implemented:

[0038] Assume that the power load change rate is ΔL(t), the distributed energy fluctuation rate is RE(t) and the prediction error is EP(t);

[0039] Define the comprehensive fluctuation index I(t), which integrates the power load change, distributed energy fluctuation and prediction error: Among them, α, β, and γ are weight parameters, which reflect the sensitivity to power load changes, distributed energy fluctuations, and prediction errors, respectively. The highest value in history;

[0040] Assume that the dynamic acquisition frequency is f(t), and the data acquisition frequency f(t) is dynamically adjusted according to the comprehensive fluctuation index I(t): Where fmin represents the minimum acquisition frequency, fmax represents the maximum acquisition frequency, k represents the smoothness of the control response of the adjustment parameter, and δ represents the nonlinear adjustment index.

[0041] Furthermore, in the data acquisition module, the execution steps of the adaptive sampling strategy include:

[0042] Initialization and parameter setting, set the initial acquisition frequency f(t) = fmin, set the weight parameters α, β, γ and the values ​​of δ and k;

[0043] Initialize historical data and calculate

[0044] At each sampling time t, the power load change rate ΔL(t), distributed energy fluctuation rate RE(t) and prediction error EP(t) are calculated in real time;

[0045] Calculate the comprehensive fluctuation index I(t), and calculate the data collection frequency f(t) based on the comprehensive fluctuation index I(t);

[0046] If f(t) is higher than the current acquisition frequency, increase the sampling frequency; if it is lower than the current frequency, decrease the sampling frequency;

[0047] Frequent fluctuations in power load data (peak period): Increase α and δ to enhance sensitivity to load changes;

[0048] Unstable output of distributed energy (periods of high volatility of solar or wind energy): Increase the weight of β and sensitively adjust the sampling frequency;

[0049] The prediction model has a large error (sudden weather changes lead to inaccurate predictions): increase the weight of γ and respond dynamically to improve the accuracy of data collection;

[0050] Based on the real-time power load change rate, distributed energy volatility and prediction error, the data collection frequency is dynamically adjusted; the sampling frequency is increased during periods of severe fluctuations and reduced during stable periods, thereby improving the response speed and data processing efficiency of the overall dispatching system.

[0051] Furthermore, in the optimization decision module, the multi-dimensional features of the substation are extracted to aggregate the features of historical data and real-time data:

[0052] Perform time series embedding and sliding window feature extraction, and use sliding window feature extraction to capture changes at different time scales: Where X(t) represents the feature matrix at time t, xi(t) represents the value of feature i at time t, i refers to 1, 2, ..., m in the formula, Δt represents the step size of the sliding window, n represents the length of the sliding window, and m represents the number of dimensions of the feature;

[0053] The extracted multi-dimensional features are transformed nonlinearly: Z(t)=φ(W·X(t)+b), where Z(t) represents the nonlinear transformed features at time t, W represents the transformation matrix, b represents the bias vector, and φ(·) represents the nonlinear activation function, ReLU and sigmoid are selected;

[0054] After acquiring multi-dimensional nonlinear features, pattern recognition and anomaly detection algorithms are used to capture patterns in historical data and detect anomalies in real time;

[0055] Using latent variable-based pattern recognition Variational Autoencoder, VAE, use variational autoencoder VAE for pattern recognition, obtain the representation of latent variable space, and identify potential patterns in historical data:

[0056] z(t)=μ(X(t))+σ(X(t))·ε, where z(t) represents the latent variable representation at time t, μ(X(t)) and σ(X(t)) represent the mean and standard deviation of the feature X(t) mapped to the latent variable space, and ε represents the random noise sampled from the standard normal distribution;

[0057] Detect abnormal data points by calculating the Mahalanobis distance in the latent variable space:

[0058] Where DM(z(t)) represents the Mahalanobis distance at time t, μz represents the mean of the latent variable space, Σz represents the covariance matrix of the latent variable space, and (z(t)-μz)T represents the transpose of z(t)-μz;

[0059] When DM(z(t)) exceeds the preset threshold λ, the data point is considered an anomaly.

[0060] Furthermore, an adaptive threshold adjustment mechanism is introduced into the optimization decision module to dynamically adjust the threshold λ:

[0061] Dynamically adjust the threshold λ based on the anomaly rate of historical data: Where λ(t) represents the dynamic threshold at time t, λ0 represents the initial threshold, and Anomalyi represents the anomaly detection result at the i-th time point;

[0062] Introduce an online learning algorithm (online gradient descent or recursive least squares) to dynamically update the model parameters θ based on new data:

[0063] Where θ(t) is the model parameter at time t, η is the learning rate, and L(X(t),θ(t)) represents the reconstruction error loss function;

[0064] Highly volatile areas: For areas where power load and distributed energy output fluctuate dramatically, the nonlinearity of feature conversion is enhanced and the dimension of latent variable space is increased;

[0065] Effectively identify potential patterns in historical data and detect anomalies in real time across different substations; adapt to complex and dynamic power system environments.

[0066] Furthermore, in the personalized scheduling unit, a personalized scheduling strategy is generated for each substation based on the data mining results, and resource allocation is dynamically adjusted:

[0067] Using the results of the data mining unit, the characteristics of each substation are analyzed in depth, including:

[0068] Load characteristics, daily fluctuation patterns of load in the substation, peak and valley characteristics, and load growth trends;

[0069] Distributed energy characteristics: the distribution, output characteristics, and fluctuation patterns of distributed energy (solar energy, wind energy) within the substation area;

[0070] Historical events and abnormal patterns, abnormal events that occurred in the substation area in the past, frequent voltage fluctuations, equipment failures, and the causes and characteristics of the events;

[0071] Based on the characteristics of the substation, determine the scheduling objectives and constraints, which include:

[0072] Scheduling objectives: maximize energy efficiency and minimize operating costs;

[0073] Constraints: equipment capacity limitations, network topology, voltage and frequency stability requirements;

[0074] Different dispatch priorities are set based on the specific needs of the substation and the risk assessment obtained through data mining:

[0075] Priority for critical loads: ensuring power supply priority for users including hospitals and data centers;

[0076] Distributed energy priority: Distributed energy is dispatched first during periods of high photovoltaic output to reduce dependence on the main power grid;

[0077] Fault repair priority: When abnormal fluctuations or faults occur in a specific area, load transfer and scheduling adjustments are prioritized;

[0078] Generate a dispatching plan based on the pattern recognition results provided by data mining and the personalized needs of the substation;

[0079] In the personalized scheduling unit, when real-time scheduling adjustment is performed:

[0080] When executing the dispatch strategy, the operation status in the substation is monitored in real time, and resource allocation is adjusted dynamically according to the actual situation:

[0081] Load fluctuation response: When the load fluctuates abnormally, the dispatching strategy can be adjusted quickly;

[0082] Distributed energy fluctuation response: Real-time monitoring of distributed energy output, and dynamic adjustment of the scheduling of other power sources or energy storage devices according to their volatility;

[0083] Fault response and scheduling optimization: When a fault or potential risk is detected, load distribution is quickly adjusted, backup resources are enabled, and scheduling strategies are re-optimized based on new data.

[0084] Furthermore, in the dynamic scheduling response module, the adaptive dynamic programming ADP algorithm is used to perform dynamic scheduling adjustments based on the scheduling strategy generated by the optimization decision module and the real-time monitoring data:

[0085] Obtaining the generated preliminary scheduling strategy from the optimization decision module;

[0086] Before dispatching begins, initialize the dispatching system, including grid load, distributed energy output, and energy storage device status;

[0087] Real-time monitoring of grid load, distributed energy output, energy storage status, voltage and frequency parameters;

[0088] Based on real-time monitoring data, the adaptability of the current dispatch strategy is evaluated; if the monitored power grid status is consistent with the initial strategy prediction, the original strategy is implemented; if there is a significant deviation, the ADP algorithm needs to be activated to adjust the strategy;

[0089] The ADP algorithm uses real-time data to continuously update the state estimation of the power grid and adjust the dispatch strategy according to the latest state. The ADP algorithm adopts a rolling time domain optimization method, that is, recalculates the optimal strategy for the next step at each time step to ensure that the system can flexibly respond to real-time changes;

[0090] According to the adjusted dispatch strategy, the system issues control instructions to perform specific dispatch operations, including adjusting load distribution, starting or stopping energy storage equipment, and regulating distributed energy output;

[0091] The dynamic dispatch response module uses real-time data and adaptive dynamic programming (ADP) algorithm to dynamically adjust the dispatch strategy according to the actual operation of the power grid to ensure the stability of the power system and the optimization of resource utilization.

[0092] Furthermore, in the distributed collaborative scheduling module, localized scheduling decisions are made in real-time regions:

[0093] The localized dispatch center continuously monitors the grid status in the sub-region and collects real-time data, including power load, distributed energy output, voltage, and current;

[0094] Each sub-region independently formulates a dispatch strategy based on its own real-time data and prediction model; after the localized dispatch system formulates the dispatch strategy, it executes it directly in the sub-region;

[0095] Communication network between different autonomous sub-regions, real-time sharing of scheduling information of each region;

[0096] A global dispatch center is established to coordinate the dispatch strategies of each autonomous sub-region. The global dispatch center obtains the real-time status of each sub-region and coordinates based on global needs.

[0097] The global dispatch center monitors the status of the entire regional power grid in real time, including the supply and demand balance of each autonomous sub-region, cross-regional power flow, and grid frequency and voltage;

[0098] When a sub-region experiences an imbalance or anomaly in supply and demand, the global dispatch center coordinates other sub-regions to dispatch resources to support the region with the problem;

[0099] For example, when an overload occurs in a certain area, redundant power resources in adjacent areas are dispatched to supplement it; the global dispatch center regularly analyzes the operating data of each sub-area, optimizes resource allocation, and adjusts the global dispatch strategy to improve the efficiency and stability of the overall power grid.

[0100] Compared with the prior art, the present invention has the following beneficial effects:

[0101] The present invention adopts an adaptive sampling strategy for real-time load fluctuations and distributed energy output characteristics; when the system detects rapid changes or abnormal fluctuations in load, it automatically increases the data collection frequency to ensure that key information is captured; when the load is stable, the sampling frequency is reduced to reduce the collection of redundant data; adaptive sampling can dynamically adjust the collection frequency to ensure that important data is not lost at critical moments, while reducing the system burden.

[0102] The present invention adopts a layered data processing architecture and divides data processing into two layers: edge and cloud. The edge layer is responsible for preliminary processing of the data collected in the station area, including denoising, feature extraction, etc., to ensure data quality; only the key data that has undergone preliminary processing is uploaded to the cloud for data analysis and decision optimization; the burden of data transmission and processing is reduced, and the real-time response capability of the scheduling system is improved.

[0103] The present invention combines the actual operating characteristics of each substation with machine learning and big data analysis technology to conduct in-depth mining and analysis of the collected data.

[0104] Dynamically adjust the dispatching strategy according to the characteristics of different substations to ensure that the power supply system in each substation can achieve optimal resource allocation and fault prevention. Targeted decision optimization can significantly improve dispatching efficiency and power supply stability.

[0105] The present invention introduces a distributed energy generation prediction model based on machine learning. By analyzing meteorological data, historical power generation data and electricity demand patterns, it predicts the output changes of distributed energy in advance, and at the same time establishes a dynamic load balancing model to adjust the allocation of power resources in real time according to the prediction results.

[0106] The present invention integrates a real-time dynamic scheduling algorithm in the scheduling system, and can dynamically adjust the scheduling strategy according to the real-time output status and load demand of distributed energy in the substation.

[0107] The present invention divides the power supply network of the substation into multiple autonomous sub-areas. Each area independently makes localized scheduling decisions and coordinates at the global level through a collaborative mechanism. Combined with edge computing technology, the system can quickly respond to local fluctuations, reduce delays, and improve the agility of scheduling. The distributed architecture effectively improves the resilience and stability of the power grid, especially when facing a large number of distributed energy sources, and can achieve a more stable and efficient power supply. BRIEF DESCRIPTION OF THE DRAWINGS

[0108] Figure 1 It is a schematic diagram of the structure of the intelligent dispatching system of the power supply grid of the substation area of ​​the present invention. DETAILED DESCRIPTION

[0109] In order to enable those skilled in the art to better understand the scheme of the present invention, the technical scheme in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings 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 should fall within the scope of protection of the present invention.

[0110] It should be noted that the terms "first", "second", etc. in the specification and claims of the present invention and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchanged where appropriate, so that the embodiments of the present invention described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units that are clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0111] The present invention is further described in detail below in conjunction with the accompanying drawings:

[0112] Example 1

[0113] See also Figure 1 The present invention provides an intelligent dispatching system for a power grid in a substation area, comprising:

[0114] The data acquisition module dynamically adjusts the data acquisition frequency according to the real-time power load volatility and the output characteristics of distributed energy. It adopts an adaptive sampling strategy to collect data, and the data collected by the adaptive sampling strategy is directly transmitted to the hierarchical data processing module for preliminary processing and analysis. This enables timely capture of key data when the power load changes or the output of distributed energy fluctuates. The sampling frequency is reduced when the load is stable, thereby reducing redundant data collection and reducing the storage and processing burden of the system.

[0115] The data acquisition module includes:

[0116] Real-time monitoring unit, which monitors real-time data of power load and distributed energy output to identify rapid load changes and abnormal fluctuations;

[0117] Adaptive adjustment unit dynamically adjusts the sampling frequency according to the monitoring results, accurately captures key data and reduces redundant data;

[0118] In the data acquisition module, define the multi-factor sampling trigger conditions, and introduce the prediction error parameter when adjusting the acquisition frequency:

[0119] The power load change is defined as ΔL(t), and the rate of change of the power load is captured by the second-order derivative to capture the short-term change trend: Where PL(t) represents the power load at time t, Δt represents the time interval, ΔL(t) represents the rate of change of power load, which reflects the acceleration of load and captures the short-term fluctuation trend, d2PL(t) represents the second-order differential of function PL(t), that is, the second rate of change, and dt2 represents the square of time t. Indicates the acceleration of the power load PL(t) changing with time t; first-order derivative Represents the speed of change of the power load PL(t), that is, the speed of increase or decrease of the load; the second-order derivative The rate of change represents the speed of change;

[0120] The multi-factor sampling trigger condition also includes the distributed energy output volatility RE(t). The distributed energy output volatility takes into account the relative change and historical volatility of the output: Where PE(t) represents the output power of distributed energy at time t, represents the moving average of distributed energy output, σΕ represents the standard deviation of distributed energy output, and ε represents a small constant, which is 10-6;

[0121] The multi-factor sampling trigger condition also includes the prediction error EP(t), which is measured by weighted absolute error: Where PL,pred(t) represents the predicted value of the power load at time t, PL,actual(t) represents the actual value of the power load at time t, wi is the weight of different time windows, n is the number of time windows, and i is the specific time window;

[0122] In the data acquisition module, an adaptive sampling strategy that dynamically adjusts the acquisition frequency:

[0123] Assume that the power load change rate is ΔL(t), the distributed energy fluctuation rate is RE(t) and the prediction error is EP(t);

[0124] Define the comprehensive fluctuation index I(t), which integrates the power load change, distributed energy fluctuation and prediction error: Among them, α, β, and γ are weight parameters, which reflect the sensitivity to power load changes, distributed energy fluctuations, and prediction errors, respectively. The highest value in history;

[0125] Assume that the dynamic acquisition frequency is f(t), and the data acquisition frequency f(t) is dynamically adjusted according to the comprehensive fluctuation index I(t): Where fmin represents the minimum acquisition frequency, fmax represents the maximum acquisition frequency, k represents the smoothness of the control response of the adjustment parameter, and δ represents the nonlinear adjustment index;

[0126] In the data acquisition module, the execution steps of the adaptive sampling strategy include:

[0127] Initialization and parameter setting, set the initial acquisition frequency f(t) = fmin, set the weight parameters α, β, γ and the values ​​of δ and k;

[0128] Initialize historical data and calculate

[0129] At each sampling time t, the power load change rate ΔL(t), distributed energy fluctuation rate RE(t) and prediction error EP(t) are calculated in real time;

[0130] Calculate the comprehensive fluctuation index I(t), and calculate the data collection frequency f(t) based on the comprehensive fluctuation index I(t);

[0131] If f(t) is higher than the current acquisition frequency, increase the sampling frequency; if it is lower than the current frequency, decrease the sampling frequency;

[0132] Frequent fluctuations in power load data (peak period): Increase α and δ to enhance sensitivity to load changes;

[0133] Unstable output of distributed energy (periods of high volatility of solar or wind energy): Increase the weight of β and sensitively adjust the sampling frequency;

[0134] The prediction model has a large error (sudden weather changes lead to inaccurate predictions): increase the weight of γ and respond dynamically to improve the accuracy of data collection;

[0135] Dynamically adjust the data collection frequency based on the real-time power load change rate, distributed energy fluctuation rate and prediction error; increase the sampling frequency during periods of severe fluctuations and reduce the frequency during stable periods, thereby improving the response speed and data processing efficiency of the overall dispatching system;

[0136] The layered data processing module includes two layers, the edge layer and the cloud layer. The edge layer is responsible for the preliminary processing of the data collected in the substation area. Only the data that has been preliminarily processed is uploaded to the cloud layer for analysis and decision optimization. The processed data will be uploaded to the cloud layer, i.e. the cloud. The edge-processed data provides support for the optimization decision module based on the characteristics of the substation area.

[0137] Specifically, through multi-level optimization strategies, the stability, response speed and resource utilization efficiency of the power grid are significantly improved; by using real-time data collection and adaptive sampling strategies, key information on power load changes and distributed energy output fluctuations is effectively captured, ensuring that the sampling frequency is increased when the load fluctuates frequently, and the sampling frequency is reduced during stable periods, thereby reducing redundant data and optimizing the use of storage and processing resources;

[0138] The hierarchical data processing modules include:

[0139] The edge layer is used to perform preliminary data processing on the data;

[0140] The cloud layer, where the preliminarily processed data is uploaded to, where it is analyzed and global decision-making is optimized;

[0141] In addition, with the widespread access to distributed energy (such as solar energy and wind energy), it is difficult to maintain the balance between supply and demand in real time. The output of distributed energy is highly uncertain and intermittent, and the current scheduling system based on historical data and static models is difficult to respond to these fluctuations in a timely manner, resulting in oversupply or undersupply of electricity, affecting the stability of the power grid.

[0142] At the same time, the dispatching system relies on preset load curves and fixed strategies and cannot be flexibly adjusted according to actual conditions. Especially after a large number of distributed energy sources are connected, this rigid dispatching strategy often cannot adapt to the complex power supply environment, increasing the risk of power grid operation.

[0143] The randomness of distributed energy access points leads to voltage fluctuations and frequency deviations. The dispatching system cannot respond quickly to these problems, which easily leads to instability of the local power grid and even triggers the protection device to cause power outages. Therefore, the present invention adopts the following scheme for targeted optimization;

[0144] Specifically, the layered data processing module further improves the efficiency and accuracy of data processing through the combination of edge computing and cloud analysis; after preliminary processing at the edge layer, only key data is transmitted to the cloud layer for global decision-making, reducing the amount of data transmission and speeding up the response;

[0145] The optimization decision module uses data from edge computing and cloud analysis to deeply explore and analyze the characteristics of different substations and generate personalized scheduling strategies. The scheduling strategies generated by the optimization decision module will be directly input into the dynamic scheduling response module to perform scheduling adjustments in real time.

[0146] The optimization decision module optimizes based on the characteristics of the substation, and the dispatch strategy dynamically adjusts the dispatch based on the load characteristics of each substation and the access status of distributed energy to achieve optimal resource allocation and fault prevention;

[0147] The optimization decision module includes:

[0148] Data mining unit, which uses machine learning algorithms to analyze historical data and real-time status of each substation to identify patterns and anomalies;

[0149] The personalized scheduling unit generates personalized scheduling strategies for each substation based on data mining results and dynamically adjusts resource allocation;

[0150] In the optimization decision module, the multi-dimensional features of the substation are extracted to aggregate the features of historical data and real-time data:

[0151] Perform time series embedding and sliding window feature extraction, and use sliding window feature extraction to capture changes at different time scales: Where X(t) represents the feature matrix at time t, xi(t) represents the value of feature i at time t, i refers to 1, 2, ..., m in the formula, Δt represents the step size of the sliding window, n represents the length of the sliding window, and m represents the number of dimensions of the feature;

[0152] The extracted multi-dimensional features are transformed nonlinearly: Z(t)=φ(W·X(t)+b), where Z(t) represents the nonlinear transformed features at time t, W represents the transformation matrix, b represents the bias vector, and φ(·) represents the nonlinear activation function, ReLU and sigmoid are selected;

[0153] After acquiring multi-dimensional nonlinear features, pattern recognition and anomaly detection algorithms are used to capture patterns in historical data and detect anomalies in real time;

[0154] Using latent variable-based pattern recognition Variational Autoencoder, VAE, use variational autoencoder VAE for pattern recognition, obtain the representation of latent variable space, and identify potential patterns in historical data:

[0155] z(t)=μ(X(t))+σ(X(t))·ε, where z(t) represents the latent variable representation at time t, μ(X(t)) and σ(X(t)) represent the mean and standard deviation of the feature X(t) mapped to the latent variable space, and ε represents the random noise sampled from the standard normal distribution;

[0156] Detect abnormal data points by calculating the Mahalanobis distance in the latent variable space:

[0157] Where DM(z(t)) represents the Mahalanobis distance at time t, μz represents the mean of the latent variable space, Σz represents the covariance matrix of the latent variable space, and (z(t)-μz)T represents the transpose of z(t)-μz;

[0158] When DM(z(t)) exceeds the preset threshold λ, the data point is considered an anomaly;

[0159] An adaptive threshold adjustment mechanism is introduced into the optimization decision module to dynamically adjust the threshold λ:

[0160] Dynamically adjust the threshold λ based on the anomaly rate of historical data: Where λ(t) represents the dynamic threshold at time t, λ0 represents the initial threshold, and Anomalyi represents the anomaly detection result at the i-th time point;

[0161] Introduce an online learning algorithm (online gradient descent or recursive least squares) to dynamically update the model parameters θ based on new data:

[0162] Where θ(t) is the model parameter at time t, η is the learning rate, and L(X(t),θ(t)) represents the reconstruction error loss function;

[0163] Highly volatile areas: For areas where power load and distributed energy output fluctuate dramatically, the nonlinearity of feature conversion is enhanced and the dimension of latent variable space is increased;

[0164] Effectively identify potential patterns in historical data and detect anomalies in real time in different substations; adapt to complex and dynamic power system environments;

[0165] In the personalized scheduling unit, a personalized scheduling strategy is generated for each substation based on the data mining results, and resource allocation is dynamically adjusted:

[0166] Using the results of the data mining unit, the characteristics of each substation are analyzed in depth, including:

[0167] Load characteristics, daily fluctuation patterns of load in the substation, peak and valley characteristics, and load growth trends;

[0168] Distributed energy characteristics: the distribution, output characteristics, and fluctuation patterns of distributed energy (solar energy, wind energy) within the substation area;

[0169] Historical events and abnormal patterns, abnormal events that occurred in the substation area in the past, frequent voltage fluctuations, equipment failures, and the causes and characteristics of the events;

[0170] Based on the characteristics of the substation, determine the scheduling objectives and constraints, which include:

[0171] Scheduling objectives: maximize energy efficiency and minimize operating costs;

[0172] Constraints: equipment capacity limitations, network topology, voltage and frequency stability requirements;

[0173] Different dispatch priorities are set based on the specific needs of the substation and the risk assessment obtained through data mining:

[0174] Priority for critical loads: ensuring power supply priority for users including hospitals and data centers;

[0175] Distributed energy priority: Distributed energy is dispatched first during periods of high photovoltaic output to reduce dependence on the main power grid;

[0176] Fault repair priority: When abnormal fluctuations or faults occur in a specific area, load transfer and scheduling adjustments are prioritized;

[0177] Generate a dispatching plan based on the pattern recognition results provided by data mining and the personalized needs of the substation;

[0178] In the personalized scheduling unit, real-time scheduling adjustment: When executing the scheduling strategy, the operation status in the substation is monitored in real time, and resource allocation is dynamically adjusted according to the actual situation:

[0179] Load fluctuation response: When the load fluctuates abnormally, the dispatching strategy can be adjusted quickly;

[0180] Distributed energy fluctuation response: Real-time monitoring of distributed energy output, and dynamic adjustment of the scheduling of other power sources or energy storage devices according to their volatility;

[0181] Fault response and scheduling optimization: When a fault or potential risk is detected, load distribution is quickly adjusted, backup resources are enabled, and scheduling strategies are re-optimized based on new data;

[0182] Specifically, the optimization decision module is based on machine learning algorithms, deeply mines the historical data and real-time status of each substation, and generates personalized dispatch strategies; this module can dynamically adjust resource allocation to ensure that the operation of the power system always meets the actual needs of each substation; through pattern recognition and anomaly detection mechanisms, it can identify potential risks in advance, prevent failures, and improve power supply reliability;

[0183] The dynamic dispatch response module makes dynamic dispatch adjustments based on the dispatch strategy generated by the optimization decision module and combined with real-time monitoring data. It uses the adaptive dynamic programming ADP algorithm to adjust the dispatch strategy according to real-time conditions. It quickly responds to the fluctuations of distributed energy and load changes to ensure the stability of the power grid. The dynamic dispatch execution results are fed back to the distributed collaborative dispatch module for coordination and optimization at the global level.

[0184] The dynamic scheduling response module includes:

[0185] Real-time scheduling unit, based on ADP algorithm, dynamically adjusts scheduling strategy;

[0186] Rapid response unit, which quickly adjusts dispatch when load changes or fluctuations in distributed energy output are detected to reduce voltage fluctuations and grid risks;

[0187] In the dynamic scheduling response module, the adaptive dynamic programming ADP algorithm is used to perform dynamic scheduling adjustments based on the scheduling strategy generated by the optimization decision module and real-time monitoring data:

[0188] Obtaining the generated preliminary scheduling strategy from the optimization decision module;

[0189] Before dispatching begins, initialize the dispatching system, including grid load, distributed energy output, and energy storage device status;

[0190] Real-time monitoring of grid load, distributed energy output, energy storage status, voltage and frequency parameters;

[0191] Based on real-time monitoring data, the adaptability of the current dispatch strategy is evaluated; if the monitored power grid status is consistent with the initial strategy prediction, the original strategy is implemented; if there is a significant deviation, the ADP algorithm needs to be activated to adjust the strategy;

[0192] The ADP algorithm uses real-time data to continuously update the state estimation of the power grid and adjust the dispatch strategy according to the latest state. The ADP algorithm adopts a rolling time domain optimization method, that is, recalculates the optimal strategy for the next step at each time step to ensure that the system can flexibly respond to real-time changes;

[0193] According to the adjusted dispatch strategy, the system issues control instructions to perform specific dispatch operations, including adjusting load distribution, starting or stopping energy storage equipment, and regulating distributed energy output;

[0194] The dynamic dispatch response module uses real-time data and adaptive dynamic programming (ADP) algorithm to dynamically adjust the dispatch strategy according to the actual operation of the power grid to ensure the stability of the power system and the optimization of resource utilization;

[0195] Specifically, the dynamic dispatch response module uses the adaptive dynamic programming (ADP) algorithm to dynamically adjust the dispatch strategy based on real-time monitoring data; enabling the system to quickly respond to fluctuations in distributed energy and changes in power load, ensuring the stability of the power grid, and continuously improving the accuracy and efficiency of dispatch decisions through rolling optimization;

[0196] The distributed collaborative dispatching module divides the substation power grid into multiple autonomous sub-regions. Each region makes independent localized dispatching decisions while coordinating at the global level. This ensures the overall stability and efficiency of the power grid and supports rapid response to local fluctuations.

[0197] The distributed collaborative scheduling module includes:

[0198] Autonomous sub-region scheduling units are used to adapt to local needs, and each sub-region makes independent scheduling decisions;

[0199] The global collaborative optimization unit is used to coordinate the dispatching strategies of each sub-region to ensure the coordination and optimization of the global power grid, especially when distributed energy is widely connected;

[0200] Divide the power grid in the distributed collaborative dispatch module:

[0201] Divide the area power grid into multiple autonomous sub-areas based on geographical location, load characteristics, and distributed energy access;

[0202] Each sub-region has an independent balance of power supply and demand and makes independent dispatch decisions;

[0203] A local dispatch center is configured in each autonomous sub-region to monitor the power load, distributed energy output, and energy storage status in the sub-region in real time;

[0204] Deploy an autonomous dispatching system for each sub-region to independently make dispatching decisions based on real-time data in the region. Dispatching decisions include load management, energy storage dispatching, and distributed energy optimization.

[0205] In the distributed collaborative scheduling module, localized scheduling decisions are made in real-time regions:

[0206] The localized dispatch center continuously monitors the grid status in the sub-region and collects real-time data, including power load, distributed energy output, voltage, and current;

[0207] Each sub-region independently formulates a dispatch strategy based on its own real-time data and prediction model; after the localized dispatch system formulates the dispatch strategy, it executes it directly in the sub-region;

[0208] Communication network between different autonomous sub-regions, real-time sharing of scheduling information of each region;

[0209] A global dispatch center is established to coordinate the dispatch strategies of each autonomous sub-region. The global dispatch center obtains the real-time status of each sub-region and coordinates based on global needs.

[0210] The global dispatch center monitors the status of the entire regional power grid in real time, including the supply and demand balance of each autonomous sub-region, cross-regional power flow, and grid frequency and voltage;

[0211] When a sub-region experiences an imbalance or anomaly in supply and demand, the global dispatch center coordinates other sub-regions to dispatch resources to support the region with the problem;

[0212] For example, when a certain area is overloaded, redundant power resources from adjacent areas are dispatched to supplement it; the global dispatch center regularly analyzes the operating data of each sub-area, optimizes resource allocation, and adjusts the global dispatch strategy to improve the efficiency and stability of the overall power grid;

[0213] Specifically, the distributed collaborative dispatching module divides the substation power grid into multiple autonomous sub-regions, enabling each region to make independent localized dispatching decisions while coordinating at the global level. The advantage of this module is that it enhances the flexibility and resilience of the system, especially when distributed energy is widely accessed. Through the coordination of the global dispatching center, resources between sub-regions can be effectively dispatched to achieve the optimal configuration of the overall power grid, ensuring rapid response to local fluctuations and maintaining stable operation of the power grid.

[0214] So far, the technical solutions of the present invention have been described in conjunction with the preferred embodiments shown in the accompanying drawings. However, it is easy for those skilled in the art to understand that the protection scope of the present invention is obviously not limited to these specific embodiments. Without departing from the principle of the present invention, those skilled in the art can make equivalent changes or substitutions to the relevant technical features, and the technical solutions after these changes or substitutions will fall within the protection scope of the present invention.

Claims

1. An intelligent dispatching system for a power grid in an area, characterized in that: include: The data collection module dynamically adjusts the data collection frequency according to the real-time power load volatility and the output characteristics of distributed energy; Adopting adaptive sampling strategy to collect data, the data collected by adaptive sampling strategy is directly transmitted to the hierarchical data processing module for preliminary processing and analysis; The layered data processing module includes two layers, edge and cloud. The edge layer is responsible for the preliminary processing of the data collected in the station area; only the data that has been preliminarily processed is uploaded to the cloud layer for analysis and decision optimization; the processed data will be uploaded to the cloud layer, i.e., the cloud; Optimize the decision-making module, use data from edge computing and cloud analysis to deeply explore and analyze the characteristics of different substations, and generate personalized scheduling strategies; The dynamic scheduling response module makes dynamic scheduling adjustments based on the scheduling strategy generated by the optimization decision module and real-time monitoring data; it uses the adaptive dynamic programming ADP algorithm to adjust the scheduling strategy according to real-time conditions; The distributed collaborative dispatching module divides the substation power grid into multiple autonomous sub-regions, each of which makes localized dispatching decisions independently while coordinating at the global level.

2. The intelligent dispatching system for the power supply grid in a substation area according to claim 1 is characterized in that: The data acquisition module includes: Real-time monitoring unit, which monitors real-time data of power load and distributed energy output to identify rapid load changes and abnormal fluctuations; Adaptive adjustment unit, dynamically adjusts the sampling frequency according to the monitoring results; The hierarchical data processing modules include: The edge layer is used to perform preliminary data processing on the data; The cloud layer, where the preliminarily processed data is uploaded to, where it is analyzed and global decision-making is optimized; The optimization decision module is optimized based on the characteristics of the substation, and the dispatch strategy is dynamically adjusted based on the load characteristics of each substation and the access status of distributed energy; The optimization decision module includes: Data mining unit, which uses machine learning algorithms to analyze historical data and real-time status of each substation to identify patterns and anomalies; The personalized scheduling unit generates personalized scheduling strategies for each substation based on data mining results and dynamically adjusts resource allocation; The dynamic scheduling response module includes: Real-time scheduling unit, based on ADP algorithm, dynamically adjusts scheduling strategy; Rapid response unit, which quickly adjusts dispatch when load changes or fluctuations in distributed energy output are detected to reduce voltage fluctuations and grid risks; The distributed collaborative scheduling module includes: Autonomous sub-region scheduling units are used to adapt to local needs, and each sub-region makes independent scheduling decisions; A global collaborative optimization unit, used to coordinate the scheduling strategies of each sub-area; Divide the power grid in the distributed collaborative dispatch module: Divide the area power grid into multiple autonomous sub-areas based on geographical location, load characteristics, and distributed energy access; Each sub-region has an independent balance of power supply and demand and makes independent dispatch decisions; A local dispatch center is configured in each autonomous sub-region to monitor the power load, distributed energy output, and energy storage status in the sub-region in real time; An autonomous dispatching system is deployed for each sub-region to make independent dispatching decisions based on real-time data in the region. The dispatching decisions include load management, energy storage dispatching, and distributed energy optimization.

3. The intelligent dispatching system for the power supply grid in a substation area according to claim 2 is characterized in that: In the data acquisition module, define the multi-factor sampling trigger conditions, and introduce the prediction error parameter when adjusting the acquisition frequency: The power load change is defined as ΔL(t), and the rate of change of the power load is captured by the second-order derivative to capture the short-term change trend: Where PL(t) represents the power load at time t, Δt represents the time interval, ΔL(t) represents the rate of change of power load, which reflects the acceleration of load and captures the short-term fluctuation trend, d2PL(t) represents the second-order differential of function PL(t), that is, the second rate of change, and dt2 represents the square of time t. Indicates the acceleration of the power load PL(t) changing with time t; first-order derivative Represents the speed of change of the power load PL(t), that is, the speed of increase or decrease of the load; the second-order derivative The rate of change represents the speed of change; The multi-factor sampling trigger condition also includes the distributed energy output volatility RE(t): Where PE(t) represents the output power of distributed energy at time t, represents the moving average of distributed energy output, σΕ represents the standard deviation of distributed energy output, and ε represents a small constant, which is 10-6; The multi-factor sampling trigger condition also includes the prediction error EP(t), which is measured by weighted absolute error: Where PL,pred(t) represents the predicted value of the power load at time t, PL,actual(t) represents the actual value of the power load at time t, wi is the weight of different time windows, n is the number of time windows, and i is the specific time window.

4. The intelligent dispatching system for the power supply grid in a substation area according to claim 3 is characterized in that: In the data acquisition module, an adaptive sampling strategy that dynamically adjusts the acquisition frequency: Assume that the power load change rate is ΔL(t), the distributed energy fluctuation rate is RE(t) and the prediction error is EP(t); Define the comprehensive fluctuation index I(t), which integrates the power load change, distributed energy fluctuation and prediction error: Among them, α, β, and γ are weight parameters, which reflect the sensitivity to power load changes, distributed energy fluctuations, and prediction errors, respectively. The highest value in history; Assume that the dynamic acquisition frequency is f(t), and the data acquisition frequency f(t) is dynamically adjusted according to the comprehensive fluctuation index I(t): Where fmin represents the minimum acquisition frequency, fmax represents the maximum acquisition frequency, k represents the smoothness of the control response of the adjustment parameter, and δ represents the nonlinear adjustment index.

5. The intelligent dispatching system for the power grid of a substation area according to claim 4, characterized in that: In the data acquisition module, the execution steps of the adaptive sampling strategy include: Initialization and parameter setting: set the initial acquisition frequency f(t) = fmin, and set the values ​​of weight parameters α, β, γ, δ and k; Initialize historical data and calculate At each sampling time t, the power load change rate ΔL(t), distributed energy fluctuation rate RE(t) and prediction error EP(t) are calculated in real time; Calculate the comprehensive fluctuation index I(t), and calculate the data collection frequency f(t) based on the comprehensive fluctuation index I(t); If f(t) is higher than the current acquisition frequency, the sampling frequency is increased; if it is lower than the current frequency, the sampling frequency is reduced.

6. The intelligent dispatching system for the power grid of a substation area according to claim 5, characterized in that: In the optimization decision module, the multi-dimensional features of the substation are extracted to aggregate the features of historical data and real-time data: Perform time series embedding and sliding window feature extraction, and use sliding window feature extraction to capture changes at different time scales: Where X(t) represents the feature matrix at time t, xi(t) represents the value of feature i at time t, i refers to 1, 2, ..., m in the formula, Δt represents the step size of the sliding window, n represents the length of the sliding window, and m represents the number of dimensions of the feature; The extracted multidimensional features are transformed nonlinearly: Z(t)=φ(W·X(t)+b), where Z(t) represents the nonlinear transformed features at time t, W represents the transformation matrix, b represents the bias vector, and φ(·) represents the nonlinear activation function; After acquiring multi-dimensional nonlinear features, pattern recognition and anomaly detection algorithms are used to capture patterns in historical data and detect anomalies in real time; Using latent variable-based pattern recognition Variational Autoencoder, VAE, use variational autoencoder VAE for pattern recognition, obtain the representation of latent variable space, and identify potential patterns in historical data: z(t)=μ(X(t))+σ(X(t))·ε, where z(t) represents the latent variable representation at time t, μ(X(t)) and σ(X(t)) represent the mean and standard deviation of the feature X(t) mapped to the latent variable space, and ε represents the random noise sampled from the standard normal distribution; Detect abnormal data points by calculating the Mahalanobis distance in the latent variable space: Where DM(z(t)) represents the Mahalanobis distance at time t, μz represents the mean of the latent variable space, Σz represents the covariance matrix of the latent variable space, and (z(t)-μz)T represents the transpose of z(t)-μz; When DM(z(t)) exceeds the preset threshold λ, the data point is considered an anomaly.

7. The intelligent dispatching system for the power grid of a substation area according to claim 6, characterized in that: An adaptive threshold adjustment mechanism is introduced into the optimization decision module to dynamically adjust the threshold λ: Dynamically adjust the threshold λ based on the anomaly rate of historical data: Where λ(t) represents the dynamic threshold at time t, λ0 represents the initial threshold, and Anomalyi represents the anomaly detection result at the i-th time point; Introduce an online learning algorithm to dynamically update the model parameters θ based on new data: Where θ(t) is the model parameter at time t, η is the learning rate, and L(X(t),θ(t)) represents the reconstruction error loss function.

8. The intelligent dispatching system for the power grid of a substation area according to claim 7, characterized in that: In the personalized scheduling unit, a personalized scheduling strategy is generated for each substation based on the data mining results, and resource allocation is dynamically adjusted: Using the results of the data mining unit, the characteristics of each substation are analyzed in depth, including: Load characteristics, daily fluctuation patterns of load in the substation, peak and valley characteristics, and load growth trends; Distributed energy characteristics: the distribution, output characteristics, and fluctuation patterns of distributed energy within the substation area; Historical events and abnormal patterns, abnormal events that occurred in the substation area in the past, frequent voltage fluctuations, equipment failures, and the causes and characteristics of the events; Based on the characteristics of the substation, determine the scheduling objectives and constraints, which include: Scheduling objectives: maximize energy efficiency and minimize operating costs; Constraints: equipment capacity limitations, network topology, voltage and frequency stability requirements; Different dispatch priorities are set based on the specific needs of the substation and the risk assessment obtained from data mining: Critical load priority: including users such as hospitals and data centers; Distributed energy priority: Distributed energy is dispatched first during periods of high photovoltaic output; Based on the pattern recognition results provided by data mining and combined with the personalized needs of the substation, a scheduling plan is generated.

9. The intelligent dispatching system for the power grid of a substation area according to claim 8, characterized in that: In the dynamic scheduling response module, the adaptive dynamic programming ADP algorithm is used to perform dynamic scheduling adjustments based on the scheduling strategy generated by the optimization decision module and real-time monitoring data: Obtaining the generated preliminary scheduling strategy from the optimization decision module; Before dispatching begins, initialize the dispatching system, including grid load, distributed energy output, and energy storage device status; Real-time monitoring of grid load, distributed energy output, energy storage status, voltage and frequency parameters; Based on real-time monitoring data, the adaptability of the current dispatch strategy is evaluated; if the monitored power grid status is consistent with the initial strategy prediction, the original strategy is implemented; if there is a significant deviation, the ADP algorithm needs to be activated to adjust the strategy; The ADP algorithm uses real-time data to continuously update the state estimation of the power grid and adjust the dispatch strategy according to the latest state. The ADP algorithm adopts a rolling time domain optimization method; According to the adjusted dispatching strategy, the system issues control instructions to perform specific dispatching operations, including adjusting load distribution, starting or stopping energy storage equipment, and regulating distributed energy output.

10. The intelligent dispatching system for the power grid of a substation area according to claim 9, characterized in that: In the distributed collaborative scheduling module, localized scheduling decisions are made in real-time regions: The localized dispatch center continuously monitors the grid status in the sub-region and collects real-time data, including power load, distributed energy output, voltage, and current; Each sub-region independently formulates a scheduling strategy based on its own real-time data and prediction model; After the localized scheduling system formulates the scheduling strategy, it executes it directly in the sub-area; Communication network between different autonomous sub-regions, real-time sharing of scheduling information of each region; A global dispatch center is established to coordinate the dispatch strategies of each autonomous sub-region. The global dispatch center obtains the real-time status of each sub-region and coordinates based on global needs. The global dispatch center monitors the status of the entire regional power grid in real time, including the supply and demand balance of each autonomous sub-region, cross-regional power flow, and grid frequency and voltage; When a sub-region experiences an imbalance or anomaly in supply and demand, the global dispatch center coordinates the dispatch resources of other sub-regions to support the problematic area.

Images