Search

CN-121566533-B - Demand side response power utilization management method based on power Hongmon architecture

CN121566533BCN 121566533 BCN121566533 BCN 121566533BCN-121566533-B

Abstract

The application provides a demand side response application power management method based on an electric power HongMong framework, which relates to the technical field of information, and particularly relates to a distributed intelligent communication framework for a novel electric power system, wherein multi-source electricity and environment data are collected through electric power HongMong edge computing nodes of a transformer substation, and three kinds of electricity main body classification results of industry, business and residents are obtained through cleaning, feature extraction and clustering, and the operation of a node is guaranteed by carrying a computer and auxiliary equipment repair adaptation module; the classified results are encrypted and transmitted to a trusted execution environment, a power consumption behavior model is built by fusing multiple factors, a dynamic power price strategy and a response rule are generated and pushed to a terminal, an intelligent operation and maintenance mechanism is combined with a computer and auxiliary equipment to repair an area response strategy, a final management instruction is generated through load balancing judgment, and after execution, optimized data are obtained, and the model and the strategy are dynamically updated. The application realizes accurate regulation and control of electricity consumption, balances the load of the power grid and improves the electricity consumption efficiency.

Inventors

  • HUANG MINGLEI
  • DU ZIPENG

Assignees

  • 广东电网有限责任公司珠海供电局

Dates

Publication Date
20260512
Application Date
20260120

Claims (9)

  1. 1. The utility model provides a demand side response power utilization management method based on electric power hong Monte framework, which is characterized in that the method comprises the following steps: collecting multi-source electricity consumption related data, cleaning and integrating, then combining the environmental data, extracting electricity consumption, load values and power factor characteristic values, and clustering to obtain three types of electricity consumption main body classification results of industry, business and residents; Encrypting and transmitting the classification result to a safe execution environment, fusing environment data and electricity utilization characteristics of electricity utilization main bodies, and constructing electricity utilization behavior models of different types of electricity utilization main bodies; Judging electricity consumption peaks according to model output, generating dynamic electricity price strategies and automatic response rules which are adapted to different electricity consumption main body types, and considering electricity consumption requirements and regulation and control targets; pushing a dynamic electricity price strategy to an electricity consumption main body terminal, collecting terminal response data and associating an electricity price information optimization strategy; Monitoring regional load in real time, if the load is abnormal, generating a regional demand side response strategy by combining the classification result of the electricity consumption main body and the environmental data, and further optimizing according to the feedback of the electricity consumption main body; Calculating the load distribution of the power grid again according to the adjusted comprehensive strategy, generating a final management instruction after judging the balance, issuing and executing the final management instruction, acquiring load optimization effect data, analyzing the influence of environmental factors, and dynamically updating the power consumption behavior model and the response strategy; The method for collecting the related data of the multi-source electricity consumption, cleaning and integrating the related data, and then combining the related data with the environmental data to extract the electricity consumption, the load value and the characteristic value of the power factor comprises the following steps: collecting original electricity data through an electric power hong edge computing node deployed in a transformer substation, cleaning the original electricity data through the electric power hong edge computing node, and if an abnormal value or a missing value is found, identifying and replacing the abnormal value through a long-short-period memory network abnormality detection module built in the electric power hong edge computing node, and complementarily repairing the missing value through adjacent electric power hong node data to obtain a cleaned data table; Performing Z-score standardization treatment on the cleaned data table by combining environmental values including temperature, humidity and air pressure to obtain a standard value table; extracting the electricity consumption and the load value from the standard value table, and calculating a power value and a power factor characteristic value according to the formula p=ui and the power factor characteristic value=active power/apparent power, wherein P is electric power, U is voltage, and I is current; classifying the electricity consumption and the load value by adopting a K-means clustering algorithm, and dividing an electricity consumption mode; analyzing the relation between the power value and the power factor characteristic value through a linear regression algorithm, and establishing a power factor prediction model; And combining the power consumption mode with the power factor prediction model to generate a power factor characteristic value.
  2. 2. The method of claim 1, wherein the collecting the related data of multi-source electricity, cleaning and integrating, and then combining the related data of environment, extracting the power consumption and load values and the characteristic value of power factor, and obtaining the classification result of three types of electricity consumption main bodies of industry, business and residents through clustering and grouping comprises the following steps: The method comprises the steps of obtaining the electricity consumption and load values and the power factor characteristic values of all electric power hong edge computing nodes, constructing electricity consumption main body characteristic vectors according to the electricity consumption and load values and the power factor characteristic values, combining the electricity consumption main body characteristic vectors with reputation weight factors introduced by a dynamic reputation value evaluation module in an electric power hong architecture according to an Euclidean distance formula, and computing the similarity between the electricity consumption main body characteristic vectors to obtain a similarity matrix; Aiming at the similarity matrix, a principal component analysis method is used for obtaining feature distribution, a K-means clustering algorithm is calculated by utilizing a heterogeneous accelerator of a power Hongmon protocol stack based on the feature distribution, and a power utilization main body is clustered to obtain a contour coefficient; If the contour coefficient is smaller than or equal to a preset contour coefficient threshold, the clustering effect is invalid, the similarity and the contour coefficient are recalculated, if the contour coefficient is larger than the preset contour coefficient threshold, the clustering effect is determined to be valid, and the Davies-Bouldin index is adopted to further verify the clustering quality, so that a clustering result is generated; and dividing the electricity utilization main body into three types of industry, business and residents through t-SNE dimension reduction visual analysis according to the clustering result to obtain an electricity utilization main body classification result.
  3. 3. The method according to claim 1, wherein the encrypting the classification result to the secure execution environment, fusing the environment data with the electricity consumption characteristics of the electricity consumption main body, and constructing electricity consumption behavior models of different types of electricity consumption main bodies includes: The method comprises the steps of encrypting and transmitting a power utilization main body classification result to a power hong-Monte protocol stack, decrypting the power utilization main body classification result in the power hong-Monte trusted execution environment to obtain load values of three power utilization main bodies of industry, business and residents in the power utilization main body classification result, analyzing the load values of the three power utilization main bodies, and extracting peak-to-valley ratio and load rate of the load values of the three power utilization main bodies; carrying out correlation analysis on the peak-valley ratio and the load rate and the temperature value and the humidity value in the weather forecast data, and determining the influence degree of the temperature value and the humidity value on the load value; Extracting environmental impact factors from holiday arrangement and emergency information, and quantifying the environmental impact factors into a numerical form; Taking peak-valley ratio, load factor characteristics, temperature value and humidity value and environmental influence factors as input variables, and constructing electricity consumption behavior models of electricity consumption main bodies of different types; if the error value of the electricity consumption behavior model of the electricity consumption main body of the different types of electricity consumption main bodies exceeds a preset error threshold value, adjusting the parameters of the electricity consumption behavior model of the electricity consumption main body of the different types of electricity consumption main bodies by adopting a grid search method, and retraining the model; if the error value is within the preset error threshold, determining that the model is available; And calculating the power consumption load change trend within 24 hours in the future through power consumption main body power consumption behavior models of different types of power consumption main bodies, and generating a load prediction result.
  4. 4. The method of claim 1, wherein the determining the electricity consumption peak according to the model output, generating a dynamic electricity price policy and an automatic response rule adapted to different electricity consumption main body types, taking into account the electricity consumption requirement and the regulation and control target, comprises: According to the load prediction results output by the electricity consumption behavior models of the electricity consumption main bodies of different types, respectively comparing the load prediction results with preset high peak values, and if the load values predicted by the load prediction results exceed the high peak values, determining that the electricity consumption peak exists; Generating electricity price strategies of different time periods based on a preset dynamic electricity price adjustment rule according to a determination result of electricity consumption peaks; Extracting electricity price information of each time period from an electricity price strategy, and applying a preset automatic response rule of an electricity consumption main body by combining the type of the electricity consumption main body to generate a data set of response behaviors of the electricity consumption main body; preprocessing a data set of the response behavior of the power utilization main body, including data cleaning and feature extraction, and generating a feature set for model training; predicting the electricity consumption behavior of the electricity consumption main body under the dynamic electricity price by using a feature set training based on a random forest classification model to obtain predicted electricity consumption behavior data of the electricity consumption main body; according to the predicted electricity consumption behavior data of the electricity consumption main body, adjusting an electricity price strategy, and optimizing a dynamic electricity price adjustment mechanism; And generating an adjusted dynamic electricity price strategy through an optimized dynamic electricity price adjustment mechanism, and updating an automatic response rule of the electricity utilization main body.
  5. 5. The method according to claim 1, wherein pushing the dynamic electricity price policy to the electricity consumption body terminal, collecting terminal response data and associating the electricity price information optimization policy, comprises: Acquiring the adjusted dynamic electricity price policies in different time periods, encrypting and broadcasting the acquired adjusted dynamic electricity price policy data by using an electric power hong equipment management protocol, and sending the encrypted and broadcasted data to an electric power consumption main body terminal, wherein the electric power consumption main body terminal needs to decrypt and verify the data integrity and the policy signature in an electric power hong credible execution environment when receiving; Judging whether the electricity price information exceeds a preset electricity price threshold value according to dynamic electricity price strategy data acquired from an electricity consumption main body terminal through a preset rule; If the electricity price information exceeds the preset electricity price threshold value, triggering an instruction to close the high-power electric appliance or delay starting the high-power electric appliance; performing association analysis on electricity price information and response behaviors of the electricity consumption main body terminal to generate a response behavior data set; Training a power utilization main body response behavior prediction model by using a decision tree algorithm; Applying the trained prediction model to new electricity price strategy data to generate a power utilization main body response behavior prediction result; according to the power utilization main body response behavior prediction result, using an optimization algorithm to adjust a preset power price threshold value and a pushing mechanism in a dynamic power price strategy, and optimizing a power utilization main body terminal response rule; And pushing the adjusted electricity price strategy to the electricity consumption main body terminal again to complete closed loop optimization.
  6. 6. The method of claim 2, wherein the real-time monitoring of the regional load, if the load is abnormal, generating a regional demand side response strategy in combination with the electricity consumption subject classification result and the environmental data, further optimizing according to the electricity consumption subject feedback, comprises: Acquiring encrypted area load data from each electric power Hong Meng Bianyuan node in real time, after decryption in an electric power Hongzhong trusted execution environment, comparing the encrypted area load data with a preset load threshold value, and judging whether the current area load exceeds the preset load threshold value; if the regional load exceeds a preset load threshold, generating a demand side response strategy adjustment scheme by adopting a decision tree algorithm according to the environmental data and the electricity consumption main body classification result; aiming at the electricity price adjustment, calculating to obtain an electricity price change range and an adjustment period by adopting a linear regression model according to the exceeding degree of the load and the historical electricity price data; Aiming at limiting electricity consumption, determining specific time and power range of limiting electricity consumption according to a time period with exceeding load and a pre-acquired electricity consumption habit of an electricity consumption main body; pushing the electricity price change range, the adjustment period, the electricity consumption limiting time and the power range to at least one electricity consumption main terminal; According to response data of a contralateral response strategy adjustment scheme obtained from each power utilization subject terminal, updating power utilization subject power utilization behavior models of different types of power utilization subjects by adopting a K-means clustering algorithm, and obtaining updated power utilization subject power utilization behavior models of different types of power utilization subjects; and re-evaluating and optimizing the next demand side response strategy based on the updated electricity consumption behavior models of the electricity consumption main bodies of different types to obtain an adjusted demand side response strategy.
  7. 7. The method of claim 1, wherein the real-time monitoring of the regional load, if the load is abnormal, generating a regional demand side response strategy by combining the electricity consumption subject classification result with the environmental data, further optimizing according to the electricity consumption subject feedback, further comprises: acquiring load data of all power hong Meng Bianyuan nodes in an area, inputting the load data of all power hong Meng Bianyuan nodes into an adjusted demand side response strategy, and recalculating power grid load distribution by adopting a load distribution algorithm in the strategy; Inputting the recalculated load distribution into a preset balance judgment model, and judging whether the load of each node is in a preset balance range or not through the model; If the load distribution is balanced, generating a final demand side management instruction comprising electricity price adjustment and limiting electricity utilization time and power range according to the recalculated load distribution; if the load distribution is unbalanced, optimizing the load distribution by adopting a genetic algorithm, inputting the optimized load distribution into a preset electricity habit model, and determining the updated specific time and power range for limiting electricity consumption; And (3) for electricity price adjustment, recalculating an electricity price change range and an adjustment period by adopting a linear regression model, and combining a calculation result with the updated limited electricity consumption time and power range to generate an updated final demand side management instruction.
  8. 8. The method according to claim 1, wherein the recalculating the power grid load distribution according to the adjusted comprehensive strategy, generating a final management instruction after determining the balance, and issuing the final management instruction to be executed, and obtaining the load optimization effect data includes: Issuing a final demand side management instruction to each electricity consumption main body terminal; each electricity consumption main body terminal receives the final demand side management instruction and executes load optimization operation; for load optimization operation, optimizing load distribution by adopting a genetic algorithm, wherein the specific implementation process comprises the steps of initializing population, calculating fitness, selecting, crossing and mutating to obtain optimized load distribution; Inputting the optimized load distribution into an equalization judging model, and judging whether the load is in a preset equalization range or not based on a preset equalization threshold value by the equalization judging model; If the load distribution is unbalanced, calculating the electricity price change range and the adjustment period by adopting a linear regression model, wherein the specific implementation process comprises data collection, model training and prediction; according to the electricity price change range and the adjustment period, the electricity consumption limiting time and the power range are combined, and the final demand side management instruction is updated; and transmitting the updated final demand side management instruction to each power consumption main body terminal, and re-executing the load optimization operation to obtain load optimization effect data.
  9. 9. The method according to claim 1, wherein the recalculating the power grid load distribution according to the adjusted comprehensive strategy, generating a final management instruction after judging the balance, issuing and executing the final management instruction, obtaining load optimization effect data, analyzing environmental factor influence, and dynamically updating the power consumption behavior model and the response strategy comprises: Inputting the power grid load data and the load optimization effect data into a pre-established load monitoring model according to the power grid load data and the load optimization effect data, and obtaining processed data output by the load monitoring model; processing the processed data by adopting a time sequence analysis method, identifying holiday effect and temperature sensitivity in the processed data, and extracting electricity consumption data of electricity consumption main bodies of different types from time sequence analysis results; Inputting the electricity consumption data of the electricity consumption main body into a behavior model training module, and updating and training the updated electricity consumption main body electricity consumption behavior models of different types of electricity consumption main bodies by adopting a gradient descent method in the behavior model training module to obtain the latest electricity consumption main body electricity consumption behavior models of different types of electricity consumption main bodies; and inputting the electricity consumption behavior models of the latest electricity consumption main bodies of different types into a preset strategy generation module, and combining the adjusted demand side response strategy in the strategy generation module, optimizing by adopting a decision tree algorithm, and updating the demand side management strategy.

Description

Demand side response power utilization management method based on power Hongmon architecture Technical Field The invention relates to the technical field of information, in particular to a demand side response application electric management method based on an electric power hong Monte framework. Background In power systems, the surge in power demand during peak hours is a long standing problem. With the acceleration of the urban process and the continuous increase of industrial electricity, the power grid is subjected to huge supply pressure in a specific period. Traditional power dispatching methods mainly rely on resource allocation at the supply side, such as increasing the output of a generator set or calling a standby power supply. However, such unilateral adjustment is often difficult to accurately match the change in demand, and may result in wasted resources or unstable power grids. Particularly in extreme weather conditions, the fluctuation of power demand is more severe, and the traditional scheduling mode is worry about. Demand side management is an emerging regulation means aimed at relieving the grid pressure by regulating the electricity consumption behavior of the electricity consumption main body end. However, current demand side management techniques face multiple challenges in practical applications. First, the diversity and randomness of the electricity consumption behavior of the electricity consumption main body makes it difficult to ensure the accuracy of demand prediction. Secondly, the willingness of the electricity consumption main body to participate in the management of the demand side and the response speed are uneven, so that the regulation and control effect is unstable. In addition, the existing demand side management platform often lacks intelligent capability, and cannot analyze mass data in real time and make accurate decisions, so that efficient resource allocation is difficult to achieve. Aiming at the problems, the prior art provides an intelligent management method based on a power budgeting framework. The distributed intelligent communication framework for the novel power system is characterized by comprising (1) a distributed intelligent computing unit (power hong Meng Bianyuan node) with data acquisition, edge computing and security cooperation core functions, (2) a power full scene protocol framework (power hong Mongolian protocol stack) supporting a long-short term memory network anomaly detection module, a dynamic reputation value evaluation module, a heterogeneous accelerator and a power hong Mongolian equipment management protocol, and (3) a security computing enclave based on hardware isolation, wherein the security computing enclave is used for providing a security capability environment (power hong Mongolian trusted execution environment) with three-in-one of data privacy protection, code integrity verification and physical attack prevention for power core business (such as load prediction and electricity price policy generation). The method is characterized in that the electricity consumption behavior of the electricity consumption main body is monitored and analyzed in real time through advanced data analysis and machine learning technology, and the response strategy of the demand side is dynamically adjusted. However, how to realize accurate prediction and quick response in complex and changeable electricity use scenarios is a technical problem to be solved urgently. Particularly, in the aspect of fusion and processing of multi-source heterogeneous data, how to ensure the real-time performance and accuracy of the data and how to realize efficient regulation and control on the premise of ensuring the electricity consumption experience of an electricity consumption main body are required to be studied and innovated. Disclosure of Invention The invention provides a demand side response application power management method based on a power budgeting framework, which mainly comprises the following steps: collecting multi-source electricity consumption related data, cleaning and integrating, then combining the environmental data, extracting electricity consumption, load values and power factor characteristic values, and clustering to obtain three types of electricity consumption main body classification results of industry, business and residents; Encrypting and transmitting the classification result to a safe execution environment, fusing environment data and electricity utilization characteristics of electricity utilization main bodies, and constructing electricity utilization behavior models of different types of electricity utilization main bodies; Judging electricity consumption peaks according to model output, generating dynamic electricity price strategies and automatic response rules which are adapted to different electricity consumption main body types, and considering electricity consumption requirements and regulation and control targets; pushing a dy