CN-122022315-A - Comprehensive energy intelligent regulation and control auxiliary decision-making method and system

CN122022315ACN 122022315 ACN122022315 ACN 122022315ACN-122022315-A

Abstract

The application provides a comprehensive energy intelligent regulation and control auxiliary decision-making method and a system, which relate to the technical field of intelligent regulation and control, wherein the method comprises the steps of collecting various target influence data, constructing an influence characteristic vector according to the target influence data, constructing an LSSVM model, searching an optimal parameter combination of the LSSVM model by adopting an SSA algorithm to obtain an optimized LSSVM model, and inputting the influence characteristic vector into the optimized LSSVM model to obtain a load prediction result; and constructing a multi-target comprehensive energy system planning optimization model, and solving the multi-target comprehensive energy system planning optimization model by adopting an ant colony-genetic algorithm based on a load prediction result to obtain and feed back equipment configuration and an electric power output strategy. The method can improve the accuracy and reliability of the load prediction of the comprehensive energy system.

Inventors

XU XIAOLONG
WANG YANG
SONG LIANG
LI GUOLIANG
WANG CHUANG
WU GUOJING
WANG TAO

Assignees

国网山东省电力公司枣庄供电公司

Dates

Publication Date: 20260512
Application Date: 20260128

Claims (10)

1. The intelligent comprehensive energy regulation and control auxiliary decision-making method is characterized by comprising the following steps of: Collecting various target influence data, constructing an influence feature vector according to the target influence data, constructing an LSSVM model, searching an optimal parameter combination of the LSSVM model by adopting an SSA algorithm to obtain an optimized LSSVM model, and inputting the influence feature vector into the optimized LSSVM model to obtain a load prediction result; And constructing a multi-target comprehensive energy system planning optimization model, and solving the multi-target comprehensive energy system planning optimization model by adopting an ant colony-genetic algorithm based on a load prediction result to obtain and feed back equipment configuration and an electric power output strategy.
2. The integrated energy intelligent regulation and control aid decision making method according to claim 1, further comprising: calculating a prediction error based on a load prediction result and a load theoretical result, arranging all the prediction errors according to time sequence to obtain a prediction error sequence, calculating an accumulated error of the prediction error sequence by using a Page-Hinkley inspection algorithm, and searching the optimal parameter combination of the LSSVM model again by using an SSA algorithm when the accumulated error is larger than a preset accumulated error threshold.
3. The intelligent comprehensive energy regulation and control auxiliary decision-making method according to claim 2, wherein the sliding window is adopted to conduct flow preprocessing on each target influence data respectively to obtain target window data, and the recursive least square method of RBF kernel function is adopted to update parameters of the LSSVM model on line.
4. The integrated energy intelligent regulation and control aid decision making method according to claim 3, further comprising: Acquiring weather forecast data, calculating the similarity of the weather forecast data and historical weather data, sequencing the historical weather data according to the sequence from big to small of the similarity, determining corresponding sliding windows based on the first n historical weather data, marking the sliding windows as target sliding windows, acquiring load data in each target sliding window, drawing load curves, carrying out weighted averaging on the n load curves to obtain target curves, and generating a base line predicted value through a Prophet algorithm; Processing target window data by adopting an LSTM network to obtain a short-term load prediction feature vector, splicing the short-term load prediction feature vector and a load prediction result to obtain a spliced vector, inputting the spliced vector into a LightGBM model, outputting an error correction term, and taking a base line predicted value and the error correction term as a load theoretical result.
5. The integrated energy intelligent regulation assistance decision making method according to claim 2, wherein when the accumulated error is greater than a preset accumulated error threshold, the method further comprises: extracting error time sequence characteristics from a historical prediction error sequence, reconstructing the error time sequence characteristics by adopting a variation self-encoder, calculating a reconstruction error, identifying an abnormal mode based on the reconstruction error, acquiring parameter combinations of the LSSVM model in the abnormal mode, marking the parameter combinations as target parameter combinations, and taking the target parameter combinations as an initialization population when the optimal parameter combinations of the LSSVM model are searched again by adopting an SSA algorithm.
6. The integrated energy intelligent regulation and control aid decision making method according to claim 5, wherein in the process of re-searching for the optimal parameter combination of the LSSVM model using SSA algorithm, the method further comprises: Three different individuals in the population are randomly selected, variant individuals are generated through a difference strategy, the variant individuals and target individuals are subjected to cross operation to generate test individuals, the fitness values of the test individuals and the target individuals are compared, and the individual with the smallest fitness value is selected to enter the next generation population.
7. The integrated energy intelligent regulation assistance decision making method of any one of claims 1-6, further comprising, after constructing the LSSVM model: it is determined whether historical data of each target influence data is recorded, If yes, index screening is carried out on the target influence data by adopting an analytic hierarchy process; If not, the processing is not performed.
8. The integrated energy intelligent regulation and control aid decision making method according to claim 7, further comprising: And acquiring a historical load corresponding to the historical data, constructing a training set according to the screened target influence data, the historical data and the historical load, training the LSSVM model by adopting the training set to obtain a trained LSSVM model, and processing the influence feature vector by adopting the trained LSSVM model to obtain a load prediction result.
9. The integrated energy intelligent regulation and control aid decision making method of claim 8, further comprising applying a perturbation factor based on a rapid gradient symbology to training data in the training set, generating antagonistic training sample data, and adding the antagonistic training sample data to the training set.
10. The comprehensive energy intelligent regulation and control auxiliary decision-making system is characterized by comprising a processor and a memory which is in communication connection with the processor; a computer readable storage medium is arranged in the memory, and a computer program is stored on the computer readable storage medium; The processor, when processing a computer program stored on the computer readable storage medium, implements the method according to any of claims 1-9.

Description

Comprehensive energy intelligent regulation and control auxiliary decision-making method and system Technical Field The application relates to the technical field of intelligent regulation, in particular to an auxiliary decision method and system for intelligent regulation of comprehensive energy. Background In the current society, the comprehensive scenes such as an industrial park, a traffic scene and the like are taken as important areas of energy consumption, and the energy demand of the comprehensive scenes presents diversified, complicated and dynamically changing characteristics. Many different types of enterprises are converged in the industrial park, the industrial park covers a plurality of fields of manufacturing industry, service industry and the like, the production process, production time and the like of each enterprise are different, so that energy consumption is unevenly distributed in time and space, and great difference exists in the requirements of various energy sources such as electric power, heat power, natural gas and the like. In a traffic scene, along with acceleration of the urban process and increase of travel demands of people, the quantity of various vehicles (such as electric automobiles, rail transit and the like) is increased sharply, the energy consumption structure is also changed deeply, and higher requirements are put on the stability, the high efficiency and the cleanliness of energy supply. The application publication number CN118779664A discloses a multi-element load prediction method, a device and equipment of a comprehensive energy system, firstly, historical data of multi-element loads of the comprehensive energy system and influence factor data corresponding to the historical period are obtained, then, based on correlation among the multi-element loads and different influence factors, specific data composition forms of sample data to be adopted are selected, and under the condition of fully considering the correlation coupling among the multi-element loads and different influence factor data, multi-element load prediction of the comprehensive energy system is carried out through a multi-element load prediction model. Although the SSA algorithm is adopted to optimize initial parameters, a parameter dynamic update mechanism is not established, so that the model gradually deviates from the actual working condition in long-term operation, and the reliability is gradually reduced. Disclosure of Invention In order to improve the accuracy and reliability of comprehensive energy system load prediction, the application provides a comprehensive energy intelligent regulation and control auxiliary decision-making method and a system. In a first aspect, the application provides an intelligent comprehensive energy regulation and control auxiliary decision-making method, which adopts the following technical scheme: An intelligent regulation and control auxiliary decision-making method for comprehensive energy comprises the following steps: Collecting various target influence data, constructing an influence feature vector according to the target influence data, constructing an LSSVM model, searching an optimal parameter combination of the LSSVM model by adopting an SSA algorithm to obtain an optimized LSSVM model, and inputting the influence feature vector into the optimized LSSVM model to obtain a load prediction result; And constructing a multi-target comprehensive energy system planning optimization model, and solving the multi-target comprehensive energy system planning optimization model by adopting an ant colony-genetic algorithm based on a load prediction result to obtain and feed back equipment configuration and an electric power output strategy. According to the method, various factors influencing the load can be comprehensively captured from multiple dimensions by collecting multiple target influence data, and abundant data sources are helpful for grasping the rule and trend of the load change more accurately, so that the accuracy and reliability of load prediction are improved. And then, the application adopts an LSSVM (least squares support vector machine) model to convert the inequality constraint into the equality constraint, and converts the quadratic programming problem into the linear equation system solving problem, thereby reducing the computational complexity and improving the solving speed. And then, an SSA algorithm is adopted to search the optimal parameter combination of the LSSVM model, so that blindness and limitation of manual parameter adjustment can be effectively avoided, and the prediction precision and stability of the model are further improved. The application inputs the influence feature vector into the optimized LSSVM model to obtain a load prediction result, and the load prediction result can help an energy system operator to know the future energy demand in advance, reasonably arrange energy production and distribution and optimize the equipm