Research On Short-term Load Forecasting In Intelligent Distribution Network Environment

Accurate short-term forecasting is the basis of power system operation planning,it is an important basis for formulating unit start-stop plans,economic dispatch plans and maintenance plans.With the large number of renewable energy sources accessing traditional distribution networks,it may cause system fluctuations,making it more and more difficult to control the operation of traditional power grids.At the same time,it will also cause a significant increase in the dimension a nd frequency of power data collection,resulting in a large accumulation of power load data.It is very difficult to find a specific expression between historical data and influencing factors according to traditional load forecasting methods.Make the trad itional forecasting method gradually develop intelligently.This dissertation discusses in detail the problems of large-scale access of renewable energy to traditional distribution networks,causing system fluctuations and difficulties in predicting loads.In the smart distribution network environment,the least squares support vector machine,the chaotic optimization particle swarm optimization least squares support vector machine and the artificial fish swarm optimization particle swarm optimization least square support vector machine are used for short-term load forecasting.Expe riments show that the short-term load forecasting algorithm based on artificial fish swarm optimization and least squares support vector machine has higher prediction accuracy.The main work of this article is as follows:(1)Identify and preprocess abnormal data in historical load data;In order to facilitate calculation,normalize the factors that affect the load,introduce the short-term load forecasting model of least squares support vector machine.(2)Since the parameters in the least squares support vector machine model are chosen empirically,there is no theoretical basis and there is a large blindness,so this paper first established a short-term load forecasting model of particle swarm optimization least squares support vector machine.However,in practical applications,we found that in the process of parameter optimization,the particle swarm algorithm can easily calculate the local optimal value,that is,the phenomenon of premature emergence occurs,which leads to inaccurate prediction accuracy.Therefore,in order to solve this problem,this dissertation proposes a chaos-optimized particle swarm optimization least squares support vector machine method for short-term load forecasting and establishes a corresponding forecasting model.(3)By further studying the experimental and theoretical knowledge,we found that the chaos optimization algorithm has nothing to do with the choice of the initial value size in theory,but it is found that they are related to each other in the actual prediction.Ther efore,this dissertation makes use of the characteristics of artificial fish-swarm algorithm for low initial value,and improves the shortcomings that the algorithm can only apply to large-scale search.The artificial fish-swarm optimization particle swarm least squares support vector machine short-term load forecasting model is established.Finally,the three forecasting models are applied to the power system network in a certain province in northwest C hina.It is found that the artificial fish algorithm optimized particle swarm least squares support vector machine method short-term load forecasting model has higher prediction accuracy.