Research On Load Forecasting Of Support Vector Machine Based Oil Intelligent Algorithm Optimization

Load forecasting is an important part of safe and stable operation of power system.High precision load forecasting can also provide data support for power plant construction,power grid supply and demand balance,power grid dispatching,generator start and stop and other related work.With the progress of China’s economic strength and the development of industrial production,power load data is becoming increasingly complex.The accuracy of load forecasting is related to the economy and stability of power grid operation,which makes load forecasting more and more important in the development of power system,and puts forward higher requirements for the accuracy of load forecasting.Support vector machine(SVM)is one of the machine learning algorithms,which is widely used in load forecasting and has achieved high prediction accuracy.In this paper,a load forecasting model based on support vector machine and group optimization algorithm is proposed to forecast power loadSupport vector machine is a machine learning algorithm,which has a solid theoretical foundation.Because of its strong generalization ability and high accuracy in forecasting or classification,support vector machine is selected as the main body of load forecasting model.However,due to the difficulty in parameter selection of SVM,improper selection will affect the prediction accuracy of the model.Swarm algorithm is widely used in various optimization problems or model solving problems.Aiming at the optimization efficiency of swarm optimization algorithm in load forecasting,this paper introduces a new wolf swarm algorithm and particle swarm optimization algorithm to optimize support vector machine for load forecasting.Wolf colony optimization algorithm is a new group algorithm,which has the advantages of high efficiency and relatively not easy to fall into local optimum.In this paper,the support vector machine optimized by Wolf colony algorithm is selected for load forecasting.At the same time,it is pointed out that the wolf colony algorithm has its inherent shortcomings.The fixed step size factor is used in the head wolf calling process,which makes the search efficiency in the gathering process low.In order to solve this problem,this paper proposes an adaptive step size factor,which can improve the search success rate of wolves in the process of aggregation.Particle swarm optimization(PSO)is widely used in various optimization situations because of its simple structure.Particle swarm optimization is also easy to fall into the local optimum,and it is difficult to escape from the local optimum.In view of this shortcoming,this paper introduces the idea of using improved step size factor in wolf swarm algorithm to the original particle swarm algorithm,and transforms the particle swarm algorithm into adaptive particle swarm algorithm,which is used in support vector machine load forecasting model.In addition,some improvements such as chaos initialization and differential mutation are added to make the improved algorithm have better results in the early,middle and late stages of the search,and it is not easy to fall into the local optimum.Wolf pack algorithm and particle swarm optimization algorithm are selected to optimize the parameters that are difficult to be specified in support vector machine,and the prediction is analyzed.Finally,the prediction results of Adaptive Wolf Swarm Optimization Support Vector Machine(AWPA-SVM),Wolf Swarm Optimization Support Vector Machine(WPA-SVM),Adaptive Particle Swarm Optimization Support Vector Machine(APSO-SVM),Particle Swarm Optimization Support Vector Machine(PSO-SVM)and Unoptimized Support Vector Machine(SVM)are compared through examples.Experimental results show that the proposed idea can improve the prediction accuracy of the algorithm better.