Application Of Improved Particle Swarm Optimization In Power Load Forecasting

With the progress of science and technology,the power industry has achieved rapid and sustainable development,the grid structure has being more and more complex,accurate and reliable power load forecast has been an important guarantee for safe and economic operation of power grid.Also,the correctness of power system load forecasting is related to whether the national economy and social production can continue to operate safely and stably.As China's economic development has entered a new normal,the power production structure and consumption growth situation also presents new normal characteristics.In order to meet the national economic industrial restructuring,industrial transformation and people's living needs,a power system load forecast with higher requirements on accuracy and reliability is required.Firstly,the paper introduces the basic situation of power load forecasting and algorithm application,which discusses the characteristics of power load forecasting,analyzes the factors affecting the power load change,and establishes the basic mathematical model of power load forecasting.Studying on the basic principle of particle swarm optimization(PSO),with analyzing the parameters involved in the algorithm.The basic concepts and principles of BP are also included.Secondly,analyzing the main characteristics of PSO,and introducing the improvement of the existing PSO,especially the IPSO-BP,with the improved algorithm of classical PSO,combined with BP and the flow of IPSO-BP are analyzed and introduced in detail;IPSO-BP solves the problem of local extremes and low precision.The algorithm also has advantages over BP in training time and convergence speed.Finally,IPSO-BP is applied to power load forecasting,which determines the basic parameters of the application of IPSO-BP in Jining power load forecasting,and analyzes the main parameters such as structural parameters,inertia weight and learning factors of neural network in detail.After predicted the 2016-2025 power load in Jining by using 2003-2014 power load data,analyzing the prediction data,and comparing the IPSO-BP with the classical BP and the classical PSO,the results show that the IPSO-BP has the advantages of good output stability,fast convergence speed and high prediction accuracy,and the accuracy of load forecasting can be controlled within 3%.