Research On Power Load Forecasting Based On QGA And SVM

The operation planning of electric power industry is inseparable from the accurate forecast of load.Load forecasting has many advantages,such as ensuring the smooth operation of power system,controlling the cost of power generation,and making power generation more economic.Therefore,it is necessary to introduce a new machine learning method,which must have nonlinear learning ability.After comparing various prediction algorithms,support vector machine(SVM)has many advantages and is more suitable for power load prediction algorithm.Therefore,this paper adopts the support vector machine method for power load prediction.Based on the internal law of power load data,this paper analyzes the outer factors and build the load forecasting model,in view of account the precision and reliability respectively.Then the forecasting model is optimized to moreover improve the precision of load forecasting.This paper has completed the following main work:(1)Taking the external factors such as temperature,climate and date type into consideration,a prediction model based on load point is established.(2)A self-adjusting quantum gate evolution method and a modified quantum state premature convergence strategy are proposed to solve the defects of the original quantum evolutionary algorithm which is premature convergence and falling into the local optimum.Based on the deviation of individuals from the optimal individuals in the population,the algorithm self-adjusts to generate quantum gates.For the selection of the reference Angle of the quantum rotation Angle,the exponential periodic attenuation strategy is adopted,and the modified quantum state immature concentration tactic is imposed to fulfill the algorithm to shorten the potential of the arithmetic running into the part-prepreerence.(3)For SVM exhaustive CSL parameters value conceptions and Time-cost conundrum,it come up with the modified quantum evolutionary algorithm(QGA)enhance the above mode.QGA-SVM neural net forecast model is defined,from the perspective of the laboratory finding,the devised improved message model can build up the aiming of electrical load prognoses.(4)Based on the successful improvement of the load prediction model,a prediction system based on the quantum evolutionary algorithm is developed to optimize the SVM.From the final operation results,the developed load prediction system has preliminarily completed the related work of power load prediction.