The Short-Term Electricity Load Forecasting Based On Chaos Theory

Short-term electricity load forecasting is important base of safe diapatch and economical operation in power system. The forecasting precision will directly affect the reliability, profit and quality of power system. Power system is a nonlinear system and appears as chaotic behavior. Electricity load is influenced by many factors and its time series is probably not stochastic but chaotic. New chaos theory provides a new method to solve the forecasting problem of electricity load.In this dissertation, many problems about chaos science are discussed and analyzed, such as basic chaos concepts, research information and applications, discrimination of chaotic time series, the theory of Phase Space Reconstruction, parameters selecting and computation of Lyapunov exponent.A combined model of monthly electricity load is studied in this dissertation. Monthly electricity load is divided into trend series and residual series because of its obvious trend. Analysis of residual series shows that it is chaotic. Short-term electricity load could be forecasted through chaos theory.A forecasting method is proposed based on Lyapunov exponent method with rule of acceptance and rejection. Practical example proves that rule of acceptance and rejection influences forecasting precision deeply and the forecasting model of monthly electricity load based on this method is satisfactory.A forecasting model of monthly electricity load based on chaotic optimization is presented by means ergodic characteristic of chaotic motion and analysis of chaotic optimal searching process. Practical results show that this forecasting model can forecast monthly electricity load more rapidly and accurately.