The Method Of Optimum Combined Power Load Forecasting Based On Differential Evolution Optimization Weight

Short-term power load plays a very important role in the power system,which is mainly used for reasonable arrangements for production planning,fuel procurement plans,maintenance plans and security analysis of the system.However,power load forecasting is susceptible to date type,climatic factors,major social events and other uncertain elements,so it is difficult to predict accurately.Hidden Markov Model(HMM),as a probability model for describing the statistical properties of dual random processes,has a strong ability to predict the structure of the data,but the setting of the initialization matrix has a great influence on the prediction results.In this paper,the influence of parameter setting on training precision is analyzed by a single variable method.The setting principle of initial matrix and observation state number and the optimization method of hidden state number based on information entropy are proposed.In this paper,the problem of data underflow in iterative operation is solved by introducing the scale factor correction model training algorithm.BP neural network has been widely used in short-term load forecasting and has achieved very good results.BP neural network has a strong learning ability,but there are shortcomings such as slow convergence and easy to fall into local optimum.In this paper,we propose an improved particle swarm algorithm to train BP neural network and optimize the network parameters.On this basis,the date and temperature factors that affect the load forecast are taken as the network input,and the short-term power load forecasting model is established.In order to further improve the prediction accuracy,this paper proposes a short-term load forecasting method based on the optimization weight coefficient of the differential evolution algorithm combined with the hidden Markov model and the BP neural network prediction model,which overcomes the shortcomings of the single prediction method and obtains the performance more stable.A higher accuracy prediction model.The method is applied to the power load forecast of Tianjin Power Supply Bureau,which verifies the validity and practicability of the method.The improved method of Hidden Markov model and BP neural network prediction model has greatly improved the prediction accuracy.On this basis,the combination forecasting method is proposed,which further improves the predictionaccuracy and has better theory Significance and application prospects.