Research On Short-term Load Forecasting Of Power System Based On Lianyungang

Load forecasting is one of the important jobs of the power dispatch, electric power, electrical plan, electricity scheduling departments, and the level of power load forecasting is an important symbol to measure the degree of development of modern power system. To improve the level of power load forecasting, is conducive to planning power management, saving primary energy and reducing the cost of power generation, and improving the economic and social benefits of the power system.This paper first digs into the actual load case based on Lianyungang city systematic. It shows that all the factors what influence the load forecasting such as historical load data、temperature and weather conditions must be considered in load forecasting model. The paper does a lot of pretreatment to the load data and other training sample in order to improve the forecast accuracy and make sure that the smooth data can be identified by model easy.Then the paper introduces the structure and principle of error back propagation algorithm (BP algorithm). The BP algorithm is used in load forecasting, simple、 efficient and feasible, but it takes a long time due to the convergence of the algorithm. Also, it falls into local minimum points easily. So the new optimization algorithm using the particle swarm algorithm to optimize the BP neural network is used. The algorithm can be targeted to optimize the weight value and threshold value of the network structure, in case of constant iteration, it works in the direction of reducing the forecast error. After that, the prediction accuracy is improved greatly so that it can satisfy the basic requirements for load forecasting.