The Research Of Load Forecasting In ChaoZhou Power Grid

Load forecasting has forward-looking significance for the planning, production and dispatching of power system, and grid load forecasting technology also draw many studies. In order to improve the precision of short-term load forecasting, this paper presents a short-term load forecasting model based on waveletBP neural network, which can improve the average accuracy rate from93.19%to96.53%.This paper overviews the research status of load forecastingat home and abroad, and compares the advantages, disadvantages and applicability of various methods. The basic principle and process are also introduced in this paper.For the purpose of avoiding the disadvantage of BP neural network, this paper proposes a waveletBP neural network to improve the forecasting precision which only uses neural network. Multi-scale analysis based on wavelet analysis is used to decompose the signal sequences into different frequency signals, that is the low frequency signals and the high frequency signals. Low-frequency signal corresponds to the load which is affected by the long-term stability and high frequencycorresponds to the loadwhich is affected by short and random factors. A waveletBP neural network model is proposed, which uses the exponential smoothing method to predict the high frequency signals and uses BP neural network to predict the low frequency signals. And it canwell avoid the situation cross affected by different factors. The model proposed in this paper is used to Chaozhou power grid short-term load forecasting, and the results show thatthis model is an effective forecasting model, and its accuracy is up to96%.