Research Of Power Load Forecasting For Power Based On Neural Network And Software Developing

Short-term power system load forecasting relates to the power system scheduling and safety in production. As the reform of electricity market-oriented continues to deepen, the traditional load forecasting methods have been difficult to meet the requirements of the electricity companies, and there is no truly computerized. Therefore, there is important application value to develop a short-term load forecasting software to meet the production requirements of power supply enterprises in cities, applying a reliable prediction method to improve power system short-term load forecasting accuracy.In this paper, the principle, characteristics, research actuality and development trend of short-term power system load forecasting were generalized firstly. Then, it summed up the advantage and problems of neural network short-term load forecasting, which is the popular for the study of intelligent forecasting methods in recent years in the prediction. This paper took the research of short-term load forecasting power sub-system for Hunan Chen Zhou Power International Development Co., Ltd. as the background, established an improved neural network load forecasting model. Pre-screened by smoothing out the representative of input data as training samples, using the momentum VLBP heuristic algorithm reduced the network of local details of the sensitivity of the error surface, effectively inhibit the network caught in local minimum. Finally, the power load forecasting system software was worked out for Hunan Chen Zhou Power International Development Co., Ltd.. The software has been stable operating in the field for half a year, and the on-site operation results showed that the short-term electric load forecasting accuracy significantly superior to the traditional load forecasting methods, which greatly improved the power system operation efficiency,and reduced energy loss of electricity.