Research Of Power System Short-term Load Forecasting

With the construction of the smart grid, in order to guarantee the safe and economical operation of the power grid, to improve the short-term load forecasting speed and accuracy have put forward higher requirements. For example, with an annual electricity consumption of 29 billion kwh of medium-sized cities, if the prediction accuracy is improved by 1%, then the annual income can increase 145 million yuan. Using the current load forecasting method to improve the accuracy is quite difficult, and poor versatility, the reason is not consider comprehensive enough to load characteristics. Strong uncertainty factors used in forecasting, like wind speed, the accuracy is not high, so the utilization rate has been low. In the extensive use of intelligent algorithm at the same time prediction efficiency also is gradually reduced. In this paper, solve the problem of uncertainty based on cloud model to improve the efficiency of prediction, at the same time, use cloud computing technology to improve alculation efficiency. Finally, corresponding simulations are made to validate the conclusions.This article refinement the load samples and historical data from the start,against short-term load forecasting, this paper established a comprehensive sample system, to evaluate the factors for load forecasting by analyzing the impact of each factor and load data, and select the most appropriate factors. In-depth study the accumulation effect of temperature, proposed the concept of dislocation samples, to join the training sample to improve the prediction accuracy.Research and validate the feasibility of cloud model optimizing LSSVM in load forecasting. Cloud model optimizing LSSVM can greatly improve the prediction accuracy of uncertainty problems. This article establish a comprehensive predictive model, determined by uncertainty, the influencing factors into uncertainties and determinants, Use cloud model to optimize LSSVM to improve the prediction accuracy of uncertainties, use PSO-LSSVM to improve the prediction accuracy of determinants, get the final forecast result by the weighted.Finally, in order to improve the efficiency of load forecasting, use cloud computing technology to solve long running time and dimension disasters by complex models, and other issues. The article takes the load data and meteorological data of Zhuji City, Zhejiang Province as an example for simulation, the experimental results showed that the method could improve the accuracy and efficiency of loadforecasting.