Research And Application Of Short-term Load Forecasting Method For Power System

The power industry is an important basic industry in the national of social and economic development.Short-term load forecasting is an important part of power system,which is of great significance to the safe,economical and stable operation of the power grid.With the rapid development of social economy,people’s quality of life is improving and climate have been greatly affected.The impact of climate change on the load is growing.The power system puts forward higher requirements for the accuracy and rationality of load forecasting.With the development of artificial intelligence technology,smart grid has become a new direction of power system development.It is of great significance to apply artificial intelligence technology to short-term load forecasting to improve the performance of prediction.This paper takes the prefecture-level city load in coastal areas of Zhejiang Province as the analysis object.The short-term load forecasting method of power system is studied and analyzed.Firstly,the load characteristics of the object are analyzed,and it is found that the load has the characteristics of intrinsic law and external factors.The internal law concludes that the load has periodicity.External factors are temperature,humidity,weather,precipitation and other meteorological factors have different effects on the load curve.Therefore,the input parameters of the load forecasting are combined with periodic load data and meteorological data.To prepare for load forecasting,load data and meteorological data are preprocessed,at the same time,the abnormal data is removed and the missing data is deleted.In recent years,artificial neural networks have been widely used in short-term load forecasting.BP neural network is a common method for load forecasting.BP neural network is a common method for load forecasting.As a static network,BP neural network through the self-learning method to correct the weight to achieve recognition error requirements.However,BP neural network is easy to fall into local minimum.Elman neural network is a recurrent neural network,compared with the BP network more than a feedback link,that is,to undertake the layer.This layer can store the output of the memory hidden layer before the moment,and can improve the processing power of network dynamic information.T-S fuzzy neural network combines neural network with fuzzy control,both the ability to express qualitative knowledge and good learning ability.In this paper,BP neural network,Elman neural network and fuzzy neural network are used to establish the short-term load forecasting model of power system,and the load data of the analysis object are predicted.The results show that the performance of BP neural network,Elman network and fuzzy neural network is better than that of the former.Aiming at the ability of T-S fuzzy control to have the function of continuous function mapping and the dynamic information processing of Elman neural network,this paper proposes a short-term load forecasting model for power system based on T-S fuzzy Elman neural network.The model not only has the ability to express qualitative knowledge,but also has the ability of recursive neural network to deal with dynamic information,and can express the dynamic mapping relationship between network input and output well.The model is applied to the load forecast of the analysis object.It is found that the prediction effect is superior to the three networks mentioned above,and the system has high prediction accuracy and good application prospect.At the end of this paper,a short-term load forecasting software system is built by combining the good interface capability of C# and the computing power of MATLAB.It provides a simple and good software environment for staff of short-term load forecasting.