Research On Short Term Load Forecasting Of Power System Based On Random Forest

With the rapid development of China’s economy and electric power industry,the demand for electricity is increasing dramatically.Howerve,short term load forecasting of power system is one of the most important tasks in power sector,which affects scheduling,formulation and implementation of power generation and power storage in power system.In the long time,Scholars both at home and abroad are devoted to study how to improve the accuracy of load forecasting,which is critical to the capital and safety of power grid operations.Therefore,it is significant for load forecasting to propose a load forecasting method with higher feasibility and precision.In this paper,based on the previous research,the random forest(RF)algorithm in data mining is introduced into load forecasting,and two load forecasting methods are proposed:Short-time load forecasting based on fuzzy clustering and random forest;Short-term load forecasting method based on wavelet transform and random forest regression.Firstly,this paper introduced the classification and characteristics of factors of power load forecasting,mainly analyzes the influence of real-time meteorological factors on the load and summarizes the research status and development of domestic and foreign load forecasting.And this paper expounds the basic theory of data mining and random forest.and analyzes the advantages of random forest algorithm.Secondly.the short-term load forecasting based on fuzzy clustering and random forest is introduced in detail.Combined with fuzzy clustering,random forest is introduced in load forecasting in this paper.And a method of combination of fuzzy clustering and random forest for load forcasting is proposed in this paper.On the other hand,various feature of the periodical load and the similarity of input samples are considered in the proposed method.Input samples are clustered depending on similarity.Then load forecasting model is established based on random forest algorithm and similar data are selected as training samples.The final results rely on the historical loads in Anhui for hourly load forcasting.Thirdly.Short term load forecasting based on wavelet analysis and stochastic forest is introduced in detail.According to the principle of wavelet analysis,the load sequence can be decomposed into high frequency and low frequency components;the random forest algorithm is used to establish the prediction model of the decomposed wavelet components by analyzing the characteristics of wavelet components;the components of different frequency bands are reconstructed.The historic load data is selected from a certain area of Anhui province in example.Compared with the traditional BP neural network,support vector machine.and not improved RF,the result shows that the proposed method has higher forecasting accuracy in this paper.At last,the paper is summarized,and the content of the paper is prospected.