Research On Load Forecasting And Ordered Power Management Based On Similar Days And DBN

Energy Internet has become the current trend of power grid development.The larger the power grid scale,the more significant the security issue becomes.High-precision load forecasting and effective orderly power management measures can not only ensure the safe transmission of power in the grid,but also can effectively improve the operating efficiency of the power grid company.At present,the method of load forecasting is divided into two kinds:the traditional data statistics method and the modern artificial intelligence method.Because the traditional data statistics method does not consider the actual operating factors of the power grid,the results are often poor.In the current intelligent algorithm,the most used load forecasting is the BP neural network.However,the traditional neural network cannot realize the training of the multi-layer network and thus affects the prediction accuracy.At the same time,effective input and output variables have a great influence on the prediction accuracy of the power system load forecasting model.Based on this,in order to improve the prediction accuracy of the short-term load of the power system,the main contents of this paper are as follows:(1)The Deep Belief Network(DBN)is composed of multi-layer limited Boltzmann Machine(RBM)stacks.As an effective feature processing method,RBM can perform power system load forecasting to solve high-dimensional,complex,and nonlinear problems.At the same time,DBN uses a non-supervised greedy layer-by-layer training algorithm to form a multi-hidden layer perceptron structure and exhibits excellent performance in regression prediction analysis.Based on this,a load forecasting model is established.(2)Training data is often very large.If you do not directly process the training of the network,it will often increase the complexity of the network model,and affect the accuracy of the results.Based on the above analysis,this paper adopts a load forecasting model based on a combination of similar day and deep belief networks.When selecting training data,the model performs a similar day processing for huge data,and the similarity day is in the level of load consumption and forecast.The date is equivalent,which not only increases the modeling speed but also increases the accuracy of the prediction.(3)Analyzed and researched the current situation of orderly electricity use management in China and the current measures for orderly electricity consumption management in Jiangsu Province.Combined with the actual situation of Jiangsu Electric Power Grid,the industry analyzed the load characteristics of Nantong City,and put forward the corresponding orderly electricity consumption measures.And suggestions.