Non-Linear Statistical Modeling And Risk Assessment Of Demand Side Electric Energy

Recently,the system decision based on power data analysis and processing is continuously advancing the process of power system optimization,which has been initially applied and practiced in production,transmission and consumption.The optimization and upgrading of the power system "Internet +" urgently requires the technical support of data analysis and energy measurement modeling to make more accurate decisions.Power load data of demand side is random and time-varying.Modeling of traditional structural and electrical parameter has more parameters.It leads to a lot of identification work,which has the problems of slow calculation and low accuracy.Non-mechanical models have the advantages of fewer parameters,less identification workload and high simulation degree in the Current background.Based on this,this paper uses Markov model,extreme value mode and artificial neural network as the theoretical basis.First,using the mechanism conversion to explore the characteristics of peak and valley power consumption.Second,building a long short-term memory model for power load prediction.Finally,the extreme value theory is used to evaluate power grid risks and provide a reference for accurate decision-making.The main work of this article is divided into four parts:(1)Using Markov transformation autore-gressive model to discuss the peak-valley state transformation of electricity load.Through the results of nonlinear test,breakpoint test and stationarity test,the MS(2)-ARIMA((1,4),1,0)model is established.It is predicted that the electricity load will be in a state of reduced electricity load in the near future.(2)For peak-valley load prediction,first use empirical mode decomposition to decompose the sequence into 7 intrinsic mode functions and 1 residual sequence.Then compare the sample entropy values,reorganize each component into random component,detail component and trend component.Finally,the long short-term memory network is used to predict daily resident electricity load.(3)In order to quantify the electricity load operating conditions under extreme conditions,we chose to establish a generalized Pareto model.According to the analysis of the hill graph and the average excess function graph,the threshold value is selected as6.8.And the model is tested,it is found from the test results that the model has a good fit degree and no obvious heteroscedasticity.(4)Calculating the grid overload risk probability by calculating the risk value of different data,and compare the grid operation risk under the three types of data.It isfound that the risk value based on the extreme value distribution is a reasonable overload threshold of the power grid,which can effectively avoid the waste of electric energy.