Research On Wind-electric-heat Correlation Analysis Method Based On Copula Function

In the "Three North" area of China,heat power and wind power coexist.There is a certain correlation between the thermal load,electrical load and wind power output in the same area.The correlation between the three will affect the power system scheduling and engineering calculation,and then affect the accuracy of decision-making.Especially in the economic dispatching of power system,it is necessary to consider the correlation among heat load,power load and wind power output,so as to make a reasonable dispatching plan and improve the wind power consumption capacity.Based on this,Copula fiinction is used to analyze the correlation among wind power heat.First of all,based on the analysis of the characteristics of heat load,electric load and wind power output,preprocess and normalize the three kinds of data,use Pearson correlation coefficient to analyze the correlation degree of heat load,electric load and wind power output data,and lay a theoretical foundation for the later analysis.Secondly,Build a wind power heat correlation model based on Copula function theory and analyze the correlation among the three.Kernel density estimation is used as the method to determine the marginal distribution of random variables.According to the binary frequency histogram,the appropriate copula function is initially selected,and the maximum likelihood method of distribution is used for parameter estimation.The advantages and disadvantages of each copula model are evaluated by Euclidean distance.The analysis of experimental examples shows that the proposed method can well reflect the correlation structure between variables,and is superior to the traditional Pearson correlation coefficient method in describing the correlation,which provides a strong support for the data processing of heat load,electric load and wind power output in the future,and has a very good engineering practical significance.Finally,in view of the shortcomings of the graph observation method in selecting copula function model,such as subjectivity,limitations in selecting copula function and inaccurate parameter estimation results,a multivariate copula analysis toolbox is proposed,Mvcat)is used to analyze the correlation between wind,electricity and heat.It includes copula falilies with different complexity.According to the Bayesian framework of residual Gauss likelihood function,Copula parameters are inferred and potential uncertainties are estimated.Markov chain Monte Carlo with mixed evolution is used,MCMC)method is used to generate copula parameters and estimate the posterior distribution.The copula function is sorted by goodness of fit index and the optimal copula function model is selected.The results of numerical examples show that the commonly used local optimization methods for copula parameter estimation tend to fall into local minima.However,the proposed method solves this limitation and improves the description of related structures.Mvcat can also evaluate the uncertainty relative to the record length,which is very important for the wide application of multivariate frequency analysis.