Research On Load Fluctuation And Risk Prediction Of Extreme Load Fluctuation Based On Recurrence Interval Analysis

With the development of economy and society,power users pay more and more attention to the reliability and security of power supply.For power-consuming enterprises,how to avoid abnormal operating conditions and maintain safe operation is the focus of attention.Most of the current research on load focused on electricity consumption or power forecasting,and there is a lack of research on load fluctuations in specific industries.Extreme power load fluctuations often correspond to abnormal operating conditions of the enterprise,which may cause equipment damage,directly affecting the normal production safety of the power-consuming enterprise.Therefore,it is necessary to deeply study the power load fluctuations and the risk of extreme load fluctuation of the power-consuming enterprise in specific industries.The main work and research results of this article are as follows:(1)This paper takes two typical electronics companies in the electronic information industry as research objects,and the recurrence interval analysis method is introduced to study the probability distribution of load fluctuation recurrence intervals.Through comparative analysis,the recurrence law of power load fluctuation of typical electronic enterprises is obtained.The mean ratio of the load fluctuation recurrence interval is introduced to test the scaling behavior.It proves that there is no scaling behavior of load fluctuations in electronic enterprises.Then,the conditional probability density function is introduced to test the shortterm memory effect and analyze its effect on the load fluctuation recurrence interval,which proves that the load fluctuation recurrence interval of electronic enterprises have short-term memory effect.(2)The multifractal detrended fluctuation analysis method is used to verify the long-term memory effect of load fluctuations,which proves the existence of the long-term memory effect.At the same time,the multifractal property and causes of load fluctuations are analyzed.It is found that load fluctuations have complex multifractal property through the comparative analysis of the multifractal parameters of load fluctuations with different amplitude.With the degree of load fluctuations intensifies,the multifractal strength of load fluctuations increases,the risk of power load fluctuations increases.(3)A risk probability prediction model of recurrence of load fluctuations is built in this paper.The change law of the risk probability of recurrence of load fluctuations with different thresholds and different time intervals is analyzed.It is found that with the increase of the time interval,the risk probability of recurrence of load fluctuations within next sampling interval showed a decreasing trend.At the same time,the value-at-risk method is combined with the loss probability density function to quantify the risk loss probability of load fluctuations.It is found that the risk loss probability decreases with the degree of load fluctuations intensifies,and its variation trend is close related to the characteristics of load fluctuations of the enterprise.Finally,surrogate data is utilized to replace the load data selectively and randomly on the basis of verifying the long-term memory effect of power load fluctuations.It is possible to determine the critical value that has no effect on the original power load data sequence by comparing and analyzing the change of the Hurst index after data replacement,thereby determining the risk early warning threshold of extreme load fluctuations.