Anomaly detection of large data stream in energy internet based on high-order statistical features

Abstract

In the energy Internet integration mechanism, any abnormal data behaviour may affect the operation security of the network system. In order to accurately detect the abnormal transmission behaviour of data stream samples, this paper studies the abnormal detection method of the large data streams in energy Internet based on high-order statistical features. According to the principle of higher-order statistics, this paper defines statistical indicators. Combined with the characteristic function expression, the range of phase space parameters is determined, and the verification of abnormal large data stream information is realized. Hadoop distributed detection framework is set, and the accurate calculation result of abnormal scheduling coefficient is obtained by solving the vector of the high-dimensional large data stream, and the design of abnormal detection method for large data stream of energy Internet based on high-order statistical characteristics is completed. The experimental results show that under the high-order statistical principle, the detection accuracy of abnormal information of data stream samples by energy Internet hosts is improved, which has outstanding value in maintaining the running security of the network system.