Real-time Electric Power Telecontrol Transmission Anomaly Detection Based On Anytime Density Clustering

At present,China's economy is developing rapidly,the various sectors are more dependent on the electric power system,whose safe and stable operation is very important.Electric Power Telecontrol Transmission anomaly detection is the guarantee of the power network operation safely,to a certain extent can reduce or even avoid the harm caused by abnormal power network.It is the main difficulty to detect anomaly in telecontrol transmission system by acquiring useful information from a large number of telecontrol transmission data and carrying out anomaly detection analysis in real time.In this paper,we focus on the real-time tele-transmission anomaly detection method based on Anytime density clustering to provide an effective and feasible method for anomaly detection system.This paper firstly introduces the research purpose and significance of this topic.At the same time,the development status and trends of anomaly detection ? data mining and Anytime algorithm are introduced and summarized at home and abroad.Secondly,considering the characteristics of tele-transmission data,a novel prediction method of telecontrol transmission based on dadptive scale wavelet packet transform is proposed.Combined with the sliding window technique,this algorithm decompose the remote data and calculate the mean ratio and variance ratio of the wavelet coefficients in the historical time window and the detection time window.Then,it can carry out the adaptive decomposition,and detect anomalies by comparing with the given threshold value.Then,this paper proposes an algorithm based on Anytime density clustering algorithm(AnyDBSCAN algorithm)to analyze the telecontrol data.The AnyDBSCAN algorithm is based on the DBSCAN algorithm and incorporates the idea of the Anytime algorithm.The algorithm can be terminated at any time,and return to the current one of the best feasible solution.it is used to solve the actual needs of the project.The effectiveness of the algorithm is verified by simulation analysis.Finally,a real-time online anomaly detection system is designed based on Spark,a distributed computing framework.According to the calculation performance of each slave node,the system allocates reasonable resources to each job through the scheduling algorithm,and then uses the AnyDBSCAN algorithm to calculate each job,finally combines and summarizes the final result,And the anomaly detection is achieved by monitoring whether the anomaly class of the model is changed or not.The simulation results show that the system can detect the anomaly better,and through parallel processing,it can greatly improve the computing time and reduce the computing delay.