Research On Abnormal State Real-time Detection Technology Of Massive Power Quality Data

In order to solve the complex power quality problems among different areas,the State Grid has built a centralized Grid Harmonic Analysis System with about 10,000 monitoring terminal devices among 26 provinces as the data collection terminal and the State Grid big data center as the computing center.The system is built to monitor the running status of the power grid real-time so that the problems of the power grid occurs under it's running process can be found in time.Quickly and accurately recognize the abnomal state of the power grid is the key to ensure the grid running normally.Traditional power quality anomaly data detection methods have poor adaptability in anomaly detection for data with periodic variation and large fluctuation range.The limited storage and computing capacity of centralized architecture results in data accumulation at data acquisition and processing terminals,and makes data calculation delay increase linearly.Therefore,the following studies have been carried in this paper.Firstly,facing the problem of poor adaptability of traditional anomaly data detection methods such as cloud model and anomaly detection method based on control chart to the data with periodic variation regularity and large fluctuation range characteristic,the traditional control chart is improved,and an online detection method of power quality anomaly data based on improved control chart is proposed,which improves the accuracy of anomal detection of power quality data.Secondly,aiming at the problems of data accumulation,large data calculation delay and uneven distribution of data in different provinces caused by centralized structure of"power grid harmonic analysis system",a real-time detection method of power quality abnormal data based on distributed data center architecture is proposed.By task decomposition and reasonable resource scheduling,the accuracy of the results is guaranteed and the computational efficiency is improved at the same time.Thirdly,the distributed data center management system has been designed and implemented,In this system,the automatic configuration,unified publication and status monitoring of tasks are provided and the deployment and implementation of real-time detection program of power quality abnormal data under the framework of distributed data center are introduced.Experiments show that the proposed real-time detection method of power quality anomaly data based on improved control chart and distributed data center architecture has high computing efficiency and accuracy in real-time detection of power quality anomaly data,and has certain application value.