PMU‐Based Wide Area Monitoring With Machine Learning to Prevent Blackouts in Bangladesh Power System

ABSTRACT

The electrical power system must be trustworthy and secure enough to provide a continuous supply to meet the power demand. With the complexity of the electricity system growing, the likelihood of blackouts and outages is rising. Therefore, an effective control system is required to increase the power system's safety, effectiveness, and reliability. Numerous new possibilities have been made possible by recent advancements in measurement, communications, and analytical technologies. Particularly, wide area measurement systems (WAMS) have gained attention for addressing anomaly operations. The fundamental component of WAMS is a Phasor Measurement Unit (PMU), which offers a dynamic view of the power system since GPS gives a timestamp and synchronized phasor. Combining these synchronized measurements in a central protection system (CPS), a wide area control, protection, and optimization platform is created by means of optical fiber communication. Since PMU devices are more efficient than standard SCADA (Supervisory Control and Data Acquisition) systems at capturing the rapid dynamics of the power system with high‐accuracy measurement, this technology is becoming more and more popular in the utility industry. The suggested concept is designed to protect big power transmission networks with PMUs while overcoming conventional limitations. The proposed technique can identify faults within a few milliseconds, which is very rapid, and this technology has been tested in the Bangladesh Power System of the Chattogram region. The deployment of PMUs can produce massive quantities of data, and by analyzing large dataset, machine learning (ML) techniques can be very useful in preventing disasters caused by unexpected outages by promptly identifying them. This study has provided the classification methods for K Nearest Neighbors, the Logistic Regression method and the Support Vector Classifier. These results indicate that they are quite accurate at identifying anomalies in the data provided by PMUs. To increase the accuracy of WAMS with ML, a rectangular window feature is integrated in the generated data of PMUs. The window feature with ML shows a significant improvement in WAMS. A Unified Real‐time Dynamic State Measurements (URTDSM) system with PMU and Phasor Data Concentrator (PDC) deployment plan has also been proposed for the Bangladeshi power system (BPS) using Wide Area Monitoring, Protection and Control (WAMPAC), which appears to be the most advanced technology for Bangladesh.