Tuar&#x2014;Transitory&#x2014;Instabilit&#x00E0; (<bold><italic>T</italic></bold>&#x00B2;<bold><italic>i</italic></bold>): An ML-Based Framework to Predict Transient Instability in a 7-Power Plant Network

Abstract

With the expansion in recent power systems, the boost in renewable energy resources with multiple interconnections, overloading of existing power networks, increased load growth, and equipment failures, power system transient instability issues have exponentially increased. Transient stability in power systems is of particular importance even while performing the steady state analysis of a power system. Power system transient stability assessment is mandatory regularly for power system operation and has a major impact on power system planning. Potential risks of blackouts and failures in power systems can be avoided or minimized with the prompt prediction of transient instability. This research work presents a predictive approach for power system transient instability. The case study is a 735kV, 29-bus, 7-powerplant network involving precise modeling of generators. In particular, the proposed method is represented by different machine learning models through extraction and regression, in which the variables of the power-generating units are used as primary sources/features. Data cleaning and sorting out techniques are being used for refining data. Feature extraction has also been implemented for further cleansing of the data. The result is in the form of concrete and robust classifier models that can overcome power systems’ instability concerns through prompt prediction.

Keywords

• Renewable energy sources
• large-scale integration
• power system transients
• transient analysis
• power system stability
• machine learning