Extreme Learning Machine Based Transient Stability Assessment: An Energy Function Approach

Abstract

One of the most challenging task at Energy Control Centre is the prediction of transient stability. Deregulation and ever increasing integration of renewables has made it complex and more demanding in terms of frequent evaluations. Transient Stability Assessment is computationally intensive task and longer computation cycle brings in obsolete results as system operating state evolves continuously. Extreme learning machine (ELM) offers prediction results comparable to other state-of-art learning methods and has much quicker learning curve. WSCC 3-machine 9-bus system and IEEE 39-bus New England test power system are considered to demonstrate the superiority of ELM as a predictive model. ELM is trained using data points generated through transient energy function (TEF) approach. Transient energy terms along with real and reactive power and fault location are used as input attributes for ELM to quickly and accurately determine transient stability margin (TEM).