Application of Machine Learning and Hyper-Parameter Optimisation for Efficient Prediction of Transient Stability

This paper focuses on the assessment of the feasibility of application of two representative Machine Learning Techniques (MLTs) for the regression task of predicting a Composite index (COMI) developed and proved to be sufficient in the past study. Hyper-parameter Optimisation (HPO) techniques have also been used for optimising the regression models. Based on the COMI calculated for a large number of probabilistic Monte Carlo (MC) simulations, all the HPO techniques have been tested for two most representative MLTs, Artificial Neural Network (ANN) and Support Vector Regression (SVR), respectively. After the best HPO techniques for the task are chosen, the optimised MLT models are applied to COMI calculated with different integration periods. The results show that with appropriate tuning, the MLTs are adequate for solving the regression task, and the ANN is performing better than the SVR.