ABSTRACT
In data analytic area, various imputation methods have been proposed in the literature to handle the missing-data issue. Some examples are mean imputation, maximum likelihood imputation, and multiple imputation. This chapter highlights the missing-data issue in power system Stability Assessment (SA), and presents four novel data-driven Intelligent Systems (ISs) to enhance the on-line and real-time power system SA/stability prediction by robustness against the missing-data events. The main feature of these ISs is that they can maintain a high accuracy when the measurement data is only partially available. The chapter uses evolutionary computing to optimize hidden nodes and new parameter in single random vector functional link (RVFL) training to pursue the best RVFL performance. To improve the on-line SA robustness against missing-data, the chapter develops a robust ensemble IS that can sustain on-line SA accuracy under PMU missing conditions. The chapter aims to counteract different missing-data issues in post-disturbance short-term voltage stability prediction.
