Simulation of Event-Based Technique for Harmonic Failures

An event-based technique to determine the optimum performance of power plants is proposed. Also known as Event Modeller, the technique is borrowed from the Event-Tracker principle [1] and Event Clustering method [2], to build on cause-effect relationships of events that potentially link many system inputs to many system outputs. Various signal processing methods use Feature Extraction, Feature Selection and Classifier to detect, solve, and mitigate Power Quality (PQ) Disturbances problems. However, most of the solutions are limited to its internal system parameters and require high computational effort. The Event Modeller, on the other hand, considers both internal and external or environmental parameters, which have the potential to fill the gap to overcome limitations of the PQ problems. The algorithm is very selective as it only focuses on the uniqueness of the real-time data, making it an alternative classification technique to detect the PQ problem. A real-time simulation of a Continuous Ship Unloader (CSU) machine is presented to demonstrate the applicability of the Event Modeller in detecting the harmonic failure which considers the dynamics and the interactions within the plant machinery.