Unbalanced Fault Diagnosis in Transmission Networks Using Multiple Model Filters

Abstract

This paper provides a scalable solution to the diagnosis of unbalanced transmission faults. The solution hinges on our previous work on diagnosis of balanced transmission faults using multiple model filters (MMF). The purpose of diagnosis is to support a secondary protection that capitalizes on new sensing technologies to cost-effectively mitigate cascading failures caused by protection misoperations. The essence of the MMF-based diagnosis approach is to probabilistically determine which assumed transmission circuit model is most likely to have generated the observed data among the multiple models of circuit topologies of anticipated contingencies. The key to the ready applicability for the diagnosis of unbalanced faults lies with the ability of the MMF to directly process high-rate sampled waveforms in time domain without any reliance on the decomposed symmetric components of unbalanced phasors. To explain the diagnosis process, the problem formulation and the solution to diagnosis of unbalanced faults are detailed for a small 3-machine 9-bus test system. It is shown that diagnosis delays for unbalanced faults remain under 50 milliseconds at a probabilistic decision threshold of above 95%. The paper applies the concept of the sensor-defined partition of a transmission network model to reduce computation and communication complexity while maintaining the performance for the diagnosis of both balanced and unbalanced faults.