Distribution system reliability worth analysis with the customer cost model based on RBF neural network

Abstract

Reliability worth analysis is an important tool for distribution systems planning and operations. The interruption cost model used in the analysis directly affects the accuracy of the reliability worth evaluation. In this paper, two interruption cost models including an average or aggregated model (AAM), and a probabilistic distribution model (PDM) are proposed by using the radial basis function (RBF) neural network with orthogonal least-squares (OLS) learning method. The residential and industrial interruption costs in AAM and PDM were integrated by the proposed neural network technique. A Monte-Carlo time sequential simulation technique was adopted for worth assessment. The technique is tested by evaluating the reliability worth of a Taipower system for the installation of disconnected switches, lateral fuses, transformers, and alternative supplies. The results show that the two cost models result in very different interruption costs, and PDM may be more realistic in modeling the system.