Applicability of ANN for Reliability Analysis of Distribution Network

Reliability, in simple words, can be considered as the capability of system to survive. Presently, consumers are demanding reliable and cheaper power supply with reduced interruption duration. The failure rate of components increases with aging, which will degrade the reliability of the network and make the system more unreliable. Being the only link between consumers and utility, it becomes important to analyse and enhance the reliability of distribution network. An effective method to evaluate the distribution system reliability in terms of Customer Damage Function (CDF) using Artificial Neural Network (ANN) is proposed in this paper. The feed forward type Neural Network (NN) is designed and trained with Bayesian Regularization learning rule. This developed NN is used to predict the Composite Customer Damage Function (CCDF) for various time intervals using the historical data. A prediction of CCDF for proper calculation of interruption cost is necessary to evaluate its impact on system reliability and to identify the aging impact of components. The radial distribution network of the Roy Billionton Test System (RBTS), which is connected at bus-2 is utilized as a case study, where various types of loads (Residential, Commercial, Small users and Govt. and Inst. Buildings) are connected.