Prediction of Vehicle Reliability using ANN

Abstract

ANNs are usually very effective as computational tools and have found extensive utilisation in solving many complex real-world problems. The attractiveness of ANNs comes from their remarkable information processing characteristics pertinent mainly to nonlinearity, high parallelism, fault and noise tolerance besides its learning and generalisation capabilities. The aim of this paper is to familiarise with ANN-based computing (neuro-computing). The predicted and observed vehicle reliability using trained ANN is very close as compared to Weibull probability distribution. The methodology adopted is demonstrated with the help of a case study which includes collection, sorting and grouping of vehicle failure data. Then distribution parameters are estimated and best fitting probability distribution is identified for predicting vehicle reliability. Subsequently the trained ANN (using SLP model) is used to predict the vehicle reliability. Suitability of a RDBMS (Oracle) for training ANN and predicting vehicle reliability is also presented. The developed methodology has been able to predict reliability of vehicle very close to its observed values.