It is essential to estimate safety and dependability of complex and massive scaled system. Fault analysis has been broadly used to determine the reliability of the complicated system. It is a legitimate and diagrammatic method for judging the occurrence of an event resulting from continuances and combinations of failure events. The fault analyzer defines an accident type and explains the connection between the failure of components and discovered system and the possibility of a top event or an undesired event is a function of the failure possibility of the system. In traditional fault analysis system, the failure possibilities of elements were considered as correct values. However, it is usually hard to predict precise failure possibility of the elements due to inadequate data. Hence, in the inadequacy of accurate data, it might be necessary to work with rough assessments of probabilities and the failure probabilities are employed as random variables with identified probability distributions. In this research work, the data is collected from “Guru Gobind oil refinery process plant” and the fault is identified. The failure mode may be like leakage of pipeline, breaking pipe is considered. The fault analyzed in one year is considered and the neural network is trained as per the collected data. The dataset values are stored in the excel sheet and these values are named as ground values. GUI of the proposed work is designed in MATLAB tool. The test data is uploaded for evaluating the fault. The results obtained after comparing the predicted fault analyzed by suing neural network with ground value are analyzed