A Fault Prediction Approach for Process Plants using Fault Tree Analysis in Sensor Malfunction

In this paper, a fault prediction approach for process plants using fault tree analysis is presented in the presence of no or false information of certain sensor. The fault propagation model is constructed by causal relationships from fault tree analysis (FTA). Knowledge about system failure, which is obtained from the fault propagation model, is represented as abnormality patterns in process variables and stored in the knowledge base. The prediction system can identify the cause of system malfunction considering no or false information of sensors by matching the pattern data from process plants with the abnormality pattern in the knowledge base. From unavailability of basic events and sensors, the estimated rates are provided for sequence checking in fault prediction. The proposed approach is applied successfully to a reactor control and protection system (RCPS)