System availability enhancement using computational intelligence–based decision tree predictive model

Abstract

System availability is a key performance measure in the process industry. It ensures continuous operation of facilities to meet production targets, personnel safety and environmental sustainability. Process machinery condition assessment, early fault detection and its management are vital elements to ensure overall system availability. These elements can be explored and managed effectively by extracting hidden knowledge from machinery vibration information to improve plant availability and safe operations. This article describes a decision tree–based computational intelligence model using machinery vibration data to detect machinery faults, their severity, and suggests appropriate action to avoid unscheduled failures. Vibration data for this work were collected using a machinery simulator and real-world machine to show the applicability of the proposed model. Later, the data were analyzed to detect faults using decision tree–based model that was developed in MATLAB. Fault detection classification accuracies of 98% during training and 93% during testing showed excellent performance of the proposed model. The model also revealed that the proposed formulation has capability of detecting faults correctly in the range of 98%−99%. The results showed that the proposed decision tree–based model is effective in evaluating the condition of process machinery and predicting unscheduled equipment breakdowns with better accuracy and with reduced human effort.