Predicting Faults in Heavy Duty Vehicles using the Mahalanobis-Taguchi Strategy

This paper presents a Mahalanobis-Taguchi Strategy (MTS) based system for predicting faults in heavy duty vehicles. Costs associated with heavy duty vehicle breakdown in a large fleet while in operation can be significantly reduced if faults leading to these breakdowns are predicted and prevented. Fifty-one attributes on the vehicles are monitored in real-time and the data fed into the system. MTS is used to develop a scale to measure the degree of abnormality of these measurements from the vehicles compared to normal measurements. The Mahalanobis distances (MD) for the attributes are calculated, orthogonal arrays (OA) and signal-to-noise (S/N) ratio are used to identify attributes of importance. By reducing the dimensionality, less attributes are tracked which reduces the cost of the system. Criteria for classifying fault measurements are defined based on these variables of importance and the MD scale. The result is a real-time monitoring system that predicts faults in the vehicles thereby preventing breakdowns during operation. The information obtained can also assist in creating an effective preventive maintenance schedule for the vehicles in the fleet