Results using trend analysis for predicting automotive maintenance needs

This organization supports the US Army Test Measurement and Diagnostic Equipment Product Manager in a project called ADIP, Army Diagnostic Improvement Program. We provide engineering support by gathering and analyzing data from several vehicles. Such measurements as oil pressure, fuel pressure, compression unbalance, starter current, battery internal resistance, etc, are made. Our aim is to predict changes in these parameters to prevent unscheduled downtime and to fix small problems before they become larger. Explanation of the math used will be given. Use of the requirement tolerance placed as plus or minus bounds around the last measurement and the extrapolated trend line to produce time to exceedance will be described. Using some preliminary data it will be shown how to establish sampling intervals. Using these techniques we have found that for the data available the sample interval is about four (4) months. A number of problems that tend to obscure trends have been identified, studied, and solutions found. A few examples will illustrate some of these. Since the trend analysis is based on historical data from individual vehicles the results apply to that vehicle only. As more vehicles from that family are studied certain family traits emerge hence one may establish general family features.