Abstract Comparison of Selected Model Evaluation Criteria for Maintenance Applications.pdf

1 of 26 Comparison of Selected Model Evaluation Criteria for Maintenance Applications Ranganath Kothamasu1, J. Shi1*, Samuel H. Huang1 and H. R. Leep2 1Intelligent CAM Systems Laboratory University of Cincinnati, Cincinnati, OH 45221 2Department of Industrial Engineering University of Louisville, Louisville, KY 40292 Abstract Model Based Preventive Maintenance relies on creating models that can either predict future operating states or upcoming failures directly. Since no modeling algorithm can guarantee a best solution in every situation, it becomes necessary to evaluate the solutions generated by these techniques. This paper reviews some popular criteria traditionally employed in model evaluation. Several evaluation criteria proposed in the literature are restricted in their applicability because of their assumptions about the modeling process/data. Some evaluation criteria are tested on two artificial datasets. The results from our tests indicate that Akaike Information Criterion (AIC) has superior performance. The conclusion has been used and verified in one industrial monitoring application. Keyword: Maintenance, Model Selection, PRESS, AIC and 2R * Corresponding author: J. Shi, 630 Rhodes Hall, University of Cincinnati, Cincinnati, OH 45221. Tel: 513 221- 2885. Fax: 513 556-3390. Email: shijh@. 2 of 26 1. INTRODUCTION Maintenance can be defined as the set of activities performed on a system to preserve or sustain its ability to render service in an efficient manner. Formally, it can be defined as the management, control, execution and quality of activities which ensure optimum levels of availability and overall performance of the plant, to meet business objectives (Rao, 1996). Maintenance in the form of Failure Diagnosis and Detection (FDD) has been influenced and defined by various operational strategies. Initial practices were confined to reactive strategies where maint