AN INTEGRATED MONITORING AND DIAGNOSTIC SYSTEM FOR ROLLER-BEARINGS

This paper discusses a machine fault diagnostic system which integrate s three techniques: 1. An autoregressive model, which compresses digit ised vibration signals and preserves the information carried in the or iginal signal. 2. A supervised artificial neural network for fault cla ssification. 3. A fuzzy logic-based ''hypothesis and test'' program, w hich, when the artificial neural network fails to provide any suggesti on, is able to provide the human diagnostician with some initial ''edu cated'' guesses of machine conditions. This integrated machine diagnos tic system was developed on a 486 personal computer. Throughout the co urse of this development, the program has been tested with three types of vibration signal: 1. Vibration signals created using bearing physi cal models. 2. Vibration signals collected from two laboratory experim ents ruing accelerometers. 3. Vibration signals collected from real pr oduction machine tools. In this article, the authors discuss the under lying theory of those three techniques. Experimental apparatus is intr oduced. Performance statistics are provided. For those conditions it w as designed and developed to diagnose, the program demonstrated remark ably dependable performance.