Web-based monitoring and fault diagnostics of machinery

This paper presents online Web-based machine fault monitoring, prediction and diagnosis system. In addition, remote and local clients have some limited control function such as on/off are added. Online video images are provided via a Web camera. Neural networks and fuzzy computing techniques are used to analysis, diagnose and predict the machine behavior. An experimental laboratory structure consists of a rotor with two rigid discs and a flexible shaft was designed, installed and tested. A v notch crack was introduced at various depths and locations along the shaft. Frequency response measurements were collected for the various combinations of crack depth and location. A neural network and fuzzy logic algorithms were designed to learn part of the test results while predicting the others. Frequency and time responses as well as the soft computing results were published over the World Wide Web in real time. Users can access the system anytime using local networks and the system URL.