Smarter maintenance through internet-based condition monitoring with indirect sensing, novelty detection, and XML

Abstract

In engineering, combining a number of solutions and technologies can result in more effective systems than using only one approach on its own. In particular, it has been shown that in condition monitoring (CM), smarter maintenance systems may be obtained by integrating various sensors together. This paper extends this idea by integrating various non-homogeneous technologies horizontally. The proposed system is an internet-based condition monitoring (e-CM) prototype that can identify abnormal tension in moving belts. It is shown that by applying a classification technique, known as novelty detection, it is possible to decide the status of belt tension by processing the belt vibration signals from an optical sensor (i.e. an indirect sensing approach). A novel method for industrial network communication using XML to create a single standard format for sensor information is also used to link the sensor to the process controller via the internet using the flexible CAN bus technology; this is used together with low-cost microcontrollers with a built-in ethernet link for data acquisition and transmission. The resulting integrated approach is more efficient because: (a) it can reduce waste by minimizing process interruptions caused by direct belt inspection methods while obtaining high detection accuracy (99.67 per cent) and (b) it can provide on-line remote CM that is cost-effective, simple, standardized, and scalable across a wide area and for a relatively large number of sensors. This improvement is especially important when applied to bottleneck processes and critical components.