This award is for support to investigate the current use of intelligent systems to monitor the health of machinery and other systems. This support is in the form of a research assistantship for a graduate student. An intelligent system is one that is capable of making limited decisions based on the analysis of current information it is receiving and also perhaps based on historical data that it has access to. In recent years there has been a rapid growth in the research effort in the area of expert and knowledge based systems, the latest component of this research being the design and use of neural networks. The principal investigator is doing research on monitoring the condition of valves in high pressure water pumps. The aim of the research is to identify parameters in vibration signals taken on the pump that are sensitive to the valve conditions. Measurements are being taken regularly on several similar pumps at the plant and will be taken over a period of twelve to eighteen months. These measurements will give a history of the vibration signals as the valves deteriorate and also allow comparison between similar pumps. Within two years most of the pumps will need some kind of maintenance overhaul. The types of parameters that will be investigated will be signal power in particular frequency bands, signal statistics, and shape characteristics of the forcing signal at the valve seat, estimating using deconvolution techniques. Having identified these parameters the next stage would be to build some portable device that could pick up the vibration signals from the pumps and rate the condition of the valves. This device could then be used by an unskilled operator to monitor the valve condition.
