Fault Detection and Diagnosis for A Multi-Actuator Pneumatic System

of the Dissertation Fault Detection and Diagnosis for A Multi-Actuator Pneumatic System by Kunbo Zhang Doctor of Philosophy in Mechanical Engineering Stony Brook University 2011 In pneumatic actuating systems, various kinds of faults are key factors in degrading system performance and increasing air consumption. It is therefore valuable to monitor pneumatic systems and implement predictive maintenance based on fast detection and diagnosis of fault conditions. This research investigates effects of leakages on a PLC control industrial multi-actuator (9 cylinders) pneumatic system. Leakages at 8 different levels and 9 different places are introduced in experimental tests. The dynamic models of three major parts in a pneumatic system, actuators, control valves and tubes, are discussed and extracted as system performance features in a quantitative study of leakage fault. Due to nonlinear properties of compressed air and friction, derived dynamic model alone is not able to effectively indicate fault location and level with an expected accuracy. On the other hand, new qualitative methods using processed sensory information for recognizing fault and estimating its levels are devised. The reliability of fault detection and diagnosis solution in a pneumatic system offered by the mathematical tools is found to be highly dependent on the successful selection of input features those are extracted from original signals and the relationship between those extracted features. Finally we present the multi-actuator based