Investigation of Fault Detection Techniques for an Industrial Pneumatic Actuator Using Neural Network: DAMADICS Case Study

The objective of this work was to develop the novel approach for fault detection using neural network in industrial actuator for DAMADICS benchmark case. This neural network model has the ability to produce effective result for fault detection. In this paper, a model-based technique is proposed for the residual generation which results from the deviation of fault-free behavior of actuator from faulty behavior on actuator. The actuator is the multi-input-multi-output (MIMO) system which is designed using four kinds of neural network architectures (NNARX, NNARMAX, NNRARX, and feed forward), and best structure is chosen based on performance indices. The actuator faults can be grouped using k-means clustering technique. This technique is applied to the DAMADICS benchmark case.