Dr Gopinath O. Chandroth recently visited the Lodge and gave a seminar on Fault Diagnosis of Machinery using Neural Networks, the subject of his recent doctoral research at the University of Sheffield.

Gopi is a graduate in Marine Engineering at the University of Calcutta where he received his BSc and MSc degrees and has more recently been award a PhD from the Department of Computer Sciences at the University of Sheffield. He is currently a Marine Surveyor at Lloyds Register of Shipping where he works in the Technical Investigations Department on Software Project Management with particular applications to condition monitoring.

The title of the Seminar was "Neural Networks in Machinery Health Monitoring". The lecture described how incipient faults in a diesel engine were diagnosed using neural nets trained on sensorial information. Solutions were investigated for their independence using statistical models and the most diverse of them were combined to produce the best overall diagnostic package.

Gopi studied the problem of machinery fault diagnosis and, in particular, tried to find ways of detecting and classifying the fault using engineering data such as vibrations and pressure. He investigated two types of failure, i.e. leading exhaust valve and worn injector nozzle orifice for the case of marine engineering.

Data acquired experimentally was processed using diverse techniques and several Artificial Numerical Networks (ANN) were applied to identify the signals. Combining the diagnostic decision of the ANN classification improves the diagnostic performance and enhances the robustness and reliability of the failed systems.

Gopi has made a significant contribution to a difficult topic with important practical applications. He has found out that it is possible to detect and classify minor faults using engineering vibration signatures, He has also shown that the system performance can be improved by combining the decision of several systems. The results of Gopi's research are original and he has put forward a series of new ideas which have been validated by his results.

Gopi's research can have important application in a wide variety of fields. His Seminar was followed with great interest by the WIT researchers whose questions resulted in a lively discussion.