Automated Fault Diagnosis at Philips Medical Systems
As our machines get faster, better and cheaper every day the increase in complexity of these systems is huge. This is no different for the medical systems developed and serviced by Philips Medical Systems (PMS). Fault diagnosis is an essential key to keep these systems dependable. Currently, most fault diagnosis practices in industry are based on manual effort. An area that is not readily explored and exploited by industry, but could offer improvement, is automated fault diagnosis. Although, many useful mechanisms inside the PMS systems exists, so far, there has not been any research about how to set up a diagnostic system. This work is a first exploration of the benefits that such a technique could have for the diagnosis of the Philips Cardio-Vascular X-Ray System. This work defines the goals and qualities of a diagnostic approach in industry. Model-Based fault Diagnosis (MBD) is a reasoning technique
for finding root causes of failures based upon a model. MBD seems to suit the goals and qualities the best, because it is able to utilize all relevant information. Shannon’s entropy is used as a heuristic to quantify the uncertainty of the diagnoses. By means of a case study of a subsystem, it is shown that a Model-Based approach is able to achieve lower uncertainty in its diagnoses than other automated approaches.