A generic approach to fault diagnosis in multi-channel clinical analysers

This paper presents an overview of the approach adopted in solving problems connected with fault diagnosis, preventive maintenance, and remote or telemaintenance in the context of a workstation complementing the operation of a multichannel clinical analyser. Such analysers are complex and are viewed as ´processes´ with both discrete and continuous time data. The workforce are largely highly educated in laboratory medicine but, critically, are not skilled in the fault diagnosis or maintenance of the larger type of analyser found in most laboratories today. Further, their level of expertise in computer systems seldom exceeds the use of wordprocessors and spreadsheets, so they are not equipped to play a major role in the maintenance of modern clinical analysers. This paper outlines the approach used in providing fault diagnosis services in an instrument workstation for use with clinical analysers. The importance of modelling the problem domain is emphasised and the role that various methods from artificial intelligence have to play is discussed. An example from a very different domain, namely environmental control of commercial mushroom tunnels, is taken where the same approach to fault diagnosis is considered. Finally. a brief conclusion is presented in which the main points learnt from building such systems is given