ABSTRACT
Diagnosis is the process of determining the nature of a fault or malfunction, based on a set of symptoms. Input data, that is, the symptoms, are interpreted and the underlying cause of these symptoms is the output. Diagnosis is, therefore, a special case of the more general problem of interpretation. There are many circumstances in which we may wish to interpret data, other than diagnosing problems. Examples include the interpretation of images (e.g., optical, x-ray, ultrasonic, electron microscopic), meters, gauges, and statistical data (e.g., from surveys of people or from a radiation counter). This chapter will examine some of the intelligent systems techniques that are used for diagnosis and for more general interpretation problems. The diagnosis of faults in a refrigerator will be used as an illustrative example, and the interpretation of ultrasonic images from welds in steel plates will be used as a detailed case study.
