A diagnosis system based on fuzzy classification

Diagnosis is seen from a cybernetical point of view because it is supposed that the system under investigation is a large scale one. Experimental results form the information basis. They are measuring results or linguistic expressions given by an expert about the history of the system (new state, working conditions, observations) and actual informations from tests, investigations and so on. The theoretical basis is classification as a fundamental ordering operation. Classes may be defined as semantic ones by an expert (teacher) or by means of cluster procedures as natural classes. The additional uncertainty of elementary and global informations is taken into consideration. One theoretical basis is, therefore, that of fuzzy sets. Data structure is described by an effective concept of a uniform parametric membership function. The structure and strategy of the computer assisted diagnosis system are shown. Some information about software instruments for digital- and microcomputers is given. An illustrative practical example demonstrates the use of the system.