Abstract:

The probabilistic causal model, based on the parsimonious covering theory, is a universal fault diagnosis model and its shortcomings and some improvements are firstly described. And then a new improved model based on a genetic algorithm is put forward to solve the quality fault diagnosis problem in mechanical machining process, in which some steps are introduced in detail,and an individual fitness evaluation function is developed by analyzing the relationship between the machining error and error resource and using fault tree analysis. Finally, some implementation techniques are described and the feasibility of the method is demonstrated by means of CASE.

Key words: mach ining p rocess, quality cont ro l, fault diagno sis, genet ic algo rithm