摩擦圧接工程設計支援エキスパートシステムの構築 : 第2報, ファジィ推論による候補圧接条件の修正システム

The conditions for friction welding are determined based on trial and error. Even an expert often needs a large amount of time to determine the optimum conditions. Hence, we have developed an expert system which selects the proper welding conditions for various materials. To store and access complex data based on experiments and literature, knowledge of friction welding is represented in production rules which are extended to the frame structures. This system contains the case-based reasoning method which utilizes selecting and modifying functions. In the case in which the input factors do not correspond to the observed data, this system selects the candidate welding conditions automatically and modifies those conditions based on the difference between input and observed data. Modification quantities are estimated by fuzzy reasoning; thus the estimates can be derived from the minimum number of parameters for all input factors. By some experiments, the system is proven to be capable of assisting process planning of friction welding for joining of carbon steel.