國內近年來受少子化的影響,人力成本逐漸上升,尤其保修人員之雇用量不可能隨產品銷售量作正比成長。於是當人力供應不穩定,且專業養成不易之當下,亟需將售後服務工作從密集仰賴專人服務之方式中改變,而須藉由導入資訊化科技與技術,希望盡可能取代專人之智慧經驗,而使售服工作可以全程自主化地運作。 本研究是以工具機保修為背景來探討,以機台組件之實體幾何為基礎,建立其本體論(Ontology)模型,用來表述其實體拓樸並歸納機能異常之類型、性質與相互關係。而歸納資料則彙整自過去對產品保修所累積之專業經驗,成為一個機台組件與異常現象之因果關係通用架構,並可做為一可作搜尋推理之模型。此架構尚可藉由累積診斷經驗來逐步修正並確立其架構中之參數內容。如此除可使用人工智慧數位化地推導出同類產品之維修重點外,並可記載個別機台的保修特徵,且歸納發現產品潛在的設計與製造弱點供參考。
Post-sale services were traditionally done by experienced human workers, but labor shortage problem has presented an immediate danger to the production industry. Moreover, the work force for post-sale service shouldn’t increase in proportion to the amount of sales, otherwise the service sector will quickly become a burden. Nowadays, an intelligent information system may be designed to provide human knowledge, judgement and so the presence, thus greatly reduce human efforts. This research will focus on the maintenance of mechanical parts and machine assemblies only. The method implements ontology to represent part geometry and their working behaviors. The abnormal instances from service history are also tagged to the topology and become a holo-structure of the service tree. This structure may serve as an inference mechanism and a statistical bin for intelligent search among similar mechanical designs. The quality of diagnosis will improve as the service case increases, meanwhile, the weakness of the part connectivity will also become explicit.