类小脑神经网路於梭织机故障诊断之应用.pdf

類小腦神經網路於梭織機故障診斷之應用 CMAC neural network application on fault diagnosis of loom system 洪清寶 王孟輝羅永昌 郭英哲 勤益技術學院電機系 Department of Electrical Engineering National Chin-Yi Institute of Technology Taichung, Taiwan, R.O.C E-mail: cbhong@chinyi.ncit.edu.tw ABSTRACT 前 言 1. In this paper, a CMAC neural network application 紡織工業曾是我國工業發展的主力，而隨著工 on fault diagnosis of loom is proposed. This novel fault 業環境的變遷，早期藉由勞力密集、低廉工資而發跡 diagnosis system contains an input layer, quantization 的廠商相繼歇 業或轉移到東南亞、大陸地區尋求發 layer, binary coding layer, and fired up memory addresses 展。在東南亞、大陸等地區挾持龐大勞力及低廉成本 coding unit. Firstly, we construct the configuration of diagnosis system depending on the fault patterns. 的競爭之下，紡織業的生存之道唯有透過技術密集， Secondly, the known fault patterns were used to train the 品質提昇、自動化生產及有效管理著手，才能在業界 neural network. Finally, the diagnosis system can be used 中佔有一席之地。其中在織布品質的提昇便在整個生 to diagnose the fault type of loom. By using the 產成本的環節上佔有著重要的地位。 characteristic of self-learning, association and 一般織布機生產的布匹係透過專業驗布人員的 generalization, like the cerebellum of human being, the 檢驗，觀查布匹的平整性、鬆緊度、異常紗、棉粒、 proposed CMAC neural network fault diagnosis scheme 斷經、繼緯…等現象，據以對布匹的品質作嚴格的控 enables a powerful, straightforward, and efficient fault 管。而當產生異常的現象時，則必須迅速找出故障的 diagnosis. Furthermore, the following merits are obtained: 1) High learning and diagnosis speed. 2) High noise 原因，以進行故障的排除。這些煩瑣的檢驗工作，除 rejection ability. 3) Alleviate the dependency to expert’s 了容易因人為的因素造成誤判，在資深驗布人員的經 expertise. 4) Eliminate