本文以一個設備工程師對於想掌握自身機台健康狀況的角度為出發點，以機台振動的資料提前了解機台狀態以進行有效的維護保養作為、提早安排保養排程及內容，以致對產能、良率及保養效率的提升。 由於許多生產工廠已生產多年，面臨到零部件以屆運轉壽命，生產機台無預警故障停機的情況常常發生。本文以A公司傳送設備為例，利用振動原理及設備上加裝的感測器蒐集資料建立決策樹及羅吉斯迴歸模型，再以此兩模型的結果及比較，藉此找出在機台發生故障或異常的預兆；本文案例指出預測設備健康的其中兩個模型(clean及adjust)都認為Z軸方向的振動為關鍵的因子，模型準確率分別83.55%及96.45%，建議A公司在此關鍵因子出現時進行清潔、調整，減少人員花費進行日巡檢的時間，並可即時監測設備狀況。 ;The article takes an equipment engineer′s perspective on wanting to grasp the health status of his own machine as the starting point, and uses the vibration data of the machine to understand the state of the machine in advance to carry out effective maintenance, and arrange the maintenance schedule and content in advance, so as to improve the production capacity, yield and maintenance efficiency. Because many production plants have been producing for many years and face the end of the operating life of parts, it is often the case that the production machines are shut down without warning of failure. This paper takes the transmission equipment of company A as an example, uses the vibration principle and the sensors installed on the equipment to collect data to establish a decision tree and a Logistic regression model, and then compare the results of the two models to find out the difference between the machine. The omen of failure or abnormality; The examples in this paper point out that two of the models that predict device health (clean and adjust) consider the vibration of Z-axis as the key factor. The model accuracy of 83.55% and 96.45% respectively, it is recommended that equipment engineers of Company A could clean and adjust the machines when these key factors appears, reduce the time spent on daily inspection, and can monitor the equipment status in real time.
