半導體設備昂貴，設備成本佔製造成本比例相當高，因此設備機台維護保養預測不準確將影響預防保養計畫之安排。若不能妥善安排人力準時維護機台時，機台容易發生當機情況，在打開機台密閉反應艙(Tool Chamber)時，容易造成毒性氣體外洩至無塵室(clean room)，造成維修人員健康危害，影響無塵室之潔淨度，導致生產晶片時，晶片遭受汙染，讓晶片生產良率下降。本研究結合灰色預測及評估診斷方法以建構一套設備機台維護保養預測模式，能有效預測及判別各類設備機台維護保養之未來情況，並將可妥善安排機台維護人力，提升維護安全及減低勞工健康危害。

Semiconductor equipment is very expensive and constitutes a large proportion of semiconductor manufacturing costs. An inaccurate forecast of equipment maintenance will affect preventive maintenance plans. If there is no arrangement for manpower to maintain the tools on time, they may break down unexpectedly and then, when the crew opens the tool chamber, it can easily result in toxic gas leaking into the clean room and affecting plant safety and production rate. Gray prediction and evaluation diagnosis was used in this study to construct an equipment repair and maintenance prediction model for effective prediction and determination of the future repair and maintenance status of various types of equipment and machinery. The gray prediction model can provide for proper arrangements for machinery repair and maintenance manpower in order to improve plant safety and minimize maintenance crew health hazards.