今日工業的競爭除了依賴精實生產之外，還得依賴能夠提供顧客在產品生命週期中具一定水平價值之責任性支援的能力。為了改進客服反應力及修配用零件市場之適業效率，已有許多公司採納新系統與產品的服務商業模型以使未規劃之停工期達到近乎零的境界。這種轉變使得能夠在失效可能發生前，進行預測及預防作業之智慧型預測工具變得十分必要。本論文介紹一種使用watchdog代理人來機器退化及失效預測的創新方法。Watchdog代理人的方法使用多重感應器資訊，用於產品之績效退化評估與生命預測。特亮是邏輯迴歸法，用於電梯之垂直通路中馬達速度描述之反覆追蹤。特徵值如加速時、減速時間及平均最大速度當成邏輯迴歸法做績效評估的輸入值。實際應用結果顯示邏輯迴歸法是在做機器退化評估中一種非常大有可為的方法。

　　　　　Today’s competition in industry depends not just on lean manufacturing, but also on the ability to provide customers with accountable life-cycle support for sustainable value. To improve customer service responsiveness and aftermarket business efficiency, new service business model for enable products and systems to achieve near-zero unscheduled downtime has been adopted by many companies. This transformation necessitates the deployment of smart prognostics tools to predict and prevent possible failures before they occur. This paper introduces an innovative approach in using Watchdog Agent TM for machine degradation and failure prognostics. The methods of Watchdog Agent TM are developed to use multi-sensor information form product for performance degradation assessment and life prediction. Specially, the logistic regression (LR) method is introduced and applied to an elevator hoistway performance assessment. The logistic regression model is designed for repayable tracking of motor speed profile of elevator hoistway movement for on-line monitoring and prognostic purposes. Features such as acceleration time (interval from triggering elevator to reaching maximum speed), deceleration time (from maximum speed to elevator stop), as well as average maximum speed were used as inputs to Logistic Regression tool for performance assessment. The real application results show that the LR is a very promising methodology for machine degradation assessment.