光電式安全防護裝置之失效分析與可靠度評估應用

Abstract

摘 要 傳統上，分析加速衰退試驗的數據係利用統計的方法，先求得產品壽命的分佈，然後應用壽命分佈的特性來預估產品的累積分佈函數或可靠度。但是當隨機參數超過一個或衰退路徑呈非線性時，這個方法就變得複雜且難以計算。因此，針對多隨機參數的問題，本研究論文應用直接數值積分法，以為上述問題之解決對策。 文中以俱有多隨機參數之安全光幕為例，首先針對其系統之組成、功能及操作環境等因素進行失效分析，找出適當之失效模型，建立系統可靠度之評估程序。其次針對其中之光束單元，依據投光器-IRLED之發光強度及受光器-Phototransistor之光接收能力的衰退特性，利用強度-應力干涉模型建構光束單元之極限狀態方程式。並且藉由Arrehenius Law建構其中之參數與作用應力間的關係式，以求得正常操作應力條件下之參數預估值。然後應用直接數值積分法對極限狀態方程式之失效或安全區域進行積分，求得正常操作條件下光束單元之失效機率或可靠度。最後，並以蒙地卡羅模擬法進行驗證，確認所提方法的可行性與有效性。 研究結果顯示，直接數值積分法和蒙地卡羅模擬法之結果相比較，當可靠度值大於0.5以上時，預估可靠度之最大誤差可以控制在5%以內，驗證本文所提方法之準確性與可行性。而本研究之可靠度評估方法不僅可用以解決安全光幕之可靠度評估問題，同時也提供一個解決高可靠度產品之可靠度評估問題的有效方法。

ABSTRACT Traditionally when analyzing test data of accelerated degradation, it is used to analyze the distribution of product life through statistical methods. It needs to find the distribution of product life first, and then apply its characteristics to assess its cumulative distribution function or reliability. If the number of random parameters is more than one or it’s degraded path is nonlinear, the above-mentioned method may become too complicated to calculate it. Therefore, this study offers a method, direct numerical integrated method, to solve the problem of multi-random parameters. The reliability analysis of a safety light curtain, which has multiple random variables, is used as an example in this study. The procedure of reliability assessment of the system is established via the analyses of the composition, function and operational conditions of the system. Secondly, it makes use of the strength-stress interference model to construct the limit state equation of light beam unit according to the degraded characteristics of light power of IRLED and Phototransistor. Also it makes use of the relationship between random parameters and temperature based on Arrehenius Law to obtain the predicted values of parameters at normal operational condition. Then, it applies direct numerical integrated method to integrate the failure or safety area of the limit state equation to obtain the failure probability or reliability of light beam unit at normal operational condition. Finally, it also verifies the proposed method by using Monte Carlo simulation. The results show that the maximum predicted error between the proposed method and Monte Carlo simulation can be controlled in 5% as the reliability is higher than 0.5. This can verify the accuracy and validity of the proposed method. This study not only can be used to assess the reliability of safety light curtain but also provide an effective solution to the highly reliable products.