數據驅動之工業機台或設備健康狀態評估

Abstract

在競爭激烈環境下，降低工業機台或設備之維護成本並提高其效率對企業至關重要，而藉助各種感測器監控機台或設備之健康狀態並依監控所得數據釐定維護策略是大家所能想到的作為。本研究針對故障偵測和診斷提出了一種綜合考慮多個數據源和多個評估指標、可應用於工業機台或設備健康狀態評估的模型建構流程，該流程透過建立評估指標與數據的綜合分析，能夠準確評估機台設備的健康狀態，並預測故障時間點。本研究所提流程屬於數據驅動，無需深入瞭解機台或設備的運作原理，也無需標記故障數據，可降低專業人力成本。本研究以收集到的航空引擎衰退模擬數據進行案例分析和驗證，證明所提流程的有效性。

In a highly competitive environment, reducing maintenance costs and improving the efficiency of industrial machinery or equipment is crucial for businesses. Monitoring the state-of-health for machinery or equipment through various sensors and determining maintenance strategies based on the monitored data is a widely considered approach. Within the context, this study proposes a comprehensive state-of-health assessment model for industrial machinery or equipment by considering multiple data sources and evaluation metrics. Through the integration of data and metrics, the state-of-health and failure time of a machine or a piece of equipment can be predicted. The proposed model is a data-driven one which does not require the in-depth understanding of operational principles of the machine or equipment. Neither does it has to label failure data. The operational cost of the machine or equipment can thus be reduced. The effectiveness of the proposed model is demonstrated and validated through a case study consisting of a batch of degradation data monitored for a certain type of aero-engine.