智慧型診斷維修系統平台建置

Development of an Intelligent Diagnosis and Maintenance System

陳慶順(Ching-Shun Chen)

指導教授 ： 黃漢邦

國立臺灣大學/工學院/機械工程學研究所/碩士(2005年)

https://doi.org/10.6342/NTU.2005.10246

摘要

在製造工業，如何提升機台使用率及產能以達到訂單達交是主要的需求。本論文的目的在於發展一智慧型診斷維修系統來維持機台的高使用率。其中，PDA與RFID技術整合所建構出的行動式診斷也納入此系統的應用，以期能協助現場工程師對機台做出即時的診斷及維修。
首先，為了有效率地監控機台狀況，本論文提出一個以PCA為基礎的統計製程控制方法作為機台是否異常的偵測，可藉由觀察少數的控制圖即可發現變異，進而採取立即的處置。在機台異常行為的診斷方面，採用倒傳遞類神經網路、支向量機器與分類回歸決策樹三種方法來執行判別，經實驗其準確率最高分別可達到97.78%、98.21%和97.44%。診斷出的結果經決策系統處理並分等級後，藉由自行建立的訊息傳送平台，包含簡訊、e-mail和即時交談等，通知相關工程師進行預防性維修。而為了平衡計算的工作量，採行分散式架構，在監控及診斷兩伺服器間，以訊息佇列的方式來達到兩伺服器間溝通。
行動式診斷方面，以WinCE.NET為嵌入式作業系統的PDA整合RFID CF卡後即可讀取存於RFID Sensors所收集的機台資料，用於進一步的機台狀況分析。為了取得更可靠及正確的資訊，本論文採用多感測融合技術，利用模糊邏輯來完成多個感測器是否正常運作的篩選和整合，建立決策樹來分類機台的異常行為，並使用Xbar 管制圖觀察運作是否正常。除此之外，透過無線網路 (IEEE 802.11b)，亦可傳送資料至診斷伺服器做更深入的分析；同時，當診斷伺服器發現有機台異常時，會主動傳送警告訊息給持有PDA的工程師執行立即的修復，並藉由PDA與遠端監控人員進行即時意見交談來解決問題。

關鍵字

主成分分析 ； 統計製程控制 ； 到傳遞類神經網路 ； 支向量機器 ； 分類回歸決策樹 ； 無線射頻識別 ； 多感測融合技術

並列摘要

In manufacturing industry, how to promote the equipment’s utility and then keep high throughput to achieve the order fulfillment is the primary requirement. The purpose of this thesis is aimed at developing an intelligent diagnosis and maintenance system to keep high availability of a machine. Moreover, accommodating the necessary of the immediate diagnostics for shop floor engineers, PDA based on the traditional and emerging technology RFID is employed for constructing mobile diagnosis system.
First of all, for effective and sensitive monitoring operation condition of a machine, PCA-based SPC with Nelson Rules is proposed. The experiment results support the feasibility of this approach. BPNN, SVM and CART are used to fault diagnosis. The highest accurate rates reach to 98.3%, 98.21%, and 97.44% respectively. The diagnostic ranked by the decision support system is sent to the related engineers to perform predictive maintenance with the message-passing platform, including SMS, e-mail, and on-line dialogue. MSMQ is applied to message transmission between monitoring server and diagnosing server for the computation workload balance.
In the mobile diagnosis system, PDA equipped with RFID CF Reader Card can easily acquire data characterizing the status of a machine from RFID sensors. In order to get more reliable and accurate information, the multi-sensor fusion algorithm is applied to integrate the sensor information. First, Fuzzy Logic is adopted for sensor validation and Decision Tree will make rudimentary diagnosis. Xbar Chart is used to observe the operation condition of the machine. Subsequently, a further analysis can be executed over remote diagnosis server, if necessary. Besides, by IEEE 802.11b wireless network, diagnosing server can send active warning notification to PDA and engineers can hold an on-line dialogue for troubleshooting.

並列關鍵字

Principal Component Analysis ； Statistical Process Control ； Back Propagation Neural Network ； Support Vector Machine ； Classification and Regression Tree ； Radio Frequency Identification ； Multi-Sensor Fusion Algorithm

參考文獻

[1] H. Albazzaz and X. Z. Wang, “Statistical Process Control Charts for Batch Operations Based on Independent Component Analysis,” Industrial and Engineering Chemistry Research, vol. 43, no. 21, pp. 6731-6741, October 2004.

[2] M. Ayoubi, “Fuzzy Systems Design Based on A Hybrid Neural Structure and Application to the Fault Diagnosis of Technical Processes,” Control Engineering Practice, vol. 4, no. 1, pp. 35-42, January 1996.

[4] M. Boukraa and S. Ando, “A Computer Vision System for Knowledge-Based 3D Scene Analysis Using Radio-Frequency Tags,” IEEE International Conference on Multimedia and Expo, vol. 2, pp. 245-248, August 2002.

[5] L. M. Camarinha-Matos, L. S. Lopes, and J. Barata, “Integration and Learning in Supervision of Flexible Assembly System,” IEEE Transactions on Robotics and Automation, vol. 12, no. 2, April 1996.

[6] G. A. Carpenter and S. Grossberg, “ART2 Self-Organization of Stable Category Recognition Codes for Analog Input Patterns,” Applied Optics, vol. 26, pp. 4919-4930, December 1987.