發展可調式加工參數於端銑表面粗糙度品質控制系統之研究

Implementing adaptive machining parameters for quality control system of surface roughness in end milling operations

古孟晨(Gu-Meng chen)

指導教授 ： 黃博滄

中原大學/電機資訊學院/工業與系統工程研究所/碩士(2011年)

https://doi.org/10.6840/cycu201100984

摘要

隨著產業自動化的趨勢，業者為了提高競爭力，不僅要維持產量之外，產品的品質也是客戶關注的重要課題。品質會因製造形式的不同有所影響，其中機械加工的部分，有一個重要的品質指標為表面粗糙度，許多研究為了改善加工工件的表面粗糙度，提出對表面粗糙度做先行的預測，但都會造成停工或是為了維持產品的品質而影響到產量。因此，本研究將透過調整CNC機器加工參數的方法，控制工件表面粗糙度的品質指標，在維持生產力之下，讓表面粗糙度能符合客戶的要求。

為了能製造出符合要求的工件品質，本研究在考量表面粗糙度的參數規劃，除了機器本身的加工參數之外，將加入力量感測器(force sensor)來偵測製程中無法控制的切削力與切削過程中震動對產品表面粗糙度有所影響的參數，以達到系統能準確地控制表面粗糙度。本實驗將利用QP1620 CNC加工機進行資料的蒐集，並採用類神經網路來建置預測與控制表面粗糙度的決策系統，探討加工參數與表面粗糙度之間的關係。

系統中的實驗模組，由四個子系統所組成，分別為第一子系統預測加工條件、第二子系統預測表面粗糙度、第三子系統調整主軸轉速與第四子系統調整進刀速度所組成。每個子系統皆由類神經網路作為決策系統，為了能有效減少實驗次數，結合田口方法輔助尋找最佳的網路參數組合。由客戶要求的表面粗糙度品質指標值為系統執行的起始點，以客戶要求的表面粗糙度來預測其加工條件，再由第二子系統來預測該條件的表面粗糙度，最後將兩者之間的差值透過調整CNC機台兩個主要加工參數(主軸轉速與進刀速度)，將表面粗糙度控制在一個品質範圍內。

本研究發展智慧型表面粗糙度品質控制系統中，各個子系統之類神經網路的誤差函數分析皆低於0.0011，且以30筆數據資料進行系統驗證，透過本實驗模組調整機器參數結果，具有87%的驗證資料可調整至符合客戶要求的產品品質。

關鍵字

類神經網路 ； 表面粗糙度 ； 加工參數 ； 田口方法

並列摘要

As the tendency to industry automation, to increase the competitiveness of the industry, the factory are not only maintain the productivity, but also to enhance the quality of product for customers. Quality is affected by different types. One important index that can represent quality characteristic is the surface roughness. Many studies were proposed to improve the surface roughness of processing work-pieces by predicting. But they would cause the stop of continuous production, and were too hard to find a balance between quality and productivity. Therefore, the purpose of this study is to have a good control of surface roughness by adjusting the parameters of CNC machine, and to meet the client’s requirements and maintain the productivity simultaneously.

In order to produce qualified work-pieces, this research not only takes processing parameter into consideration, but also uses force sensors to detect influences from uncontrolled cutting force and vibration that can affect the surface roughness during cutting process, to achieve the system can accurately control the surface roughness. Data is collecting from Surftest QP1620 CNC machine and then trained by neural network to build a decision making system that investigate the relation between processing parameter and surface roughness.

This intelligent surface roughness quality control system is consist of four sub-systems. The first subsystem is to make a forecasting machine parameters of the processing condition. The second one is a forecasting of the surface roughness. The third subsystem is to adjust the parameter of spindle speed. And the fourth subsystem is to adjust the feed rate. Every subsystem uses neural network as decision making system. To decrease the training time, the Taguchi method to find the best network parameter combination. Surface roughness quality indicated by clients was asked as a starting point for system implementation, and use the indicators to predict the processing condition in the first subsystem, adding the second subsystem to predict the surface roughness under this condition. Then, two major parameters (spindle speed, feed rate) of the machine CNC can be adjusted through the subtraction of the results between the first and second system. The surface roughness can be controlled in an acceptable quality range.

In this study, a intelligent surface roughness quality control system. The analysis of the error function are less than 0.0011 for subsystems, 30 data was used to validate the system, the results of the experiment model adjust the processing parameters can be found up to 87% data to meet customer requirements.

並列關鍵字

neural network ； Taguchi method ； machining parameter ； surface roughness

參考文獻

賴柏辰，運用主成份田口類神經模組建立端銑表面粗糙度可調式控制系統，中原大學工業與系統工程學系碩士學位論文，2010。

Huang, B. P., Chen, J. C., Li, Y., Artificial-neural-networks-based surface

Huang, B. P., Chen, J. C., Lin C. J., A Taguchi-Neural-Based In-process Tool

Kirby, E. D., Chen, J. C., Zhang, J. Z., Development of a fuzzy-nets-based

Aggarwal, A., Singh, H., Kumar, P., Singh, M., Optimization of multiple quality characteristics for CNC turning under cryogenic cutting environment using desirability function, Journal of Materials Processing Technology, Vol.205, pp.42-50, 2008.