智能化工具機即時加工檢測與伺服系統調機

Real-time Surface Estimation and Servo System Optimization of Intelligent Machine Tools

江孟桓(Meng-Huan Jiang)

指導教授 ： 張文陽

國立虎尾科技大學/工程學院/機械與電腦輔助工程系碩士班/碩士(2015年)

https://doi.org/10.6827/NFU.2015.00149

摘要

本研究為智能化工具機即時加工檢測與伺服系統調機，即以機器視覺自動化辨識銑削加工件之表面粗糙度並即時調整加工參數達到最佳化加工表現。研究內容大致分為二部分，影像特徵辨識以及伺服調機分析，影像處理技術包含影像灰階化、二值化及刀痕邊緣偵測。然而，刀痕邊緣偵測可能因外在環境光源及系統光源亮度影響造成表面特徵顯示不完整，且無法經由調整影像閾值得到理想結果，此刻必須瞭解環境之光源及系統光源特性並適當調整，故本研究中使用光譜儀量測工件表面及周邊環境照度之強度與均勻性並適度調整系統參數或隔絕影響因素。在工具機系統中，伺服調機的優劣可能影響到整體系統穩定度，本研究中伺服調機可分為伺服系統調機及共振抑制分析，伺服調機是以馬達方塊圖之參數增益調整，並經由馬達特性曲線及馬達響應圖，得到馬達響應參數最佳化數據，並且經由實際加工驗證系統之最佳化參數，另外藉由伺服系統共振頻譜響應圖分析，將明顯之共振頻段過濾並消除，並結合敲擊槌、加速度規及頻譜分析儀量取加工結構之自然頻率，使整體實驗在最穩定之系統中執行，以避免不必要誤差產生。目前影像表面粗糙度檢測範圍由Rz1.6~Rz50之間，而在儀器與影像系統較驗分析中，平均誤差量在1.75um以內，以此驗證本系統具有一定準確度；在自然頻率共振分析中，經由衝擊槌給予工具機衝擊，所產生之主軸共振頻譜響應圖中，其響應訊號分為X、Y及Z三軸向，響應峰值(peak)明顯產生於2929 Hz及6500 Hz處，但是因主軸最高轉速為12000 rpm，經公式(8)換算後所得之頻率值為200 Hz，故可忽略所量得的高頻共振訊號，並深入討論頻率由0至250 Hz之頻譜響應，在低頻區域量得之頻譜圖中，可發現在20Hz及125Hz處有較高之X軸訊號值，而59 Hz處則是有較高的Y軸及Z軸響應訊號，在伺服調機之共振訊號中，也有與這相近的共振點位置，59 Hz及125 Hz處有相符的峰值，可驗證自然共振頻率分析與軟體伺服調機分析之可靠性，在伺服系統中可將此部分之共振點經由調整增益值方式濾除，而在實體主軸運轉時，則必須避開3540 RPM及7500 RPM之加工轉速，以獲得最穩定之加工環境。

關鍵字

Labview ； 自動化檢測 ； 表面粗糙度 ； 伺服調機 ； 共振分析

並列摘要

This study is servo system optimization and real-time surface roughness estimation using machine vision. The automatic estimation surface roughness of milling workpieces is investigated using machine vision and the regulating machining parameters are obtained from the optimal processing performance. In the process, combination of vision auto-focusing and Taguchi methods can make the system more stability and integrated. This study is divided into two parts, the image feature recognition and servo tunning analysis. For the image feature recognition, the image processing techniques include image pre-process, edges detection and auto-focusing. However, the edges detection of the tool mark feature causes image defect of surface charateristics due to intensity of ambient light. Before the operation of image recognition, the luminances of workpiece surface and enviroment are measured using the spectrometer for adjusting parameters. Besides, auto-focusing techniques carried out on Fibonacci search. That is conbinated with image process and servo position control. It means fixing the CCD camera on the activity block whth moving around and image caculating to achieve the best focusing position automatically. In the system of machining tools, servo tunning can deside machining stability. Thus, the study of servo tunning is distinguished to servo system tunning, resonance suppression analysis . Furthermore, according to frequency response diagram that can remove resonance zone and eliminating the errors of servo system. The results of reserchment show that the ranges of surfaces roughness estimation are located in Rz1.6 to Rz50 , between Instrument calibration and image system, the deviation can be Controlled below 1.75um. In the natural resonance frequency analysis, resonance response graph of spindle shows the peak is located in 2929 Hz and 6500 Hz. However, the maximum rotating speed is 12000 rpm, according equation 8, the frequency is 200 Hz. Therefore, the high frequency signal can be ignored and depth discussion of frequency from 0 to 250 Hz response. In low frequency graph, there are higher signals at 20 Hz and 125 Hz from X axis, and 59 Hz from Y and Z axis. Resonance signals in the servo conditioners have the same response in 59 Hz and 125 Hz. From above discussion, the reliability of natural resonant frequency analysis and software servo Tuning can be verified. To get the most stable of the working environment, apart from filter the resonance point with gain value adjustment, avoid 3540 RPM and 7500 RPM in the process of operation is the other factor.

並列關鍵字

Labview ； Automatic Inspection ； Surface Roughness ； Servo Tunning ； Resonance analysis

參考文獻

[4]曾冠瑋，2013，“端銑刀銑削沉頭孔之路徑模式及加工條件探討”， 國立台北科技大學 ，碩士學位論文

[6]賴建宏，2002，“馬達臨界轉速之測量及模態測試”，中原大學 ，碩士學位論文

[8]游泰和，2012，“圓柱楔形超音波馬達之研發”，國立交通大學 ，博士學位論文

[9]張文順，2015，“多體系統動力學分析與控制”，國立虎尾科技大學 ，博士學位論文

[14]楊承翰，2014，“基於FPGA之物件影像自動化辨識與檢測系統”，國立虎尾科技大學，碩士學位論文