臺灣博碩士論文加值系統

本文主要在研究如何有效的提升線切割式放電加工的切削速率, 於粗加
工時能夠高速且穩定的進行. 文中利用放電頻率偵測與控制系統, 監測放
電過程,計算出總放電頻率與不正常放電比, 再加上放電時間, 作為判定
加工惡化的依據, 並於線上即時調整床台進給速率, 避免惡化發生或增加
加工速率. 文中利用倒傳遞類神經網路作為偵測的工具, 將上述三種放電
資訊輸入, 透過網路計算推論出當時加工惡化的程度. 並根據經驗來設定
目標區, 控制床台進給速率使得惡化程度值能落在目標區內, 使同時兼固
速率與穩定性. 為了擴大本網路的適用性, 還採用不同厚度的工件進行加
工, 來添作網路的學習範例, 期望能處理厚度變化的問題. 實驗證明, 以
少數範例所架設的網路, 能成功的偵測大部份的加工情形; 且目標區的設
定是唯一的, 無須根據加工條件的不同而做調整, 使整個控制系統見更為
簡易.

Improvement of the machining speed of wire electrical
discharge machining(WEDM) process under various machining
conditions is studied. Neural network is employed to monitor the
degree of deterioration of the process includingwire rupture and
back-stroke of the feed. Pulse on time, total sparking
frequency, and abnormal ratio defined as the ratio of the total
numbers ofthe abnormal sparks to the total numbers of the sparks
in a sampling periodare chosen as the inputs of the network. The
back propagation neural networkis used, and output is a
numerical number between 0 and 1. The larger its value is, the
closer the machining process approaches dangerous states.
Asimple feedrate control strategy is proposed. It is adjusted
regularly according to the output of the network. Experimental
results confirm that the degree of the deterioration of the
process can be detected, and the stable high speed machining can
be achieved under most of the machiningconditions