Neural Networks have been used on line to classify drilling conditions into three categories: safe, caution and danger. This approach allows changing a drill just before failure, a valuable benefit. Drill size, feed rate, spindle speed and other operational features are used as inputs of neural networks. The output of the neural network indicates the drilling condition. An on line drill condition detection system can achieve a success rate over 95%. Numerous values of initial weights, learning rates, and smoothing factors were systematically explored and tested to achieve high reliability.
Ⅰ. Introduction<BR>Ⅱ. Neural Networks<BR>Ⅲ. Back Propagation Neural Networks<BR>Ⅳ. Training Neural Networks<BR>Ⅴ. Online Detection of Drilling Conditions<BR>Ⅵ. Comparisons and Discussion of Results<BR>Ⅶ. Conclusion<BR>References<BR>
(0)
(0)