Flatness pattern recognition model based on recurrent neural network

Abstract

The flatness pattern recognition is the key link of the flatness control. The traditional flatness pattern recognition has some shortcomings, such as the poor recognition precision and the poor anti-interference ability. With the complexity of the data regression tasks increasing, the classification algorithm based on the deep learning has been used for many tasks such as the data classification, the image processing, the pattern recognition and the feature extraction. A deep learning can achieve a complex function approximation by learning a kind of deep nonlinear network structure. Based on this background, a flatness pattern recognition model based on the recurrent neural network was proposed. The results showed that the flatness pattern recognition model based on RNN could achieve the training of large flatness data, and the recognition accuracy and the generalization ability of the model were very high. It provides a new method for the further improvement of the accuracy of the flatness control.