Intelligent integrated framework towards high-accuracy machining

Abstract

The thermal error is the main factor hindering the realization of the high-accuracy machining. The intelligent integrated framework towards the high-accuracy machining, which is expected to realize improve the machining accuracy, is scarcely reported so far. The self-perception is difficult due to the limited bandwidth of the industrial Internet, and the thermal error model with a strong robustness and high prediction accuracy has not been established. To address the above challenges, in this study, an intelligent integrated framework is designed towards the high-accuracy machining. The intelligent gateway is designed to conduct the data transfer and edge computing. Then the data perception and transfer are conducted with the ZigBee wireless sensor network and cluster-type network structure. Then the data computation is carried out to realize the preliminary training and fine training of the thermal error model. The Bi-directional Long Short-Term Memory Network (BILSTM) is combined with a new Residual Network (RESNET) to propose the BILSTM-RESNET (BLRNET) model, and then the Convolutional Neural Network (CNN)-BLRNET model is proposed, and the spatiotemporal features in the thermal error data are extracted. Finally, the CNN-BLRNET model is embedded into the intelligent integrated framework, and the thermal error control service is provided by the intelligent integrated framework. With the implementation of the intelligent integrated framework, the measured contour error is reduced by 41.2%, and the roundness error of the machined workpiece is reduced by 67.5%.