Modeling method and transfer learning method for deep residual LSTM network and thermal error prediction model

Abstract

The invention discloses a deep residual LSTM network, and the network is characterized in that the network comprises the following layers in sequence: an input layer; a convolutional layer; a pooling layer; a remodeling layer; an LSTM layer; a Dense layer; an output layer; wherein n pre-activated residual blocks are arranged between the LSTM layer and the dense layer, and n is greater than or equal to 1; the pre-activated residual block comprises a first BN layer, a first weight layer, a first convolution layer, a second BN layer, a second weight layer and a second convolution layer which are arranged in sequence; the first BN layer and the second BN layer are used for solving the problem that the network cannot converge; the first weight layer and the second weight layer are used for extracting features; and activation functions for reducing interdependence between parameters are respectively arranged between the first BN layer and the first weight layer and between the second BN layer and the second weight layer. The invention further discloses a modeling method and a transfer learning method of the thermal error prediction model. According to the method, the problem of prediction precision saturation caused by network depth increase can be avoided, and the prediction precision and robustness can be effectively improved.

Description

technical field [0001] The invention belongs to the technical field of mechanical error analysis, in particular to a deep residual LSTM network and a modeling method and a migration learning method of a thermal error prediction model. Background technique [0002] As the key equipment to realize high-precision machining of complex parts, precision machine tools are widely used in aviation, aerospace, nuclear power and other fields. However, thermal errors will significantly reduce the machining accuracy of the machine tool. It has been shown that thermal errors are a major part of the total errors. Therefore, reducing or avoiding thermal errors is extremely important to ensure the geometric errors of machined parts. An unbalanced temperature field is the main cause of thermal errors. Internal and external heat sources cause an imbalance in the temperature field. Internal heat sources include, but are not limited to, servo motors, bearings, ball screws, rolling guides, et...

AI Technical Summary

Problems solved by technology

Previous studies considered the critical temperature as an input to a data-based model, and clustering methods were used to select the critical temperature, but collinearity between temperatures could not be completely eliminated, resulting in reduced robustness

Moreover, the prediction accuracy of the traditional model is not high enough, because the error model has no self-learning and self-updating ability

Furthermore, empirical correlations and data-based error models do not accurately reflect error mechanisms

Images