Study On Optimization Design Of Lathe Spindle Based On Improved BP Neural Network

As the key part of the machine tool, spindle system’s static and dynamic performance has a direct influence on the machining quality and cutting efficiency. With the continuous improvement of machining speed and precision of modern machine tools, the spindle components are also put forward higher design and manufacturing requirements. In this paper, CWZ61200 heavy duty machine tool is chosen as a research object. Finite element analysis, genetic algorithm, neural network and orthogonal test are combined to apply to the structure optimization design of the spindle, achieving the goal of reducing the weight of the spindle while ensuring good static and dynamic performance of the machine tool’s spindle.Firstly, in order to solve some defects of the standard BP neural network in the actual application, genetic algorithm with the characteristics of global optimization is used to optimize the initial weights and thresholds of BP neural network. Improved neural network with better performance is obtained.Secondly, the spindle’s static and dynamic performance parameters are obtained by the finite element analysis with the Workbench ANSYS and Solidworks software. The spindle’s structure parameters and performance parameters are respectively selected as input and output samples according to the table designed by orthogonal test method. By constantly modifying the 3D model of the spindle and the finite element analysis, the training samples of neural network are obtained.Then, the samples obtained by the above method are used to train spindle performance prediction model based on improved BP neural network, and the forecast effect is compared with the standard BP neural network model, finally the improved neural network model are choosed to obtain the nonlinear relationship between the structure parameters and the performance of the spindle.Finally, selecting spindle’s structure parameters as design variables, spindle’s performance parameters as constraint conditions, the nonlinear relationship between which is obtained by the improved BP neural network mapping, spindle’s quality as the objective function to establish optimization model, which is solved by MATLAB optimization toolbox, and the best combination of structural parameters is obtained. The rationality of the optimization method is verified by comparing the static and dynamic characteristics of the machine tool spindle before and after optimization.In the paper, the improved BP neural network is applied to establish nonlinear mapping relationship between the spindle’s structure parameters and its static and dynamic performance, which is applied to the optimization design of the spindle. The optimization goal of weight reduction is achieved on the basis of good static and dynamic performance of the spindle.