Model Modification of the Mill Transmission System Based on the PSO-BP Neural Network

• Published in: Jixie Chuandong Vol. 48; pp. 48 - 53
• Main Authors: Tao Zheng, Bao Xianle, Guo Qintao, Zhou Tianyang
• Format: Journal Article
• Language: Chinese
• Published: Editorial Office of Journal of Mechanical Transmission 01.02.2024
• Subjects: Model modification;Neural network;Modal analysis;Similar design;Hierarchical correction
• ISSN: 1004-2539

Aiming at the complexity of the mill transmission system structure, the uncertainty of the constraint conditions among the components and the nonlinearity, a finite element model correction method based on the PSO-BP neural network is proposed in this study. This method approximates the nonlinear mapping relationship between the two by improving the back propagation (BP) neural network, combines with the actual structural response, and uses the generalization property of the neural network to obtain the numerical value of the model design parameters. After the correction, the frequency error is reduced from a maximum of 18% to about 4%, and the error range of the correction coefficient is all within 0.5%, while obviously improving the accuracy of the finite element model. Meanwhile, it does not need a large number of iterative solving steps, avoids the complex nonlinear optimization process of the traditional inverse problem model modification method, improves the efficiency, verifies the feasibility of the P