Details

Autor(en) / Beteiligte

• Xin, Zhang
• Wenbin, Wang
• Zihan, Song
• Haoyue, Xu
• Chenyue, Cai

Titel

Research of Deep Learning Neural Network Based on Regression Analysis in Numerical Simulation Analysis of Motor Stress

Ist Teil von

• The Proceedings of the 9th Frontier Academic Forum of Electrical Engineering, 2021, Vol.743, p.541-550

Ort / Verlag

Singapore: Springer Singapore Pte. Limited

Erscheinungsjahr

2021

Link zum Volltext

Quelle

Alma/SFX Local Collection

Beschreibungen/Notizen

• This paper studies a learning method based on deep learning neural network, and studies the numerical analysis of motor stress simulation, which can obtain the minimum stress value of the required position in the shortest time. A method is proposed to perforate the stator of the motor and fill it with negative magnetostrictive material, change the position and radius of the hole, and find the best position to reduce the noise so as to suppress the vibration and noise of the motor. Through finite element analysis, the stress values on each point of the permanent magnet synchronous motor stator corresponding to different punch positions and radii are obtained as training samples. We established a multiple regression model with 3 fully connected layers, two inputs and one output, and optimized the algorithm to better perform regression analysis on the motor stress value to achieve motor noise optimization.