Structural parameter optimization of micro harmonic flexspline based on neural network
SHAO Weilong
LI Luyang
YE Wenli
WANG Qishan
Abstract：[Objective]Little research has been conducted on the rational parameter setting of micro harmonic reducers.To improve the force condition and transmission performance of micro harmonic flexsplines,a neural network-based parameter optimization method was proposed.[Methods]Firstly,a flexspline simulation model was established,and optimization parameters were screened using local sensitivity analysis method.Then,the artificial neural network was optimized by the genetic algorithm,and a mapping model between the optimization parameters and flexspline stress as well as flexspline stiffness was constructed.Finally,through model analysis,the global sensitivity of the optimization parameters was obtained,and the relation between the optimization parameters and flexspline stress as well as flexspline stiffness was revealed.[Results]The calculation results show that the optimization of flexspline structural parameters based on neural network-based global sensitivity analysis can effectively alleviate the stress concentration of the flexspline,the stiffness of the flexspline is improved,and the transmission performance of the harmonic reducer is enhanced.
Keywords：Artificial neural networkGlobal sensitivityMicro harmonic reducerParameter optimization
Publication Date：2025-09-15
Online Publishing Date：2026-09-12(First online date of this platform, not the publication date of the document)
Pages：8( 47-54 )