Skip to main navigation Skip to search Skip to main content

基于神经网络和改进粒子群算法的电涡流传感器探头参数优化

Translated title of the contribution: Optimization of eddy current sensor probe parameters based on neural network and improved particle swarm optimization algorithm
  • 熊 伟
  • , 王 琪*
  • *Corresponding author for this work
  • 湖北第二师范学院物理与机电工程学院

Research output: Contribution to journal › Article › peer-review

Abstract

Aiming at the issues of insufficient range and low accuracy of eddy current sensors in mechanical equipment condition monitoring,a parameter optimization method based on neural network and improved particle swarm optimization algorithm is proposed.Firstly,simulation experiments were conducted on the performance indicators of eddy current sensor coils,including inductance,resistance,and magnetic induction intensity.Through comparative experiments,the material,shape,and cross-sectional shape of the coils were determined,resulting in a data set of the geometric parameters and their corresponding performance indicators of 729 sets of coils.Subsequently,the data set was divided into training and testing sets,and a deep neural network(DNN)model was trained to fit the relationship between coil geometric parameters and performance indicators,achieving high-precision prediction of the coil performance indicators.Finally,an improved Sigmoid particle swarm optimization algorithm(Sigmoid-PSO)was used to optimize the coil parameters,and the optimal coil geometric parameters and their corresponding performance indicators were obtained.This method solves the problems of high time cost and insufficient optimization precision existing in the parameter optimization methods based on finite element simulation.Training a deep learning model with a small amount of simulation data can obtain a high-precision sensor index regression model.Additionally,the introduction of the Sigmoid-PSO algorithm can prevent unnecessary mutations during the parameter update process,and can find the global optimum solution of the parameters more effectively.The optimization effects of the PSO algorithm and the Sigmoid-PSO algorithm,as well as the coil performance parameters before and after optimization were compared and analyzed.The results proved the reliability of the proposed method,and the optimized coil had better performance indicators.
Translated title of the contributionOptimization of eddy current sensor probe parameters based on neural network and improved particle swarm optimization algorithm
Original languageChinese
Pages (from-to)999-1007
Number of pages0
Journal武汉大学学报:工学版
Volume58
Issue number6
DOIs
StatePublished - 1 Jun 2025

Funding

湖北省教育厅科学技术研究计划指导性项目(编号:B2023187);湖北本科高校省级教学改革研究项目(编号:2023492)。

Indexed by

  • CSCD

Fingerprint

Dive into the research topics of 'Optimization of eddy current sensor probe parameters based on neural network and improved particle swarm optimization algorithm'. Together they form a unique fingerprint.

Cite this