基于DWPSO-SVM的sEMG手势动作识别  

sEMG gesture recognition based on DWPSO-SVM

作　　者：王宇春[1] 王敏[1] 袁东华 赵化启[1] WANG Yuchun;WANG Min;YUAN Donghua;ZHAO Huaqi(College of Information and Electronic Technology,Jiamusi University,Jiamusi Heilongjiang 154007,China)

机构地区：[1]佳木斯大学信息电子技术学院,黑龙江佳木斯154007

出　　处：《智能计算机与应用》2023年第12期158-164,共7页Intelligent Computer and Applications

基　　金：黑龙江省省属本科高校基本科研业务费科研项目(2018-KYYWF-0943);黑龙江省卫生健康委立项科研课题(2019-287);佳木斯大学优秀学科团队项目(JDXKTD-2019008);佳木斯大学教育教学改革研究项目(2021JY1-49)。

摘　　要：为了提高表面肌电信号(surface Electromyographic signal, sEMG)手势动作识别的准确率,本文提出基于双权重粒子群算法(Particle Swarm Optimization, PSO)优化支持向量机(Support Vector Machine, SVM)的分类模型(DWPSO-SVM)。针对传统PSO在参数寻优时易陷入“早熟”问题,进一步提高粒子寻优能力,本文在标准PSO的基础上引入约束因子结合同向更新策略用于速度约束,有效的提高了粒子的寻优能力并缓解了“早熟”现象;其次,分析了多种权重更新策略对惯性权重和约束因子的影响;最终,采用非线性更新策略结合DWPSO优化SVM模型构建特征分类模型。实验表明,本文提出的DWPSO-SVM模型能够有效完成sEMG手势动作识别任务。In order to improve the accuracy of surface electromyographic signal(sEMG)gesture recognition,this paper proposes a classification model(DWPSO-SVM)based on Particle Swarm Optimization(PSO)to optimize Support Vector Machine(SVM).In response to the problem of premature convergence in parameter optimization of traditional PSO,and further improving the particle optimization ability,this paper introduces a constraint factor and a directed following strategy based on the standard PSO for speed constraints,effectively improving the particle optimization ability and alleviating the phenomenon of premature convergence;Secondly,the impact of various weight update strategies on inertia weights and constraint factors was analyzed;Finally,a non-linear update strategy combined with DWPSO optimization SVM was used to construct a feature classification model.The experiment shows that the DWPSO-SVM model proposed in this article can effectively complete the sEMG gesture action recognition task.

关 键 词：SEMG 粒子群算法 支持向量机 手势动作识别 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]