基于PSO-BP神经网络的微流体惯性开关接触电阻预测研究  

Research on Predicting Contact Resistance of Microfluid InertialSwitch Based on PSO BP Neural Network

作　　者：袁君鑫 聂伟荣[1] 席占稳[1] 曹云[1] YUAN Junxin;NIE Weirong;XI Zhanwen;CAO Yun(School of Mechanical Engineering,Nanjing University of Science and Technology,Nanjing 210094,China)

机构地区：[1]南京理工大学机械工程学院,江苏南京210094

出　　处：《机械与电子》2024年第12期18-24,共7页Machinery & Electronics

摘　　要：微流体惯性开关可以避免机械惯性开关的磨损、变形和振动等问题,因此被广泛应用于汽车、军事和医疗等行业。由于开关本身电阻会对电信号传递产生较大影响,针对这一问题,对微流体惯性开关的接触电阻进行理论分析与仿真并建立模型以预测接触电阻大小。采用数值仿真法,以接触电阻为研究对象,通过改变电极宽度、两电极间距、电极与侧壁距离、惯性力作用时间和正应力5个参数得到580组样本数据,根据样本数据建立随机森林、BP神经网络和基于粒子群优化的BP神经网络3种预测模型。通过对建立的3种预测模型对比分析,发现基于粒子群优化的BP神经网络模型的预测效果最好。为了验证基于粒子群优化的BP神经网络预测模型的合理性,以接触电阻最小值为优化目标,利用遗传算法对该预测模型拟合的映射函数的参数在给定范围内寻优,发现遗传算法得到的电阻最优值与理论仿真值相差10.41%,结果表明可以借助基于粒子群优化的BP神经网络预测模型估计开关接触电阻。Microfluidic inertial switches can avoid problems such as wear,deformation,and vibration of mechanical inertial switches,and are therefore widely used in industries such as automobiles,military,and healthcare.Due to the significant impact of switch resistance on electrical signal transmission,the contact resistance of microfluidic inertial switches is theoretically analyzed and simulated in this paper,and a model is established to predict the magnitude of contact resistance.The numerical simulation method is used,with contact resistance as the research object.580 sets of sample data are obtained by changing five parameters:electrode width,distance between two electrodes,distance between electrodes and sidewalls,inertia force action time,and normal stress.Based on the sample data,three prediction models are established:random forest,BP neural network,and BP neural network based on particle swarm optimization.Through comparative analysis of the three established prediction models,it is found that the BP neural network model based on particle swarm optimization has the best prediction performance.In order to verify the rationality of the BP neural network prediction model based on particle swarm optimization,with the minimum contact resistance as the optimization objective,a genetic algorithmis used to optimize the parameters of the mapping function fitted by the prediction model within the given range.It is found that the optimal resistance value obtained by the genetic algorithm was 10.41%different from the theoretical simulation value.The results show that the BP neural network prediction model based on particle swarm optimization can be used to estimate the switch contact resistance.

关 键 词：惯性开关 接触电阻 预测模型 粒子群优化 遗传算法 

分 类 号：TM564[电气工程—电器]