基于PSO-BP的磁致伸缩式静力水准系统温度补偿研究  

Research on Temperature Compensation of Magnetostrictive Hydrostatic Leveling System Based on PSO-BP

作　　者：陆俊宇[1] 秦世伟[1] LU Junyu;QIN Shiwei(Department of Civil Engineering,School of Mechanism and Engineering Science,Shanghai University,Shanghai 200444,China)

机构地区：[1]上海大学力学与工程科学学院,上海200444

出　　处：《计算机测量与控制》2022年第12期251-256,263,共7页Computer Measurement &Control

基　　金：上海市2021年度“科技创新行动计划”社会发展科技攻关项目(21DZ1204202)。

摘　　要：静力水准系统(HLS)被广泛应用于基坑、桥隧和建筑物结构等工程的沉降监测;针对环境温度变化引起HLS测量误差的问题,对温度、温度变化速率和温度梯度3种温度误差影响因素进行详细分析,并提出了一种基于粒子群优化(PSO)算法和反向传播神经网络(BPNN)的温度补偿模型;通过BP神经网络对钵体液位值建立修正模型,利用PSO算法优化神经网络参数来提高拟合准确度与模型泛化能力;试验结果表明,利用多个影响因素建立的PSO-BP模型对HLS温度补偿后,使钵体液位值的均方误差和最大误差相比于补偿前平均降低80%以上,从而有效减少计算沉降的误差,极大提高HLS测量精度。Hydrostatic Leveling System(HLS)is widely used in the settlement monitorings of foundation pits,bridges,tunnels and building structures.Aiming at the problem of HLS measurement errors caused by changes in ambient temperature,three temperature error influencing factors of temperature,temperature change rate and temperature gradient are analyzed in detail,and a temperature compensation model based on particle swarm optimization(PSO)and back-propagation neural network(BPNN)is proposed.The BP neural network is used to establish a correction model for the vessels liquid level value,and the PSO algorithm is used to optimize the neural network parameters to improve the fitting accuracy and ability of model generalization.The test results show that,after the HLS temperature compensation of the PSO-BP model is established by multiple influencing factors,the mean square error and the maximum error of the vessels liquid level value are reduced by more than 80% on average.The calculated settlement error is effectively reduced,which greatly improves the HLS measurement accuracy.

关 键 词：静力水准系统 温度补偿 粒子群优化 BP神经网络 环境温度 

分 类 号：TH70[机械工程—仪器科学与技术]