摘要
为了实现对复合封装高量程加速度计的频域动态校准,在BP神经网络和粒子群优化算法相结合的基础上,采用惯性权重线性递减、学习因子非线性变化的方法,对加速度计的动态模型进行了系统辨识。数值仿真和实验结果表明,该算法辨识出了加速度计的三阶模态谐振频率(其中最高一阶谐振频率达478.02 kHz),与实验数据FFT计算结果相比较,相对误差均小于5%,且在幅值误差5%和10%内的工作频带相对误差分别为8.59%和6.11%。通过系统辨识得到了较精确的加速度计频域动态参数,有较强的准确性和可靠性。
In order to realize the frequency-domain dynamic calibration of composite packaged high range accelerometer,the method of linear decrease of inertia weight and nonlinear change of learning factor was proposed,which was based on the combination of BP neural network and particle swarm optimization algorithm.The dynamic model was systematically identified.Numerical simulation and experimental results show that the algorithm can identify the third-order modal resonance frequency of the accelerometer(the highest first-order resonance frequency is 478.02 kHz).Compared with the FFT calculation results of the experimental data,the relative error is less than 5%,and the relative error of the working frequency band within 5%and 10%of the amplitude error is 8.59%and 6.11%respectively.Through the system identification,more accurate frequency-domain dynamic parameters of the accelerometer are obtained,which has strong accuracy and reliability.
作者
温晓杰
石云波
赵锐
曹慧亮
王彦林
张娟娟
WEN Xiaojie;SHI Yunbo;ZHAO Rui;CAO Huiliang;WANG Yanlin;ZHANG Juanjuan(Science and Technology on Electronic Test&Measurement Laboratory,North University of China,Taiyuan Shanxi 030051,China;Xi'an Institute of Electromechanical Information Technology,Xi'an Shaanxi 710065,China)
出处
《传感技术学报》
CAS
CSCD
北大核心
2021年第11期1475-1481,共7页
Chinese Journal of Sensors and Actuators
基金
国家自然科学基金面上项目(52175524)
山西省重点研发计划项目(202003D111004)。
关键词
高量程加速度计
动态校准
系统辨识
粒子群算法
high range accelerometer
dynamic calibration
system identification
particle swarm optimization(PSO)
