基于LM_BP神经网络模型的MEMS加速度计温度补偿方法研究  

Research on Compensation Method of Temperature Characteristics for MEMS Accelerometer Based on LM_BP Neural Network

作　　者：杨志梅[1] 周小龙[1] 徐大诚[1] 

机构地区：[1]苏州大学微纳传感技术研究中心,江苏苏州215100

出　　处：《仪表技术与传感器》2015年第11期30-33,共4页Instrument Technique and Sensor

基　　金：国家科技支撑计划(2015BAZ05750)

摘　　要：随着MEMS加速度计应用领域的不断广泛,其温度性能越来越受到重视。在研究扭摆式硅微加速度计结构与温度特性的基础上,采用改进LM_BP神经网络来构建MEMS加速度计的补偿模型,通过实时温度变化优化出温度补偿模型参数,进而实现实时温度补偿。实验结果表明,通过该方法补偿后的标度因数温度系数和全温零偏稳定性分别由252 ppm/℃和16.62 mg/h减小为100 ppm/℃和2.30 mg/h,证明了该温度补偿方法的有效性和可行性。With the application fields of MEMS accelerometer being extended , more and more attention is paid to the tempera-ture performance .The temperature compensation model was established by LMBP neural network based on analyzing the structure of pendulous micro machined silicon accelerometer and temperature characteristic .Through the realt-ime temperature change , the temperature parameters for the model were optimized and then the real -time temperature compensation was realized .Experimental re-sults show the temperature coefficient of scale factor and the full -temperature zero bias stability were decreased from 252 ppm/℃and 16.62 mg/h to 100 ppm/℃and 2.30 mg/h respectively , proving that the temperature compensation method is correct and feasible .

关 键 词：扭摆式硅微加速度计 神经网络 温度补偿 标度因数 零偏稳定性 

分 类 号：U666[交通运输工程—船舶及航道工程]