A Novel Temperature Drift Error Precise Estimation Model of MEMS Capacitive Accelerometers Using MTAM

Abstract

AbstractDue to incomplete Temperature Correlated Quantities (TCQ) of Si-based material and the complex nonlinearity between Temperature Drift Error (TDE) and TCQ, the conventional precise estimation model is unable to decouple Si-based material’s temperature dependence, which reduces environmental adaptability of MEMS Capacitive Accelerometer (MCA) in different complex conditions. So, a novel TDE precise estimation model of MCA based on Microstructure Thermal Analysis Methodology (MTAM) is proposed. First, Si-based material’s temperature dependence is analyzed in microstructure using thermal expansion theory, which extracts new TCQ determining structural deformation, including temperature variation ΔT and its square ΔT2. Second, precise TDE test is formed using heat conduction analysis in the thermal chamber, and temperature experiment is designed. Then, a novel model is implemented with Radical Basis Function Neural Network (RBFNN), and its accuracy is evaluated by Mean Square Deviation (MSD). At last, the conventional model based on Back Propagation Neural Network (BPNN) and the novel one are compared in temperature experiment. The experimental results demonstrates MCA’s accuracy is improved by 60% with the novel model, which means temperature dependence of Si-based material is decoupled remarkably and environmental adaptability of MCA is enhanced significantly. Furthermore, the novel model is able to expand MCA’s application in complex conditions and universal for the diverse systems.KeywordsMCATemperature drift errorMTAMPrecise TDE testRBFNN