The Research Of MEMS Inertial Sensors Parameter Identification And Error Compensation Technology

Low cost MEMS inertial sensors has many advantages such as small volume,light weight, low power and high reliability etc. But the measurement precision of lowcost MEMS inertial sensors is much lower, so improving the measurement precision isthe key technology for Micro Guidance Navigation and Control system. This thesisaims to solve the problem of low cost MEMS inertial sensors in low measurementprecision and carry on the error calibration and compensation sensors technologyresearch.First, according to the work environment of low cost MEMS inertial sensors andthe shortcomings of the traditional filter, we determined the wavelet analysis as aprefilter of measurement unit. After the comparison of different bases in differentdecomposition layers mean square error and SNR, we determined a suitable waveletbasis and decomposition level; meanwhile, the dynamic oscillation signal has beensubjected to the threshold filter. The denoising results show that the filter method canreduce the noise of the inertial sensors effectively.Second, this paper analyses the error mechanism of low-cost MEMS inertialsensors, establishes the mathematical model of the error, according to test thesix-position static of accelerometer and the dynamic rate of the gyroscope, combinedwith the analytic methods to understand the error coefficient operator, achievecompensation sensors finally.Again, this paper establish kalman filter equation for deterministic errorcoefficient, use the method of Allan variance to identify the the inertial sensor dataerror term, and apply the detailed characterization and identification to the kalmanfilter equation error matrix. Modeling the zero bias and scale factor with temperatureand speed, identify the parameters in the model using nonlinear regression. Finally,compensate for the sensor in the range of temperature and speed test. Finally, modeling the inertial sensors of random drift by using the method oftime series analysis, using linear kalman filter, extended kalman filter, adaptive kalmanfilter eventual realization dynamic oscillation signal filter, compared inertial sensorscalibrated in the test-wide and random error filter compensation with MTi, the resultsshowed that the overall effect of compensation is much better than MTi.