Research On The Technology Of Calibration And Temperature Compensation For The MIMU

Measurement errors are the most important factors to affect inertial navigationsystem’s accuracy in the MIMU(Micro Inertial Measurement Unit) of micro inertialnavigation systems. Micro inertial navigation systems’ navigation accuracy can beimproved effectively through founding micro inertial measurement unit’s error models,using error compensation algorithm to gain error parameters and compensating harmfulerrors. This issue is researched for measurement errors from the following two aspects:First of all: Micro Inertial Measurement Unit’s CalibrationCalibration error models for the micro inertial measurement unit are set up throughanalyzing error sources in the micro inertial measurement unit. The micro gyroscopes’accuracy is not ideal because of gyroscopes’ scale factors’ nonlinear elements seriously.Non-linear compensation methods is proposed in this article. Nonlinear interpolation andneural network methods based on BP technology are compensating the nonlinearcharacteristic through analyzing gyroscopes’ scale factors’ nonlinear properties.Compensation effectiveness is verified through swing tests. Results prove this twocompensation means both effectively. But the effect of neural network technology based onBP is more ideal than nonlinear interpolation and it reflects nonlinear ability of BP.Secondly: Compared with laser(fiber optical) inertial measurement units, microinertial measurement units has a low accuracy and a high sensitivity for temperature.Consequently, how to reduce the temperature’s affecting the micro inertial measurementunits’ accuracy is a very popular technology. Temperature drift has not only to do withtemperature the rate of change, but also to do with temperature values. It is necessary tointroduce temperature’s rate of change to the temperature’s error compensated models.Complete range temperature error compensated models is set up through researching statictemperature experiments and dynamical temperature experiments’ output database andreferring the Calibration results at normal atmospheric temperature(30℃). This method isto unify the zero error, scale factors, misaligned errors and establish errors’ compensatedmodels. It changes the traditional method which scale factors and zero error areindependently to make temperature compensation. Two means that polynomial fitting andRBF neural network method are to establish models for micro inertial measurement units’ temperature compensation. By using gauss nonlinear function of exponential decay in theRBF neural network, input and output mapping is approached and RBF network andtraditional polynomial fitting are compared through test.Finally, that calibration error technology and temperature compensation technology ofMIMU can improve micro inertial navigation systems’ positioning accuracy is testedthrough collecting static and dynamic test database in the systems. Experimentsdemonstrate that this new method has very important engineering significance and practicalvalue.