Research On IMU Error Modeling And Temperature Compensation Technology Based On MEMS Sensor

MIMU(Micro Inertial Measurement Unit)is consisted of micro gyroscope and micro accelerometer,which is widely used in the field of inertial navigation technology and becomes the research focus in recent years,its accuracy has a direct impact on the system.Compared with the conventional optical fiber assembly and laser assembly,it has the advantages of smaller size,lower price,larger measurement range and more conveniences.It is the core part of the navigation system frame.But this kind of micro inertial device of temperature sensitivity is high,small changes in temperature will influence device's sensitivity and the ability of sensitive data.Under the influence of the environment,it produces a random drift,so to put forward appropriate error model and temperature compensation method can improve the whole system precision.Firstly,to understand the working principle of the micro gyroscope and accelerometer,to analyze error source and parameter characteristics of the micro inertial measurement unit with low precision.The main error sources are divided into three categories,which are the deterministic error,the temperature error which is related to the temperature changes and random variation.To analyze causes of these three types of errors.Then according to the selected MIMU,to build error models of the micro gyroscope and micro accelerometer by using six different positions of the positive and reverse rotation method,To calibrate the gyroscope and accelerometer in the turntable on the input angular velocity and to calculate the main deterministic error parameters.To collect the static drift data of MIMU and to use the Allan variance method to list the main random error components.Then calculate the five kinds of error coefficient.The wavelet threshold method to reduce noise is used to deal with this part of random drift.Through the selection of different wavelet basis compares the noise reduction effect.And a new improved wavelet threshold noise reduction function is proposed,which is proved by the theoretical proof and the actual noise reduction effect.Combining the linear theory of ARMA model and the nonlinear idea of BP neural network,the ARMA-BPNN combination model is proposed to fit the random drift error,and the model is more accurate.Finally,according to the temperature sensitivity of micro inertial measurement unit,to analyze the temperature drift under static conditions.Judge correlation degree between the temperature drift and the temperature value,the temperature change rate,the difference between MIMU temperature output value and the temperature control system display.And then to select the parameters that need to be analyzed.Using multivariate nonlinear theory to fit the relationship between temperature and temperature drift,and establish the formula of temperature change rate and temperature drift.The PSO particle swarm optimization algorithm is proposed to deal with the temperature drift,and the temperature compensation is realized through the network training.