Research On Temperature Compensation Algorithm For MEMS Accelerometer

MEMS accelerometers are micro-inertial devices designed and manufactured by using Micro Electro Mechanical Systems（MEMS）technology.They have the advantages of high integration,good reliability,and easy mass production with IC technology.Therefore,they are widely used in military defense and consumer electronics.However,the accelerometer output drifts with temperature changed,resulting in a loss of accuracy,requiring temperature compensation of its output.Now,MEMS accelerometer temperature compensation technologies include optimizing the structure and process in the design and manufacturing stage of the accelerometer to reduce temperature drift,and performing compensation based on temperature compensation algorithms in the accelerometer application stage.In practical applications,the temperature compensation of the accelerometer output based on the temperature compensation algorithm is cheaper,simpler to test,and more practical.Therefore,the research on the temperature compensation algorithm of MEMS accelerometer is extremely valuable in engineering applications.In this paper,the temperature compensation algorithm of MEMS accelerometer is mainly researched.Firstly,the working principle and temperature characteristics of capacitive MEMS accelerometers are analyzed,which lays a theoretical foundation for the subsequent research on temperature compensation algorithms;Secondly,based on the principle of SPI and serial communication,a data acquisition and transmission scheme is designed,a temperature experimental test system is built,and a four-position static temperature experiment and a multi-temperature point centrifugal experiment are designed;Finally,based on the data obtained from temperature experiments,the temperature compensation algorithm with independent modeling of bias and scale factor and a unified modeling temperature compensation algorithm are researched.The temperature compensation algorithm for independent modeling is researched,fit the bias based on the generalized neural network,fit the scale factor based on the second-order polynomial,mathematical model of the bias and the scale factor changing with temperature is established,and introduce the mathematical model into the acceleration in the input-output relationship of the accelerometer,the temperature compensation of the accelerometer measurement value is realized.The temperature compensation algorithm for overall modeling is researched,base Genetic algorithm optimization BP neural network,particle swarm optimization optimization support vector machine and fuzzy c-means clustering optimization TS type fuzzy neural network to model the overall modeling of accelerometer temperature output voltage,accelerometer acceleration output voltage and centrifuge input acceleration,directly get compensated accelerate,the temperature compensation of the accelerometer measurement value is realized.The research results show that both types of temperature compensation algorithms can effectively compensate for the influence of temperature changes on the accelerometer output.The temperature compensation algorithm with unified modeling has higher compensation accuracy than the independent modeling compensation method.The bias position and scale factor of the compensated accelerometer are closer to the design value,and the temperature coefficient of bias position and scale factor is relatively smaller,which is more advantageous.