Application of wavelet neural network in signal processing of MEMS accelerometers

摘要

A new method with wavelet neural network is described to optimize MEMS accelerometers for temperature independent sensitivity. Linear accelerations are measured and compensated by a thermocouple the fast algorithm which is used to deal with the nonlinearity error. The simulation results show that MEMS accelerometers with compensation is characterized by an excellent temperature stability of the sensitivity with less than 0.1% variation for a temperature range −40–100°C, while the variation of acceleration without compensation is 8%. The proposed algorithm can be useful for realization of high accuracy miniature gyroscope systems based on MEMS technology.