Temperature Compensation Method and Embedded System Design for MEMS Accelerometers

The tethered underwater profiler utilizes MEMS accelerometers to detect its motion state, and the temperature change in the profiler cavity will lead to the measurement misalignment of the micro-inertial measurement unit inside the accelerometer. In this paper, we experimentally obtain the triaxial acceleration values under different attitudes <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$(-90\sim 90^{\circ})$</tex> with temperature <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$(25\sim 50^{\circ}\mathrm{C})$</tex>, and establish the multivariate linear fitting model, the BP neural network model, and the PSO-BP neural network model after the experimental data are processed by low-pass filtering, respectively. The embedded system is designed to write the temperature compensation model of the accelerometer into the microcontroller STM32F103 for operation and output the compensated results. The R-value of the BP model and the PSO-BP model are both 1, the three-axis RMSEs of the multivariate linear fitting model are 1.02, 0.927, and 6.2, and that of the three-axis RMSEs of the BP model are 0.1293, 0.0229, 0.0156, and the three-axis RMSE of the PSO-BP model is 0.0576, 5.1297e-5, 0.0031, respectively. The experimental results show that the PSO-BP model has a good effect of compensating the temperature drift error of the accelerometers, and can effectively reduce the measurement error of the MEMS accelerometers introduced by the temperature change and improve the performance of the tethered accelerometers. Measurement error and improve the measurement accuracy of the inertial measurement unit measurement of the bollarded underwater profiler.