基于MEMS的水下触觉力传感器及BP网络标定实现

Underwater Tactile Force Sensor Based on MEMS and BP Network Calibration Realization

由于受到水压强影响,水下触觉力测量易产生交叉灵敏度,造成触觉力信号分辨困难的问题。针对此问题设计了基于MEMS的水下触觉力传感器。该触觉力传感器呈胶囊外形且为差压式结构,上侧感知水压强与触觉力信息,下侧感知水压强信息,通过在MEMS上下侧矢量叠加实现触觉力测量。设计了基于STM32的传感器数据采集电路及BP网络数据融合算法。通过构建4层BP网络,触觉力信号、温度信号、深度信号作为网络输入,真实触觉力值作为网络输出训练网络,将训练好的神经网络发送至STM32通过数据处理实现真实触觉力输出,传感器融合精度5%。文中设计的触觉力传感器具有的差压式结构能够消除水静态压强影响,并且结合BP网络数据融合算法实现了真实触觉力信号输出,对水下环境用触觉力测量具有重要意义。

Due to the influence of hydraulic pressure, the underwater haptic force measurement is prone to cross-sensitivity, which causes difficulty in distinguishing tactile signals.A MEMS-based underwater tactile force sensor was designed for this problem.The tactile force sensor had a capsule shape and a differential pressure structure, the upper side sensed the water pressure and tactile force information, the lower side sensed the water pressure information, and the tactile force measurement was realized by the vector superposition on the upper and lower sides of the MEMS.The sensor data acquisition circuit and BP network data fusion algorithm based on STM32 were designed.By constructing a four-layer BP network, the tactile force signal, temperature signal, and depth signal were employed as network input, and the real tactile force value was employed as network output to train the network.The trained neural network was sent to STM32 to realize real tactile force output through data processing, and the sensor fusion accuracy was 5%.The differential pressure structure of the tactile force sensor designed in this paper eliminates the influence of water static pressure, and the BP network data fusion algorithm realizes the real tactile force signal output, which is of great significance for the measurement of tactile force in the underwater environment.