Real-Time Detection of Lower Limb Training Stability Function Based on Smart Wearable Sensors

Abstract

The research of smart wearable sensors in limb training has great application significance. In the face of real-time detection requirements, this paper proposes a hardware solution for the stability function of lower limb training based on the theory of intelligent wearable sensors. For the specific implementation circuit of the device, considering the reliability of the system, the system implements antijamming design for the hardware circuit from three aspects: adding decoupling capacitors, optimizing layout and wiring, and rationally grounding the hardware circuit, and performs moving average filtering on the collected sensor data to remove noise, which solves the problem of sensor data precision issues. During the simulation process, by analyzing the changes of acceleration, angular velocity, and attitude angle under different lower limb training activities and different wearing positions, the characteristics of stability combined acceleration, combined angular velocity, and attitude angle were constructed, and the stability mean, variance, and attitude angle were extracted. The experimental results show that the extracted 57 feature dimensions are first reduced to 21 dimensions by the principal component analysis algorithm, and then, the optimal feature subset is selected by the encapsulation method, and the dimension is reduced to 9. The proposed multifeature fusion algorithm has higher accuracy, and the maximum has increased by 6.5%, effectively improving the accuracy of the lower limb training stability function detection algorithm.