Joint Angle Estimation of Lower Limb Based on PFA-LSTM-Attention Algorithm

In order to explore the applicability of surface EMG signal in continuous motion estimation of lower limbs, this paper proposes a method for optimizing weight allocation, introduces the attention mechanism in Long short-term memory networks, and optimizes the network model through Pathfinder algorithm to achieve continuous angle estimation of lower limb joints. Four motion modes were set up and eight characteristic muscles were selected to collect the original signals. After filtering and noise reduction, it is used as the input of the regression prediction model. The experimental results show that this method has high prediction accuracy and model stability in the angle prediction of multiple motion modes of the hip and knee joints, which provides conditions for the compliance control of the lower limb exoskeleton auxiliary clothing in the later stage.