Study and analysis the flexion moment in active and passive knee prosthesis using back propagation neural network predictive

The purpose of this study was to develop an artificial neural network (ANN) model for predicting the flexion moment in knee prosthesis during the gait cycle. It was decided to employ the feedforward backpropagation (BP) algorithm as an adaptive method in ANN. Two types of prosthetic knee joints were employed in this study, a mechanical model (3R60) and a microprocesses model (C-Leg4). Three inputs parameters were used in building the ANN model. This involved, the vertical ground reaction force (vGRF), hip angle and knee angle. The vGRF was measured on the treadmill using Zebris FDM-T system. While, the data of knee angle and hip angle were estimated for prosthetic knee joints using Kinovia program. The flexion moment in knee joint was selected as target parameter for ANN model. Finally, the verification results demonstrate the feasibility and efficiency of ANN model in prediction the flexion moment in different types of knee prosthesis. Furthermore, there is a good compatible in data of knee flexion moment between microprocessor-controlled prosthetic knee joints with the intact limb which help in improve many of gait features and reduce hip work production.