Lower Limb Joint Motion Prediction Under Adaptive Neurofuzzy Model Inference

Traditional methods overly rely on static or linear models, which cannot fully consider the dynamic and nonlinear relationships, resulting in inaccurate prediction of lower limb joint motion. This article combines the nonlinear modeling ability of neural networks and the fuzzy reasoning ability of fuzzy logic to improve the prediction performance of lower limb joint motion by better adapting to dynamic and nonlinear relationships. Firstly, this article collects lower limb motion data through various sensors, preprocesses the collected data for feature extraction, and uses Gaussian function for feature blur. It performs fuzzy inference through the combination of fuzzy rules, and finally outputs the results of lower limb joint motion prediction through ambiguity. A self-adaptive neuro fuzzy inference system is proposed to improve the predictive performance of lower limb joint motion by combining neural networks and fuzzy logic. ANFIS performs excellently in predicting lower limb joint movement, with an average mean square error of only 0.010, providing strong support for objective evaluation of movement disorders and disease diagnosis and treatment. This innovative method has made significant contributions to the fields of biomechanics and rehabilitation.