Abstract: Aiming at the problem of poor self-adaptation of the traditional freezing of gait recognition method for Parkinson’s patients， the freezing of gait recognition method based on variational mode decomposition is proposed. Firstly， the variational mode decomposition is used instead of the traditional time-frequency analysis method to fully adaptively decompose the freezing of gait signal. Secondly， in order to improve the recognition accuracy and recognition speed of the algorithm， the CART model is selected as the base classifier of the ensemble classifier and the feature dimension reduction process is performed. Finally， aiming at the problem of unbalanced data set and limited performance of single classifier， the data sampling-ensemble classifier is designed and the recognition algorithm is optimized by Bayesian optimization. The experimental results show that compared with Adaboost， Tomeklinks-Adaboost， and ROS-Adaboost ensemble algorithm， RUSBoost ensemble algorithm can complete the freezing of gait recognition task more efficiently.

Key words: freezing of gait; feature extraction; variational mode decomposition; RUSBoost; Bayesian optimization