Attitude Estimation Algorithm of Abnormal Gait Recognition with Dynamic Step Conjugate Gradient

In this paper, a conjugate gradient hybrid filtering fusion algorithm for multi-sensor data fusion is proposed to solve the problems of gyroscope drift and cumulative error in gait analysis system based on inertial sensors. The algorithm improves the attitude resolution algorithm for abnormal gait recognition effectively. A conjugate gradient optimization algorithm with dynamic step length is proposed to solve the problem that the traditional conjugate gradient algorithm can only tackle the problem of the settlement accuracy under the constant motion rate. The step length is proportional to the actual physical angular velocity length of human motion, which meets the requirements of abnormal gait recognition under different motion rates. As for the problem of the inaccurate attitude calculation angle caused by motion acceleration generated from the non-uniformity of human motion, a motion acceleration suppression algorithm is proposed, which separates the gravity acceleration from motion acceleration to improve the accuracy of the gait analysis system. By comparing this fusion algorithm of conjugate gradient hybrid filter with other algorithms, we can conclude that the proposed algorithm can reduce the accumulated error of gyroscope and the interference of motion acceleration on the solution accuracy and improve the static performance and dynamic performance of the abnormal gait recognition system effectively.