The Analysis And Research About Wearable Robot Sensor Signal Prediction Algorithm Based On Particle Filter Optimization

"Baby boom" caused by the war II causes the aging problem, which quicklybecomes a hot issue. The old man’s physiological function degrades due to organfailure, resulting in a series of problems such as mobility and self-confidence decline.It is undoubtedly helpful for the older if we can provide a wearable exoskeleton legpower robot to help them to normal operation.Wearable power robot is a kind of assistive mechanical robot, and it is a sort ofexoskeleton mechanical power device that can help people to extend the legmovement ability. It senses and feedbacks body movement information through thebooster perception system in it, and then it can predict the next human bodymovement intentions according to these information. After that, the robot can drivehuman movement by controlling the motor installed at the hip and knee, finallyrealizing the power effect. The implementation of robot assisted effect is mainlycomposed of two parts: sport intention prediction and the drive control of thecontroller. This article mainly focuses on the device prediction algorithm. Our paperimproves MWDAR time series model and introduces the particle filter algorithm,besides, we employ frequency multiplication to ensure real time, which can help tosolve the prediction delay problem, since sensor response frequency is significantlylower than the human nervous system’s response frequency, which causes responsedelay, and weight calculation of particle filter optimization algorithm also bringsprediction delay.This paper first analyzes the features of the wearable power robots and relevantalgorithm of time series forecasting model, and introduces the characteristics of avariety of time series prediction algorithm and their implementation. Secondly, wefocus on MWDAR analysis model and optimization algorithm of particle filter, andcombining the characteristics of the power robot, we put forward a new algorithm ofoptimal prediction model--PF_MWDAR. Finally, we set up small experimentplatform, collect sensor signal data by using static torque, and verify the feasibilityand validity of the algorithm through MATLAB simulation.