A personalized exercise training evaluation method based on multi-modal perception

By constructing a joint energy transfer matrix and an improved TS2Vec model, the shortcomings of existing technologies in identifying individual dynamic differences and dynamic temporal changes are addressed, enabling personalized and continuous assessment of users' motor abilities and improving the accuracy and stability of sports training assessment.

CN121862308BActive Publication Date: 2026-07-24XIAMEN OCEAN VOCATIONAL & TECH COLLEGE
View PDF 2 Cites 0 Cited by

Patent Information

Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
XIAMEN OCEAN VOCATIONAL & TECH COLLEGE
Filing Date
2026-03-19
Publication Date
2026-07-24

AI Technical Summary

Technical Problem

Existing sports training assessment methods are unable to accurately characterize individual dynamic differences, cannot identify differences among users in terms of force application, kinetic chain coordination, and energy utilization, and lack in-depth extraction of dynamic temporal changes in multimodal sensing data and ability evolution analysis over continuous training cycles.

Method used

By constructing a joint energy transfer matrix and using an improved TS2Vec model to encode features of multimodal time series data, a user's motion capability state space is formed and a motion capability evolution field is generated, enabling a comprehensive evaluation of the user's movement quality, force application rationality, movement stability, and training effect trends.

Benefits of technology

It can more accurately depict the individual dynamic characteristics of human movement, improve the objectivity and accuracy of sports training assessment, enhance the ability to express complex movement behaviors, and realize continuous modeling and personalized assessment of users' training capabilities.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN121862308B_ABST
    Figure CN121862308B_ABST
Patent Text Reader

Abstract

The application discloses a kind of personalized sports training evaluation methods based on multi-modal perception, comprising: collecting multi-modal perception data, pre-processing generates multi-modal motion dataset;Calculate joint angular velocity and torque, construct energy transfer matrix and extract energy flow expression;Build multi-modal sequence data, input improved TS2Vec model coding, obtain time series representation vector;Construct user movement ability state space, map to generate state vector;Calculate state change relationship, construct movement ability evolution field;Based on movement ability evolution field, analyze action quality, force reasonable and training trend.The application realizes deep time series modeling of multi-modal motion data and individual dynamics difference identification by constructing human movement energy flow expression, improved TS2Vec time series representation model and movement ability evolution field, so as to realize personalized, continuous sports training evaluation.
Need to check novelty before this filing date? Find Prior Art