A motor load posture observation method based on multi-source heterogeneous signal fusion

CN121710764BActive Publication Date: 2026-06-05NANJING UNIV OF SCI & TECH
View PDF 2 Cites 0 Cited by

Patent Information

Application Number
CN202610211716.9
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2026-02-13
Publication Date
2026-06-05
Estimated Expiration
2046-02-13

AI Technical Summary

Technical Problem

Existing motor load observation technologies suffer from problems such as nonlinear errors in the transmission mechanism, low visual sampling frequency and high delay, and difficulty in parameter adjustment of traditional Kalman filtering under complex working conditions, which result in the load-side positioning accuracy failing to meet high-precision control requirements.

Method used

A multi-source heterogeneous signal fusion method is adopted, which combines visual encoders and signal encoders to extract the absolute attitude and motion vector of the load. Real-time observation is carried out through a neurodynamic model and Kalman gain, and a neuro-Kalman observation framework is constructed to compensate for the nonlinear backlash of the transmission chain and visual time delay, thereby achieving high-precision full closed-loop control.

Benefits of technology

It significantly improves the load-side observation accuracy and dynamic reliability under complex operating conditions, eliminates cumulative drift, ensures the real-time response and robustness of the control loop, can identify and suppress visual interference, and enhances the operational safety of the observer in variable industrial environments.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN121710764B_ABST
    Figure CN121710764B_ABST
Patent Text Reader

Abstract

The application discloses a motor load posture observation method based on multi-source heterogeneous signal fusion, comprising the following steps: extracting the absolute position prediction value of the load end and its observation uncertainty through a visual encoder; synchronously extracting the dynamic characteristics from the electric signal of the driving motor by using a signal encoder, and performing integral deduction based on physical constraints to obtain the prior trajectory of the load end; combining the timing characteristics, observation uncertainty and innovation dynamic to generate Kalman gain, and realizing nonlinear correction of the prior trajectory of the load end. In addition, the application introduces a fault detection mechanism based on normalized innovation square, which automatically adjusts the gain under the condition of visual occlusion or interference, and ensures the observation robustness. The application effectively solves the pain points of visual processing delay, nonlinear error of the load end transmission and large cumulative error of the traditional observer, and significantly improves the precision, real-time performance and anti-interference ability of the load end position observation.
Need to check novelty before this filing date? Find Prior Art

Citation Information

Patent Citations

  • Motor model prediction control method and device

    CN118694225A

  • Gyroscope-based brushless motor attitude detection and balance control method and system

    CN120110250A