Multi-sensor fusion high-speed industrial unit trajectory tracking intelligent state estimation method

By employing a multi-sensor fusion and deep neural network adaptive gain matrix approach, the problems of unstable sensor data and noise interference in high-speed industrial unit trajectory tracking are solved, achieving high-precision and low-cost state estimation and improving trajectory tracking performance and robustness.

CN122151625AInactive Publication Date: 2026-06-05NANTONG UNIV
0 Cites 0 Cited by

Patent Information

Application Number
CN202610158419.2
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2026-02-04
Publication Date
2026-06-05
Estimated Expiration
Not applicable · inactive patent

AI Technical Summary

Technical Problem

In existing high-speed industrial unit trajectory tracking, unstable data from a single sensor, severe noise interference, insufficient adaptive capability of fixed-gain Kalman filtering, and poor robustness result in poor trajectory tracking performance.

Method used

A multi-sensor fusion method is adopted to acquire data through machine vision sensors, inertial measurement units, and motion encoders, establish a unified heterogeneous multi-source observation framework, dynamically adjust sensor weights, utilize a deep neural network to adaptively fusion gain matrix, and combine robust weight matrix and confidence evaluation index to achieve adaptive and robust state estimation.

Benefits of technology

It improves the accuracy and robustness of state estimation under complex working conditions, reduces computational costs, significantly improves trajectory tracking performance and anti-interference ability, and meets the high precision and reliability requirements of high-speed industrial applications.

✦ Generated by Eureka AI based on patent content.
Patent Text Reader

Abstract

The invention discloses a multi-sensor fusion high-speed industrial unit trajectory tracking intelligent state estimation method, and belongs to the technical field of industrial motion control. The method comprises the following steps: constructing a heterogeneous multi-source observation framework and a quality evaluation mechanism, respectively extracting noise time sequence characteristics and system dynamic characteristics of observation deviation by using a double-flow gating time sequence memory network, and dynamically allocating weights based on observation quality for fusion; a self-adaptive Kalman gain matrix is generated through a lightweight deep neural network, and intelligent fusion and high-precision state estimation of multi-source data are realized; further introducing a robust weight matrix based on a Grommet loss function, and updating an abnormal threshold by combining the absolute deviation of the median of the sliding window, so as to effectively inhibit outlier observation; and state estimation quality is evaluated and guaranteed in real time through confidence evaluation and a saturation amplitude limiting mechanism. According to the method, the estimation precision, the robustness and the system stability of the high-speed industrial unit under the complex working condition are remarkably improved.
Need to check novelty before this filing date? Find Prior Art