A global agent computing power optimization system and method based on physical homology

By establishing a unified physical homogeneity benchmark and logical-physical consistency verification, invalid calculations are identified and eliminated, achieving full-domain computing power optimization, solving the problems of low computing power utilization and high power consumption, and improving computing power utilization and device performance.

CN122387684APending Publication Date: 2026-07-14HEBEI WUYUAN INSPECTION & TESTING CO LTD
View PDF 2 Cites 0 Cited by

Patent Information

Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
HEBEI WUYUAN INSPECTION & TESTING CO LTD
Filing Date
2026-05-07
Publication Date
2026-07-14

AI Technical Summary

Technical Problem

Existing technologies for intelligent agents, cloud computing centers, energy storage systems, and industrial equipment suffer from low computing power utilization, high power consumption, and difficulty in implementing optimization solutions. The lack of a unified physical benchmark makes it impossible to determine the homogeneity, synchronization, and consistency of computing tasks. Traditional solutions rely on sensors and hardware modifications, resulting in complex deployment and poor security.

Method used

Using a pure software approach, a unified physical homogeneous benchmark is established. By utilizing transfer function calibration and logical-physical consistency verification, invalid calculations are identified and eliminated, achieving global computing power optimization. A dual threshold system is used for accurate judgment and optimization, supporting the consistency judgment of computing power performance across heterogeneous platforms and performing silent automatic optimization.

Benefits of technology

It achieves a 15%~35% increase in computing power utilization, an 8%~22% reduction in device power consumption, a 10%~30% increase in computing response speed, and a 10%~25% extension in device lifespan, thereby reducing device power consumption and improving system stability and security.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN122387684A_ABST
    Figure CN122387684A_ABST
Patent Text Reader

Abstract

The application discloses a global agent computing power optimization system and method based on physical homology, which is suitable for multiple types of agents and global computing devices. By constructing a unified physical homology benchmark, combining transfer function calibration, coherence coefficient calculation and logic-physical consistency verification, invalid, repeated, empty running and redundant computing power load are identified and removed from the bottom physical mechanism. The system is deployed with pure software, without sensors, hardware modification, data permission and external network access. It runs silently and automatically optimizes. It can handle 25%-40% of redundant load, increase computing power utilization rate by 15%-35%, reduce power consumption by 8%-22%, and increase response speed by 10%-30%. The system realizes efficient, low-consumption and stable operation of global computing power.
Need to check novelty before this filing date? Find Prior Art

Citation Information

Patent Citations

  • ICE load balancing system suitable for heterogeneous computing platform

    CN114039921A

  • Intelligent dynamic load balancing method and system based on Clico-VPP

    CN121967421A