Collaborative inference method and system for power edge intelligence, and electronic device

By employing deep neural networks and deep Q-network algorithms in the power edge intelligent system, the system dynamically determines task offloading strategies, solving the problems of limited computing power and inference latency. This enables adaptive task allocation and load balancing, improving the system's real-time performance and robustness.

CN122240209APending Publication Date: 2026-06-19BEIJING SMARTCHIP MICROELECTRONICS TECHNOLOGY CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
BEIJING SMARTCHIP MICROELECTRONICS TECHNOLOGY CO LTD
Filing Date
2026-05-25
Publication Date
2026-06-19

AI Technical Summary

Technical Problem

In power edge intelligent systems, existing technologies suffer from limited computing power and unpredictable inference latency of computationally intensive network algorithm models, which affects the accuracy and real-time performance of business perception decisions and makes it difficult to rationally allocate inference tasks under dynamic changes in network status and task requirements.

Method used

By predicting the computational complexity of tasks through edge intelligent nodes, a deep neural network model based on complexity factors and a deep Q-network algorithm are adopted to dynamically determine the task offloading strategy, offloading tasks to multiple edge intelligent nodes for distributed collaborative inference, thereby optimizing inference latency and load balancing.

Benefits of technology

It achieves adaptive task offloading, reduces overall processing latency, improves the collaborative efficiency of edge computing networks, adapts to the dynamic changes of power business data, and improves resource utilization efficiency and system stability.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN122240209A_ABST
    Figure CN122240209A_ABST
Patent Text Reader

Abstract

This invention relates to the field of the Internet of Things (IoT) for power systems, providing a collaborative reasoning method, system, and electronic device for power edge intelligence. The method includes: predicting the computational complexity of a task based on acquired task data by edge intelligent nodes; determining whether to offload the task to multiple edge intelligent nodes in an edge computing network for distributed collaborative reasoning based on the task's computational complexity; when determining to offload the task to multiple edge intelligent nodes for distributed collaborative reasoning, acquiring the status information of the currently available edge intelligent nodes in the edge computing network, and determining a task partitioning strategy among the currently available edge intelligent nodes based on the status information; offloading the task to the currently available edge intelligent nodes according to the task partitioning strategy; and completing the task's reasoning computation collaboratively through the collaboration of the currently available edge intelligent nodes. This invention enables power edge cluster collaboration, optimizing inference latency and load balancing.
Need to check novelty before this filing date? Find Prior Art