Intelligent electric meter signal enhancement method and system based on reinforcement learning

By adaptively dividing the frequency band and conducting multi-dimensional evaluation, combined with deep neural network prediction, and optimizing the filtering strategy, the local distortion problem of signal enhancement methods in existing technologies is solved, achieving higher signal fidelity and reliability.

CN121984538BActive Publication Date: 2026-06-16HANGZHOU HUALONG ELECTRONIC TECH CO LTD
View PDF 2 Cites 0 Cited by

Patent Information

Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
HANGZHOU HUALONG ELECTRONIC TECH CO LTD
Filing Date
2026-04-07
Publication Date
2026-06-16

AI Technical Summary

Technical Problem

Existing reinforcement learning-based power line carrier communication methods, when evaluating filtering effects, rely solely on the overall signal-to-noise ratio, which can lead to excessive distortion or information loss in local frequency bands. This approach fails to effectively improve the reliability and fidelity of signal enhancement.

Method used

By adaptively dividing the frequency bands, the reward participation degree of each frequency band is calculated. Combining the time-domain and frequency-domain effect values, an adaptive reward function is designed. A deep neural network is used to predict the future signal state and optimize the filtering strategy to achieve signal enhancement.

Benefits of technology

By precisely targeting the frequency bands most affected by noise, the fidelity and reliability of signal enhancement are improved, local distortion is prevented, and the robustness of the system during continuous communication is enhanced.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN121984538B_ABST
    Figure CN121984538B_ABST
Patent Text Reader

Abstract

The present application relates to the field of data processing, more particularly, the present application relates to a smart meter signal enhancement method and system based on reinforcement learning, the method comprising: collecting the history and current noisy carrier signal of the meter and the history noise-free reference signal; based on the adaptive division of the frequency band of the signal spectrum difference; for each frequency band, calculate the reward participation; the candidate filter action is applied to the current signal, and the time domain and frequency domain enhancement effect value of each frequency band is calculated; based on the reward participation, the two-dimensional effect value is weighted and summed and the action complexity penalty is deducted to obtain the adaptive reward function value; combined with the discounted reward obtained by the future state prediction model, the comprehensive return value is calculated to determine the optimal filter action vector, and finally the signal enhancement is realized. The present application effectively avoids the signal distortion problem caused by the traditional overall signal-to-noise ratio evaluation, and significantly improves the reliability and signal fidelity of the power line carrier communication.
Need to check novelty before this filing date? Find Prior Art

Citation Information

Patent Citations

  • CN121090915A

  • CN121485808A