Resistance modeling method, system, electronic device, and storage medium

By adopting a phased training strategy based on neural networks, the problems of insufficient computational complexity and fitting accuracy in the parameter extraction process of the resistance model are solved, and efficient and accurate resistance characteristic modeling is achieved. In particular, it performs well in the nonlinear relationship of small-sized resistors, which improves the support for integrated circuit design and simulation.

CN120542334BActive Publication Date: 2026-07-03GUANGZHOU ZENGXIN TECH CO LTD
View PDF 2 Cites 0 Cited by

Patent Information

Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
GUANGZHOU ZENGXIN TECH CO LTD
Filing Date
2025-05-13
Publication Date
2026-07-03

AI Technical Summary

Technical Problem

Existing technologies are computationally complex and time-consuming in the process of extracting resistance model parameters. Traditional methods are inefficient and cannot accurately reflect nonlinear relationships. In particular, the fitting accuracy is insufficient when the number of blocks is small and the resistor size is small. Furthermore, the advancement of process nodes increases the difficulty of model establishment.

Method used

A phased training strategy based on neural networks is adopted. First, the size characteristics of the resistor are learned under predefined environmental conditions to generate an initial model. Then, the model is trained using the full dataset, and the parameters are adjusted to achieve resistance characteristic prediction under multiple environmental conditions. The model accuracy is ensured through validation.

Benefits of technology

It improves the prediction accuracy and efficiency of resistance models under different environmental conditions, reduces manual parameter adjustments, and performs particularly well in fitting nonlinear relationships of small number of blocks and small-sized resistors, providing reliable resistance model support.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN120542334B_ABST
    Figure CN120542334B_ABST
Patent Text Reader

Abstract

This application discloses a resistance modeling method, system, electronic device, and storage medium based on neural networks. The method achieves resistance characteristic modeling by acquiring a test dataset, constructing a neural network model, training and verifying the model's accuracy in stages. A staged training strategy is employed: first, the model learns the resistance size characteristics in a predefined environment to form an initial model; then, it optimizes the performance in multiple environments using full data; and finally, the accuracy is verified. This method solves the problems of computational complexity, time consumption, and low fitting accuracy of traditional models. It particularly improves the fitting accuracy for nonlinear relationships of small-block and small-size resistors, reduces manual adjustment workload, and enhances prediction accuracy and modeling efficiency under different environments, providing a reliable resistance model for integrated circuit design.
Need to check novelty before this filing date? Find Prior Art