Three-phase grid-connected inverter admittance identification method based on complex valued neural network

By constructing a three-phase grid-connected inverter admittance identification method based on complex-valued neural networks, and combining small-signal flow graphs and pole-residual expansion, the problems of large workload and uninterpretable fitting results in traditional methods are solved, and efficient and fast admittance characteristic identification is achieved.

CN122334044AActive Publication Date: 2026-07-03SICHUAN UNIV
3 Cites 0 Cited by

Patent Information

Application Number
CN202610796457.0
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2026-06-04
Publication Date
2026-07-03
Estimated Expiration
2046-06-04

AI Technical Summary

Technical Problem

Traditional methods for obtaining the frequency domain admittance characteristics of three-phase grid-connected inverters are labor-intensive and inefficient, making it difficult to meet the needs of rapid analysis under multiple operating conditions. Furthermore, the fitting results based on neural networks have poor physical interpretability.

Method used

A complex-valued neural network is used as the modeling framework for port admittance. By combining small-signal flow graphs and pole-residual expansions, a loss function for the complex-valued neural network is constructed. The network is trained using electromagnetic simulation data to learn the mapping relationship between operating conditions and pole-residuals, thereby compressing network complexity and improving physical interpretability.

Benefits of technology

It achieves low computational load, high identification efficiency, and rapid identification under multiple operating conditions. The fitting results have high physical interpretability and can quickly and accurately identify the port admittance characteristics of three-phase grid-connected inverters.

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

Abstract

The invention discloses a three-phase grid-connected inverter admittance identification method based on a complex valued neural network, and relates to the technical field of power electronics. The method comprises the following steps: in a synchronous coordinate system, constructing a small signal flow diagram of a three-phase grid-connected inverter, constructing a port admittance matrix under a steady-state working point, determining a corresponding characteristic polynomial, and performing factorization to obtain a unified pole set; constructing a pole-residual expansion of the port admittance matrix based on the unified pole set, and determining a transfer function of elements in the port admittance matrix; constructing a complex valued neural network and a loss function thereof; training a complex-valued neural network through a loss function; and inputting the working condition information of the grid-connected point and the frequency point into the complex valued neural network to obtain the port admittance under the current working condition. The method has the technical advantages of being low in calculation amount, high in identification efficiency, capable of supporting multi-working-condition rapid identification and high in fitting result physical interpretability.
Need to check novelty before this filing date? Find Prior Art