Skip to main navigation Skip to search Skip to main content

A Novel Neural-Networks-Enhanced Linear Active Disturbance Rejection Current Control Strategy for Variable-Flux Machine

  • Mingqiao Wang*
  • , Xinyu Dong
  • , Yunlong Wu
  • , Yong Liu*
  • , Ping Zheng
  • , Yi Sui
  • , Boru Jia
  • *Corresponding author for this work
  • College of Electrical and Electronic Engineering
  • Beijing Institute of Technology
  • National Key Laboratory of Multi-Perch Vehicle Propulsion Systems

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

Abstract

This paper proposes a novel neural-networkenhanced linear active disturbance rejection (LADR) current control strategy for variable-flux machines (VFMs) to address the challenge of significant inductance parameter variations during magnetization state (MS) regulation. The developed solution integrates a backpropagation neural network (BPNN) with the LADR control strategy, where the BPNN dynamically predicts the inductance parameters of the machine using real-time MS and d-axis current, enabling adaptive parameter adjustment of the LADR controller. Implemented on variable-magnetic-circuit series-parallel VFM (VMC-SPVFM), comparative simulations are established to demonstrate that the proposed strategy achieves superior operational performance, particularly in dynamic current tracking under varying MS conditions, compared with conventional proportional-integral (PI) control.

Original languageEnglish
Title of host publicationICEMS 2025 - 28th International Conference on Electrical Machines and Systems
PublisherInstitute of Electrical and Electronics Engineers Inc.
Pages2104-2109
Number of pages6
ISBN (Electronic)9788986510232
DOIs
Publication statusPublished - 2025
Externally publishedYes
Event28th International Conference on Electrical Machines and Systems, ICEMS 2025 - Busan, Korea, Republic of
Duration: 16 Nov 202519 Nov 2025

Publication series

NameICEMS 2025 - 28th International Conference on Electrical Machines and Systems

Conference

Conference28th International Conference on Electrical Machines and Systems, ICEMS 2025
Country/TerritoryKorea, Republic of
CityBusan
Period16/11/2519/11/25

Keywords

  • Variable-flux machine (VFM)
  • current control
  • linear active disturbance rejection control (LADRC)
  • magnetization state (MS) regulation
  • neural network

Fingerprint

Dive into the research topics of 'A Novel Neural-Networks-Enhanced Linear Active Disturbance Rejection Current Control Strategy for Variable-Flux Machine'. Together they form a unique fingerprint.

Cite this