Title: Modelling of a magneto-rheological fluid dual clutch with BP neural network

Authors: Jin Zhao; Haiping Du; Donghong Ning; Huan Zhang; Lei Deng; Weihua Li

Addresses: School of Electrical, Computer and Telecommunications Engineering, University of Wollongong, Wollongong, NSW 2522, Australia ' School of Electrical, Computer and Telecommunications Engineering, University of Wollongong, Wollongong, NSW 2522, Australia ' College of Engineering, Ocean University of China, Qingdao, China ' School of Electrical, Computer and Telecommunications Engineering, University of Wollongong, Wollongong, NSW 2522, Australia ' School of Electrical, Computer and Telecommunications Engineering, University of Wollongong, Wollongong, NSW 2522, Australia ' School of Mechanical, Materials and Mechatronic Engineering, University of Wollongong, Wollongong, NSW 2522, Australia

Abstract: In this paper, a backpropagation (BP) neural network model for a novel magneto-rheological fluid dual-clutch (MRFDC) is presented. The MRFDC is a complicated system with high nonlinearity and strong hysteresis, and the conventional parametric modelling methods are based on parameter identification and optimisation. Thus, the modelling work is usually difficult, and the performance of conventional models is usually not good enough for the MRFDC. In contrast, the proposed BP neural network model in this work is easily obtained and able to precisely describe the input and output relationship of the MRFDC. To be specific, the proposed BP neural network model approximates the dynamic behaviours of the MRFDC regarding dynamic input currents and rate-dependent hysteresis. The model input variables are selected considering the working mode of the MRFDC and its rate-dependent dynamic magnetic hysteresis. Then, the BP neural network is trained by the input and output datasets obtained from experiments. The model performance is validated by experiments, and experimental results show that the proposed model is able to predict the output torque capacity of the MRFDC precisely with dynamic input currents.

Keywords: BP neural network; magnetorheological fluid; magnetorheological clutch.

DOI: 10.1504/IJPT.2023.134747

International Journal of Powertrains, 2023 Vol.12 No.3, pp.227 - 239

Received: 20 Nov 2021
Received in revised form: 27 Sep 2022
Accepted: 15 Nov 2022
Published online: 09 Nov 2023 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article