Title: Life prediction method of automobile electromagnetic relay based on dual self-attention one-dimensional convolutional neural network
Authors: Zhan Li; Lan Chao; Zhao Hao; Guo Ji-feng
Addresses: Forestry and Woodworking Machinery Engineering Technology Center, Northeast Forestry University, Harbin, 150040, China ' Forestry and Woodworking Machinery Engineering Technology Center, Northeast Forestry University, Harbin, 150040, China ' Forestry and Woodworking Machinery Engineering Technology Center, Northeast Forestry University, Harbin, 150040, China ' Forestry and Woodworking Machinery Engineering Technology Center, Northeast Forestry University, Harbin, 150040, China
Abstract: Automobile electromagnetic relay is an electrical component that is widely used in the field of automobile manufacturing. It plays an important role in the control, regulation, and protection of low-voltage electrical systems of vehicles. Because of their frequent use and relatively poor working environment, they may fail and be damaged after being corroded by sand, dust, oil, and other pollutants over a long duration. Consequently, a method for predicting the life of automotive relays based on dual self-attention convolutional neural networks is proposed. The network is composed of global and local time series convolution. Two convolutional neural networks are connected in parallel to extract the global and local features of the automotive electromagnetic relay life data, and the extracted data features are determined to predict the life of the relay. Finally, the root mean square error was used to evaluate the prediction results.
Keywords: convolutional neural network; automotive electromagnetic relay; life prediction; root mean square error; self-attention.
DOI: 10.1504/IJMIC.2021.123369
International Journal of Modelling, Identification and Control, 2021 Vol.38 No.3/4, pp.312 - 319
Received: 08 Dec 2020
Accepted: 25 Jan 2021
Published online: 13 Jun 2022 *