WaveSpectro-XAF: Dual-branch deep learning with STFT spectrograms for automotive brushed DC motor noise classification

Willy Dharmawan; Haitham Saleh; Nur Hamid; Rahmayati Alindra; Sami El-Ferik; Choesnul Jaqin

doi:10.1016/j.icte.2026.01.004

WaveSpectro-XAF: Dual-branch deep learning with STFT spectrograms for automotive brushed DC motor noise classification

Willy Dharmawan
, Haitham Saleh^*
, Nur Hamid
, Rahmayati Alindra
, Sami El-Ferik
, Choesnul Jaqin

^*Corresponding author for this work

Research output: Contribution to journal › Article › peer-review

Abstract

This paper presents WaveSpectro-XAF, a dual-branch deep learning model that fuses raw waveforms and STFT spectrograms through cross-attention for DC-motor noise classification. Using 4008 recordings across OK, NG High Noise, NG Vibration, and NG Abnormal classes, the model was evaluated under three STFT windowing schemes. WaveSpectro-XAF achieved the highest performance among all baselines, reaching 0.981 mean F1, 0.989 best F1, and above 99.9% AUC. Statistical tests (ANOVA, Kruskal–Wallis, post-hoc analyses) confirm significant robustness. Results demonstrate the model’s suitability for automated, reliable Industry 4.0 motor noise diagnostics.

Original language	English
Journal	ICT Express
DOIs	https://doi.org/10.1016/j.icte.2026.01.004
State	Accepted/In press - 2026

Bibliographical note

Publisher Copyright:
Copyright © 2026. Published by Elsevier B.V.

UN SDGs

This output contributes to the following UN Sustainable Development Goals (SDGs)

SDG 9 Industry, Innovation, and Infrastructure

Keywords

Automated quality control
Cross-attention fusion
Dual-branch deep learning
Motor noise classification
STFT spectrogram

ASJC Scopus subject areas

Software
Information Systems
Hardware and Architecture
Computer Networks and Communications
Artificial Intelligence

Access to Document

10.1016/j.icte.2026.01.004

Cite this

@article{7da4e740149d4eb2ad346e0f11a642f2,

title = "WaveSpectro-XAF: Dual-branch deep learning with STFT spectrograms for automotive brushed DC motor noise classification",

abstract = "This paper presents WaveSpectro-XAF, a dual-branch deep learning model that fuses raw waveforms and STFT spectrograms through cross-attention for DC-motor noise classification. Using 4008 recordings across OK, NG High Noise, NG Vibration, and NG Abnormal classes, the model was evaluated under three STFT windowing schemes. WaveSpectro-XAF achieved the highest performance among all baselines, reaching 0.981 mean F1, 0.989 best F1, and above 99.9\% AUC. Statistical tests (ANOVA, Kruskal–Wallis, post-hoc analyses) confirm significant robustness. Results demonstrate the model{\textquoteright}s suitability for automated, reliable Industry 4.0 motor noise diagnostics.",

keywords = "Automated quality control, Cross-attention fusion, Dual-branch deep learning, Motor noise classification, STFT spectrogram",

author = "Willy Dharmawan and Haitham Saleh and Nur Hamid and Rahmayati Alindra and Sami El-Ferik and Choesnul Jaqin",

note = "Publisher Copyright: Copyright {\textcopyright} 2026. Published by Elsevier B.V.",

year = "2026",

doi = "10.1016/j.icte.2026.01.004",

language = "English",

journal = "ICT Express",

issn = "2405-9595",

publisher = "Korean Institute of Communications and Information Sciences",

}

TY - JOUR

T1 - WaveSpectro-XAF

T2 - Dual-branch deep learning with STFT spectrograms for automotive brushed DC motor noise classification

AU - Dharmawan, Willy

AU - Saleh, Haitham

AU - Hamid, Nur

AU - Alindra, Rahmayati

AU - El-Ferik, Sami

AU - Jaqin, Choesnul

PY - 2026

Y1 - 2026

N2 - This paper presents WaveSpectro-XAF, a dual-branch deep learning model that fuses raw waveforms and STFT spectrograms through cross-attention for DC-motor noise classification. Using 4008 recordings across OK, NG High Noise, NG Vibration, and NG Abnormal classes, the model was evaluated under three STFT windowing schemes. WaveSpectro-XAF achieved the highest performance among all baselines, reaching 0.981 mean F1, 0.989 best F1, and above 99.9% AUC. Statistical tests (ANOVA, Kruskal–Wallis, post-hoc analyses) confirm significant robustness. Results demonstrate the model’s suitability for automated, reliable Industry 4.0 motor noise diagnostics.

AB - This paper presents WaveSpectro-XAF, a dual-branch deep learning model that fuses raw waveforms and STFT spectrograms through cross-attention for DC-motor noise classification. Using 4008 recordings across OK, NG High Noise, NG Vibration, and NG Abnormal classes, the model was evaluated under three STFT windowing schemes. WaveSpectro-XAF achieved the highest performance among all baselines, reaching 0.981 mean F1, 0.989 best F1, and above 99.9% AUC. Statistical tests (ANOVA, Kruskal–Wallis, post-hoc analyses) confirm significant robustness. Results demonstrate the model’s suitability for automated, reliable Industry 4.0 motor noise diagnostics.

KW - Automated quality control

KW - Cross-attention fusion

KW - Dual-branch deep learning

KW - Motor noise classification

KW - STFT spectrogram

UR - https://www.scopus.com/pages/publications/105041731146

U2 - 10.1016/j.icte.2026.01.004

DO - 10.1016/j.icte.2026.01.004

M3 - Article

AN - SCOPUS:105041731146

SN - 2405-9595

JO - ICT Express

JF - ICT Express

ER -

WaveSpectro-XAF: Dual-branch deep learning with STFT spectrograms for automotive brushed DC motor noise classification

Abstract

Bibliographical note

UN SDGs

Keywords

ASJC Scopus subject areas

Access to Document

Fingerprint

Cite this