Adaptive Multi-Feature Fusion for Vehicle Micro-Motor Noise Recognition Considering Auditory Perception

• DOI: 10.32604/sv.2023.044203
• Publication Date: 2023-12-21T10:30:29Z
• Authors: Ting Zhao; Weiping Ding; Haibo Huang; Yudong Wu
• Keywords: Feature (linguistics); Sensor Fusion

The deployment of vehicle micro-motors has witnessed an expansion owing to the progression in electrification and intelligent technologies. However, some may exhibit design deficiencies, component wear, assembly errors, other imperfections that arise during or manufacturing phases. Consequently, these might generate anomalous noises their operation, consequently exerting a substantial adverse influence on overall comfort drivers passengers. Automobile diverse array structural variations, leading manifestation multitude distinctive auditory irregularities. To address identification forms abnormal noise, this research presents novel approach rooted utilization vibro-acoustic fusion-convolutional neural network (VAF-CNN). This method entails distinct branches, each serving capture disparate features from multi-sensor data, all while considering perception traits inherent human system. intermediary layer integrates concept adaptive weighting features, thus affording calibration mechanism for hailing multiple sensors, thereby enabling further refinement within branch network. For optimal model efficacy, feature fusion is implemented concluding layer. substantiate efficacy proposed approach, paper initially employs augmented data methodology inspired by modified SpecAugment, applied dataset noise samples, encompassing scenarios both with without in-vehicle interior noise. serves mitigate issue limited sample availability. Subsequent comparative evaluations are executed, contrasting performance founded upon single-sensor against models reliant data. experimental results suggested yields heightened recognition accuracy greater resilience interference. Moreover, it holds notable practical significance engineering domain, as furnishes valuable support targeted management emanating micro-motors.