Cloud-Based IoT/Data-Driven NVH Monitoring for Electric Drive Gears: Honing-Induced Waviness, Transferred Vibrations, and DMC-Based Part Matching—Literature Review, Feasibility Study, and Concept Design

This work investigates a smart manufacturing approach to monitor gear noise, vibration, and harshness (NVH) in high-speed electric drivetrain gears. We focus on how micro-geometry errors introduced by the honing process can imprint waviness on gear teeth that causes persistent gear whine (including non-integer “ghost” noise orders). Compounding this challenge, vibrations can propagate through factory structures, making source identification difficult when multiple machines operate in proximity. We propose a cloud-based Industrial IoT architecture: a dense network of low-cost accelerometers synchronized via Precision Time Protocol (PTP IEEE 1588) collects vibration data across the plant. Each measurement is tagged via Data Matrix Code (DMC) and work-order integration to link it to the specific gear and process. Big Data infrastructure (time-series database, object storage) combined with real-time stream processing enables anomaly detection (using models like Isolation Forest and XGBoost) and root-cause analysis with explainable AI (SHAP values). A feasibility study outlines requirements (accuracy, latency, security) and compares design options (wired vs wireless sensors, PTP vs NTP sync, MQTT/OPC UA protocols, edge vs cloud processing). We present a 12-month pilot implementation plan and a conceptual system architecture. The solution aims to reduce scrap and rework, lower warranty risks, enable predictive maintenance, and support smart factory initiatives by providing early-warning NVH quality insights for each produced gear.