Optimization of the End-of-Line Bench Detection System for Gear Teeth Surface Imperfections Identification

This thesis presents an advanced methodology for detecting gear tooth surface defects in the reduction system on an electric drive unit (EDU), with a specific focus on the transfer box (T-Box) integrated into the eCD-140 model, developed by FPT Industrial. The study was conducted on an end-of-line test bench, with the primary objective of establishing an accurate noise threshold to identify defective units. The initial defect detection approach led to an excessive number of false discards, necessitating a reassessment of the selection criteria. To address this issue, a refined methodology was developed through a systematic analysis of previously discarded serial numbers, thereby highlighting a final comparison respect to the initial technique. The implementation of this enhanced methodology significantly reduced the rate of false rejections, and the accuracy of the selection process is increased. By integrating the validated threshold into the end-of-line test bench logic, the proposed analysis demonstrates an improvement in the automation and effectiveness of the quality control process. This improvement minimizes unnecessary production halts and optimize the overall post-processing procedure. As a result, only truly defective units are identified, ensuring that fully functional ones proceed seamlessly to the market release. To understand the engineering requirements and the fundamental role of the NVH characteristic in the development of electric vehicles, it was necessary to explore the state of the art of electric vehicles, allowing for a detailed analysis of the noise and vibration characteristics that affect the systems and subsystems of electric vehicles. Additionally, it was essential to examine the architecture of the end-of-line test bench, as well as the data acquisition and processing process necessary to perform an accurate analysis of the electric axle in terms of NVH.