Abstract
Electrical connections in consumer products are typically made manually rather than through automated assembly systems due to the high variety of connector types and connector positions, and the soft flexible nature of their structures. Manual connections are prone to failure through missed or improper connections in the assembly process and can lead to unexpected downtime and expensive rework. Past approaches for registering connection success such as vision verification or Augmented Reality have shown limited ability to verify correct connection state. However, the feasibility of an acoustic-based verification system for electrical connector confirmation has not been extensively researched. One of the major problems preventing acoustic based verification in a manufacturing or assembly environment is the typically low signal to noise ratio (SNR) between the sound of an electrical connection and the diverse soundscape of the plant. In this study, a physical means of background noise mitigation and signature amplification are investigated in order to increase the SNR between the electrical connection and the plant soundscape in order to improve detection. The concept is that an increase in the SNR will lead to an improvement in the accuracy and robustness of an acoustic event detection and classification system. Digital filtering has been used in the past to deal with low SNRs, however, it runs the risk of filtering out potential important features for classification. A sensor platform is designed to filter out and reduce background noise from the plant without effecting the raw acoustic signal of the electrical connection, and an automated detection algorithm is presented. The solution is over 75% effective at detecting and classifying connections.
Tool geometry analysis for plunge milling of lead-free CuZn-alloys
