Expert diagnostic system for moving-coil loudspeakers using nonlinear modeling

Abstract

This work aims at the development of an expert diagnostic system for moving-coil loudspeakers. Special emphasis is placed on the defects resulting from loudspeaker nonlinearities. As a loudspeaker operates in the large signal domain, nonlinear distortions may arise and impair sound quality. Analysis of nonlinear responses can shed light on potential design faults of a loudspeaker. By exploiting this fact, this expert diagnostic system enables classification of design faults using a defect database alongside an intelligent fault inference module. Six types of defects are investigated in this paper. A large signal model based on electromechanical analogous circuits is employed for generating the defect database, through which a neural-fuzzy network is utilized for inferring the defect types. Numerical simulations and experimental investigations were undertaken for validating the loudspeaker diagnostic system.