کنفرانس بین المللی ماشین و محرکه های الکتریکی

صفحه اصلی / The 5th International Conference on Electrical Machines and Drives

An Explainable AI Framework for High-Accuracy Fault Diagnosis and Sound Quality Engineering in Electric Motors

نویسندگان :

Amirhossein Ghadamossoltani¹ Mohammadali Sandidzadeh² Mohammad Mosavi Gazafroodi³ Farzaad Soleimaani⁴

1- دانشگاه علم و صنعت ایران 2- دانشگاه علم و صنعت ایران 3- دانشگاه علم و صنعت ایران 4- دانشگاه علم و صنعت ایران

کلمات کلیدی :

Acoustic Digital Twin،Deep Learning،Electric Motor،Fault Diagnosis،Sound Quality،Explainable AI (XAI)،Condition Monitoring

چکیده :

The hitherto exponential growth in the use of electric motors across important industries, ranging from domestic devices and sophisticated manufacture to EVs, has spawned an ever-increasing need for sophisticated condition monitoring and sound quality engineering methods. Conventional fault diagnostics are expensive and intrusive, while sound quality engineering is a labor-intensive discipline. The present work presents a novel and comprehensive paradigm called the Acoustic Digital Twin (ADT), which builds upon earlier digital twin theory. The ADT has been designed particularly for high-fidelity simulation and modeling of sound profiles that are attendant upon electric machines. It has two chief constituents: (1) an Acoustic Simulation Engine that includes deep generative neural networks and is proficient at generating raw audio signals for a specified set of operational parameters, and (2) a Diagnostic and Interpretation Engine that readily identifies faults by cross-validation between actual and anticipated audio data. By making use of Explainable AI (XAI) techniques, the latter engine doesn't only assign the nature of the fault but also returns an interpretative graphical visualization underlying its physical origin. The software is able to deliver fault diagnosis accuracy above 98% while simultaneously shortening the sound engineering design cycle by a maximum of 50%.