Lista zgłoszonych artykułów

The turbocharger regeneration process requires precise assessment of component condition both before and after repair to ensure reliability, performance, and compliance with manufacturer specifications. Traditional diagnostic methods, such as visual inspection or geometric measurements, are time-consuming and often insufficient for detecting micro-damages or dynamic deviations. In response to these limitations, this paper proposes the use of vibroacoustic signal analysis supported by Large Language Models (LLMs) in the context of turbocharger regeneration. The study involved analyzing vibration and acoustic signals collected from turbochargers tested before and after the regeneration process. After preprocessing and extracting relevant features (e.g., resonance bands, amplitude and frequency coefficients), the data were interpreted using LLMs capable of processing complex numerical and descriptive datasets. Models trained on historical data enabled the recognition of typical anomalies, identification of assembly errors, and evaluation of regeneration quality. The application of LLMs not only improved diagnostic accuracy but also allowed for the automation of technical condition reporting in the form of clear, human-readable summaries. The results indicate that integrating vibroacoustic methods with advanced artificial intelligence significantly enhances the quality control of regenerated turbochargers and shows strong potential for industrial applications and specialized service operations.