Health Assessment of Cigarette Factory Drying Machine Equipment by Integrating SimCLR Framework and Wavelet Transform

Abstract

In the process of cigarette production, the drying machine, as the core equipment, has low operation and maintenance efficiency and lacks correlation analysis between macro health status and micro fault roots, making it difficult to meet the needs of intelligent manufacturing. To address this challenge, the study aims to construct a dual-layer intelligent health assessment framework that combines macroscopic overall health status assessment of equipment with microscopic key component fault diagnosis. At the macro level, a dynamic health assessment model based on game theory combined with weighting is proposed, which optimizes the subjective weights of the fuzzy analytic hierarchy process and the objective weights of the entropy weight method through Nash equilibrium. At the micro level, a simple framework for unsupervised fault diagnosis of key components, such as rolling bearings, is constructed by integrating continuous wavelet transform and representation contrastive learning. Comparative learning is used to extract strong, robust fault features from unlabeled data. The experimental results showed that the mean absolute error of the macro evaluation model during the state transition stage was only 3.15, and the average evaluation accuracy reached 97.2%, which was significantly better than the traditional weighting model. The accuracy of the micro diagnostic model remained at 90.2% in strong noise environments, and the missed diagnosis rate in variable load testing was less than 2.7%. This dual-layer framework effectively 
achieves a precise closed-loop from overall monitoring to component diagnosis, providing a theoretically valuable solution for predictive maintenance and intelligent operation of complex industrial equipment.