Predictive analysis of industrial safety based on noise, vibrations and machinery reliability

This study develops a predictive approach for assessing industrial safety risk through the integrated analysis of physical indicators associated with machinery operation, specifically noise, vibration, and mechanical reliability. The research was conducted in industrial environments characterized by the continuous operation of rotating equipment, including motors, pumps, compressors, and transmission systems. A dataset of approximately 18,000 operational records collected over a 12-month period was analyzed, incorporating acoustic measurements, vibration parameters, machinery condition, and records of potentially unsafe operating states. Equivalent sound pressure level (Leq), Root Mean Square (RMS) acceleration, and crest factor were calculated as the main dynamic indicators, and these variables were normalized and integrated into an Industrial Risk Index (IRI) designed to represent the operational safety state of the equipment. Subsequently, a logistic regression model was developed to classify operating conditions into safe or risk states. The results showed that the combined use of acoustic and vibration indicators improves the identification of hazardous conditions compared with isolated metrics. The predictive model achieved strong classification performance, with an accuracy of 0.88, sensitivity of 0.86, specificity of 0.84, and an AUC-ROC of 0.90, demonstrating a high capacity to distinguish safe operation from risk scenarios. Sustained increases in noise and vibration, particularly when associated with signs of mechanical degradation, were found to precede unsafe states. The findings confirm that integrating dynamic condition monitoring with predictive analytics strengthens failure anticipation and supports preventive decision-making, providing a technically interpretable basis for more proactive industrial safety management systems.