Vibration-based wear diagnostics of jaw crusher steel plates using streaming data analysis

This paper presents a methodology for assessing the wear of crushing plates in jaw crushers processing ferrosilicomanganese, based on vibration spectrum analysis. The study addresses the critical need for continuous equipment condition monitoring without production stoppages. Conventional methods (visual inspection, geometric measurements) fail to provide predictive diagnostics, leading to increased operational costs.A novel approach is proposed, integrating: real-time vibration signal analysis (sampling rate: 2.5 ms); spectrum classification using RBF neural networks; correlation of vibration parameters with actual plate wear (error ≤15%). Algorithms were developed to: quantify wear levels (5 grades: 0–25%, 26–35%, 36–45%, 46–50%, >50%); calculate transition times between wear stages; predict remaining service life.The method was tested on a laboratory setup (DSch120×200 crusher, dunite material) and under industrial conditions (Chelyabinsk Electrometallurgical Plant, DSch-1 crusher, ferrosilicomanganese). Results demonstrated high accuracy (R2 ≥ 0.85) when compared with direct plate thickness measurements.Practical implications: reduced equipment downtime, optimised plate replacement schedules, and improved crushing efficiency.The study paves the way for automated monitoring systems in mining and metallurgical industries.