Dynamic Inclination Identification Methods for Mine-Use Monorail Crane Transport Robots Under Dual Operating Conditions

Abstract

Monorail cranes are essential in auxiliary transportation within deep mines. In order to ensure the stability and safety of the monorail cranes under different traveling conditions of many curved rails, it is necessary to improve the recognition accuracy and reliability of the dynamic inclination angle of the monorail cranes. Therefore, this article proposes a novel dual operating condition dynamic inclination angle joint estimation (DOCDIJE). First, based on the working condition recognition module, the data collected by the multisource sensing equipment are recognized, establishing a solid foundation for further accurate recognition and analysis of dynamic inclination. Second, based on the identification results, variational Bayes and adaptive unscented Kalman filter (VB-AUKF) models, convolutional neural networks, and gated recurrent units with attention mechanisms (CNN-GRU-ATT) models are used to analyze different driving conditions. Under normal operating conditions, the dynamic inclination is computationally determined in real time using an improved VB-AUKF algorithm grounded in the inclination calculation principles revealed by the established dynamic inclination model. During special operating conditions, the CNN-GRU-ATT algorithm predicts the current dynamic inclination in real time by accessing the historical distance-sequence dynamic inclination data stored in the data memory. Eventually, the dynamic inclination data of all working conditions are output in a time-sequential manner. Experimental tests demonstrate that the proposed algorithm error analysis results are significantly smaller than the traditional algorithm, and its dynamic inclination recognition accuracy can reach 95.76%, indicating that the DOCDIJE algorithm has good accuracy and reliability under different operating conditions of the monorail crane.