The Models for Predictive Maintenance of Robotic Agricultural Vehicles

Abstract

AbstractA cyber-physical system is considered, including a robotic agricultural vehicle system and a complex simulation model for predictive maintenance based on continuous monitoring of the technical condition. Equipping a robotic vehicle with a measuring subsystem, including many sensors and implementing wireless access, provides the transfer of the necessary data for assessing the technical condition in real-time. Continuous monitoring makes it possible to move from preventive maintenance to predictive maintenance, a component of the Industry 4.0 concept, and uses the Internet of Things (IoT) technology. When building an intelligent diagnostic system, the authors used a model-based approach. Together with digital twins of vehicle units, it is possible to identify deterioration processes and predict equipment defects and failures. The general structure of the diagnostic system for robotic agricultural vehicles is presented. The hierarchical structure of simulation models for a robotic vehicle is described. The analysis of existing works in the field of modeling the maintenance for various objects is carried out. A timed colored Petri net model is proposed. The model belongs to the class of stochastic Petri nets and making it possible to evaluate predictive maintenance's effectiveness for a given failure rate of vehicle units and aggregates. A formal description of a robotic vehicle has been developed in the form of the primary parameter multisets. The logical conditions of events in the simulation model are determined. The experimental results have confirmed the model adequacy. The results were applied to study a robotic vehicle group's work in agricultural fields under difficult operating conditions.KeywordsRobotic vehiclesCyber-Physical systemPredictive maintenanceSimulationPetri nets