The management of a fleet of autonomous vehicles involves a number of tasks, including the distribution of vehicles to operations of the production and logistics process and maintenance organizations. In conditions of uncertainty of the impact of the external environment, the problem of timely provision of spare parts arises. Predictive maintenance is based on multi-factor analysis to compare operating scenarios and process dynamics simulation. Multi-factor analysis is performed at the virtual testing stage for early planning of maintenance and repair activities. The paper proposes to verify the results of multi-factor analysis using simulation models on Petri nets. A set of autonomous vehicles is considered, which is divided into three groups: active vehicles in operation, reserve vehicles, and vehicles for implementing the cannibalization strategy. This strategy consists of removing spare parts from a given cannibalized vehicle while significantly delaying the delivery of spare parts lots. The structure of the production and logistics system is formed and the graph of its states is built in the form of a finite state machine. An algorithm for providing autonomous vehicle maintenance with spare parts is developed taking into account the probabilistic characteristics of the processes. A hierarchical simulation model based on stochastic timed colored Petri nets was built. The model contains a main module and three lower-level modules for monitoring the operation time, simulating the maintenance process, ordering and delivering spare parts, and generating a random flow of requests for maintenance and repair. Statistical simulation experiments were conducted to control the deadlines for completing work and making decisions on reservation and cannibalization operations. Conducting virtual tests on a simulation model allows formulating requirements and recommendations for designing maintenance systems for autonomous and unmanned vehicles.
Orlov et al. (Tue,) studied this question.