Autonomous vehicle fleet system models for virtual tests in organizing maintenance

The article considers the problem of managing a fleet of agricultural autonomous vehicles (AV) and unmanned combine harvesters. The maintenance system of such a fleet should ensure the implementation of various strategies for the operation, maintenance, and repair of vehicles. To minimize costs and reduce design time, virtual tests based on digital twins in the form of mathematical and simulation models are widely used. A set of system models used in the virtual testing of AV at the stage of virtual commissioning is proposed. The complex includes two optimization models and a simulation model of the agrotechnical production-logistics system. The problem of finding the optimal calendar network schedule for agrotechnical work and the problem of assigning AV to production work are formulated and solved. The simulation model is designed as a stochastic timed colored Petri net. The structure of the hierarchical simulation model is given and the maintenance model based on the Petri net is described. Virtual tests are carried out as a statistical experiment on a simulation model. There is given an example of the resulting graphs of the AV operating rate, which allows determining the periods between preventive maintenance. The choice of the apparatus of stochastic colored Petri nets as the basis of simulation models is due to the convenience and possibilities of setting probabilistic laws for production processes and the processes of AV units' degradation and deterioration. A formal description of AV in the form of multisets with components corresponding to time characteristics and cost parameters is proposed. Conducting virtual tests on the developed models makes it possible to formulate requirements and recommendations for organizing the operation and maintenance of AV.