FIELD: transportation.
SUBSTANCE: to assess the main parameter, that determines the level and nature of the load for the diagnosis, especially critical parts of vehicles placed sensors, the number and combination of which is chosen individually on the host vehicle, make measurements of different frequency signals form the input parameters for the neural network performed on based on the Nvidia CUDA parallel computing technology, produce initial training of the neural network with the operation of the facility, assess the level of load on the vehicle unit, taking into account complex influence of the individual parameters.
EFFECT: invention provides a reliable vehicle diagnostics.
1 dwg, 1 tbl
