FIELD: physics; control.
SUBSTANCE: invention relates to control. The method involves creation an initial fuzzy product rule base for a plurality of scenarios of developments and preliminary training of the control system to execute tasks associated with movement of the mobile object under varying ambient environment conditions. The system is trained using a first genetic algorithm which generates the best version of the fuzzy product rule base for each scenario of developments, and a second genetic algorithm which generates the best version of a neural network for each output linguistic variable for each scenario of developments. Data with the generated best version of the rule base are fed into a fuzzy logic unit and replace the initial rule base for a given scenario. Data generated with the best version of the neural network are fed into a unit with a neural network and replace the initial neural network. The fuzzy logic unit generates control signals for actuating devices depending on the situation based on the changed rule bases and corrected membership function graphs.
EFFECT: broader functional capabilities of the control system and controlling movement of a mobile object in different situations without human involvement.
2 cl, 5 dwg
