METHOD FOR AUTOMATED CLASSIFICATION OF SEARCH OBJECTS IN MULTICHANNEL MAGNETOMETRIC SYSTEM BASED ON ARTIFICIAL NEURAL NETWORK Russian patent published in 2023 - IPC G06N3/02 

FIELD: neural networks.

SUBSTANCE: method for automated classification of search objects in a multichannel magnetometric system based on an artificial neural network. In this method, a multichannel magnetometric system, driven and controlled by a foot operator, measures the amplitudes of magnetic induction over the search object with flux-gate sensors placed in four linearly located magnetometric sensitive elements and records the data obtained in the data array controller, where the measured values are used to calculate the values of the depth, magnetic moment, and the distance between the measured amplitude maxima determines the length of the object, and then they are transmitted to the input nodes of the neural network, where, after training the neural network, the measured values of the signals are classified according to three parameters of ferromagnetic objects based on belonging to the objects of search such as unexploded ordnance.

EFFECT: more robust classification of hidden ferromagnetic objects, such as unexploded ordnance, based on the features of the structural relationships of the search object parameters.

1 cl, 5 dwg