FIELD: hydro acoustics.
SUBSTANCE: invention relates to the field of hydro acoustics and can be used for the detected in the direction finding mode hydro acoustic signals sources recognition and classification expert intelligent systems development. Hydro acoustic signals detection and their neural network classification system, containing the analog-digital converter, to which input an input signal is supplied, and which output is connected to the recirculator input, which output is connected to the M narrowband filters inputs. M narrowband filters outputs are connected to the Μ multipliers pairs first inputs, which outputs are connected to the M integrators pairs inputs, which outputs are connected to the Μ quadrators pairs inputs. Μ quadrators pairs outputs are pairwise connected to the Μ adders inputs, which outputs are connected to the M square root calculators inputs, which outputs are connected to the M delay devices inputs, which outputs are connected to the adder M inputs, which output is connected to the threshold device input, permanent storage device 2M outputs are connected to the M multipliers pairs second inputs. Control device outputs are connected to the analog-digital converter, recirculator, read-only memory and threshold devices control inputs. Principal difference from the prototype is that the targets neural network recognition and classification circuit is additionally introduced, which contains a target class recognition unit by the amplitude-frequency characteristic, by the feedback covered with the learning unit. At that, the threshold device output is connected to the target class by the amplitude-frequency characteristic recognition unit input, which output generates a signal by the target type according to the studied spectral area to the classification object belonging degree.
EFFECT: enabling of detected in the direction finding mode hydro acoustic signals surface and underwater sources automatic recognition and classification.
1 cl, 5 dwg
