SU546910A1 - Signal recognition device - Google Patents

Description

3

In the drawing, the disclosed device is shown.

The device consists of three parts: a linear threshold clockframe 1, a weight analog signal generator 2 and a threshold nanometer 3 generator. Linear threshold classifier 1 contains a multiplication unit 4, an integrator 5, an adder 6, and a comparator 7. Weight analogue shaper signal 2 comprises a start-up 8, a linear saw-tooth voltage generator 9, a first non-linear converter 10, a second non-linear generator II, an inverter 12, and an adder of polarity signals 13.

The lgscale threshold classifier 1, with the help of which the set of input implementations is bwh. (O is divided into two classes, it implements the function

| Gj

out Sign. () O (t) dt - 0 I,

where a (t) is an analog weighting function;

7 - the follow-up period when recognizing iodiodic signals or the time of signal analysis, set by the developer, when recognizing individual sections of the signal;

UQ is a constant threshold threshold.

The input signal to be classifi- cated on a real-time scale is fed to one of the inputs of the block multiplied by 4-and-3 starting elements, which selects a short sync pulse that coincides with the beginning of the follow-up period of the input signal. The sync pulse triggers the generator of the linear interpolation accumulator 9, from which voltage is supplied to the first nonlinear converter 10 and the second nonlinear converter 11.

Nonlinear converters 10 and 11 are made of diode-resistor cells, the number of which is determined by the number of node points of the simulated curve approximation. As a result, the arithmetic mean is used, which is obtained previously on the ensemble of realizations from the representatives of the trainings1, their samples of each class.

The difference between the signals from the nonlinear converters 10 and II is generated using an inverter 12 and a diverter of different-polar signals4

13, in the form of a weighing aialogue signal, enters the second input of the wm. 4 unit, the output of which impulses the impulse of a complex shape to the integrator 5. The constant voltage from the output of the integrator 5 is fed to one of the inputs of the adder 6. To the second adder input 6 post} constant threshold voltage.

The magnitude of this voltage is easy to determine, for example, if you form a half-sum of the voltages coming from the outputs of nonlinear converters 10 and 11, multiply the total voltage by the weight function and the resultant integrate

during T.

The creation of the described device allows the creation of small-sized and easy-to-use instruments for the distribution of analog signals in real maoptab

time that improves the accuracy of the device.

Claims (2)

1. A device for recognizing signals comprising a multiplication unit connected to the input of the device, the adder connected to the output of the threshold voltage generator and to the input of the comparator, the output of which is connected to the output of the device, characterized in that, in order to improve the accuracy of scattering and simplification of the device, it contains a weight analog analog generator that is connected to the input of the unit multiply both to the input of the device, and the integrator connected with the output of the multiplication block and the input of the adder. 2. Device io and. 1, characterized in that 1, in which the shaper of the analogue weight siala contains a serially connected starting element, a linear sawtooth generator, a first pelinear converter, an inverter, a summator of different polarity signals, and a second nonlinear converter connected to the generator of a linear sawtooth voltage and to the input of the adder of different polarity signals. The source of information irrespective of the examination: 1. Fu K. Inconsistent methods in pattern recognition and learning. M., Science, 1971, p. 18. vs, .ii} L / x |