SU478318A1 - Computational Diagnostic Device - Google Patents

Description

one

The invention relates to automation and computing and can be used in the development of recognition devices.

A computational diagnostic device is known that contains an associative memory block with a sequential interrogation by words and bits, a dispenser, AND circuits, counters.

When solving most problems, the device operation speed is low due to the need to perform additional calculations in these cases, for example, using nomograms.

The purpose of the invention is to increase the speed of work.

This is achieved due to the fact that a weighted summation block, a controlled attenuator, a recording block and a coefficient setting unit (| attenuator dividing factor) are inserted into the i-BO device (| input controlled attenuator, the output of which is connected to the registration unit and the measuring input unit setting the division factor of the attenuator, the control input of this unit is connected to the control output associative storage unit, and the output of this unit is connected to the control input of the controlled, attenuator.

This vyloplenie gives the opportunity to speed up the operation of the device.

The proposed device allows diagnostics (ctssssification) by determining the probability by the formula:

. Hepju

PU ,, ... Si, ..., sJ

 Zftqt) NqCDj / $ ,, ..., Si, ..., Sj о с о-) - probability m: the fact that if the object under study has X signs 3, ..., ... Sn, it belongs to the class Dj; - measure of non-success of the studied object with any object of the training sample; fCO) decreasing function q; Nq (Dj / S ,, ..., $ i.v.MSrv) is the number of objects of the training sample from the class Dj, which have a measure of mismatch with the object under study equal to O; j is the class number / 1. 2. i is the recognition number / 1. 2. The drawing shows the block diagram of the device being considered. The proposed device contains an associative storage unit 1 with sequential polling by words and bits. The output of the main information of the associative storage unit is connected to the counting input of the distributor 2, the output of the beginning of words - to the installation input of the distributor, the output of the end of words - to the first inputs of two-input circuits I 35; control output - to the control input of the unit for setting the attenuator division factor 6. Distributor 2 outputs are connected to the second inputs of two-input circuits AND 3-5, the outputs of which are connected to the inputs of counters 7-9. The installation inputs of the counters are connected to the output of the associative storage unit 1. The counting outputs, tweaks 7-9 are connected to the measuring inputs of the weighted summing unit 10, and its output is connected to the measuring input of the controlled attenuator 11. The output of the controlled attenuator 11 is connected to the registration unit 12 and to the measuring input of the unit for setting the division factor of attenuators, the output of which is connected to the control input of the controlled attenuator 11. The proposed device operates as follows. First, an interrogation word code is introduced into the associative memory unit 1, which represents the features of the object under study. Then the operation of the proposed device goes in two cycles. The first cycle is intended to take into account the denominator of the above formula by setting the appropriate 184 division factor of the controlled attenuator 11 when the input signal is fed corresponding to the denominator. In the second cycle, for each class, the probability is found as a fraction of the signal corresponding to the numerator for this class, with respect to the sum of the signals for all classes, i.e., the denominator. In the first case, the associative memory unit 1 sequentially compares the code of the polling word with the codes of the words of the training sample. At each such comparison, the output of the distributor 2 is moved and one of the counters 7-9 registers. After comparing all the codes of any particular group of words (corresponding to one class) of the training sample, the sum of the numerator of the above formula will be generated at the output of the cumulative summation unit. In the first cycle, the impulse to the installation inputs of the counters 7-9 is not applied, give. this, in the process of comparing codes, these counters accumulate the numbers of the form tn D (Di / S ,, ..., SL, ..., Sn), the formulas in the denominator, and the output of the block of the summation block 10 is outputted to a signal corresponding to the whole denominator formulas. At the end of the first cycle, the associative memory unit 1 supplies the control input of the unit 6 for setting the dividing factor of the attenuator to a pulse, under which this unit controls the dividing ratio of the controlled attenuator 11 until the output voltage of the attenuator reaches a fixed value (conditional unit, or 100% probability ) if there is a signal on the attenuator measurement input corresponding to the denominator of the formula. In the second cycle, in contrast to the first, counters 7-9 are set to zero after comparing polling words with words of each class. In this case, the block of summation 10 generates signals corresponding to the numerator of the formula, sequentially for all classes. When the attenuator division factor is set in the first cycle, the registration unit 12 in the second cycle registers for each class a probability value. The subject of the invention is a computational diagnostic device comprising an associative memory unit connected to the valve,. the first inputs AND and the installation inputs of the meters, the outputs of the distributor are connected to the second inputs of the AND circuits, the outputs of which are connected to the inputs of the meters, characterized by the fact that, in order to increase the speed of the device, a weighted summation block was inserted , the registration unit and the unit for setting the division factor of the attenuator, and

meter strokes are connected to the measuring inputs of a weighted summation unit, the output of which is connected to the measuring input of a controlled attenuator, the output of the controlled attenuator is connected to the recording unit and the measuring input of the attenuator division setting unit, the other input of which is connected to the control output of the associative memory unit and with a control input of a controlled attenuator.