SU922806A1 - Device for determining current value of likehood function of autoregrassion random process - Google Patents

Description

The invention relates to the technical cybernetics and can be used for pattern recognition, for example, in technical diagnostics.

A device for determining the current value of the likelihood function of an autoregressive random process is known, which contains a delay unit whose outputs are connected to the inputs of an adder with weight inputs, the remaining input of which is connected to the input of the delay unit, and the output is connected to the input of a quadrator, the output of which is connected to the input of an adder, and there is also an analog-code converter, four adders, two quadrators, a control unit, three multiplication blocks, a block for calculating the logarithm of the product of determinants, a memory register and a selection block maximum Ia (1].

However, the known device has insufficient resolution 2.

The accuracy and accuracy of measuring the likelihood function of an autoregressive random process described by an autoregressive model is higher than the second order, since the correlation of the signal sample at a given time with its previous samples lagging more than two quantization steps in time is not taken into account.

The closest in technical essence is a device for determining the value of the likelihood function of an autoregressive random process containing a delay unit, the output of which is connected to one of the inputs of the adder with weight inputs, the other input of which is connected to the input of the delay unit, and the output the mentioned quadrant is connected to one of the inputs of the additional adder, the output of which is connected to the input of the averaging circuit,

922806

the output of which is connected to the input of the classifier. In addition, the device contains additional quadrants, the inputs of which are connected to the inputs of an adder with weight inputs 5, and the outputs are connected to the inputs of an additional adder [2].

However, the known device has insufficient resolution and accuracy of measurement of the current value of the Rare Similarity Function of an autoregressive random process described by an autoregressive model higher than the first order, since the correlation of the signal count at a given time with its previous values is not taken into account. readings that are lagging more than one step of the delay in time and therefore very valuable information is not taken into account in the device.

Ego, as a rule, does not allow to achieve high resolution and accuracy.

The purpose of the invention is to increase 25 the resolution of measuring the current value of the likelihood function of an autoregressive random process.

This goal is achieved by the fact that, in a device for determining the current value of the likelihood function, an autoregressive random process contains a delay unit whose input is the input of the device ₃₅ and is connected to the first input of the first multi-input adder, the other inputs of which are connected to the corresponding outputs of the delay unit the output of the first multi-input adder is connected to the input of the quad, the output of which is connected to the first input of the second adder, the output of which is connected to the input of the averaging unit, vv There is one reference source, the output of which is connected to the second input of the adder, and the third input of which is connected to the output of the averaging unit.

In addition, the purpose of the invention is to improve accuracy. The resolution of the device is improved because the device takes into account the correlation between the signal readout at a given time and its previous readings, which are lagging behind each other in time by the delay step. For the same reason

accuracy is also improved, especially if the signal readout at a given time is strongly correlated with the signal readings that are delayed by several delay steps.

The drawing shows a block diagram of a device for measuring the current value, likelihood function of a fourth-order autoregressive random process.

The device comprises a delay unit 1 consisting of delay elements 2 with a delay step. The outputs of the delay unit 1 are connected to the weight inputs of the multi-input adder 3, the first input of which is connected to the input of the delay unit 1. The output of the adder is connected to the input of the quadrant 4. The output of the quadrant 4 is connected to one of the inputs of the adder 5, the output of which is connected to the input of the averaging unit 6, the output of which is connected to another input of the adder 5. The third input of the adder 5 is connected to the reference voltage source 7.

The device works as follows.

An analog signal representing a centered autoregressive random process is fed to the input ₍ delay unit 1 and delayed by each element 2 by delay line by delay step%). From the input and outputs of the delay unit 1, signals are sent to the weight inputs of the adder 3. From the input of unit 1 the delay is taken the time-delayed signal, from the outputs of the elements 2 of the block 1 delay, located from left to right, the signals Χ ^ -ΐ,., are taken, respectively;

where p is the order of the autoregression model (it is equal to the number of 2 delay elements in the 1 delay block). The output signals of adjacent elements 2 delays are shifted in time by a delay step. In the adder 3 with the weight inputs, each input signal is multiplied by the weight coefficient ^{50 the} square of this input K ^, where

2, ..., p and after that all works are summed up. As the output of the adder 3 with the weight inputs, the value K ₀ X ^ + K ₁ X | ^. + K < ₁ X _{£ is} obtained. <^ - +

⁵⁵ + .., + ΚρΧ ^ .ρΐ, which is fed to the input of the quadrant 4. Quadrator 4 squares this value. From the output of the quad, 4 signal is received5 922806

6

em on one of the inputs of the adder 5.

At the other input of the adder $ through block 6 averaging signal from the output of the same adder. To the third input of the adder 5 from the source 7 5

a constant voltage of magnitude C is supplied. From the output of the adder 5, the current value of the likelihood function of the autoregressive random process is removed ¹⁰

ί <p. _τ -C- (Κ ₀ Χ ^ + Κ _ή + K.Hh.p _G +

+ ... K _{p Xp7} ) ¹ , ^d

moreover, 1 - ^ - o = O and X ^. ₁ = 0, when where I. t is the current value of the function.

the likelihood of an autoregres- ^{, 5-} zion random process;

ά - averaging constant (А ^ 1),

. c £ = e ^t , 20

where T is the averaging time;

C - weighing constant:

C = 1n (72l b), where 31 = 3.14,

Lt- is the variance of the forecast;

1f "T" value of the function dvdopodopo- ²⁵

biography of an autoregressive random process at a time;

К _в , К ₄ , К „, ..., К _Р - weight coefficients ι .. д г 30

cents; ,

where is a; - autoregression model coefficients;

the order of the autoregression model;

- delay step. ^one

In comparison with the known, the proposed device is more convenient in recognizing the random 40 process, since it immediately gives out

the current value of the function of the likelihood of an autoregression random process and thus it is not necessary to recognize additional processes that require a large amount of computation to recognize a random process.

Claims (1)

Claim A device for determining the current value of the likelihood function of an autoregressive random process containing a delay unit, whose input is the device input and connected to the first input of the first multi-input adder, other inputs of which are connected to the corresponding 'outputs of the delay unit, the output of the first multi-input adder is connected to the quad input, output which is connected to the first input of the second adder, the output of which is connected to the input of the averaging unit, characterized in that, in order to increase p Resolution, the device has a voltage source, the output of which is connected to the second input of the second adder, the third input of which is connected to the output of the averaging unit.