SU849211A1 - Random process generator - Google Patents

Description

 The invention relates to computing and can be used as a unit of a specialized electronic computer for obtaining random processes, as master equipment for reproducing random vibrations in the study of the reliability and proper functioning of products for various purposes using vibration shakers. Random process generators are known that contain band-pass filters in which the output is controlled and the output level is controlled. The generators contain devices for generating random processes containing a feedback loop. In these devices, a random process with a certain type of spectrum is obtained by loading a definable output signal level at the output of each filter and then summing them on the accumulator ri. However, the instability of the elements (L, C-elements) affects the stability of the main characteristics of the random process generator. , and spir; Difficulties arise in the search for infra-low frequency random processes. In addition, the process of setting the required type of spectrum is laborious and time consuming. . The closest in technical essence to the invention is a random process generator comprising a block of clock frequencies, the outputs of which are connected to the inputs ni of multiplexers, the outputs of which are connected to the inputs of m noise sources, and m digital filters, the second inputs of digital filters are connected to the outputs of noise sources, and the outputs through m sources of the reference signal are connected to the inputs of switches, the other inputs of which are combined with the inputs of multiplexers and through m decoders are connected to outputs m of code registers, and outputs ommutatora connected to inputs of n low-pass filters, whose outputs are connected to inputs of an adder 2. The disadvantage of the known oscillator consists in reheating the error range specifying desired that an Ob cd nnoyapproksimatsiey piecewise constant power spectral density. The most significant is the effect of piecewise constant.

. approximations on the accuracy of the spectra when simulating broadband random vibroprocesses, which are characterized by the presence of 1 dips in the spectral characteristic of resonant bursts.

The purpose of the invention is to improve the accuracy of the generator.

The goal is achieved by a J4TO into a generator containing a block of clock frequency generators, the outputs of which are connected to the inputs of a group of multiplexers, respectively, whose inputs are combined with the first inputs of the corresponding switches and connected to the outputs of the corresponding decoders, whose inputs are connected to the outputs of the corresponding code registers, the outputs of the multiplexers connected to the inputs of the respective noise sources and the first inputs of the corresponding digital filters whose outputs are connected to the first inputs The respective reference sources, the outputs of which are connected to the second inputs of the respective switches, the outputs of which are connected to the corresponding inputs of the filters of the low Fc1stot, the outputs of which are connected to the inputs of the adder, respectively, entered the first and second groups of elements AND, the group of elements OR-NOT, the group of adders modulo two, a group of delay elements, a group of probabilistic binary elements and a group of control flip-flops, the unit and zero outputs of which are connected respectively to the first inputs The signals of the elements of the IS-NOT group and, the elements AND of the second group, the outputs of which are connected to the first inputs of the corresponding modulators are two groups, the outputs of which are connected to the inputs of the corresponding digital group filters and the delay elements of the group respectively, the outputs of which are connected to the second inputs of the corresponding elements AND The second group and the elements of the I – HH group, the outputs of which are connected to the first inputs of the elements AND of the first group, correspond to the outputs of which are connected to the second inputs of the corresponding adders modulo two groups, the outputs of the noise sources of the group are connected to the inputs of probability binary elements of the group, respectively, the first and second outputs of which are connected to the secondary inputs of the AND elements of the first group and the sources of the reference signal, respectively.

FIG. 1 shows a flowchart of a random process generator in FIG. 2, Za and 5a - functional diagrams of generator blocks; in fig. 36, Sv, 56 and 5a are timing diagrams in FIGS. 4, 6 and 7 — dependencies of frequency characteristics.

The generator contains a block of 1 clock generators, m multiplexers 2, m sources of noise 3, m digits OBfcJx filters 4, m sources 5 of the opted signal m switches b, t. Decoders 7, m registers 8 code n filters 9 low frequencies, adder

10, L1 probabilistic binary elements 11 of the group, m elements 12 And the first group, m adders 13 modulo two, m elements 14 And the second group, m elements 15 IS-NOT groups

tons of elements 16 group delays and m triggers 17 management troupe.

In block 1 of clock frequencies, a grid of central frequencies is generated, ensuring the overlapping of a given frequency range, and fed to the inputs of multiplexers 2, which, in accordance with the signals received by ia, their input from the decoders of digital noise and digital filters 4 required frequencies, which are determined Codes are listed on registers of 8 codes, the outputs of which are connected to the inputs of the decoders, from the outputs of which an allowable signal arrives at the first inputs of m switches b. Signals with a uniform spectrum from the outputs m of sources of noise 3 are fed to the inputs of probabilistic binary elements

11, where the probability of occurrence of a unit on an output in each channel is varied. The output signals from the elements 11, the passage through the elements 12 And the first and the adders 13

modulo two, is fed to the inputs of digital filters 4. In addition, control signals from m probability binary elements 11 are fed to inputs m of sources 5 of the reference signal, the second inputs of which are connected to outputs m of digital filters 4, and outputs of sources 5 connected to the second inputs of m switches b.

The outputs m of the adders 13 are connected to the inputs m of the delay elements 16, the outputs of which are connected to the second inputs of elements 15 AND-NOT and elements 14 AND of the second group, and zero and single inputs m are connected to the first inputs m of elements 15 AND-NOT and elements 14 I control triggers 17, respectively. Outputs m ele, cops 15 AND-NOT connected to the secondary inputs m elements 12 And the first group, and the outputs of elements 14 And connected to the second input1M gp adders 13, n outputs m switches b connected by the i-th inputs n filters 9 low frequencies, outputs which are connected to the inputs of the adder 10.

Blocks 1-10 perform similar functions, as do similar blocks in a known generator. The validity binary element 11 is a device that allows to generate binary sequences.

with variable probabilities of occurrence 1. Moreover, in this case, element 11 allows you to adjust the probability of occurrence

in wide

limits from O to 1. Structurally, such a device is implemented according to well-known standard schemes. Elements 12-15 perform the functions of the two-input AND circuits, a modulo-two adder, AI and NAND, respectively, and the delay element 16 performs the function of delaying information by one cycle (Fig. 2).

The device works as follows.

In order to obtain a given power spectral density, the 8 code registers are assigned frequency codes in such a way that wider filters are in areas with a small change in spectral density, and narrower ones are in resonant bursts and dips. The clock code determines the center frequency of the filter. The clock generators block 1 generates a full frequency grid, providing for any recombination of filters with different bandwidths.

All clock frequencies from block 1 are fed to multiplexers 2. Decoders 7, in accordance with the code entered in register 8, give permission for passing noise and filters 4 of a certain clock frequency to the inputs of sources 3. Register codes 8 are not repeated. Total clock frequencies n, and simultaneously dialed codes m p.

Further, white noise from the outputs of the noise source 3 under the action of clock synchronizing pulses is fed to the input of probabilistic binary elements 11, which regulate the probability of occurrence of a unit at the first outputs of elements 12 And (Fig. 1 and 2). Depending on the type of spectral density power in each band generated by the digital filter 4, the control triggers 17 are set to one or zero state. In case the spectral power density in a given specific band has a convex appearance, the control trigger is set to zero. In this case, the functional circuit (FIG. 2) is automatically converted to the circuit shown in FIG. Behind. In case the spectral power density is concave, the control trigger is set to one. In this case, the circuit shown in Fig. 2, repeats the operation of the circuit shown in Fig. 6. ,

Consider the case when it is necessary to reproduce the convex view of the spectral power density in a given

: frequency band. In this case, the set of elements 12-17 (Fig. 2) of the structural scheme, by fixing the zero state on the trigger 17, is converted into the scheme shown in Fig. 3a. The input of the element And (Fig. 3a) from the output of the probabilistic binary element 11 receives a random sequence of independent random binary digits XK with probability P () P, where 1, 2, 3 ...

The input sequence hz (Fig. 4b) is converted into a sequence z (Fig. Sv) in such a way that in the new sequence there are no consecutive units. The energy of the low frequency and high frequency components is reduced. This increases the amplitude

components having a frequency of 1 / 2T. The amplitude of the component with a frequency of 1 / 2T and components close to it at Р () increases especially strongly. Analytically algorithm for transforming a sequence of independent

random binary digits x, in the sequence Z c; (Fig. 3b, c) is described by the expression

,

3 &

L -4-1- (1)

The correlation function of the sequence z; is defined by the expression.

. R (ni) P (Zj, i,). (1) Considering that P (X, 1) P and P (2,1 P (. (Z., I ,,),

and transform the expression (2), receive.

fi (m) p-r () + p.p (). (h;

The value of P (z (; 1) is obtained from the following dependency:

p (2; -r) P- (f- p). (p + pChar ar5-sp + sp 4 ...)

 R / (IR).

To find the correlation function R (m), the inhomogeneous difference equation of the first order is solved with constant coefficients (5) under the initial conditions R (o) P / (l-f-P)

iR (m) + P-R (m-i): pV (i 4p ;. (5) ;.

Applying the Laurent transform to equation (5) and using the advance term, finally

 : "). (- pr (-W The spectral power density of a random sequence r differs from the spectral power density of the input sequence x by the factor w (f; which is defined as follows:, v (, b P L (-Gt-pf b ((- (- tp) (i-vp) i-4.pcsin smt {) a. Expression (7) shows that the use of the 12-divided-modulated XK pulse sequence at the input of the elements, which has a uniform spectrum at the output of the converter from elements 12-17, in the zero state of the control trigger 17, a random process Z1 is obtained, with a variable spectrum The power density has a convex spectrum (Fig. 4). Using the sequence X |. coming from a probabilistic binary element, narrowband randomized processes can be generated with a varying shape of the spectral density, and the shape is changed by changing the probability of the input sequences x Consider the second case when it is necessary to reproduce a concave view of the spectral power density in a particular frequency band. In this case, the set of elements 12-17 (Fig. 2) by fixing on the trigger. The control unit 17 is formed into the circuit shown in FIG. 5a. Modulo two admittance (fig. 5a) at the output of the probabilistic binary element 11 is a random sequence of independent random binary digits x ,. The input sequence x (Fig. 56) is transformed into the sequence z, ((Fig. BB). Analytically, the algorithm for converting independent random binary digits Hc To sequence 1 (Fig. 5) is described by the expression 2lc-.0. The correlation function of the sequence | z, (8) is defined by the expression (2) Bearing in mind that P (.) p (7.o, y, 1) + P (g.1,) C () - P + f (1-P) 45, and transform the expression (2) get YSTN1-2p) P () 5 To find the correlation function R (m), solve a non-uniform first-order difference equation with constant coefficients (9). As a result, R (n) - ( l-2p) .Oo; The spectral power of the random sequence z differs from the spectral density of the input sequence x, by the factor W (f}, which is defined as follows: P + H-lP sin lTf Expression (11) shows that; using 12 delta elements at the input {ta-modulated pulse sequence XK, having a uniform spectrum, with a single state of flip-flop 17, a random process z is obtained with a variable power spectral density having a concave spectrum (Fig. 6). Thus, the probability changes the appearance of binary symbols X c at the outputs of the hl binary probabilistic elements 11 under the corresponding states n control triggers 17 at the outputs of the adders 13 are obtained random processes with a changeable shape having either a convex or concave appearance (Fig.4 and b). the output processes with m adders 13 are fed to the inputs m of digital filters 4, where band-pass filtering of input digital processes occurs. Digital filters 4 separate the band of frequency components from the original process with a center frequency of 1 / 2T, and for each The clock filter frequency is different; as a result, random processes are obtained at the output of digital filters, the frequency components of which lie strictly in certain non-intersecting frequency ranges. In this case, it is possible to approximate the frequency components in each range not only of the piecewise-constant as in the well-known generator, but also in a nonlinear manner by changing the probability P (xy; 1) P in elements 11. When measuring the probability P In the m elements 11 at the outputs m of the digital filters 4, the overall signal level changes. The control signal from the output of the elements 11 to the second input of the sources 5 of the reference signal adjusts the decrease or increase in the level at the outputs of the digital filters 4. In case it is necessary to reproduce uniform scores

the spectral characteristic with a constant level, the control trigger 17 is set to a single value, and at the output of the element 11, P (xc 1) 0.5 is set.

The sources 5 of the reference signal change the output signal levels in accordance with the spectrum of the specified random process. The output signal from the outputs of the sources 5 is fed to the input of a particular low-pass filter. The switching function is carried out by switches, 6, which, in accordance with the codes in registers 8, and, accordingly, the signal at the output of the decoder, connect the output of sources 5 to that other low-pass filter 9. At the output of the summing device, a resulting random process with a given spectrum is obtained. An example of a reproducible process with the required spectrum is shown in FIG. 7

A set of clock frequency codes and a mustache, the triggering of the triggers 17 is performed manually, but this function can be assigned to the computer control program. In this case, the program of distributing filters over the frequency range is executed automatically. Using this approach to implement a random process generator, it becomes possible, in the course of its operation, to quickly change the shape of the power spectral density.

The invention allows a more precise definition of the required spectrum, which is achieved by introducing a non-linear approximation. This entails the possibility of more accurate correction of resonant emissions and dips. With a smaller set of digital filters, it is possible to approximate more complex spectral characteristics,

Thus, the expansion of the functional capabilities of the device is accomplished due to the possibility of reproducing processes with a more complicated spectrum.

The proposed device is characterized by simplicity of technical implementation and relatively small hardware costs. All blocks are implemented using typical elements of computing technology, for example, 155 series integrated circuits.

The economic effect of the application of the invention is determined by a more precise definition of the necessary conditions for vibration tests, which will allow us to obtain more complete information about the reliability of individual components of the tested product and prevent the release of defective products.

Claims (2)

1. KVSSh V.T. Aptsara-. tour to play random vibrations in a reliability study. - Bulletin of mechanical engineering, 1970. 2. Authors certificate of the USSR, 631961, cl. G, p6 .F 1/02, 1977 (prototype) T 17 Id YU H and. . one . I i. . I I I I I 11 t  I .. i. I .. I. I. i. I. I t GSh 5 Jf I. I. i. . . I I I. I. r. . I i. I j 6M - I I I I I. J i ii t .five Fig.Z PC J / 4 Silk p- / g / % f % FIG. 6