CN105512496A - Automatic measurement method of geometric characteristics of randomly-bundled cable bundles - Google Patents

Abstract

The invention discloses an automatic measurement method of geometric characteristics of randomly-bundled cable bundles and belongs to the field of electromagnetic measurement. In order to solve the problem of strict application conditions of the existing method for obtaining parameters of the geometric characteristics of cable bundles, a randomly-bundled cable bundle model based on fractal theory, a cable bundle outer characteristic parameter set based on fractal coefficient, a determinacy forward model for describing relation between the fractal coefficient and impedance characteristic, and a Bayes inverse estimation model of the forward model are respectively built; with combination of a Markov Chain Monte algorithm, the Bayes inverse estimation model is solved by using the actual measurement data of the cable bundle outer impedance characteristic; the calculation is carried out to obtain optimal estimation of the fractal coefficient so as to obtain the distribution characteristic of the geometric characteristics of the randomly-bundled cable bundles. The method is applicable to measurement of the geometric characteristics of various randomly-bundled cable bundles.

Description

Tie up cable beam geometry characteristic method for automatic measurement at random

Technical field

The invention belongs to electromagnetic measurement field.

Background technology

The coupled problem of cable bundle is the typical problem faced in Aeronautics and Astronautics device EMC Design, and this problem is also just receiving increasing concern.And at the EMC Design initial stage, the geometrical property of cable bundle first to be determined, could study its Electro Magnetic Compatibility.At present uniform transmission line model be have employed for the approximate description tying up cable bundle at random, namely suppose that cable bundle xsect geometry does not change vertically between transmission circuit network node.This scheme is the one compromise process made between model complexity and accuracy.And actual conditions are, due to the restriction of cable bundle technique, even if the of the same type same batch cable bundle using same process to produce, also there is very large randomness in its geometric cross-section.In addition, because vibration causes that cable and fixed frame rub mutually, temperature variation causes the factors such as insulation course is aging in cable bundle use procedure, cable structure parameter and medium parameter are all changed a lot, and then affects Distribution electric parameter and the electromagnetic coupling effect of cable bundle is had an impact.Traditional uniform transmission line model is not enough to the change describing these geometrical properties, therefore must adopt the method for measurement, quantizes the randomness of these geometrical property parameters, to study the coupling effect of cable bundle better.

The method of current existing acquisition cable bundle external characteristics is mainly and is obtained by the method for numerical simulation, mainly contains following three kinds of methods:

(1) the uniform transmission line Cascade of Segment equivalent is used, and the cross section geometric structure random variation of adjacent transmission lines, application monte carlo method carries out the distribution character that Multi simulation running can obtain result.The advantage of this kind of method realizes simply, but the computational resource consumed is many, and arbitrary structures cascade easily causes cable discontinuous, affects the accuracy that high band is analyzed.

(2) random mid point set method: generate the randomness that one group of fractal curve comes analog cable position, and the degree using that Fractal Dimension sign cable location changes vertically.Based on RDSI algorithm, use spline interpolation techniques to keep the continuous and level and smooth of cable, use the Gaussian Distribution Parameters of set-point and cable segments to characterize the randomness of cable location.This method can save calculated amount, but the prerequisite used is the wire only comprising same model in cable bundle, does not consider the situation of multiple cable bundle.

(3) WH (Wire-Hole) modelling: the randomness characterizing cable structure by controlling the random transmission of every root cable (Wire) in predefined cable bundle location hole (Hole).This model is still only applicable to the modeling of the cable bundle be made up of single cable.

Summary of the invention

The object of the invention is the problem in order to solve existing cable bundle geometrical property parameter acquiring method model applicable elements harshness, the invention provides one and tying up cable beam geometry characteristic method for automatic measurement at random.

Of the present inventionly tie up cable beam geometry characteristic method for automatic measurement at random, described method comprises the steps:

Step one: measure one group of external impedance characteristic of tying up cable bundle at random to be measured, comprise open-circuit impedance characteristic and short-circuit impedance characteristic, according to the distribution situation of tying up cable beam impedance characteristic at random, obtains impedance operator data of tying up cable bundle at random to be measured;

Step 2: tie up cable bundle at random according to be measured, sets up and ties up cable beam geometry model at random based on fractal theory;

Step 3: according to cable kind of tying up at random in cable bundle to be measured, sets up the parameter set of the description cable bundle geometrical property based on cable fractal coefficient;

Step 4: that sets up according to transmission line theory and step 2 ties up cable beam geometry model at random, builds the mapping relations of cable fractal coefficient and impedance operator, obtains determinacy forward model;

Step 5: the determinacy forward model that the impedance operator data obtained according to step one and step 4 build, sets up Bayes against estimation model;

Step 6: adopt Markov chain Monte-Carlo algorithm to solve against estimation model the Bayes that step 5 obtains, obtain the optimal estimation of fractal coefficient;

Step 7: the optimal estimation of fractal coefficient step 6 obtained substitutes in the parameter set that step 3 obtains, and obtains the parameter of the geometrical property describing cable bundle.

Described step one, measure method of tying up the external impedance characteristic of cable bundle at random to be measured:

Two ends of tying up cable bundle at random to be measured are connected with vector network analyzer by an air plug, an interconnecting device and a switch matrix respectively successively simultaneously; PC control vector network analyzer is utilized to measure open-circuit impedance characteristic and short-circuit impedance characteristic of tying up cable bundle at random to be measured.

Described step 3, the parameter set based on the description cable bundle geometrical property of cable fractal coefficient is:

x＝{x ₁(θ),x ₂(θ),...,x _n(θ)}；

Wherein, θ={ θ ₁, θ ₂..., θ _m∈ R ^mfor cable fractal coefficient;

Wherein, θ _ibe i-th kind of cable fractal coefficient tying up cable bundle at random, m is the quantity of tying up cable bundle kind at random, R ^mfor m ties up real number space.

In described step 4, determinacy forward model is: y=F (θ)+v, and this model is set up based on transmission line theory;

Wherein, y ∈ R ^dfor tying up the impedance operator data of cable bundle at random, d is the dimension of impedance operator data, v ∈ R ^dbe that F (θ) is the forward model not considering stochastic error by the stochastic error introduced when measuring, represent the mapping function of cable fractal coefficient to impedance operator.

In described step 5, Bayes against estimation model is: $\mathrm{Pr}ob\left(\mathrm{\θ}\right|y)=\frac{\mathrm{Pr}ob\left(y\right|\mathrm{\θ})\mathrm{Pr}ob\left(\mathrm{\θ}\right)}{\∫\mathrm{Pr}ob\left(y\right|{\mathrm{\θ}}^{\′})\mathrm{Pr}ob\left({\mathrm{\θ}}^{\′}\right){\mathrm{d\θ}}^{\′}};$

Wherein, Prob (θ) is the prior estimate of the cable fractal coefficient provided based on background knowledge; Prob (y| θ) is the probability distribution of the impedance operator data acquisition tying up cable bundle by a group at random; Prob (θ | be y) likelihood distribution, be in conjunction with determinacy forward model and tie up cable bundle at random impedance operator data after Posterior probability distribution.

In described step 6, the optimal estimation θ of fractal coefficient ^*=maximizeProb (θ | y);

Wherein, ${\mathrm{\θ}}^{*}=\{{\mathrm{\θ}}_1^{*},{\mathrm{\θ}}_2^{*},...,{\mathrm{\θ}}_m^{*}\}.$

Beneficial effect of the present invention is, can quantize the randomness of the complicated cable bundle model geometric characterisitic parameter that multiple cable is formed, applicability is strong, and measuring system is simple to operate, stable and have high performance reproducibility, makes measurement result have very high stability and accuracy.

Accompanying drawing explanation

Fig. 1 adopts measuring table to measure the principle schematic of tying up cable beam geometry characteristic at random in embodiment.

Fig. 2 is the left view of cable support 1 in Fig. 1.

Fig. 3 is the T-shaped equivalent electrical circuit tying up cable bundle at random to be measured in embodiment.

Fig. 4 is the impedance measurement principle schematic diagram tying up cable bundle at random to be measured in embodiment.

Fig. 5 is the principle schematic of the cross-sectional model tying up cable bundle at random.

Embodiment

Composition graphs 1 to Fig. 5 illustrates present embodiment, cable beam geometry characteristic method for automatic measurement of tying up at random described in present embodiment realizes based on test macro, and described test macro comprises cable support 1, two air plugs 2, two interconnecting devices 3, SMA interface 50 Europe resistance 4, two switch matrix 5, vector network analyzer 6, host computer 7 and general purpose interface bus 8.

Cable bundle of tying up at random to be measured can be tied up for same item molded line cable the cable bundle that obtains at random or tie up for dissimilar cable the cable bundle obtained at random.In present embodiment, for the cable bundle of three same item type cable bundle;

One two ends of tying up cable bundle at random to be measured connect two interconnecting devices 3 respectively by air plug 2, one of them interconnecting device 3 is connected with a switch matrix 5 by SMA interface 50 Europe resistance 4, another interconnecting device 3 is directly connected with another switch matrix 5, two switch matrix 5 are by bnc interface connected vector network analyzer, be connected with host computer by USB interface, as shown in Figure 1 simultaneously.

Described measuring method comprises the steps:

Step one, control two switch matrix 5 by host computer 7, vector network analyzer 6 is utilized to measure one group of external impedance characteristic of tying up cable bundle at random to be measured, according to the impedance operator distribution situation of tying up cable bundle at random, obtain impedance operator data of tying up cable bundle at random to be measured;

In present embodiment, set up the high-frequency transmission line model of three conductor cable, consider cable kelvin effect and dielectric loss effect, the T-shaped equivalent electrical circuit as shown in Figure 3 of the RLCG parameter in basic transmission line model is substituted.According to the impedance measurement principle schematic diagram of Fig. 4, measure the external impedance characteristic of tying up cable bundle at random, open-circuit impedance curve and short-circuit impedance curve can be obtained respectively, the impedance operator Z ∈ R recorded ^drepresent.

Present embodiment needs measurement one group external impedance characteristic of tying up cable bundle at random to be measured, multiple like this impedance operator of tying up cable bundle at random to be measured can present certain probability distribution rule, and then determine impedance operator data of tying up cable bundle at random to be measured, improve accuracy.

Step 2, tie up cable bundle at random according to be measured, set up and tie up cable beam geometry model at random based on fractal theory;

In order to describe the randomness of tying up at random in cable bundle external characteristics accurately, the Diamond-Square algorithm in fractal theory is adopted to carry out modeling to wire harness Linear Segments.It is this brass tacks of continuous print that fractal theory have followed cable bundle, can ensure again the description to cable bundle randomness, adopts the spline interpolation used in RDSI method to ensure the flatness of cable bundle structure in modeling simultaneously.

Step 3, according to cable kind of tying up at random in cable bundle to be measured, set up the parameter set of the description cable bundle geometrical property based on cable fractal coefficient;

As shown in Figure 5 tie up in cable bundle cross-sectional model at random, the parameter describing cable bundle external characteristics randomness mainly contains:

Cable centerline position coordinates (x _i, y _i), Cable radius r _i, cable insulating layer thickness △ r _i, the distance s between two cables _i,j.

Except parameter described in cross-sectional model, also has cable bundle length l.

Therefore based on the parameter set of the description cable bundle external characteristics randomness of fractal coefficient be:

{(x _i(θ),y _i(θ)),r _i(θ),△r _i(θ),s _i,j(θ),l(θ)}。

What step 4, foundation transmission line theory and step 2 were set up ties up cable beam geometry model at random, builds the mapping relations of cable fractal coefficient and impedance operator, obtains determinacy forward model::

y＝F(θ)+v(5)

Wherein, y ∈ R ^dfor tying up the impedance operator data of cable bundle at random, d is the dimension of impedance operator data, v ∈ R ^dby the stochastic error introduced when measuring, the impact etc. of such as BNC connector.F (θ) is the forward model not considering stochastic error, represents the mapping function of cable fractal coefficient to impedance operator.

The determinacy forward model that step 5, the impedance operator data obtained according to step one and step 4 build, set up Bayes against estimation model:

$\mathrm{Pr}ob\left(\mathrm{\θ}\right|y)=\frac{\mathrm{Pr}ob\left(y\right|\mathrm{\θ})\mathrm{Pr}ob\left(\mathrm{\θ}\right)}{\∫\mathrm{Pr}ob\left(y\right|{\mathrm{\θ}}^{\′})\mathrm{Pr}ob\left({\mathrm{\θ}}^{\′}\right){\mathrm{d\θ}}^{\′}}---\left(6\right)$

Wherein, Prob (θ) is the prior estimate of the cable fractal coefficient provided based on background knowledge; Prob (y| θ) is the probability distribution of the impedance operator data acquisition tying up cable bundle by a group at random; Prob (θ | be y) likelihood distribution, be in conjunction with determinacy forward model and tie up cable bundle at random impedance operator data after Posterior probability distribution.

Step 6, employing Markov chain Monte-Carlo algorithm solve against estimation model the Bayes that step 5 obtains, and obtain the optimal estimation of fractal coefficient:

θ*＝maximizeProb(θ|y)(7)

Wherein, ${\mathrm{\θ}}^{*}=\{{\mathrm{\θ}}_1^{*},{\mathrm{\θ}}_2^{*},...,{\mathrm{\θ}}_m^{*}\}.$

Step 7, the optimal estimation of fractal coefficient step 6 obtained substitute in the parameter set that step 3 obtains, and obtain the parameter of the geometrical property describing cable bundle:

{(x _i(θ ^*),y _i(θ ^*)),r _i(θ ^*),△r _i(θ ^*),s _i,j(θ ^*),l(θ ^*)}。

Claims (6)

1. tie up a cable beam geometry characteristic method for automatic measurement at random, it is characterized in that, described method comprises the steps:

Step one: measure one group of external impedance characteristic of tying up cable bundle at random to be measured, comprise open-circuit impedance characteristic and short-circuit impedance characteristic, according to the distribution situation of tying up cable beam impedance characteristic at random, obtains impedance operator data of tying up cable bundle at random to be measured;

Step 2: tie up cable bundle at random according to be measured, sets up and ties up cable beam geometry model at random based on fractal theory;

Step 3: according to cable kind of tying up at random in cable bundle to be measured, sets up the parameter set of the description cable bundle geometrical property based on cable fractal coefficient;

Step 4: that sets up according to transmission line theory and step 2 ties up cable beam geometry model at random, builds the mapping relations of cable fractal coefficient and impedance operator, obtains determinacy forward model;

Step 5: the determinacy forward model that the impedance operator data obtained according to step one and step 4 build, sets up Bayes against estimation model;

Step 6: adopt Markov chain Monte-Carlo algorithm to solve against estimation model the Bayes that step 5 obtains, obtain the optimal estimation of fractal coefficient;

Step 7: the optimal estimation of fractal coefficient step 6 obtained substitutes in the parameter set that step 3 obtains, and obtains the parameter of the geometrical property describing cable bundle.

2. according to claim 1ly tie up cable beam geometry characteristic method for automatic measurement at random, it is characterized in that, described step one, measure method of tying up the external impedance characteristic of cable bundle at random to be measured:

Two ends of tying up cable bundle at random to be measured are connected with vector network analyzer by an air plug, an interconnecting device and a switch matrix respectively successively simultaneously; PC control vector network analyzer is utilized to measure open-circuit impedance characteristic and short-circuit impedance characteristic of tying up cable bundle at random to be measured.

3. according to claim 1 and 2ly tie up cable beam geometry characteristic method for automatic measurement at random, it is characterized in that, described step 3, the parameter set based on the description cable bundle geometrical property of cable fractal coefficient is:

x＝{x ₁(θ),x ₂(θ),...,x _n(θ)}；

Wherein, θ={ θ ₁, θ ₂..., θ _m∈ R ^mfor cable fractal coefficient;

Wherein, θ _ibe i-th kind of cable fractal coefficient tying up cable bundle at random, m is the quantity of tying up cable bundle kind at random, R ^mfor m ties up real number space.

4. according to claim 3ly tie up cable beam geometry characteristic method for automatic measurement at random, it is characterized in that, in described step 4, determinacy forward model is: y=F (θ)+v, and this model is set up based on transmission line theory;

Wherein, y ∈ R ^dfor tying up the impedance operator data of cable bundle at random, d is the dimension of impedance operator data, v ∈ R ^dbe that F (θ) is the forward model not considering stochastic error by the stochastic error introduced when measuring, represent the mapping function of cable fractal coefficient to impedance operator.

5. according to claim 4ly tie up cable beam geometry characteristic method for automatic measurement at random, it is characterized in that, in described step 5, Bayes against estimation model is: $\mathrm{Pr}ob\left(\mathrm{\θ}|y\right)=\frac{\mathrm{Pr}ob\left(y|\mathrm{\θ}\right)\mathrm{Pr}ob\left(\mathrm{\θ}\right)}{\∫\mathrm{Pr}ob\left(y|{\mathrm{\θ}}^{\′}\right)\mathrm{Pr}ob\left({\mathrm{\θ}}^{\′}\right){\mathrm{d\θ}}^{\′}};$

Wherein, the prior estimate of cable fractal coefficient that provides of Prob (θ); Prob (y| θ) is the probability distribution of the impedance operator data acquisition tying up cable bundle by a group at random; Prob (θ | be y) likelihood distribution, be in conjunction with determinacy forward model and tie up cable bundle at random impedance operator data after Posterior probability distribution.

6. according to claim 5ly tie up cable beam geometry characteristic method for automatic measurement at random, it is characterized in that, in described step 6, the optimal estimation θ of fractal coefficient ^*=maximizeProb (θ | y);

Wherein, ${\mathrm{\θ}}^{*}=\{{\mathrm{\θ}}_1^0,{\mathrm{\θ}}_2^{*},...,{\mathrm{\θ}}_m^{*}\}.$