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

Abstract

The invention relates to an automatic measurement method for geometric characteristics of randomly bundled cable bundles, which belongs to the field of electromagnetic measurement. In order to solve the problem that the model of the existing cable harness geometric characteristic parameter acquisition method has strict applicable conditions. The present invention respectively establishes a randomly bundled cable bundle model based on the fractal theory, a cable bundle external characteristic parameter set based on the fractal coefficient, a deterministic forward model describing the relationship between the fractal coefficient and the impedance characteristic, and the Bayesian inverse of the forward model The estimation model, combined with the Markov chain Monte Carlo algorithm, uses the measured data of the external impedance characteristics of the cable bundle to solve the Bayesian inverse estimation model, calculates the optimal estimate of the fractal coefficient, and then obtains the randomly bundled cable harness The distribution characteristics of the geometric parameters. This method is applicable to the measurement of geometric characteristics of various randomly bundled cable bundles.

Description

Automatic Measurement Method of Geometric Characteristics of Randomly Bundled Cable Bundles

technical field

The invention belongs to the field of electromagnetic measurement.

Background technique

The coupling problem of cable bundles is a typical problem faced in the electromagnetic compatibility design of aviation and spacecraft, and this problem is also receiving more and more attention. In the early stage of electromagnetic compatibility design, the geometric characteristics of the cable harness must first be determined before its electromagnetic compatibility performance can be studied. The current approximate description of randomly bundled cable bundles uses a uniform transmission line model, which assumes that the cross-sectional geometry of the cable bundle does not change axially between the nodes of the transmission line network. This solution is a compromise between model complexity and accuracy. However, the actual situation is that due to the limitation of the cable binding process, even if the same type and batch of cable bundles are produced by the same process, their geometric cross-sections still have great randomness. In addition, during the use of the cable harness, due to factors such as mutual friction between the cable and the fixed frame caused by vibration, and aging of the insulating layer caused by temperature changes, the structural parameters and dielectric parameters of the cable have certain changes, which in turn affects the distributed electrical parameters and affects the cable harness. The electromagnetic coupling effect has an influence. The traditional uniform transmission line model is not enough to describe the changes of these geometric characteristics, so measurement methods must be used to quantify the randomness of these geometric characteristic parameters in order to better study the coupling effect of cable bundles.

At present, the existing methods for obtaining the external characteristics of the cable bundle are mainly obtained through numerical simulation methods, and there are mainly the following three methods:

(1) Use segment equivalent uniform transmission line models to cascade, and the cross-sectional geometry of adjacent transmission lines changes randomly, and the distribution characteristics of the results can be obtained by applying the Monte Carlo method for multiple simulations. The advantage of this method is that it is simple to implement, but it consumes a lot of computing resources, and the cascading of any structure is likely to cause discontinuous cables, which affects the accuracy of high-frequency band analysis.

(2) Random midpoint location method: A set of fractal curves is generated to simulate the randomness of the cable position, and the fractal dimension is used to characterize the degree of change of the cable position along the axial direction. Based on the RDSI algorithm, spline interpolation technology is used to keep the continuity and smoothness of the cable, and the Gaussian distribution parameters of the set point and the number of cable segments are used to characterize the randomness of the cable position. This method can save the amount of calculation, but the premise of using it is that the cable bundle only contains the same type of wire, and the situation of multiple cable bundles is not considered.

(3) WH (Wire-Hole) model method: characterize the randomness of the wire structure by controlling the random transfer of each wire (Wire) in the predefined wire harness position hole (Hole). This model is still only suitable for modeling cable harnesses consisting of a single cable.

Contents of the invention

The object of the present invention is to solve the problem that the existing method for obtaining geometric characteristic parameters of cable bundles has strict applicable conditions. The present invention provides an automatic measurement method for geometric characteristics of randomly bundled cable bundles.

The method for automatically measuring the geometric characteristics of random bundled cable bundles according to the present invention, the method comprises the following steps:

Step 1: Measure the external impedance characteristics of a group of randomly bundled cable bundles to be tested, including open-circuit impedance characteristics and short-circuit impedance characteristics, and obtain the impedance characteristics of the randomly bundled cable bundles to be tested according to the distribution of impedance characteristics of the randomly bundled cable bundles data;

Step 2: According to the randomly bundled cable bundle to be tested, a geometric model of the randomly bundled cable bundle based on fractal theory is established;

Step 3: According to the types of cables in the randomly bundled cable bundle to be tested, establish a parameter set based on the cable fractal coefficient to describe the geometric characteristics of the cable bundle;

Step 4: According to the transmission line theory and the geometric model of randomly bundled cable bundles established in Step 2, construct the mapping relationship between the cable fractal coefficient and impedance characteristics, and obtain a deterministic forward model;

Step 5: Establish a Bayesian inverse estimation model based on the impedance characteristic data obtained in step 1 and the deterministic forward model constructed in step 4;

Step 6: Using the Markov chain Monte Carlo algorithm to solve the Bayesian inverse estimation model obtained in step 5, and obtain the optimal estimate of the fractal coefficient;

Step 7: Substituting the optimal estimate of the fractal coefficient obtained in Step 6 into the parameter set obtained in Step 3 to obtain parameters describing the geometric characteristics of the cable bundle.

The first step, the method of measuring the external impedance characteristics of the randomly bundled cable harness to be tested:

The two ends of the random bundled cable bundle to be tested are respectively connected to the vector network analyzer through an aviation plug, a transfer box and a switch matrix in turn; use the host computer to control the vector network analyzer to measure the random bundled cable bundle to be tested The open circuit impedance characteristics and short circuit impedance characteristics.

In the third step, the parameter set describing the geometric characteristics of the cable bundle based on the cable fractal coefficient is:

x _{＝ {} x1(θ), _x2 (θ),...,xn(θ)};

Among them, θ={θ ₁ ,θ ₂ ,...,θ _m }∈R ^m is the cable fractal coefficient;

Among them, θ _i is the cable fractal coefficient of the i-th randomly bundled cable bundle, m is the number of types of randomly bundled cable bundles, and R ^m is the m-dimensional real number space.

In the step 4, the deterministic forward model is: y=F(θ)+v, and this model is established based on the transmission line theory;

Among them, y∈R ^d is the impedance characteristic data of the randomly bundled cable bundle, d is the dimension of the impedance characteristic data, v∈R ^d is the random error introduced in the measurement, F(θ) is the forward direction without considering the random error Model, representing the mapping function from cable fractal coefficients to impedance characteristics.

In the fifth step, the Bayesian inverse estimation model is: $\mathrm{PR}ob\left(\mathrm{\θ}\right|\mathrm{the\; y})=\frac{\mathrm{PR}ob\left(\mathrm{the\; y}\right|\mathrm{\θ})\mathrm{PR}ob\left(\mathrm{\θ}\right)}{\∫\mathrm{PR}ob\left(\mathrm{the\; y}\right|{\mathrm{\θ}}^{\′})\mathrm{PR}ob\left({\mathrm{\θ}}^{\′}\right){\mathrm{d\θ}}^{\′}};$

Among them, Prob(θ) is the prior estimation of cable fractal coefficient based on background knowledge; Prob(y|θ) is the probability distribution obtained from a set of impedance characteristic data of randomly bundled cable bundles; Prob(θ| y) is a likelihood distribution, which is the posterior probability distribution after combining the deterministic forward model and the impedance characteristic data of the randomly bundled cable bundle.

In said step six, the optimal estimate of the fractal coefficient θ ^* =maximizeProb(θ|y);

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

The beneficial effect of the present invention is that it can quantify the randomness of the geometric characteristic parameters of the complex cable bundle model composed of various cables, has strong applicability, and the measurement system is simple, stable and highly repeatable, so that the measurement results have High stability and accuracy.

Description of drawings

Fig. 1 is a schematic diagram of the principle of measuring geometric characteristics of randomly bundled cable bundles by using a measurement platform in a specific embodiment.

FIG. 2 is a left side view of the cable support 1 in FIG. 1 .

Fig. 3 is a T-shaped equivalent circuit of a randomly bundled cable bundle to be tested in a specific embodiment.

Fig. 4 is a schematic diagram of the principle of impedance measurement of a randomly bundled cable bundle to be tested in a specific embodiment.

Fig. 5 is a principle schematic diagram of a cross-sectional model of a randomly bundled cable bundle.

detailed description

This embodiment is described in detail with reference to FIGS. 1 to 5. The method for automatically measuring the geometric characteristics of randomly bundled cable bundles described in this embodiment is realized based on a test system. The test system includes a cable bracket 1 and two aerial plugs 2 , two transition boxes 3, SMA interface 50 ohm resistors 4, two switch matrices 5, vector network analyzer 6, host computer 7 and general interface bus 8.

The randomly bundled cable bundle to be tested may be a cable bundle obtained by randomly bundled cables of the same type or a cable bundle obtained by randomly bundled cables of different types. In this embodiment, a cable bundle bundled with three cables of the same type is taken as an example;

The two ends of a randomly bundled cable bundle to be tested are respectively connected to two transition boxes 3 through aviation plugs 2, one of which is connected to a switch matrix 5 through an SMA interface 50 ohm resistor 4, and the other transition box 3 It is directly connected to another switch matrix 5, and the two switch matrices 5 are connected to the vector network analyzer through the BNC interface, and are connected to the host computer through the USB interface at the same time, as shown in Fig. 1 .

Described measurement method comprises the steps:

Step 1. Control the two switch matrices 5 through the host computer 7, use the vector network analyzer 6 to measure the external impedance characteristics of a group of randomly bundled cable bundles to be tested, and obtain the measured impedance according to the distribution of the impedance characteristics of the randomly bundled cable bundles. Impedance characteristic data of randomly bundled cable bundles;

In this embodiment, a high-frequency transmission line model of a three-conductor cable is established, and the RLCG parameter in the basic transmission line model is replaced by a T-shaped equivalent circuit as shown in FIG. 3 , considering the cable skin effect and dielectric loss effect. According to the schematic diagram of the impedance measurement principle in Figure 4, the external impedance characteristics of the randomly bundled cable bundles are measured, and the open circuit impedance curve and the short circuit impedance curve can be obtained respectively, and the measured impedance characteristics are represented by Z ∈ R ^d .

This embodiment needs to measure the external impedance characteristics of a group of randomly bundled cable bundles to be tested, so that the impedance characteristics of multiple randomly bundled cable bundles to be tested will show a certain probability distribution law, and then determine the randomly bundled cable bundles to be tested Impedance characteristic data, improve accuracy.

Step 2. According to the randomly bundled cable bundle to be tested, a geometric model of the randomly bundled cable bundle based on fractal theory is established;

In order to accurately describe the randomness in the external characteristics of randomly bundled cables, the midpoint displacement method in the fractal theory is used to model the line elements in the bundle. The fractal theory follows the basic fact that the cable bundle is continuous, and can guarantee the description of the randomness of the cable bundle. The spline interpolation used in the RDSI method is used in the modeling to ensure the smoothness of the cable bundle structure.

Step 3, according to the types of cables in the randomly bundled cable bundle to be tested, establish a parameter set based on the cable fractal coefficient to describe the geometric characteristics of the cable bundle;

In the cross-sectional model of a randomly bundled cable bundle as shown in Figure 5, the parameters describing the randomness of the external characteristics of the cable bundle mainly include:

Cable center position coordinates ( _xi , y _i ), cable radius r _i , cable insulation thickness △r _i , distance _si, j between two cables.

In addition to the parameters described in the cross-section model, there is also the cable bundle length l.

Therefore, the parameter set to describe the randomness of the cable bundle characteristics based on fractal coefficients is:

{(x _i (θ), y _i (θ)), ri (θ), △ r _i (θ), s _i , _j (θ), l(θ)}.

Step 4. Based on the transmission line theory and the geometric model of randomly bundled cable bundles established in Step 2, construct the mapping relationship between cable fractal coefficients and impedance characteristics, and obtain a deterministic forward model:

y=F(θ)+v(5)

Among them, y∈R ^d is the impedance characteristic data of randomly bundled cable bundles, d is the dimension of impedance characteristic data, and v∈R ^d is the random error introduced during measurement, such as the influence of BNC connectors, etc. F(θ) is a forward model that does not consider random errors, and represents the mapping function from cable fractal coefficients to impedance characteristics.

Step 5. Based on the impedance characteristic data obtained in step 1 and the deterministic forward model constructed in step 4, a Bayesian inverse estimation model is established:

$\mathrm{<span\; class="notranslate">\; PR</span>}\mathrm{<span\; class="notranslate">\; o</span>}\mathrm{<span\; class="notranslate">\; b</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; \&theta;</span>}<span\; class="notranslate">\; |</span>\mathrm{<span\; class="notranslate">\; the\; y</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; =</span>\frac{\mathrm{<span\; class="notranslate">\; PR</span>}\mathrm{<span\; class="notranslate">\; o</span>}\mathrm{<span\; class="notranslate">\; b</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; the\; y</span>}<span\; class="notranslate">\; |</span>\mathrm{<span\; class="notranslate">\; \&theta;</span>}<span\; class="notranslate">\; )</span>\mathrm{<span\; class="notranslate">\; PR</span>}\mathrm{<span\; class="notranslate">\; o</span>}\mathrm{<span\; class="notranslate">\; b</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; \&theta;</span>}<span\; class="notranslate">\; )</span>}{<span\; class="notranslate">\; \&Integral;</span>\mathrm{<span\; class="notranslate">\; PR</span>}\mathrm{<span\; class="notranslate">\; o</span>}\mathrm{<span\; class="notranslate">\; b</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; the\; y</span>}<span\; class="notranslate">\; |</span>{\mathrm{<span\; class="notranslate">\; \&theta;</span>}}^{<span\; class="notranslate">\; \&prime;</span>}<span\; class="notranslate">\; )</span>\mathrm{<span\; class="notranslate">\; PR</span>}\mathrm{<span\; class="notranslate">\; o</span>}\mathrm{<span\; class="notranslate">\; b</span>}<span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; \&theta;</span>}}^{<span\; class="notranslate">\; \&prime;</span>}<span\; class="notranslate">\; )</span>{\mathrm{<span\; class="notranslate">\; d\&theta;</span>}}^{<span\; class="notranslate">\; \&prime;</span>}}<span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 6</span>}<span\; class="notranslate">\; )</span>$

Among them, Prob(θ) is the prior estimation of cable fractal coefficient based on background knowledge; Prob(y|θ) is the probability distribution obtained from a set of impedance characteristic data of randomly bundled cable bundles; Prob(θ| y) is a likelihood distribution, which is the posterior probability distribution after combining the deterministic forward model and the impedance characteristic data of the randomly bundled cable bundle.

Step 6, using the Markov chain Monte Carlo algorithm to solve the Bayesian inverse estimation model obtained in step 5, and obtain the optimal estimate of the fractal coefficient:

θ*=maximizeProb(θ|y)(7)

in, ${\mathrm{\&theta;}}^{*}=\{{\mathrm{\&theta;}}_1^{*},{\mathrm{\&theta;}}_2^{*},...,{\mathrm{\&theta;}}_m^{*}\}.$

Step 7, substituting the optimal estimate of the fractal coefficient obtained in step 6 into the parameter set obtained in step 3 to obtain the parameters describing the geometric characteristics of the cable harness:

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

Claims (6)

1. An automated method for measuring geometrical characteristics of a bundle of randomly bundled cables, the method comprising the steps of:

the method comprises the following steps: measuring external impedance characteristics of a group of randomly bundled cable bundles to be tested, wherein the external impedance characteristics comprise open-circuit impedance characteristics and short-circuit impedance characteristics, and obtaining impedance characteristic data of the randomly bundled cable bundles to be tested according to the distribution condition of the impedance characteristics of the randomly bundled cable bundles;

step two: establishing a geometric model of the randomly bundled cable bundle based on a fractal theory according to the randomly bundled cable bundle to be detected;

step three: establishing a parameter set for describing the geometric characteristics of the cable bundle based on a cable fractal coefficient according to the cable type in the cable bundle to be bound randomly;

step four: according to the transmission line theory and the geometric model of the randomly bundled cable bundle established in the second step, constructing a mapping relation between a cable fractal coefficient and impedance characteristics to obtain a deterministic forward model;

step five: establishing a Bayesian inverse estimation model according to the impedance characteristic data obtained in the first step and the deterministic forward model established in the fourth step;

step six: solving the Bayesian inverse estimation model obtained in the fifth step by adopting a Markov chain Monte Carlo algorithm to obtain the optimal estimation of the fractal coefficient;

step seven: and substituting the optimal estimation of the fractal coefficient obtained in the sixth step into the parameter set obtained in the third step to obtain the parameters describing the geometric characteristics of the cable bundle.

2. The method for automatically measuring geometrical characteristics of randomly bundled cable bundles according to claim 1, wherein the first step is a method for measuring external impedance characteristics of randomly bundled cable bundles to be measured:

two ends of the to-be-tested randomly-bundled cable bundle are respectively connected with the vector network analyzer sequentially through an aerial plug, a switching box and a switch matrix; and measuring the open-circuit impedance characteristic and the short-circuit impedance characteristic of the random bundled cable bundle to be measured by utilizing the upper computer to control the vector network analyzer.

3. The method for automatically measuring the geometric characteristics of the randomly bundled cable bundle according to claim 1 or 2, wherein in the third step, the parameter set describing the geometric characteristics of the cable bundle based on the cable fractal coefficients is as follows:

x＝{x₁(θ),x₂(θ),...,x_n(θ)}；

wherein θ ═ θ₁,θ₂,...,θ_m}∈R^mIs a cable fractal coefficient;

wherein, theta_iIs the cable fractal coefficient of the ith randomly bundled cable bundle, m is the number of the randomly bundled cable bundle types, R^mIs an m-dimensional real number space.

4. The method for automatically measuring geometrical characteristics of randomly bundled cables as claimed in claim 3, wherein in the fourth step, the deterministic forward model is: f (θ) + v, the model is established based on transmission line theory;

wherein, y ∈ R^dImpedance characteristic data for randomly bundled cable bundles, d is the dimension of the impedance characteristic data, v ∈ R^dThe method is characterized in that random errors are introduced during measurement, and F (theta) is a forward model without considering the random errors and represents a mapping function from a cable fractal coefficient to impedance characteristics.

5. The method for automatically measuring the geometric characteristics of the randomly bundled cable bundle as claimed in claim 4, wherein in the step five, the Bayesian inverse estimation model is as follows: $\mathrm{Pr}ob\left(\mathrm{\&theta;}|y\right)=\frac{\mathrm{Pr}ob\left(y|\mathrm{\&theta;}\right)\mathrm{Pr}ob\left(\mathrm{\&theta;}\right)}{\&Integral;\mathrm{Pr}ob\left(y|{\mathrm{\&theta;}}^{\&prime;}\right)\mathrm{Pr}ob\left({\mathrm{\&theta;}}^{\&prime;}\right){\mathrm{d\&theta;}}^{\&prime;}};$

wherein, Prob (θ) gives a priori estimate of the fractal coefficient of the cable; prob (y | θ) is a probability distribution obtained from impedance characteristic data of a set of randomly bundled cable bundles; prob (θ | y) is a likelihood distribution, which is a posterior probability distribution after combining the deterministic forward model and the impedance characterization data of the randomly bundled cable bundle.

6. The method for automatically measuring geometrical characteristics of randomly bundled cables as claimed in claim 5, wherein in step six, the optimal estimation of fractal coefficients θ^*＝maximizeProb(θ|y)；

Wherein, ${\mathrm{\&theta;}}^{*}=\{{\mathrm{\&theta;}}_1^0,{\mathrm{\&theta;}}_2^{*},...,{\mathrm{\&theta;}}_m^{*}\}.$