Method for predicting crosstalk voltage peak value at far end of disturbed wire during cable crosstalk
By constructing a field-wire coupled electromagnetic crosstalk simulation model and an improved dual hidden layer radial basis function neural network, the time and cost problems of the distracted line voltage evaluation during cable crosstalk are solved, and fast and accurate voltage is achieved Peak prediction reflects the actual cable crosstalk situation.
Patent Information
- Application Number
- CN202411069799.X
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-08-06
- Publication Date
- 2024-11-15
AI Technical Summary
In the prior art, when studying cable crosstalk, the evaluation of the distracted line's far-end crosstalk voltage requires a lot of time and design costs, and traditional simulation methods are complex in calculations and are not suitable for fast prediction.
A field-line coupled electromagnetic crosstalk simulation model is constructed, and a modified dual hidden layer radial basis function neural network is used to predict the crosstalk voltage peak at the distracted line with cable-related parameters and interference source parameters.
It saves calculation time, improves simulation efficiency, and can quickly predict the peak of the distracted line far-end crosstalk voltage under changes in any cable-related parameters and interference source parameters, reflecting the actual cable crosstalk situation.
Smart Images

Figure CN118965997A_ABST
Abstract
Description
Technical Field
[0001] The invention belongs to the field of electromagnetic compatibility, and in particular relates to a method for predicting a far-end crosstalk voltage peak value of a disturbed line during cable crosstalk. Background Art
[0002] Cable crosstalk is a common electromagnetic interference. When a signal is transmitted on a cable, it couples with adjacent cables through inductance and / or capacitance to generate interference voltage. Crosstalk will affect the normal transmission of the signal, and in severe cases may even affect electronic equipment. The far-end crosstalk voltage refers to the crosstalk voltage detected at the receiving end of the cable that transmits the signal. It reflects the noise generated at the receiving end due to attenuation and crosstalk effects during the transmission process. If the far-end crosstalk voltage is too large, it will cause damage to the connected equipment or devices. Therefore, it is necessary to analyze and study the far-end crosstalk voltage of the disturbed line.
[0003] At present, there are three common simulation models for studying cable crosstalk: lumped parameter model, field-circuit joint analysis model and transmission line model. The lumped parameter model method is not suitable for solving the propagation characteristics of high-frequency waves; the advantage of the field-circuit joint analysis model method is that the calculation results are accurate, but the calculation is cumbersome and occupies a large amount of memory; the transmission line model method obtains the corresponding electromagnetic crosstalk by combining the terminal conditions, with simple steps, easy implementation and accurate results. The evaluation steps of the far-end crosstalk voltage of the disturbed line are usually repeated experimental testing and design, which will cost a lot of time and design costs; and the traditional simulation method requires repetitive work, long simulation time, complex calculation, and cannot meet the rapid prediction requirements of engineering applications. Summary of the invention
[0004] In view of the problems existing in the prior art, the purpose of the present invention is to propose a method for predicting the far-end crosstalk voltage peak of a disturbed line during cable crosstalk. In this method, a far-end crosstalk voltage peak prediction model for a disturbed line is constructed to realize the prediction of the far-end crosstalk voltage peak of the disturbed line during arbitrary cable crosstalk, reflecting the coupling situation of the disturbed line under real crosstalk conditions.
[0005] The technical solution adopted by the present invention to achieve the above-mentioned purpose is: a method for predicting the far-end crosstalk voltage peak value of a disturbed line during cable crosstalk, comprising:
[0006] Step 1: Construct a field-line coupled electromagnetic crosstalk simulation model by combining the existing transmission line model analysis method;
[0007] The field-line coupling electromagnetic crosstalk simulation model includes an interfering line and a disturbed line. A unit pulse signal with a set frequency and amplitude is applied to one end of the interfering line. The interference source signal on the interfering line is simulated by accumulating the energy of the pulse signal. The far-end crosstalk voltage peak of the disturbed line is obtained by simulation calculation.
[0008] Assume that the interference source signal on the interference line is f(t), and its expression is:
[0009]
[0010] Where i represents the i-th pulse signal, t represents time, n is the number of sampling points, N represents the total number of accumulated pulse signals, δ(t-nT) is the unit pulse signal, and its period is T; for the interference source signal f(t) with a sampling period of T′, its duty cycle D is related to the n-th duty cycle factor a n The relationship is expressed as:
[0011]
[0012] Step 2: Using different cable-related parameters and interference source parameters as input variables and the far-end crosstalk voltage peak value of the disturbed line as the output variable, the constructed field-line coupling electromagnetic crosstalk simulation model is used for simulation, a data set is constructed, and the data set is normalized, and the normalized data set is divided into a training set and a test set; wherein the cable-related parameters include the cable spacing L, the cable height from the ground H, and the cable radius r;
[0013] Step 3: Construct a prediction model for the peak value of the far-end crosstalk voltage of the disturbed line. The prediction model for the peak value of the far-end crosstalk voltage of the disturbed line adopts an improved double-hidden layer radial basis function neural network. The improved double-hidden layer radial basis function neural network is based on the traditional single-hidden layer radial basis function neural network, and adds a hidden layer whose radial basis function adopts Multi-Quadric function; the cable related parameters and the interference source parameters are used as the input variables of the prediction model for the peak value of the far-end crosstalk voltage of the disturbed line. The peak value of the far-end crosstalk voltage V max as the output variable of the affected line far-end crosstalk voltage peak prediction model; input the training set into the affected line far-end crosstalk voltage peak prediction model, train the affected line far-end crosstalk voltage peak prediction model, and obtain the trained affected line far-end crosstalk voltage peak prediction model; input the test set into the trained affected line far-end crosstalk voltage peak prediction model to test it and verify its performance. At this point, the affected line far-end crosstalk voltage peak prediction model for predicting the affected line far-end crosstalk voltage peak is obtained, and the affected line far-end crosstalk voltage peak prediction is completed during cable crosstalk.
[0014] Further, in step 2, the interference source is a unit pulse signal equivalent to a sine wave, a triangle wave, a sawtooth wave or a square wave signal.
[0015] When the interference source on the interference line is equivalent to a square wave signal, the interference source parameter is the rise time T rise , peak time T hold and pulse peak value V pulse .
[0016] In step 2, different cable-related parameters and interference source parameters are used as input variables, including: cable spacing L, cable height from the ground H, cable radius r, rise time T rise , peak time T hold and pulse peak value V pulse , where the cable spacing L ranges from 10mm to 50mm; the cable height H ranges from 50mm to 80mm; the cable radius r ranges from 0.05mm to 0.15mm; the rise time T rise The value range is 0.0005μs~0.0007μs; the peak time T hold The value range is 0.005μs~0.006μs; the pulse peak value V pulse The value range is 5V~7V;
[0017] The specific values of each input variable are as follows:
[0018] Cable spacing L: L = L1 + 10L0, L0 = 0, 1, 2, 3, 4, unit: mm;
[0019] Cable height from the ground H: H = H1 + 10H0, H0 = 0, 1, 2, 3, unit: mm;
[0020] Cable radius r: r = r1 + 0.05r0, r0 = 0, 1, 2, in mm;
[0021] Rise time T rise :T rise =m1+0.0002m0, m0=0,1, unit is μs;
[0022] Peak time T hold :T hold =a1+0.001a0, a0=0,1, unit is μs;
[0023] Pulse peak value V pulse :V pulse =v1+2v0, v0=0,1, unit is V;
[0024] Among them, L1, H1, r1, m1, a1 and v1 represent the input variables cable spacing L, cable height from the ground H, cable radius r, rise time T respectively. rise , peak time T hold , Pulse peak value V pulse The initial values are L1=10mm, H1=50mm, r1=0.05mm, m1=0.0005μs, a1=0.005μs, v1=5V, and then the input variables are evaluated using the above expressions.
[0025] Furthermore, when a sinusoidal wave signal is used as an equivalent interference source on an interference line, the interference source parameters are amplitude, period and phase.
[0026] Furthermore, when the interference source on the interference line is equivalent to a triangular wave signal, the interference source parameters are peak value, period, frequency, rise time and fall time.
[0027] Furthermore, when the sawtooth wave signal is used as the interference source on the interference line, the interference source parameters are amplitude, scanning time, retrace time, and scanning rate.
[0028] The mathematical relationship between the input and output of the improved double hidden layer radial basis function neural network in step 3 is:
[0029]
[0030] in:
[0031]
[0032] Where: N′ represents the termination number of the cumulative summation of the function; T″ is the upper limit of the sampling interval; X(t) is the input function vector of the network, t represents time; C j (t) is the central function vector of the j-th radial basis process neuron in the first hidden layer, j = 1, 2, ..., J, t represents time; ω jk is the connection weight between the j-th radial basis process neuron in the first hidden layer and the k-th general neuron in the second hidden layer, k = 1, 2, ..., K; θ k is the threshold of the kth general neuron in the second hidden layer; θ is the threshold of the output layer unit; v k represents the connection weight between the second hidden layer and the output layer; y represents the output;
[0033] The Gaussian function used in the first hidden layer is represented by φ1(x), that is:
[0034]
[0035] Among them, x represents the independent variable in the Gaussian function, σ j is the width coefficient of the Gaussian function;
[0036] The Multi-Quadric function used in the second hidden layer is represented by φ2(z), that is:
[0037] φ2(z)=(z 2 +α 2 ) 1 / 2
[0038] Wherein, z represents the independent variable in the Multi-Quadric function, and α is a constant greater than 0.
[0039] Through the above design scheme, the present invention can bring the following beneficial effects: the present invention proposes a method for predicting the far-end crosstalk voltage peak value of the affected line during cable crosstalk, which saves calculation time and improves simulation efficiency; and can realize the prediction of the far-end crosstalk voltage peak value of the affected line under the condition of any change in cable-related parameters and interference source parameters of the interference line, and find out the coupling magnitude of the voltage on the affected cable. It can reflect the coupling situation of the far-end crosstalk voltage of the affected line under the actual cable crosstalk situation. BRIEF DESCRIPTION OF THE DRAWINGS
[0040] The drawings described herein are used to provide a further understanding of the present invention and constitute a part of the present application. The exemplary embodiments of the present invention and their descriptions are used to understand the present invention and do not constitute improper limitations of the present invention. In the drawings:
[0041] Figure 1 A flow chart of a method for predicting the far-end crosstalk voltage peak value of a disturbed line during cable crosstalk;
[0042] Figure 2 This is a schematic diagram of a common ground cable in an embodiment of the present invention;
[0043] Figure 3 This is a schematic diagram of cable connection in an embodiment of the present invention;
[0044] Figure 4 This is a schematic diagram of the time domain waveform of the interference line in an embodiment of the present invention;
[0045] Figure 5 This is a comparison chart of the prediction results of the training set in the embodiment of the present invention;
[0046] Figure 6 This is a comparison chart of the prediction results of the test set in an embodiment of the present invention. DETAILED DESCRIPTION
[0047] In order to make the purpose, features and advantages of the present invention more obvious and easy to understand, the technical scheme in the present invention is clearly and completely described below in conjunction with the embodiments of the present invention and the accompanying drawings. Obviously, the present invention is not limited to the following embodiments, and the specific implementation method can be determined according to the technical scheme of the present invention and the actual situation. In order to avoid confusing the essence of the present invention, known methods, processes and flows are not described in detail. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work belong to the scope of protection of the present invention.
[0048] like Figures 1 to 6 As shown, the present invention provides a method for predicting the far-end crosstalk voltage peak value of a disturbed line during cable crosstalk, the method comprising the following steps:
[0049] Step 1: Construct a field-line coupled electromagnetic crosstalk simulation model by combining the transmission line model analysis method
[0050] Step 101: In CST software, use CST Cable Studio to model a group of common ground cables.
[0051] The cable model is built based on CST Cable Studio, which has a library of cable models of various types, such as single-core cables, twisted-pair cables, coaxial cables, shielded cables, etc. A set of common ground cable models is built here, and both the interfering and disturbed cables are single-core cables. The cable parameters are shown in Table 1.
[0052] Table 1
[0053] Fill Factor Number of shares Material 0.79 51 Cu
[0054] Step 102: Constructing a field-line coupled electromagnetic crosstalk simulation model
[0055] The present invention combines the existing transmission line model analysis method to construct a field-line coupled electromagnetic crosstalk simulation model. Taking a group of common ground cables as the research object, an electromagnetic pulse signal is injected into the interfering line, which is coupled to the disturbed line through the spatial field where the common ground cable model is located. The far-end crosstalk voltage peak of the disturbed line is obtained by simulation calculation.
[0056] The common ground cable diagram is as follows Figure 2 As shown, cable 1 injects interference signals, which is called the interference line. One end of the interference line is connected to the interference source E. s , and the other end is connected to a load resistor R; Cable No. 2 receives interference signals from the interference line and is called the interfered line, with both ends connected to a load resistor R. The distance between the two cables is represented by L, the height from the ground is represented by H, and the cable length is represented by X, forming a closed loop through a common grounding point.
[0057] A field-line coupled electromagnetic crosstalk simulation model was constructed in the CST cable studio. The material of the reference ground was PEC, with the length set to 1600 mm, the width to 800 mm, and the thickness to 100 mm.
[0058] Both the interference wire and the winding wire are made of single-core wire. The cable connection diagram is as follows: Figure 3 As shown, the load resistance connected to both ends of the interference line and the affected winding is 50Ω. Port 1 in the figure is used to inject the interference source into the interference line.
[0059] Step 103: Use a unit pulse signal with set frequency and amplitude to simulate the interference source signal on the interference line through the energy accumulation of the pulse signal, such as Figure 4 As shown, taking the square wave signal as an example, the interference source on the equivalent interference line is established to establish the interference source model of the interference line;
[0060] Assume that the interference source signal on the interference line is f(t), and its expression is:
[0061]
[0062] Where i represents the i-th pulse signal, t represents time, n is the number of sampling points, N represents the total number of accumulated pulse signals, δ(t-nT) is the unit pulse signal, and its period is T. For the interference source signal f(t) with a sampling period of T′, its duty cycle D is related to the n-th duty cycle factor a n The relationship is expressed as:
[0063]
[0064] The simulation uses a square wave signal equivalent to a unit pulse signal: the initial interference source waveform injected into the interference line 1 port uses a square wave signal. The present invention only uses a square wave signal as an example of an interference source on the equivalent interference line, but the present invention is not limited to this. Common sine waves (related parameters: amplitude, period, phase), triangle waves (related parameters: peak value, period, frequency, rise time, fall time), sawtooth waves (related parameters: amplitude, scanning time, retrace time, scanning rate) can also be used as equivalent unit pulse signals.
[0065] Step 2: In the CST cable studio, multiple simulation calculations are performed by changing the cable-related parameters and interference source parameters, and the far-end crosstalk voltage peaks of the disturbed line under each state are counted to construct a data set for the prediction model and process the data set;
[0066] Step 201: Perform cable crosstalk simulation with different cable-related parameters and interference source parameters in the CST cable studio to obtain the far-end crosstalk voltage peak value of the disturbed line under different parameter states;
[0067] There are many factors that affect cable crosstalk. For example, the larger the cable spacing, the smaller the far-end crosstalk voltage of the affected line; the greater the height of the affected cable from the ground, the greater the far-end crosstalk voltage of the affected line; the greater the peak value of the interference source on the interference line, the greater the far-end crosstalk voltage of the affected line. Although these are already well-known contents in the field of cable crosstalk, the existing technology studies the impact of a single factor on cable crosstalk. The present invention focuses on analyzing the impact of different cable-related parameters and interference source parameters that affect cable crosstalk simulation on the far-end crosstalk voltage peak of the affected line when they are randomly combined. Therefore, the factors affecting cable crosstalk are listed as different input conditions and simulated in the CST cable studio.
[0068] In this embodiment, the input variables include: cable spacing L, cable height from the ground H, cable radius r, rise time T rise , peak time T hold and pulse peak value V pulse, where the cable spacing L ranges from 10mm to 50mm; the cable height H ranges from 50mm to 80mm; the cable radius r ranges from 0.05mm to 0.15mm; the rise time T rise The value range is 0.0005μs~0.0007μs; the peak time T hold The value range is 0.005μs~0.006μs; the pulse peak value V pulse The value range is 5V~7V;
[0069] The specific values of each input variable are as follows:
[0070] Cable spacing L: L = L1 + 10L0, L0 = 0, 1, 2, 3, 4, unit: mm;
[0071] Cable height from the ground H: H = H1 + 10H0, H0 = 0, 1, 2, 3, unit: mm;
[0072] Cable radius r: r = r1 + 0.05r0, r0 = 0, 1, 2, in mm;
[0073] Rise time T rise :T rise =m1+0.0002m0, m0=0,1, unit is μs;
[0074] Peak time T hold :T hold =a1+0.001a0, a0=0,1, unit is μs;
[0075] Pulse peak value V pulse :V pulse =v1+2v0, v0=0,1, unit is V;
[0076] Among them, L1, H1, r1, m1, a1 and v1 represent the input variables cable spacing L, cable height from the ground H, cable radius r, rise time T respectively. rise , peak time T hold , Pulse peak value V pulse The initial values of L1 = 10 mm, H1 = 50 mm, r1 = 0.05 mm, m1 = 0.0005 μs, a1 = 0.005 μs, v1 = 5 V, and then the input variables are respectively valued using the above expressions. The purpose is to make the values of each input variable non-unique and provide a sufficient number of data sets for the training model in step 3.
[0077] It should be noted that other factors that affect the cable crosstalk simulation, such as cable line type, cable length, pulse fall time, pulse delay, etc., are all fixed values, so they are not input variables here.
[0078] The control variable method was used to assign values to the above input variables respectively. A total of 480 simulation calculations were performed in the CST cable studio, and the peak value of the far-end crosstalk voltage of the disturbed line was saved as the output variable.
[0079] Step 202: Count the far-end crosstalk voltage peak values of the victim line under different cable-related parameters and interference source parameters, use different cable-related parameters and interference source parameters as input variables, and the far-end crosstalk voltage peak values of the victim line as output variables, and construct a data set.
[0080] The peak value of the far-end crosstalk voltage V of the disturbed line is calculated under each input variable. max , and make records. Some of the collected data sets are shown in Table 2:
[0081] Table 2
[0082] L H r <![CDATA[T rise ]]> <![CDATA[T hold ]]> <![CDATA[V pulse ]]> <![CDATA[V max ]]> 1 10 50 0.05 0.0005 0.005 5 0.249 2 20 50 0.05 0.0005 0.005 5 0.097 3 30 50 0.05 0.0005 0.005 5 0.042 4 40 50 0.05 0.0005 0.005 5 0.033 5 50 50 0.05 0.0005 0.005 5 0.022 6 10 50 0.1 0.0005 0.005 5 0.238 7 20 50 0.1 0.0005 0.005 5 0.088 8 30 50 0.1 0.0005 0.005 5 0.041 9 … … … … … … …
[0083] Step 203: Normalize the data set using the Z-score method
[0084] The data set collected in step 202 is normalized. The data normalization process uses the Z-score method, and its calculation formula is as follows:
[0085] U′=(U-mean) / std
[0086] Wherein, U is the original data of the sample data set, U' is the data after normalization, mean is the mean of the original data, and std is the standard deviation of the original data. The present invention is not limited to a specific normalization processing formula, and other calculation formulas are also applicable.
[0087] Step 3: Use the improved double hidden layer radial basis function (RBF) neural network to build a prediction model for the peak voltage of the far-end crosstalk of the disturbed line, divide the data set preprocessed in step 203, train the prediction model using the divided training set, and test the prediction model using the divided test set.
[0088] After normalizing the data set collected in step 203, the pre-processed data set is randomly divided into two parts, a training set and a test set, at a ratio of 80%:20%. The cable spacing L, cable height H, cable radius r, and rise time T are concentrated in the data set. rise , peak time T hold , Pulse peak value V pulse As the input value of the improved double hidden layer radial basis function (RBF) neural network, the far-end crosstalk voltage peak value V maxAs the prediction target of the improved double hidden layer radial basis function (RBF) neural network, a prediction model for the far-end crosstalk voltage peak of the disturbed line is constructed. The prediction model is trained with the divided training set data and then tested with the divided test set data.
[0089] In the present invention, there are only 384 training samples and only 96 test samples, which are small sample data. An improved double hidden layer radial basis function (RBF) neural network is used. The reasons why RBF neural network is suitable for small sample data are mainly the following three points: efficient approximation ability, local activation function and flexible network structure. However, in practical application, there are still problems such as low approximation accuracy and poor generalization ability. In view of the above problems, the present invention adds a hidden layer of radial basis function using Multi-Quadric function on the basis of traditional single hidden layer radial basis function (RBF) neural network, and constructs an improved double hidden layer radial basis function (RBF) neural network.
[0090] The mathematical relationship between the input and output of the constructed improved double hidden layer radial basis function (RBF) neural network is expressed as follows:
[0091]
[0092]
[0093] Where N′ represents the termination number of the cumulative summation of the function; T″ is the upper limit of the sampling interval; X(t) is the input function vector of the network, t represents time; C j (t) is the central function vector of the j-th radial basis process neuron in the first hidden layer, j = 1, 2, ..., J, t represents time; ω jk is the connection weight between the j-th radial basis process neuron in the first hidden layer and the k-th general neuron in the second hidden layer, k = 1, 2, ..., K; θ k is the threshold of the kth general neuron in the second hidden layer; θ is the threshold of the output layer unit; v k represents the connection weight between the second hidden layer and the output layer; y represents the Gaussian function used to output the first hidden layer, represented by φ1(x), that is:
[0094]
[0095] Among them, x represents the independent variable in the Gaussian function, σ j is the width coefficient of the Gaussian function;
[0096] The Multi-Quadric function used in the second hidden layer is represented by φ2(z), that is:
[0097] φ2(z)=(z 2 +α 2 ) 1 / 2
[0098] Wherein, z represents the independent variable in the Multi-Quadric function, and α is a constant greater than 0.
[0099] By adding a hidden layer to the network and using the newly added hidden layer to make a secondary approximation of the relationship between the original hidden layer and the output layer, the number of weights in the network can be reduced, even if J>K, to achieve network structure optimization and improve the approximation accuracy of the predicted value to the actual value.
[0100] A large number of simulation experiments were conducted in combination with the data set of the present invention to adjust the structural parameters of the improved double hidden layer radial basis function (RBF) neural network, determine the expansion speed of the first hidden layer radial basis function and the value of the second hidden layer Multi-Quadric function α, so as to achieve the best prediction effect. The training parameters of the network are shown in Table 3:
[0101] Table 3
[0102]
[0103] Based on the trained far-end crosstalk voltage peak prediction model of the disturbed line, the cable spacing L, cable height H, cable radius r, and rise time T in the test set are calculated. rise , peak time T hold , Pulse peak value V pulse As the input value of the far-end crosstalk voltage peak prediction model of the disturbed line, the far-end crosstalk voltage peak value V max prediction.
[0104] Step 4: Analysis of the validity and accuracy of the prediction model of the far-end crosstalk voltage peak of the disturbed line
[0105] Step 401: Validity analysis of the victim line far-end crosstalk voltage peak prediction model
[0106] Using the root mean square error RMSE and correlation coefficient R of the test set data 2 , mean absolute error MAE to analyze and evaluate the effectiveness of the prediction model; the comparison between the predicted value and the actual value of the data in the test set is used to evaluate the accuracy of the prediction model. The root mean square error RMSE and correlation coefficient R of multiple groups of data in the test set 2 , mean absolute error MAE, as shown in Table 4:
[0107] Table 4
[0108] RMSE <![CDATA[R 2 ]]> MAE 0.073 0.96104 0.0348
[0109] As can be seen from the table above, the root mean square error RMSE and mean absolute error MAE of multiple groups of data in the test set are very low, and the correlation coefficient R 2is close to 1. It can be seen that this network model is suitable for predicting the peak value of the far-end crosstalk voltage of the disturbed line.
[0110] Step 402: Accuracy analysis of the victim line far-end crosstalk voltage peak prediction model
[0111] Comparison of prediction results of training set and test set, such as Figure 5 and Figure 6 As shown. Taking 95% to 105% of the true value of the test set as the range, if the predicted value falls within this range, it means that the prediction is accurate. The ratio of the number of accurate estimates M1 in the test set to the total number of test sets M is calculated to evaluate the accuracy of the prediction model. The formula for calculating accuracy is as follows:
[0112]
[0113] Calculated by the above formula, the prediction accuracy of the model is 90.6%, which is high and has good prediction performance.
[0114] On a 6-core 16G personal commercial machine, the calculation time of a transmission line method is about 8s, while the time for a prediction using the affected line far-end crosstalk voltage peak prediction model is only 20ms, saving the time of repetitive simulation. The affected line far-end crosstalk voltage peak prediction model provided by the present invention saves calculation time and improves simulation efficiency. It can also realize the prediction of the affected line far-end crosstalk voltage peak when any cable-related parameters and interference source parameters of the interference line change. It can reflect the far-end crosstalk voltage coupling of the affected line under the actual cable crosstalk situation.
[0115] Obviously, the above embodiments of the present invention are merely examples for clearly illustrating the present invention, and are not limitations on the implementation methods of the present invention. For ordinary technicians in the relevant field, other different forms of changes or modifications can be made based on the above description. It is impossible to list all the implementation methods here. All obvious changes or modifications derived from the technical solution of the present invention are still within the protection scope of the present invention.
Claims
1. A method for predicting the peak value of far-end crosstalk voltage of a disturbed line during cable crosstalk, characterized in that: include: Step 1: Construct a field-line coupled electromagnetic crosstalk simulation model by combining the existing transmission line model analysis method; The field-line coupling electromagnetic crosstalk simulation model includes an interfering line and a disturbed line. A unit pulse signal with a set frequency and amplitude is applied to one end of the interfering line. The interference source signal on the interfering line is simulated by accumulating the energy of the pulse signal. The far-end crosstalk voltage peak of the disturbed line is obtained by simulation calculation. Assume that the interference source signal on the interference line is f(t), and its expression is: Where i represents the i-th pulse signal, t represents time, n is the number of sampling points, N represents the total number of accumulated pulse signals, δ(t-nT) is the unit pulse signal, and its period is T; for the interference source signal f(t) with a sampling period of T′, its duty cycle D is related to the n-th duty cycle factor a n The relationship is expressed as: Step 2: Using different cable-related parameters and interference source parameters as input variables and the far-end crosstalk voltage peak value of the disturbed line as the output variable, the constructed field-line coupling electromagnetic crosstalk simulation model is used for simulation, a data set is constructed, and the data set is normalized, and the normalized data set is divided into a training set and a test set; wherein the cable-related parameters include the cable spacing L, the cable height from the ground H, and the cable radius r; Step 3: Construct a prediction model for the peak value of the far-end crosstalk voltage of the disturbed line. The prediction model for the peak value of the far-end crosstalk voltage of the disturbed line adopts an improved double-hidden layer radial basis function neural network. The improved double-hidden layer radial basis function neural network is based on the traditional single-hidden layer radial basis function neural network, and adds a hidden layer whose radial basis function adopts Multi-Quadric function; the cable related parameters and the interference source parameters are used as the input variables of the prediction model for the peak value of the far-end crosstalk voltage of the disturbed line. The peak value of the far-end crosstalk voltage V max as the output variable of the affected line far-end crosstalk voltage peak prediction model; input the training set into the affected line far-end crosstalk voltage peak prediction model, train the affected line far-end crosstalk voltage peak prediction model, and obtain the trained affected line far-end crosstalk voltage peak prediction model; input the test set into the trained affected line far-end crosstalk voltage peak prediction model to test it and verify its performance. At this point, the affected line far-end crosstalk voltage peak prediction model for predicting the affected line far-end crosstalk voltage peak is obtained, and the affected line far-end crosstalk voltage peak prediction is completed during cable crosstalk.
2. The method for predicting the peak value of far-end crosstalk voltage of a victim line during cable crosstalk according to claim 1, characterized in that: In step 2, the interference source is a unit pulse signal equivalent to a sine wave, a triangle wave, a sawtooth wave or a square wave signal.
3. The method for predicting the peak value of far-end crosstalk voltage of a victim line during cable crosstalk according to claim 2, characterized in that: When the interference source on the interference line is equivalent to a square wave signal, the interference source parameter is the rise time T rise , peak time T hold and pulse peak value V pulse .
4. The method for predicting the peak value of far-end crosstalk voltage of a victim line during cable crosstalk according to claim 3, characterized in that: In step 2, the input variables include: cable spacing L, cable height from the ground H, cable radius r, rise time T rise , peak time T hold and pulse peak value V pulse , where the cable spacing L ranges from 10mm to 50mm; the cable height H ranges from 50mm to 80mm; the cable radius r ranges from 0.05mm to 0.15mm; the rise time T rise The value range is 0.0005μs~0.0007μs; the peak time T hold The value range is 0.005μs~0.006μs; the pulse peak value V pulse The value range is 5V~7V; The specific values of each input variable are as follows: Cable spacing L: L = L1 + 10L0, L0 = 0, 1, 2, 3, 4, unit: mm; Cable height from the ground H: H = H1 + 10H0, H0 = 0, 1, 2, 3, unit: mm; Cable radius r: r = r1 + 0.05r0, r0 = 0, 1, 2, in mm; Rise time T rise :T rise =m1+0.0002m0, m0=0,1, unit is μs; Peak time T hold :T hold =a1+0.001a0, a0=0,1, unit is μs; Pulse peak value V pulse :V pulse =v1+2v0, v0=0,1, unit is V; Among them, L1, H1, r1, m1, a1 and v1 represent the input variables cable spacing L, cable height from the ground H, cable radius r, rise time T respectively. rise , peak time T hold , Pulse peak value V pulse The initial values are L1=10mm, H1=50mm, r1=0.05mm, m1=0.0005μs, a1=0.005μs, v1=5V, and then the input variables are evaluated using the above expressions.
5. The method for predicting the peak value of far-end crosstalk voltage of a victim line during cable crosstalk according to claim 2, characterized in that: When the interference source on the interference line is equivalent to a sinusoidal wave signal, the interference source parameters are amplitude, period and phase.
6. The method for predicting the peak value of far-end crosstalk voltage of a victim line during cable crosstalk according to claim 2, characterized in that: When the interference source on the interference line is equivalent to a triangular wave signal, the interference source parameters are peak value, period, frequency, rise time and fall time.
7. The method for predicting the peak value of far-end crosstalk voltage of a victim line during cable crosstalk according to claim 2, characterized in that: When the sawtooth wave signal is used as the interference source on the interference line, the interference source parameters are amplitude, scanning time, retrace time and scanning rate.
8. The method for predicting the peak value of far-end crosstalk voltage of a victim line during cable crosstalk according to claim 1, characterized in that: The mathematical relationship between the input and output of the improved double hidden layer radial basis function neural network in step 3 is: Where N′ represents the termination number of the cumulative summation of the function; T″ is the upper limit of the sampling interval; X(t) is the input function vector of the network, t represents time; C j (t) is the central function vector of the j-th radial basis process neuron in the first hidden layer, j = 1, 2, ..., J, t represents time; ω jk is the connection weight between the j-th radial basis process neuron in the first hidden layer and the k-th general neuron in the second hidden layer, k = 1, 2, ..., K; θ k is the threshold of the kth general neuron in the second hidden layer; θ is the threshold of the output layer unit; v k represents the connection weight between the second hidden layer and the output layer; y represents the output; The Gaussian function used in the first hidden layer is represented by φ1(x), that is: Among them, x represents the independent variable in the Gaussian function, σ j is the width coefficient of the Gaussian function; The Multi-Quadric function used in the second hidden layer is represented by φ2(z), that is: φ2(z)=(z2+α2)12 Wherein, z represents the independent variable in the Multi-Quadric function, and α is a constant greater than 0.
Citation Information
Cited By
Crosstalk measurement method and device for coaxial cable of dilution refrigerator, dilution refrigerator and superconducting quantum computing equipment
CN119827871A
Method and device for measuring crosstalk of coaxial cable of dilution refrigerator, dilution refrigerator and superconducting quantum computing device
CN119827871B
Simulation IC synchronization test optimization method and system based on crosstalk simulation
CN120995151A
Simulation ic synchronous test optimization method and system based on crosstalk simulation
CN120995151B
