High-speed cable optimization method and system based on impedance continuous compensation

Abstract

This invention discloses a high-speed cable optimization method and system based on impedance continuity compensation, relating to the field of high-speed cable optimization. The method includes: identifying the location, amplitude, and type of each impedance abrupt change point; calculating the reflection phase difference between any two abrupt change points; identifying the locations of coherent traps where reflected signals from multiple abrupt change points coherently superimpose within the Nyquist frequency range; constructing a coherence elimination objective function; determining compensation parameters for geometric pre-compensation, dielectric constant gradient compensation, and active dynamic compensation; adjusting the reflection coefficient amplitude of the abrupt change points according to the target residual impedance abrupt change amplitude; changing the reflection phase distribution according to the optimal control parameters of dielectric constant gradient compensation; and introducing an adjustable reflection coefficient to adjust the standing wave condition according to the target terminal matching impedance value. The advantages of this invention are: effectively eliminating the coherent superposition effect of multi-level reflections in high-speed cables, achieving impedance continuity and signal integrity of high-speed cables across the entire frequency band.

Description

Technical Field

[0001] This invention relates to the field of high-speed cable optimization, specifically to a high-speed cable optimization method and system based on continuous impedance compensation. Background Technology

[0002] In the current field of high-speed data transmission (vehicle Ethernet, aerospace, industrial control, 5G support, etc.), the requirements for the transmission rate, high frequency characteristics and reliability of high-speed cables are constantly increasing. High-speed cables need to meet the requirements of low reflection and low loss transmission in the Nyquist frequency range. Their full-link architecture usually includes connectors, solder joints, cable bodies, intermediate transition points and other structures. Impedance abrupt change points are easily formed at the connection of each structure due to processing and assembly errors.

[0003] Time Domain Reflectometer (TDR) is the mainstream method for impedance testing of existing high-speed cables. It can collect impedance distribution data of the entire link and identify the amplitude of impedance change points. In the industry, it is common to try to reduce the impact of impedance change on signal transmission by passively controlling manufacturing process tolerances and single-point impedance matching compensation. Impedance matching and reflection suppression have become the core technical links in the design and manufacturing of high-speed cables.

[0004] However, in existing technologies, time-domain reflectometer testing only focuses on whether the amplitude of each impedance abrupt change point is within a preset tolerance range, without incorporating the phase relationship between multiple abrupt changes into the analysis system. This makes it impossible to identify which abrupt changes combine at which frequencies result in coherent superposition of reflected signals, leading to the inability to accurately locate and eliminate coherent traps. Furthermore, compensation methods such as geometric compensation, dielectric constant adjustment, and active dynamic compensation are typically designed in isolation, lacking joint optimization among the three, and thus failing to fundamentally destroy the phase coherence of multi-level reflections. Summary of the Invention

[0005] To address the aforementioned technical problems, this paper provides a high-speed cable optimization method and system based on continuous impedance compensation. This technical solution solves the problems mentioned in the background technology, such as time domain reflectometer testing only focusing on whether the amplitude of each impedance abrupt change point is within the preset tolerance range, without incorporating the phase relationship between multiple abrupt change points into the analysis system, and the problem that compensation methods such as geometric structure compensation, dielectric constant adjustment and active dynamic compensation are usually designed in isolation.

[0006] To achieve the above objectives, the technical solution adopted by the present invention is as follows:

[0007] A high-speed cable optimization method based on impedance continuous compensation includes:

[0008] Based on the end-to-end architecture of high-speed cables, the location, amplitude, and type of each impedance abrupt change point are identified, and the reflection phase difference between any two abrupt change points is calculated.

[0009] Based on the reflection phase difference, identify the coherent trap locations where the reflected signals from multiple abrupt change points coherently superimpose within the Nyquist frequency range;

[0010] Construct a coherence elimination objective function and determine the compensation parameters for geometric pre-compensation, dielectric constant gradient compensation, and active dynamic compensation;

[0011] Based on the target residual impedance abrupt change amplitude in the compensation parameters, perform geometric pre-compensation and adjust the reflection coefficient amplitude at the abrupt change point.

[0012] Based on the optimal control parameters of the dielectric constant gradient compensation in the compensation parameters, the dielectric constant gradient compensation is performed to change the reflection phase distribution through a controlled impedance change gradient.

[0013] Based on the target terminal matching impedance value in the compensation parameters, active dynamic compensation is performed, and an adjustable reflection coefficient is introduced at the terminal to adjust the standing wave conditions.

[0014] Preferably, the end-to-end architecture based on high-speed cables, which identifies the location, amplitude, and type of each impedance abrupt change point, and calculates the reflection phase difference between any two abrupt change points, specifically includes:

[0015] Obtain the end-to-end architecture of a high-speed cable, which includes, but is not limited to: a first connector, a first solder joint, a first cable body, an intermediate transition point, a second connector, a second solder joint, and a second cable body;

[0016] The impedance distribution data of the entire link is collected by a time domain reflectometer, the time axis is converted into a distance axis, and the impedance distribution curve of the entire link is plotted. The distance coordinate is the position coordinate along the transmission path.

[0017] Identify all abrupt changes in the impedance distribution curve and record the location coordinates, magnitude, and type of each abrupt change.

[0018] Calculate the reflection phase difference between any two abrupt change points based on their location coordinates.

[0019] The reflection coefficient of each abrupt change point is calculated based on the abrupt change amplitude and the characteristic impedance on the incident side.

[0020] Based on the reflection coefficients and location coordinates of each mutation point, a full-link reflection transfer function is constructed.

[0021] Preferably, the step of identifying the coherent trap location where the reflected signals of multiple abrupt change points coherently superimpose within the Nyquist frequency range based on the reflection phase difference specifically includes:

[0022] Based on the full-link reflection transfer function, calculate the return loss in the Nyquist frequency range and plot the return loss curve.

[0023] Identify all local peaks in the return loss curve within the Nyquist frequency range, and record the frequency and amplitude corresponding to each local peak.

[0024] For each local peak frequency point, traverse all combinations of impedance abrupt points and calculate the phase value of each impedance abrupt point at that frequency point.

[0025] Select the set of impedance abrupt change points where the absolute value of the difference between phase values ​​at the specified frequency is less than a preset angle threshold. Mark the set as a coherent abrupt change point combination, mark the frequency point as a coherent trap frequency point, and mark the amplitude of the local peak as the coherent trap depth.

[0026] Preferably, the construction of the coherence elimination objective function and the determination of the compensation parameters for geometric pre-compensation, dielectric constant gradient compensation, and active dynamic compensation specifically include:

[0027] Construct a coherence cancellation objective function, which is equal to the maximum value of the magnitude of the reflection transfer function in the frequency range from zero to the Nyquist frequency.

[0028] Set an optimization variable set, which includes: the target residual impedance change amplitude at each impedance change point after geometric pre-compensation, the control parameters for dielectric constant gradual compensation, and the target terminal matching impedance value for active dynamic compensation.

[0029] With minimizing the coherence elimination objective function as the optimization objective, the optimal combination of compensation parameters is solved using the Bayesian optimization algorithm;

[0030] The Bayesian optimization algorithm includes:

[0031] Several sets of compensation parameters are randomly sampled in the optimization variable space as the initial training set, and a Gaussian process surrogate model is fitted based on the training set.

[0032] The Gaussian process surrogate model uses the Matern kernel function as the covariance function to fit the nonlinear relationship between the compensation parameters and the coherence elimination objective function;

[0033] Based on the Gaussian process surrogate model, the coherence elimination objective function for each unsampled point is used to predict the mean and standard deviation.

[0034] The next sampling point is selected using the expected improved acquisition function, which is equal to the difference between the current optimal objective function value and the predicted mean, multiplied by the standard normal cumulative distribution function, plus the predicted standard deviation multiplied by the standard normal probability density function.

[0035] Perform experiments at the newly selected sampling points, update the training set, and repeat the steps of fitting the Gaussian process surrogate model and selecting sampling points until convergence.

[0036] The convergence criterion is that the relative change of the objective function value after multiple consecutive iterations is less than a preset value, or the number of iterations reaches a preset upper limit, thus obtaining the optimal combination of compensation parameters.

[0037] The optimal compensation parameter combination includes: the target residual impedance change amplitude at each impedance change point, the optimal control parameters for dielectric constant gradual compensation, and the target terminal matching impedance value.

[0038] Preferably, the step of performing geometric pre-compensation based on the target residual impedance abrupt change amplitude in the compensation parameters and adjusting the reflection coefficient amplitude at the abrupt change point specifically includes:

[0039] For impedance abrupt change points identified as capacitive depressions, obtain the current location coordinates and current impedance abrupt change amplitude of the capacitive depression, and calculate the difference between the current impedance abrupt change amplitude and the target residual impedance abrupt change amplitude.

[0040] The depth of laser etching is determined based on the difference. The laser etching process reduces the reference ground area around the signal conductor in this region and adjusts the amplitude of the residual impedance change at this point.

[0041] For impedance abrupt change points identified as inductive spikes, obtain the current position coordinates and current impedance abrupt change amplitude of the inductive spike, and calculate the difference between the current impedance abrupt change amplitude and the target residual impedance abrupt change amplitude.

[0042] The dielectric constant variation gradient of the gradient impedance matching structure is determined based on the difference. An insulating support with a dielectric constant variation gradient is installed inside the connector housing to adjust the residual impedance change amplitude at that point.

[0043] For a composite impedance abrupt change point that simultaneously exhibits capacitive dips and inductive spikes, an inductive compensation structure is set in the pre-region of the composite impedance abrupt change point, and a capacitive compensation structure is set in the post-region to adjust the equivalent residual impedance abrupt change amplitude of the composite impedance abrupt change point.

[0044] Preferably, the step of performing gradual dielectric constant compensation based on the optimal control parameters of the gradual dielectric constant compensation in the compensation parameters, and changing the reflection phase distribution through a controlled impedance change gradient, specifically includes:

[0045] Obtain the optimal control parameters for dielectric constant gradient compensation in the compensation parameters. The optimal control parameters include: the starting point position of dielectric constant gradient, the ending point position of dielectric constant gradient, the gradient of dielectric constant change, and the target dielectric constant value of the nominal dielectric constant segment.

[0046] Starting from the point where the dielectric constant gradually changes and ending at the point where the dielectric constant gradually changes, a linear transition curve is formed from the low dielectric constant starting segment to the nominal dielectric constant segment, following the linear change of the dielectric constant gradient.

[0047] Based on the transition curve of dielectric constant distribution along the axial direction, the change in target dielectric constant at each position of the insulating layer is calculated by subtracting the original dielectric constant of the insulating layer material from the target dielectric constant value on the curve.

[0048] The change in the target dielectric constant is converted into the equivalent heat input of laser irradiation. The equivalent heat input is equal to the laser power density multiplied by the scanning time, divided by the spot diameter, and then multiplied by the material matching ratio coefficient.

[0049] The material matching ratio is determined through a pre-calibration test and is used to correct the differences in the laser energy absorption efficiency and dielectric constant modification sensitivity of different insulating layer materials.

[0050] The reflective phase distribution is altered by continuously adjusting at least one of the three parameters: laser power density, scanning time, and focused spot diameter.

[0051] Preferably, the step of performing active dynamic compensation based on the target terminal matching impedance value in the compensation parameters, and introducing an adjustable reflection coefficient at the terminal to adjust the standing wave condition specifically includes:

[0052] An active dynamic compensation module is integrated at the receiving end of the cable. The active dynamic compensation module includes an adjustable impedance matching network, a signal quality monitoring unit, and a microcontroller.

[0053] The adjustable impedance matching network includes a digital potentiometer and an adjustable capacitor array, with an adjustment range covering a preset impedance adjustment interval.

[0054] During the link training phase, the microcontroller reads the target terminal matching impedance value, generates an initial impedance adjustment command, and controls the adjustable impedance matching network to configure the terminal matching impedance to the target terminal matching impedance value.

[0055] The signal quality monitoring unit continuously collects signal quality parameters from the receiver during the link training phase. These signal quality parameters include bit error rate, eye diagram opening, jitter amplitude, and signal-to-noise ratio.

[0056] The collected signal quality parameters are compared with preset qualified thresholds to generate a signal quality score;

[0057] The microcontroller executes an adaptive gradient descent algorithm, which includes:

[0058] Starting from the current terminal matching impedance value, adjust the terminal matching impedance value in both positive and negative directions with a preset step size, and record the change in signal quality score in the two directions.

[0059] Select the direction with the greatest improvement in signal quality score as the adjustment direction, and continue to adjust in that direction at the same pace until the signal quality score reaches its peak or the number of adjustments reaches the preset limit. Lock the terminal matching impedance value when the signal quality score reaches its peak value as the optimal working impedance value.

[0060] During cable operation, the signal quality monitoring unit continuously monitors signal quality parameters at preset time intervals. When the current signal quality score is detected to be lower than the preset fluctuation threshold, the microcontroller restarts the gradient descent adaptive algorithm to search for the optimal working impedance value again.

[0061] Furthermore, this solution proposes a high-speed cable optimization system based on continuous impedance compensation to implement the high-speed cable optimization method based on continuous impedance compensation as described above, including:

[0062] The transfer function module is used to identify the location, magnitude, and type of impedance abrupt change points in the end-to-end architecture based on high-speed cables, and to calculate the reflection phase difference between any two abrupt change points.

[0063] A coherent abrupt change point identification module is used to identify the location of a coherent trap where the reflected signals of multiple abrupt change points coherently superimpose within the Nyquist frequency range, based on the reflected phase difference.

[0064] The compensation parameter module is used to construct the coherence elimination objective function and determine the compensation parameters for geometric pre-compensation, dielectric constant gradient compensation, and active dynamic compensation.

[0065] The compensation optimization module is used to perform geometric pre-compensation based on the target residual impedance change amplitude in the compensation parameters, and adjust the reflection coefficient amplitude at the change point; perform dielectric constant gradual compensation based on the optimal control parameters of dielectric constant gradual compensation in the compensation parameters, and change the reflection phase distribution through a controlled impedance change gradient; and perform active dynamic compensation based on the target terminal matching impedance value in the compensation parameters, and introduce an adjustable reflection coefficient at the terminal to adjust the standing wave conditions.

[0066] Preferably, the compensation optimization module includes:

[0067] A coefficient amplitude optimization unit is used to perform geometric pre-compensation based on the target residual impedance abrupt change amplitude in the compensation parameters, and adjust the reflection coefficient amplitude at the abrupt change point.

[0068] A phase distribution optimization unit is used to perform dielectric constant gradual compensation according to the optimal control parameters of the dielectric constant gradual compensation in the compensation parameters, and to change the reflection phase distribution through a controlled impedance change gradient.

[0069] The standing wave condition optimization unit is used to perform active dynamic compensation based on the target terminal matching impedance value in the compensation parameters, and to adjust the standing wave condition by introducing an adjustable reflection coefficient at the terminal.

[0070] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0071] This invention proposes a high-speed cable optimization method and system based on continuous impedance compensation. It collects end-to-end impedance distribution data using a time-domain reflectometer, identifies the location, amplitude, and type of each impedance abrupt change point, calculates the reflection phase difference between any two abrupt change points, and constructs an end-to-end reflection transfer function. This extends impedance analysis from traditional amplitude detection to phase distribution analysis, thereby identifying the location of coherent traps where reflected signals from multiple abrupt change points coherently superimpose within the Nyquist frequency range. This solves the technical problem of existing technologies being unable to perceive the phase relationship between multiple levels of reflection. By constructing a coherence elimination objective function, a Bayesian optimization algorithm is used to jointly optimize the compensation parameters of geometric pre-compensation, dielectric constant gradient compensation, and active dynamic compensation, overcoming the technical defects of isolated design and lack of coordination in existing compensation methods. Based on the optimized compensation parameters, geometric pre-compensation adjusts the reflection coefficient amplitude at abrupt change points, dielectric constant gradient compensation changes the reflection phase distribution, and active dynamic compensation introduces an adjustable reflection coefficient at the terminal to adjust the standing wave condition. This collaboratively eliminates the coherent superposition of multiple levels of reflection from three dimensions: amplitude, phase, and terminal matching. This method can effectively disrupt the phase-locked relationship between multiple reflections, eliminate coherence traps, and achieve impedance continuity and signal integrity of high-speed cables in the Nyquist frequency range. This effectively improves the transmission reliability and anti-interference capability of high-speed cables in high-frequency and high-speed transmission scenarios such as automotive Ethernet, aerospace, industrial control, and 5G support. Attached Figure Description

[0072] Figure 1 This is a flowchart of a high-speed cable optimization method based on continuous impedance compensation according to the present invention.

[0073] Figure 2 This is a flowchart illustrating the calculation of the reflection phase difference between any two abrupt change points according to the present invention.

[0074] Figure 3 This is a flowchart illustrating the coherent trap location method for identifying the coherent superposition of reflected signals at multiple abrupt change points within the Nyquist frequency range, as described in this invention.

[0075] Figure 4The flowchart below shows the construction of the coherence elimination objective function and the determination of compensation parameters for geometric pre-compensation, dielectric constant gradient compensation, and active dynamic compensation in this invention. Detailed Implementation

[0076] The following description is intended to disclose the invention and enable those skilled in the art to implement it. The preferred embodiments described below are merely examples, and other obvious variations will occur to those skilled in the art.

[0077] Reference Figure 1 As shown, a high-speed cable optimization method based on continuous impedance compensation includes:

[0078] Based on the end-to-end architecture of high-speed cables, the location, amplitude, and type of each impedance abrupt change point are identified, and the reflection phase difference between any two abrupt change points is calculated.

[0079] Based on the reflection phase difference, identify the coherent trap locations where the reflected signals from multiple abrupt change points coherently superimpose within the Nyquist frequency range;

[0080] Construct a coherence elimination objective function and determine the compensation parameters for geometric pre-compensation, dielectric constant gradient compensation, and active dynamic compensation;

[0081] Based on the target residual impedance abrupt change amplitude in the compensation parameters, perform geometric pre-compensation and adjust the reflection coefficient amplitude at the abrupt change point.

[0082] Based on the optimal control parameters of the dielectric constant gradient compensation in the compensation parameters, the dielectric constant gradient compensation is performed to change the reflection phase distribution through a controlled impedance change gradient.

[0083] Based on the target terminal matching impedance value in the compensation parameters, active dynamic compensation is performed, and an adjustable reflection coefficient is introduced at the terminal to adjust the standing wave conditions.

[0084] This solution, based on the end-to-end architecture of high-speed cables, identifies the location, amplitude, and type of impedance abrupt changes. It calculates the reflection phase difference between any two abrupt changes and constructs an end-to-end reflection transfer function to quantify the reflection contribution of each abrupt change into a complex form. The reflection coefficient determines the amplitude of the reflected signal, and the phase factor determines the phase. This allows the reflection contributions of each abrupt change to be vector-superimposed, extending impedance analysis from traditional amplitude detection to phase distribution analysis. When the phase factors of multiple abrupt changes are similar at a certain frequency, their reflection contributions reinforce each other, forming a coherent superposition; conversely, they cancel each other out. This provides a precise quantitative basis for subsequent identification of coherent traps. Furthermore, by identifying local peaks in the return loss curve within the Nyquist frequency range, a set of impedance abrupt changes where the absolute value of the phase difference at that frequency is less than a preset angle threshold is selected. By marking coherent abrupt change points as a combination, the location, frequency, and depth of coherent traps can be accurately located, solving the technical problem that existing technologies cannot perceive the phase relationship between multi-level reflections. Furthermore, this scheme constructs a coherence elimination objective function with the maximum value of the reflection transfer function magnitude in the frequency domain as the optimization objective. The target residual impedance change amplitude of each abrupt change point in the geometric pre-compensation, the control parameters of the dielectric constant gradual compensation, and the target terminal matching impedance value of the active dynamic compensation are used as the set of optimization variables. The optimal compensation parameter combination is solved using a Bayesian optimization algorithm. Bayesian optimization fits the nonlinear relationship between the compensation parameters and the objective function through a Gaussian process surrogate model, and combines the expected improvement acquisition function to efficiently search for the optimal solution in the global range. This overcomes the shortcomings of traditional traversal search being costly and gradient optimization being prone to getting trapped in local optima. At the same time, it realizes the joint optimization of the three types of compensation methods, solving the technical problem of isolated design and lack of coordination of existing compensation methods. At the compensation execution level, this solution, based on the optimized target residual impedance change amplitude, uses laser etching to reduce the reference ground area around the signal conductor for capacitive dips, and sets insulating supports with dielectric constant variation gradients inside the connector housing for inductive spikes. For composite change points, a composite compensation structure with front inductive and rear capacitive elements is used to achieve precise control of the reflection coefficient amplitude. Based on the optimal control parameters for dielectric constant gradient compensation, the dielectric constant of the insulating layer material is controlled to continuously and gradually change along the axial direction through laser-assisted modification, forming a controlled impedance change gradient to change the reflection phase distribution. Based on the target terminal matching impedance value, an active dynamic compensation module including an adjustable impedance matching network, a signal quality monitoring unit, and a microcontroller is integrated at the cable receiver. During the link training phase, the optimal working impedance value is searched through a gradient descent adaptive algorithm, and the signal quality is continuously monitored during operation to achieve dynamic adaptive adjustment of the terminal matching impedance.By coordinating compensation in three dimensions—amplitude, phase, and terminal matching—this solution can effectively disrupt the phase-locked relationship between multiple reflections, eliminate coherence traps, and achieve impedance continuity and signal integrity of high-speed cables in the Nyquist frequency range. This effectively improves the transmission reliability and anti-interference capability of high-speed cables in high-frequency and high-speed transmission scenarios such as automotive Ethernet, aerospace, industrial control, and 5G support.

[0085] Reference Figure 2 As shown, calculating the reflection phase difference between any two abrupt change points specifically includes:

[0086] Obtain the end-to-end architecture of a high-speed cable, which includes, but is not limited to: a first connector, a first solder joint, a first cable body, an intermediate transition point, a second connector, a second solder joint, and a second cable body;

[0087] The impedance distribution data of the entire link is collected by a time domain reflectometer, the time axis is converted into a distance axis, and the impedance distribution curve of the entire link is plotted. The distance coordinate is the position coordinate along the transmission path.

[0088] Identify all abrupt changes in the impedance distribution curve and record the location coordinates, magnitude, and type of each abrupt change.

[0089] Calculate the reflection phase difference between any two abrupt change points based on their location coordinates.

[0090] The reflection coefficient of each abrupt change point is calculated based on the abrupt change amplitude and the characteristic impedance on the incident side.

[0091] Based on the reflection coefficients and location coordinates of each mutation point, a full-link reflection transfer function is constructed.

[0092] This solution expands upon the traditional time-domain reflectometry (TD-RS) testing, which focuses solely on impedance amplitude information, to include impedance phase distribution information. Traditional TD-RS testing primarily determines whether the amplitude of each impedance abrupt change is within tolerance, neglecting the phase relationship between multiple abrupt changes. Therefore, this solution constructs a full-link reflection transfer function to quantify the reflection contribution of each impedance abrupt change in the cable's entire link into a complex form. The reflection coefficient determines the amplitude of the reflected signal, and the phase factor determines the phase. Specifically, the reflection transfer function allows for vector superposition of the reflection contributions from each abrupt change. When the phase factors of multiple abrupt changes are similar at a certain frequency, their reflection contributions reinforce each other, forming a coherent superposition. Conversely, when the phase factors are uniformly distributed, the reflection contributions cancel each other out. This provides a precise quantitative basis for identifying which combinations of abrupt changes will coherently superimpose at which frequencies, solving the problem that existing technologies cannot perceive the phase relationship between multiple levels of reflection.

[0093] The abrupt change point refers to the location in the impedance distribution curve where the impedance value deviates from the characteristic impedance on the incident side and the rate of change of impedance between adjacent sampling points exceeds a preset threshold. In other words, it is a discontinuous region in the discrete sampling sequence where the impedance value shows a significant upward peak or a downward dip. The preset threshold is determined by statistical analysis of the TDR impedance data of multiple grid cables, taking the mean of the impedance change rate of the sampling points plus a certain number of standard deviations.

[0094] The amplitude of the mutation is equal to the difference between the characteristic impedance on the side after the mutation point and the characteristic impedance on the incident side.

[0095] The mutation types include: capacitive dips where the impedance value is lower than the characteristic impedance of the incident side and inductive spikes where the impedance value is higher than the characteristic impedance of the incident side.

[0096] The reflected phase difference = -4 × π × signal frequency × distance between two points ÷ signal propagation speed in the cable;

[0097] The reflection coefficient = abrupt change amplitude ÷ (2 × incident side characteristic impedance + abrupt change amplitude).

[0098] The reflection transfer function is the sum of the products of the reflection coefficients at each abrupt change point and the corresponding phase factors, where the phase factor at each abrupt change point is an exponential function with the natural constant as the base and (-4 × imaginary unit × π × signal frequency × coordinates of the abrupt change point ÷ signal propagation speed) as the exponent.

[0099] Reference Figure 3 As shown, the locations of coherent traps where the reflected signals from multiple identified abrupt change points coherently superimpose within the Nyquist frequency range specifically include:

[0100] Based on the full-link reflection transfer function, calculate the return loss in the Nyquist frequency range and plot the return loss curve.

[0101] Identify all local peaks in the return loss curve within the Nyquist frequency range, and record the frequency and amplitude corresponding to each local peak.

[0102] For each local peak frequency point, traverse all combinations of impedance abrupt points and calculate the phase value of each impedance abrupt point at that frequency point.

[0103] Select the set of impedance abrupt change points where the absolute value of the difference between phase values ​​at the specified frequency is less than a preset angle threshold. Mark the set as a coherent abrupt change point combination, mark the frequency point as a coherent trap frequency point, and mark the amplitude of the local peak as the coherent trap depth.

[0104] This solution extracts specific coherent trap information from the full-link reflection transfer function, transforming abstract mathematical functions into actionable engineering positioning data. Return loss is a core indicator for measuring cable reflection characteristics; a lower value (larger absolute value) indicates weaker reflection and better signal transmission quality. When the return loss curve shows a sharp peak at a certain frequency (i.e., a sudden increase in value and a sudden decrease in absolute value), it indicates significant coherent reflection superposition at that frequency. The reflected signals from multiple abrupt change points superimpose in phase at that frequency, forming a concentrated resonant peak. Specifically, when the phase difference between multiple abrupt change points at that frequency is less than a preset angle threshold, they are determined to be in approximately in-phase, and their reflected signals will coherently superimpose. This solution effectively identifies which abrupt change points are involved in coherent superposition, at which frequency the superposition occurs, and the severity of the superposition. This identification provides a clear target for subsequent compensation operations, namely, what needs to be disrupted is the phase coherence between specific combinations of abrupt change points, rather than blindly compensating for all abrupt change points, thus avoiding waste of compensation resources and new problems that may be introduced by overcompensation.

[0105] The calculation of return loss within the Nyquist frequency range refers to taking the logarithm of the magnitude of the full-link reflection transfer function at each frequency point to the base 10, multiplying it by -20, and obtaining the return loss value corresponding to that frequency point.

[0106] The method of identifying all local peaks in the Nyquist frequency range of the return loss curve refers to traversing the discrete sampling sequence of return loss, comparing the amplitude of each sampling point with the amplitude of its adjacent sampling points, and determining that the point is a local peak when the amplitude of the point is greater than both the previous and next sampling points.

[0107] The preset angle threshold is determined based on the signal propagation speed in the cable, the Nyquist frequency, and the typical spacing between adjacent impedance abrupt change points in the cable. It is preferably calculated by multiplying four times pi by the typical spacing by the Nyquist frequency by the propagation speed and then multiplying by a scaling factor of 0.5 to 0.8, with a value range of 20 degrees to 40 degrees.

[0108] Reference Figure 4 As shown, the specific steps for constructing the coherence elimination objective function and determining the compensation parameters for geometric pre-compensation, dielectric constant gradient compensation, and active dynamic compensation include:

[0109] Construct a coherence cancellation objective function, which is equal to the maximum value of the magnitude of the reflection transfer function in the frequency range from zero to the Nyquist frequency.

[0110] Set an optimization variable set, which includes: the target residual impedance change amplitude at each impedance change point after geometric pre-compensation, the control parameters for dielectric constant gradual compensation, and the target terminal matching impedance value for active dynamic compensation.

[0111] With minimizing the coherence elimination objective function as the optimization objective, the optimal combination of compensation parameters is solved using the Bayesian optimization algorithm;

[0112] The Bayesian optimization algorithm includes:

[0113] Several sets of compensation parameters are randomly sampled in the optimization variable space as the initial training set, and a Gaussian process surrogate model is fitted based on the training set.

[0114] The Gaussian process surrogate model uses the Matern kernel function as the covariance function to fit the nonlinear relationship between the compensation parameters and the coherence elimination objective function;

[0115] Based on the Gaussian process surrogate model, the coherence elimination objective function for each unsampled point is used to predict the mean and standard deviation.

[0116] The next sampling point is selected using the expected improved acquisition function, which is equal to the difference between the current optimal objective function value and the predicted mean, multiplied by the standard normal cumulative distribution function, plus the predicted standard deviation multiplied by the standard normal probability density function.

[0117] Perform experiments at the newly selected sampling points, update the training set, and repeat the steps of fitting the Gaussian process surrogate model and selecting sampling points until convergence.

[0118] The convergence criterion is that the relative change of the objective function value after multiple consecutive iterations is less than a preset value, or the number of iterations reaches a preset upper limit, thus obtaining the optimal combination of compensation parameters.

[0119] The optimal compensation parameter combination includes: the target residual impedance change amplitude at each impedance change point, the optimal control parameters for dielectric constant gradual compensation, and the target terminal matching impedance value.

[0120] This scheme integrates three compensation methods—geometric pre-compensation, dielectric constant gradient compensation, and active dynamic compensation—into a unified optimization framework. It uses mathematical optimization methods to find the globally optimal combination of compensation parameters and constructs a coherence elimination objective function to achieve joint optimization. Specifically, it employs a Bayesian optimization algorithm, based on measurements using a vector network analyzer, effectively reducing the computational cost of the objective function and the dimensionality of the optimization variables. By constructing a Gaussian process surrogate model, it can find the globally optimal solution within a limited number of iterations, avoiding the high cost of exhaustive search. Secondly, there is a strong coupling relationship among the three dimensions of optimization variables: the dielectric constant distribution affects the phase difference between abrupt change points, thus affecting the target residual amplitude of geometric compensation; the terminal impedance adjusts the standing wave distribution of the entire link, thus affecting the optimization direction of the dielectric constant distribution; and the residual amplitude after geometric compensation affects the optimization range of the terminal impedance. This coupling relationship dictates that the three must be jointly optimized, rather than designed independently.

[0121] The implementation logic of the desired improved acquisition function is as follows: when the predicted mean of the unsampled points is better than the current optimal value, the first term is positive, and the point is selected first to utilize the better solution; when the predicted mean of the unsampled points is worse than the current optimal value but the prediction standard deviation is large, the second term is positive, and the point is selected in a balanced manner to explore the unknown optimization space.

[0122] The independent variables of the standard normal cumulative distribution function and the standard normal probability density function are both the ratio of the difference between the current optimal objective function value and the predicted mean to the predicted standard deviation.

[0123] The set of optimized variables specifically includes: determining the variable dimensions of the geometric pre-compensation parameters based on the location and number of impedance abrupt change points in the full-link architecture; each abrupt change point corresponds to a residual impedance abrupt change amplitude variable, with the initial value of the variable set to zero, indicating an uncompensated state; the variable's value range is determined based on the original amplitude of the abrupt change point and the minimum residual amplitude achievable by the process; discretizing the cable axial length into several continuous intervals according to a preset step size; the dielectric constant within each interval is used as an independent optimized variable, with the initial value set to the dielectric constant of the substrate; the variable's value range is determined based on material properties and process feasibility; the terminal matching impedance value of the active dynamic compensation is used as an independent optimized variable, with the initial value set to the incident side characteristic impedance; the variable's value range is determined based on the adjustment range of the adjustable impedance device; and the above-mentioned geometric pre-compensation parameters, dielectric constant gradient compensation parameters, and active dynamic compensation parameters together constitute the set of optimized variables, which serves as the input to the Bayesian optimization algorithm.

[0124] The step of performing geometric pre-compensation based on the target residual impedance abrupt change amplitude in the compensation parameters and adjusting the reflection coefficient amplitude at the abrupt change point specifically includes:

[0125] For impedance abrupt change points identified as capacitive depressions, obtain the current location coordinates and current impedance abrupt change amplitude of the capacitive depression, and calculate the difference between the current impedance abrupt change amplitude and the target residual impedance abrupt change amplitude.

[0126] The depth of laser etching is determined based on the difference. The laser etching process reduces the reference ground area around the signal conductor in this region and adjusts the amplitude of the residual impedance change at this point.

[0127] For impedance abrupt change points identified as inductive spikes, obtain the current position coordinates and current impedance abrupt change amplitude of the inductive spike, and calculate the difference between the current impedance abrupt change amplitude and the target residual impedance abrupt change amplitude.

[0128] The dielectric constant variation gradient of the gradient impedance matching structure is determined based on the difference. An insulating support with a dielectric constant variation gradient is installed inside the connector housing to adjust the residual impedance change amplitude at that point.

[0129] For a composite impedance abrupt change point that simultaneously exhibits capacitive dips and inductive spikes, an inductive compensation structure is set in the pre-region of the composite impedance abrupt change point, and a capacitive compensation structure is set in the post-region to adjust the equivalent residual impedance abrupt change amplitude of the composite impedance abrupt change point.

[0130] This solution, through precise adjustments to the physical structure, implements the target residual impedance mutation amplitude obtained from joint optimization into the actual physical structure, achieving active pre-compensation at the physical structure level and overcoming the limitations of existing technologies that only passively control manufacturing tolerances. Specifically, for capacitive depressions, the physical essence is that excessive parasitic capacitance leads to impedance reduction. In solder joints and connector pad areas, excessively large pad areas and excessively close ground plane distances introduce additional parasitic capacitance. Laser etching technology, by precisely removing excess conductor area, can effectively reduce parasitic capacitance and restore impedance. There is a definite mapping relationship between the laser etching depth and the target residual impedance mutation amplitude: the lower the target residual impedance mutation amplitude, i.e., the more thorough the compensation requirement, the larger the reference ground area that needs to be etched away. This mapping relationship is established in the following way: under the same... On test specimens of materials and structural parameters, laser etching experiments at different depths were conducted to measure the impedance change at the capacitive recesses before and after etching, and to construct linear or nonlinear regression models of etching depth and impedance increase. For inductive spikes, the physical essence is that excessive parasitic inductance leads to an increase in impedance. At the junction of connector pins and cable conductors, factors such as pin length and crimp gap introduce parasitic inductance. The gradient impedance matching structure, through an insulating support with a dielectric constant that gradually changes along the axial direction, allows electromagnetic waves to experience continuously changing wave impedance, rather than a single-point impedance step. There is a definite mapping relationship between the dielectric constant change gradient and the target residual impedance change amplitude: the lower the target residual impedance change amplitude, the larger the dielectric constant change gradient. This mapping relationship is established in the following way: a three-dimensional electromagnetic simulation model including the connector housing and the gradient insulating support is built, different dielectric constant change gradients are parametrically scanned, the residual impedance change amplitude corresponding to each gradient is simulated and calculated, and a gradient-impedance mapping curve is constructed; physical samples are made by selecting typical gradient values, and the simulation results are verified by time-domain reflectometry. The mapping curve is corrected according to the measured data to obtain the final calibrated mapping relationship; for composite impedance change points, the impedance complementarity of the inductive compensation and capacitive compensation is adapted to the hardware compensation measures for the two types of changes respectively. Among them, capacitive compensation corresponds to the laser etching process, and inductive compensation corresponds to the dielectric constant gradient insulating support in the connector housing. By deploying the inductive compensation structure in the front and the capacitive compensation structure in the back, the impedance abnormality caused by excessive inductance and excessive capacitance is simultaneously offset, and the overall impedance transition is achieved smoothly. Based on this quantitative mapping relationship, the joint optimization results can be accurately executed to ensure the controllability and repeatability of the compensation effect.

[0131] The step of performing gradual dielectric constant compensation based on the optimal control parameters of the gradual dielectric constant compensation in the compensation parameters, and changing the reflection phase distribution through a controlled impedance change gradient, specifically includes:

[0132] Obtain the optimal control parameters for dielectric constant gradient compensation in the compensation parameters. The optimal control parameters include: the starting point position of dielectric constant gradient, the ending point position of dielectric constant gradient, the gradient of dielectric constant change, and the target dielectric constant value of the nominal dielectric constant segment.

[0133] Starting from the point where the dielectric constant gradually changes and ending at the point where the dielectric constant gradually changes, a linear transition curve is formed from the low dielectric constant starting segment to the nominal dielectric constant segment, following the linear change of the dielectric constant gradient.

[0134] Based on the transition curve of dielectric constant distribution along the axial direction, the change in target dielectric constant at each position of the insulating layer is calculated by subtracting the original dielectric constant of the insulating layer material from the target dielectric constant value on the curve.

[0135] The change in the target dielectric constant is converted into the equivalent heat input of laser irradiation. The equivalent heat input is equal to the laser power density multiplied by the scanning time, divided by the spot diameter, and then multiplied by the material matching ratio coefficient.

[0136] The material matching ratio is determined through a pre-calibration test and is used to correct the differences in the laser energy absorption efficiency and dielectric constant modification sensitivity of different insulating layer materials.

[0137] By continuously adjusting at least one of the three parameters—laser power density, scanning time, and focused spot diameter—the gradient of dielectric constant change can be controlled to alter the reflection phase distribution.

[0138] The physical principle behind laser irradiation-assisted dielectric constant modification can be explained as follows: laser irradiation generates a thermal effect on the insulating material. When the material absorbs laser energy and its temperature rises above the glass transition temperature, the molecular chains rearrange and cross-link, reducing the free volume of the material and lowering the equivalent dielectric constant. When the energy density is higher, local micro-ablation or micropore formation occurs in the material, further reducing the dielectric constant. The equivalent heat input is the core parameter that determines the modification effect. It is directly proportional to the laser power density, directly proportional to the scanning time, and inversely proportional to the spot diameter. Specifically: laser power density determines the energy injection rate per unit area; the higher the power density, the faster the material heats up and the deeper the modification. Scanning time determines the time the laser acts on a unit area; the longer the scanning time, the more energy accumulates and the deeper the modification. The focused spot diameter determines the lateral resolution of the modified area; a small spot can achieve fine gradient control, but has a higher energy density per unit area. By mapping the target dielectric constant change to the required equivalent heat input, the required equivalent heat input is calculated based on the required dielectric constant change at each position on the transition curve of the dielectric constant distribution along the axial direction. This heat input is then achieved by adjusting the combination of laser power density, scanning time, and spot diameter. During implementation, the laser processing head moves along the cable axis. The control system reads the target dielectric constant change at the current position on the transition curve of the dielectric constant distribution along the axial direction in real time, and obtains the corresponding equivalent heat input by looking up the table. This dynamically adjusts the laser parameters to make the actual dielectric constant change approach the target value. The modified dielectric constant distribution is measured in real time by the online monitoring unit and compared with the transition curve of the dielectric constant distribution along the axial direction to form a closed-loop control to ensure modification accuracy. Among them, laser power density refers to the power per unit area after laser focusing at a fixed reference spot diameter, which is the main adjustment variable for dielectric constant modification. Scanning time refers to the scanning time per unit length of the laser processing head along the cable axis, which is the auxiliary adjustment variable for dielectric constant modification. The spot diameter is a fixed reference process parameter preset in the laser modification process, determined by the characteristics of the cable insulation material and the axial processing accuracy. It remains constant in the compensation process of the same batch and type of high-speed cables and is only used as a normalized reference coefficient for the equivalent heat input, which is not adjusted with the change in the target dielectric constant.

[0139] The step of performing active dynamic compensation based on the target terminal matching impedance value in the compensation parameters, and adjusting the standing wave conditions by introducing an adjustable reflection coefficient at the terminal, specifically includes:

[0140] An active dynamic compensation module is integrated at the receiving end of the cable. The active dynamic compensation module includes an adjustable impedance matching network, a signal quality monitoring unit, and a microcontroller.

[0141] The adjustable impedance matching network includes a digital potentiometer and an adjustable capacitor array, with an adjustment range covering a preset impedance adjustment interval.

[0142] During the link training phase, the microcontroller reads the target terminal matching impedance value, generates an initial impedance adjustment command, and controls the adjustable impedance matching network to configure the terminal matching impedance to the target terminal matching impedance value.

[0143] The signal quality monitoring unit continuously collects signal quality parameters from the receiver during the link training phase. These signal quality parameters include bit error rate, eye diagram opening, jitter amplitude, and signal-to-noise ratio.

[0144] The collected signal quality parameters are compared with preset qualified thresholds to generate a signal quality score;

[0145] The microcontroller executes an adaptive gradient descent algorithm, which includes:

[0146] Starting from the current terminal matching impedance value, adjust the terminal matching impedance value in both positive and negative directions with a preset step size, and record the change in signal quality score in the two directions.

[0147] Select the direction with the greatest improvement in signal quality score as the adjustment direction, and continue to adjust in that direction at the same pace until the signal quality score reaches its peak or the number of adjustments reaches the preset limit. Lock the terminal matching impedance value when the signal quality score reaches its peak value as the optimal working impedance value.

[0148] During cable operation, the signal quality monitoring unit continuously monitors signal quality parameters at preset time intervals. When the current signal quality score is detected to be lower than the preset fluctuation threshold, the microcontroller restarts the gradient descent adaptive algorithm to search for the optimal working impedance value again.

[0149] This solution aims to endow cables with dynamic adaptive capabilities, enabling them to automatically adjust impedance matching parameters based on the actual link status during operation. Specifically, the adjustable impedance matching network is the core hardware for dynamic impedance adjustment. Digital potentiometers achieve discrete resistance adjustment by changing the tap positions of the resistor network, and the adjustable capacitor array achieves discrete capacitance adjustment by controlling the number of capacitors connected via switches. The combination of these two components can fully cover the preset impedance adjustment range. In the initial configuration during the link training phase, the target terminal matching impedance value in the compensation parameters is directly compensated, providing a high-quality starting point for subsequent adaptive adjustment and significantly shortening the optimization time. The gradient descent adaptive algorithm effectively solves the technical challenge of searching for the optimal impedance value in unknown environments. By observing the trend of signal quality changes through trial adjustments, it gradually approaches the optimal matching value along the gradient upward direction. This algorithm does not require pre-constructing a specific mathematical model of the link; it can complete the online accurate search solely based on the feedback of measured signal quality. Thus, it adapts to fluctuations in operating conditions such as temperature changes, cable bending, and device aging, effectively solving the long-term reliability problem of traditional fixed impedance matching cables. Together with geometric pre-compensation and dielectric constant gradual change compensation, it forms a complete compensation system combining dynamic and static elements, comprehensively eliminating impedance coherent reflections.

[0150] The step of comparing the collected signal quality parameters with a preset pass threshold to generate a signal quality score specifically includes:

[0151] The acquired signal quality parameters are compared with the preset lower limit of the qualified threshold. If any parameter is lower than the lower limit of the qualified threshold, it is directly marked as unqualified signal quality.

[0152] If all parameters are not lower than the lower limit of the acceptable threshold, the signal quality score is calculated as follows:

[0153] For parameters where a larger value indicates better quality, the single parameter score is equal to the difference between the measured value of the parameter and the lower limit of the qualified threshold, divided by the difference between the ideal value and the lower limit of the qualified threshold, and then multiplied by 100.

[0154] For parameters where smaller values ​​indicate better quality, the single parameter score is equal to the difference between the upper limit of the qualified threshold and the measured value, divided by the difference between the upper limit of the qualified threshold and the ideal value, and then multiplied by 100.

[0155] The scores of each individual parameter are calculated by normalized weighted summation, and then a signal quality score of [0, 100] is generated based on the engineering influence weight of each parameter. The higher the score, the better the signal quality.

[0156] Furthermore, based on the same inventive concept as the aforementioned high-speed cable optimization method based on continuous impedance compensation, this solution proposes a high-speed cable optimization system based on continuous impedance compensation, comprising:

[0157] The transfer function module is used to identify the location, magnitude, and type of impedance abrupt change points in the end-to-end architecture based on high-speed cables, and to calculate the reflection phase difference between any two abrupt change points.

[0158] A coherent abrupt change point identification module is used to identify the location of a coherent trap where the reflected signals of multiple abrupt change points coherently superimpose within the Nyquist frequency range, based on the reflected phase difference.

[0159] The compensation parameter module is used to construct the coherence elimination objective function and determine the compensation parameters for geometric pre-compensation, dielectric constant gradient compensation, and active dynamic compensation.

[0160] The compensation optimization module is used to perform geometric pre-compensation based on the target residual impedance change amplitude in the compensation parameters, and adjust the reflection coefficient amplitude at the change point; perform dielectric constant gradual compensation based on the optimal control parameters of dielectric constant gradual compensation in the compensation parameters, and change the reflection phase distribution through a controlled impedance change gradient; and perform active dynamic compensation based on the target terminal matching impedance value in the compensation parameters, and introduce an adjustable reflection coefficient at the terminal to adjust the standing wave conditions.

[0161] The compensation optimization module includes:

[0162] A coefficient amplitude optimization unit is used to perform geometric pre-compensation based on the target residual impedance abrupt change amplitude in the compensation parameters, and adjust the reflection coefficient amplitude at the abrupt change point.

[0163] A phase distribution optimization unit is used to perform dielectric constant gradual compensation according to the optimal control parameters of the dielectric constant gradual compensation in the compensation parameters, and to change the reflection phase distribution through a controlled impedance change gradient.

[0164] The standing wave condition optimization unit is used to perform active dynamic compensation based on the target terminal matching impedance value in the compensation parameters, and to adjust the standing wave condition by introducing an adjustable reflection coefficient at the terminal.

[0165] In summary, the advantages of this invention are: by extending the analysis of impedance problems from amplitude to phase distribution, and combining the joint optimization of geometric pre-compensation, dielectric constant gradient compensation, and active dynamic compensation, the coherent superposition effect of multi-level reflections is effectively eliminated, thereby achieving impedance continuity and signal integrity of high-speed cables in the Nyquist frequency range.

[0166] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely principles of the invention. Various changes and modifications can be made to the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the claimed invention. The scope of protection claimed by the appended claims and their equivalents is defined.

Claims

1. A high-speed cable optimization method based on continuous compensation of impedance, characterized in that, include: Based on the end-to-end architecture of high-speed cables, the location, amplitude, and type of each impedance abrupt change point are identified, and the reflection phase difference between any two abrupt change points is calculated. Based on the reflection phase difference, identify the coherent trap locations where the reflected signals from multiple abrupt change points coherently superimpose within the Nyquist frequency range; Construct a coherence elimination objective function and determine the compensation parameters for geometric pre-compensation, dielectric constant gradient compensation, and active dynamic compensation; Based on the target residual impedance abrupt change amplitude in the compensation parameters, perform geometric pre-compensation and adjust the reflection coefficient amplitude at the abrupt change point. Based on the optimal control parameters of the dielectric constant gradient compensation in the compensation parameters, the dielectric constant gradient compensation is performed to change the reflection phase distribution through a controlled impedance change gradient. Based on the target terminal matching impedance value in the compensation parameters, active dynamic compensation is performed, and an adjustable reflection coefficient is introduced at the terminal to adjust the standing wave conditions.

2. The method of claim 1, wherein, The full-link architecture based on high-speed cables identifies the location, amplitude, and type of impedance abrupt changes, and calculates the reflection phase difference between any two abrupt changes, specifically including: Obtain the end-to-end architecture of a high-speed cable, which includes, but is not limited to: a first connector, a first solder joint, a first cable body, an intermediate transition point, a second connector, a second solder joint, and a second cable body; The impedance distribution data of the entire link is collected by a time domain reflectometer, the time axis is converted into a distance axis, and the impedance distribution curve of the entire link is plotted. The distance coordinate is the position coordinate along the transmission path. Identify all abrupt changes in the impedance distribution curve and record the location coordinates, magnitude, and type of each abrupt change. Calculate the reflection phase difference between any two abrupt change points based on their location coordinates. The reflection coefficient of each abrupt change point is calculated based on the abrupt change amplitude and the characteristic impedance on the incident side. Based on the reflection coefficients and location coordinates of each mutation point, a full-link reflection transfer function is constructed.

3. The high-speed cable optimization method based on continuous impedance compensation according to claim 2, characterized in that, The method of identifying the location of coherent traps in the Nyquist frequency range by recognizing the coherent superposition of reflected signals at multiple abrupt change points based on the reflected phase difference specifically includes: Based on the full-link reflection transfer function, calculate the return loss in the Nyquist frequency range and plot the return loss curve. Identify all local peaks in the return loss curve within the Nyquist frequency range, and record the frequency and amplitude corresponding to each local peak. For each local peak frequency point, traverse all combinations of impedance abrupt points and calculate the phase value of each impedance abrupt point at that frequency point. Select the set of impedance abrupt change points where the absolute value of the difference between phase values ​​at the specified frequency is less than a preset angle threshold. Mark the set as a coherent abrupt change point combination, mark the frequency point as a coherent trap frequency point, and mark the amplitude of the local peak as the coherent trap depth.

4. The high-speed cable optimization method based on continuous impedance compensation according to claim 3, characterized in that, The specific steps for constructing the coherence elimination objective function and determining the compensation parameters for geometric pre-compensation, dielectric constant gradient compensation, and active dynamic compensation include: Construct a coherence cancellation objective function, which is equal to the maximum value of the magnitude of the reflection transfer function in the frequency range from zero to the Nyquist frequency. Set an optimization variable set, which includes: the target residual impedance change amplitude at each impedance change point after geometric pre-compensation, the control parameters for dielectric constant gradual compensation, and the target terminal matching impedance value for active dynamic compensation. With minimizing the coherence elimination objective function as the optimization objective, the optimal combination of compensation parameters is solved using the Bayesian optimization algorithm; The Bayesian optimization algorithm includes: Several sets of compensation parameters are randomly sampled in the optimization variable space as the initial training set, and a Gaussian process surrogate model is fitted based on the training set. The Gaussian process surrogate model uses the Matern kernel function as the covariance function to fit the nonlinear relationship between the compensation parameters and the coherence elimination objective function; The Gaussian process surrogate model outputs the coherence elimination objective function, predicting the mean and standard deviation for each unsampled point; The next sampling point is selected using the expected improved acquisition function, which is equal to the difference between the current optimal objective function value and the predicted mean, multiplied by the standard normal cumulative distribution function, plus the predicted standard deviation multiplied by the standard normal probability density function. Perform experiments at the newly selected sampling points, update the training set, and repeat the steps of fitting the Gaussian process surrogate model and selecting sampling points until convergence. The convergence criterion is that the relative change of the objective function value after multiple consecutive iterations is less than a preset value, or the number of iterations reaches a preset upper limit, thus obtaining the optimal combination of compensation parameters. The optimal compensation parameter combination includes: the target residual impedance change amplitude at each impedance change point, the optimal control parameters for dielectric constant gradual compensation, and the target terminal matching impedance value.

5. The high-speed cable optimization method based on continuous impedance compensation according to claim 4, characterized in that, The step of performing geometric pre-compensation based on the target residual impedance abrupt change amplitude in the compensation parameters and adjusting the reflection coefficient amplitude at the abrupt change point specifically includes: For impedance abrupt change points identified as capacitive depressions, obtain the current location coordinates and current impedance abrupt change amplitude of the capacitive depression, and calculate the difference between the current impedance abrupt change amplitude and the target residual impedance abrupt change amplitude. The depth of laser etching is determined based on the difference. The laser etching process reduces the reference ground area around the signal conductor in this region and adjusts the amplitude of the residual impedance change at this point. For impedance abrupt change points identified as inductive spikes, obtain the current position coordinates and current impedance abrupt change amplitude of the inductive spike, and calculate the difference between the current impedance abrupt change amplitude and the target residual impedance abrupt change amplitude. The dielectric constant variation gradient of the gradient impedance matching structure is determined based on the difference. An insulating support with a dielectric constant variation gradient is installed inside the connector housing to adjust the residual impedance change amplitude at that point. For a composite impedance abrupt change point that simultaneously exhibits capacitive dips and inductive spikes, an inductive compensation structure is set in the pre-region of the composite impedance abrupt change point, and a capacitive compensation structure is set in the post-region to adjust the equivalent residual impedance abrupt change amplitude of the composite impedance abrupt change point.

6. The high-speed cable optimization method based on continuous impedance compensation according to claim 5, characterized in that, The step of performing gradual dielectric constant compensation based on the optimal control parameters of the gradual dielectric constant compensation in the compensation parameters, and changing the reflection phase distribution through a controlled impedance change gradient, specifically includes: Obtain the optimal control parameters for dielectric constant gradient compensation in the compensation parameters. The optimal control parameters include: the starting point position of dielectric constant gradient, the ending point position of dielectric constant gradient, the gradient of dielectric constant change, and the target dielectric constant value of the nominal dielectric constant segment. Starting from the point where the dielectric constant gradually changes and ending at the point where the dielectric constant gradually changes, a linear transition curve is formed from the low dielectric constant starting segment to the nominal dielectric constant segment, following the linear change of the dielectric constant gradient. Based on the transition curve of dielectric constant distribution along the axial direction, the change in target dielectric constant at each position of the insulating layer is calculated by subtracting the original dielectric constant of the insulating layer material from the target dielectric constant value on the curve. The change in the target dielectric constant is converted into the equivalent heat input of laser irradiation. The equivalent heat input is equal to the laser power density multiplied by the scanning time, divided by the spot diameter, and then multiplied by the material matching ratio coefficient. The material matching ratio is determined through a pre-calibration test and is used to correct the differences in the laser energy absorption efficiency and dielectric constant modification sensitivity of different insulating layer materials. The reflective phase distribution is altered by continuously adjusting at least one of the three parameters: laser power density, scanning time, and focused spot diameter.

7. The high-speed cable optimization method based on continuous impedance compensation according to claim 6, characterized in that, The step of performing active dynamic compensation based on the target terminal matching impedance value in the compensation parameters, and adjusting the standing wave conditions by introducing an adjustable reflection coefficient at the terminal, specifically includes: An active dynamic compensation module is integrated at the receiving end of the cable. The active dynamic compensation module includes an adjustable impedance matching network, a signal quality monitoring unit, and a microcontroller. The adjustable impedance matching network includes a digital potentiometer and an adjustable capacitor array, with an adjustment range covering a preset impedance adjustment interval. During the link training phase, the microcontroller reads the target terminal matching impedance value, generates an initial impedance adjustment command, and controls the adjustable impedance matching network to configure the terminal matching impedance to the target terminal matching impedance value. The signal quality monitoring unit continuously collects signal quality parameters from the receiver during the link training phase. These signal quality parameters include bit error rate, eye diagram opening, jitter amplitude, and signal-to-noise ratio. The collected signal quality parameters are compared with preset acceptable thresholds to generate a signal quality score; The microcontroller executes an adaptive gradient descent algorithm, which includes: Starting from the current terminal matching impedance value, adjust the terminal matching impedance value in both positive and negative directions with a preset step size, and record the change in signal quality score in the two directions. Select the direction with the greatest improvement in signal quality score as the adjustment direction, and continue to adjust in that direction at the same pace until the signal quality score reaches its peak or the number of adjustments reaches the preset limit. Lock the terminal matching impedance value when the signal quality score reaches its peak value as the optimal working impedance value. During cable operation, the signal quality monitoring unit continuously monitors signal quality parameters at preset time intervals. When the current signal quality score is detected to be lower than the preset fluctuation threshold, the microcontroller restarts the gradient descent adaptive algorithm to search for the optimal working impedance value again.

8. A high-speed cable optimization system based on continuous impedance compensation, characterized in that, The method for implementing the high-speed cable optimization method based on impedance continuous compensation as described in any one of claims 1-7 includes: The transfer function module is used to identify the location, magnitude, and type of impedance abrupt change points in the end-to-end architecture based on high-speed cables, and to calculate the reflection phase difference between any two abrupt change points. A coherent abrupt change point identification module is used to identify the location of a coherent trap where the reflected signals of multiple abrupt change points coherently superimpose within the Nyquist frequency range, based on the reflected phase difference. The compensation parameter module is used to construct the coherence elimination objective function and determine the compensation parameters for geometric pre-compensation, dielectric constant gradient compensation, and active dynamic compensation. The compensation optimization module is used to perform geometric pre-compensation based on the target residual impedance change amplitude in the compensation parameters, and adjust the reflection coefficient amplitude at the change point; perform dielectric constant gradual compensation based on the optimal control parameters of dielectric constant gradual compensation in the compensation parameters, and change the reflection phase distribution through a controlled impedance change gradient; and perform active dynamic compensation based on the target terminal matching impedance value in the compensation parameters, and introduce an adjustable reflection coefficient at the terminal to adjust the standing wave conditions.

9. A high-speed cable optimization system based on continuous impedance compensation according to claim 8, characterized in that, The compensation optimization module includes: A coefficient amplitude optimization unit is used to perform geometric pre-compensation based on the target residual impedance abrupt change amplitude in the compensation parameters, and adjust the reflection coefficient amplitude at the abrupt change point. A phase distribution optimization unit is used to perform dielectric constant gradual compensation according to the optimal control parameters of the dielectric constant gradual compensation in the compensation parameters, and to change the reflection phase distribution through a controlled impedance change gradient. The standing wave condition optimization unit is used to perform active dynamic compensation based on the target terminal matching impedance value in the compensation parameters, and to adjust the standing wave condition by introducing an adjustable reflection coefficient at the terminal.

Images