A network communication coaxial cable detection method based on intelligent sensor

Abstract

The application relates to the technical field of cable nondestructive testing, and discloses a network communication coaxial cable detection method based on an intelligent sensor, which comprises the following steps: collecting a double-frequency reflected voltage signal and mapping the double-frequency reflected voltage signal to a complex analytic domain; using mutual power spectrum to offset the common-mode phase shift caused by the temperature gradient of the cable; and extracting an amplitude-weighted phase gradient which avoids the singularity of an inverse tangent function; meanwhile, calculating transient cumulative energy to construct an adaptive exponential compensation weight, inversely modulating the phase gradient feature to amplify the weak fault feature at a remote end; further, extracting a dimensionless transient feature probability density and local instantaneous information entropy based on the feature, extracting a transient cumulative entropy proportion according to the dimensionless transient feature probability density and the local instantaneous information entropy, constructing a dynamic penalty factor to cut off the extremely remote noise background which is out of control due to exponential amplification; finally, combining the nominal transmission speed of high-frequency electromagnetic waves to output the physical distance of a fault point; the problems of phase winding and remote noise out of control are solved, and the precision positioning of weak faults is realized.

Description

Technical Field

[0001] This invention relates to the field of non-destructive testing technology for cables, specifically a method for testing network communication coaxial cables based on intelligent sensors. Background Technology

[0002] Network communication systems use coaxial cables extensively for high-frequency data transmission. However, cables are prone to minor damage such as micro-oxidation of the shielding layer and slight deformation of the insulation layer. Accurately detecting these deep-seated minor faults is crucial to ensuring communication quality.

[0003] Currently, most common coaxial cable detection technologies are based on high-frequency reflection methods. A transmitter injects a continuous test wave into the cable, and sensors collect the reflected signals for time-domain or frequency-domain analysis. However, existing technologies face many limitations in practical applications. First, long-distance cables are often in complex external environments, and uneven temperature gradients along the cable can cause electromagnetic wave transmission delays and drift. Existing technologies struggle to effectively counteract this temperature-induced common-mode phase shift interference, leading to distorted detection benchmarks. Second, when extracting the phase characteristics of reflected signals, traditional algorithms heavily rely on the arctangent function. When encountering high-frequency noise, the arctangent function is prone to severe phase entanglement, resulting in mathematical singularities in derivative operations, generating false pulses, and severely interfering with subsequent feature extraction. Third, high-frequency signals exhibit inherent energy attenuation during propagation in cables, making it easy for genuine, weak fault characteristics at the far end of the cable to be masked. If conventional technologies employ a global compensation strategy, while enhancing weak signals at the far end, it inevitably leads to the exponential amplification of the pure noise floor at the extremely far end, causing it to completely spiral out of control. The rapidly expanding noise peaks are easily misjudged as serious faults by the system, resulting in severe false alarms.

[0004] Therefore, a network communication coaxial cable detection method based on intelligent sensors is needed to overcome temperature common-mode drift, avoid phase entanglement singularities, and accurately cut off the noise background at the far end while amplifying the weak features at the far end in reverse, so as to achieve high-precision fault location. Summary of the Invention

[0005] This invention provides a method for detecting coaxial cables used in network communication based on intelligent sensors, which helps to solve the problems mentioned in the background art.

[0006] This invention provides the following technical solution: a method for detecting coaxial cables used in network communication based on intelligent sensors, comprising the following steps:

[0007] Initial operation and data acquisition steps: Connect a high-frequency intelligent voltage sensor and a dual-frequency continuous wave transmitter to the test end of the coaxial cable under test; control the transmitter to synchronously inject the first frequency continuous test wave and the second frequency continuous test wave into the cable; control the intelligent voltage sensor to continuously acquire the time domain data of the reflected voltage at the cable end within the set objective observation time window; extract the first frequency reflected voltage signal and the second frequency reflected voltage signal respectively.

[0008] Dual-frequency signal analytical domain conversion steps: The first frequency reflected voltage signal is mapped to the complex analytical domain using a transformation algorithm to obtain the first frequency analytical voltage signal; the second frequency reflected voltage signal is mapped to the complex analytical domain using the same transformation algorithm to obtain the second frequency analytical voltage signal.

[0009] The steps for extracting the cross power spectrum and phase gradient are as follows: Calculate the product of the complex conjugate of the first frequency analytic voltage signal and the second frequency analytic voltage signal to obtain the transient cross power spectrum; calculate the product of the complex conjugate of the transient cross power spectrum and its own first-order time derivative; extract the imaginary part of the above product result to obtain the amplitude-weighted phase gradient.

[0010] The energy-reverse exponential adaptive modulation steps are as follows: Calculate the transient accumulated energy based on the first frequency analytical voltage signal and the second frequency analytical voltage signal; construct an adaptive exponential compensation weight based on the transient accumulated energy; multiply the adaptive exponential compensation weight by the amplitude-weighted phase gradient to obtain the phase gradient characteristics after exponential modulation.

[0011] The steps of the dimensionless entropy-driven dynamic penalty truncation are as follows: extract the dimensionless transient feature probability density based on the exponentially modulated phase gradient features; calculate the local instantaneous information entropy based on the dimensionless transient feature probability density; extract the transient cumulative entropy ratio based on the local instantaneous information entropy; and construct the dynamic penalty factor based on the transient cumulative entropy ratio.

[0012] The steps for locating and outputting weak cable faults are as follows: Multiply the dynamic penalty factor by the phase gradient feature after exponential modulation to obtain the final cable fault characteristic index; combine the nominal transmission speed of the high-frequency electromagnetic wave of the coaxial cable with the final cable fault characteristic index to calculate and output the physical distance of the fault point.

[0013] Optionally, the dual-frequency signal analytical domain conversion step includes:

[0014] Perform Hilbert principal value integral transform on the first frequency reflected voltage signal to construct the first frequency analytical voltage signal;

[0015] The same analytic domain mapping operation is performed on the second frequency reflected voltage signal, that is, Hilbert principal value integral transform is performed to construct the second frequency analytic voltage signal.

[0016] Optionally, the step of calculating the transient accumulated energy includes:

[0017] The absolute amplitudes of the first frequency analytical voltage signal and the second frequency analytical voltage signal are obtained respectively;

[0018] Calculate the sum of squares of the two absolute amplitudes to obtain the transient total energy density;

[0019] The transient cumulative energy is obtained by performing time-axis integration on the total transient energy density.

[0020] Optionally, the steps of constructing the adaptive exponential compensation weights and obtaining the exponentially modulated phase gradient features include:

[0021] The total test energy gained when accumulating points to the end of the time limit;

[0022] Calculate the ratio of transient cumulative energy to the total test energy, and use this ratio as the independent variable of the natural exponential function to obtain the adaptive exponential compensation weight.

[0023] The amplitude-weighted phase gradient is multiplied by the adaptive exponential compensation weight to complete the modulation, thus obtaining the exponentially modulated phase gradient characteristics.

[0024] Optionally, the step of extracting the dimensionless transient feature probability density includes:

[0025] Within the entire objective observation time window, extract the global maximum value of the absolute value of the exponentially modulated phase gradient feature;

[0026] Extract the absolute value of the instantaneous exponentially modulated phase gradient feature;

[0027] Divide the instantaneous absolute value by the aforementioned global maximum value and perform extreme value normalization to obtain the dimensionless transient characteristic probability density.

[0028] Optionally, the step of calculating the local instantaneous information entropy includes:

[0029] Calculate the natural logarithm of the dimensionless transient characteristic probability density;

[0030] Multiply the dimensionless transient characteristic probability density by the natural logarithm;

[0031] Taking the negative value of the above product result yields the local instantaneous information entropy.

[0032] Optionally, the step of extracting the transient cumulative entropy ratio includes:

[0033] Perform time-axis integration on the local instantaneous information entropy within the current time range to obtain the integration result at the current moment;

[0034] The local instantaneous information entropy is integrated over the entire objective observation time window to obtain the total time window integration result.

[0035] Divide the current integral result by the total time window integral result to obtain the transient cumulative entropy ratio.

[0036] Optionally, the step of constructing the dynamic penalty factor includes:

[0037] Obtain the known system entropy sensitivity coefficient that is set directly;

[0038] Calculate the system entropy sensitivity coefficient raised to the power of the transient cumulative entropy ratio;

[0039] The dynamic penalty factor is generated by subtracting the power value obtained from the above calculation from the number one.

[0040] Optionally, the steps of obtaining the final cable fault characteristic index and outputting the physical distance of the fault point include:

[0041] The final cable fault characteristic index is obtained by multiplying the exponentially modulated phase gradient feature with the dynamic penalty factor.

[0042] Extract the specific time point that causes the final cable fault characteristic index to reach its maximum value;

[0043] Multiply the nominal high-frequency electromagnetic wave transmission speed of the coaxial cable by the specific time point to obtain the reference distance span value;

[0044] Divide the baseline distance span value by the number two to output the physical distance from the weak fault point of the cable to the test head.

[0045] The present invention has the following beneficial effects:

[0046] 1. This technical solution addresses the specific long cable environment of high-frequency network communication coaxial cables, which are subject to complex temperature gradients and where high-frequency signals are prone to exponential attenuation. It proposes a weak fault detection algorithm based on dual-frequency signal analysis and entropy-driven truncation. The core of the solution lies in extracting the cross-power spectrum phase gradient without singularities, combining it with transient accumulated energy for inverse exponential adaptive modulation, and using information entropy to construct a dynamic penalty factor for output truncation. In actual coaxial cable testing, temperature gradients can cause severe common-mode phase shift interference, and the subtle fault characteristics deep within the cable can be severely attenuated and masked by dielectric loss. If only the weak signal at the far end is amplified exponentially, the pure noise background at the far end will inevitably explode out of control, causing serious false alarms in the system. In this specific environment, this solution not only effectively cancels the temperature drift and avoids the phase entanglement singularity caused by the traditional arctangent function, but also utilizes the difference in information entropy between the real signal and the noise. While successfully amplifying the deep weak fault characteristics in reverse, it accurately suppresses and cuts off the pure noise background at the far end that is out of control due to exponential amplification. Thus, it greatly improves the sensitivity of the system to detect weak faults at the far end while ensuring an extremely low false alarm rate.

[0047] 2. By calculating the transient cross-power spectrum of two complex analytic voltage signals and taking the imaginary part by multiplying the complex conjugate of the cross-power spectrum with its first-order time derivative, the transient phase gradient feature weighted by the square of the amplitude is extracted. The stable and continuous phase gradient is directly obtained by using the product differentiation rule of complex plane of analytic signals. This eliminates the arctangent function operation that is heavily relied upon in traditional detection algorithms, thereby avoiding the severe phase entanglement phenomenon that is easily triggered under high-frequency noise interference from the underlying operation logic, as well as the mathematical singularity of derivative operation and false interference pulses caused by it. This improves the feature analysis stability and fidelity of weak fault reflection signals in high-frequency harsh electromagnetic environments.

[0048] 3. By synchronously injecting dual-frequency continuous test waves at the test head of the coaxial cable and collecting the reflected voltage signal, and then using Hilbert principal value integral transform to map the collected time-domain reflected voltage signal to the complex analytical domain, the analytical voltage signal of the corresponding frequency is obtained. This not only effectively eliminates the interference of redundant negative frequency components in the physical signal in the real domain, and accurately and without distortion extracts the transient amplitude and transient phase characteristics, but also provides a precisely aligned reference data structure for subsequent use of frequency difference characteristics to remove the transmission delay phase shift interference caused by the uneven temperature gradient along the cable. This effectively ensures the basic detection accuracy and reference data reliability of the weak cable damage detection system under complex temperature fluctuation environment.

[0049] 4. By calculating the transient total energy density of the dual-frequency analytical voltage signal at each moment in real time and performing integral calculation along the time axis to obtain the transient cumulative energy, and then combining the total test energy of the whole time period to construct an adaptive exponential compensation weight, the objective distribution law of the exponential energy attenuation of the high-frequency test signal in the long-distance cable medium with the passage of time and distance is quantified. This can provide a suitable adaptive gain benchmark for the reverse compensation of the subsequent deep weak fault signals, and solve the problem of the detection blind zone where the extremely small amplitude of the high-frequency detection signal caused by the dielectric loss at the far end of the cable is compressed twice by the strong signal at the front end, resulting in the masking of the weak echo of the real fault.

[0050] 5. By extracting the adaptive exponential compensation weights dynamically constructed based on the energy attenuation law, and directly applying them to the amplitude-weighted phase gradient features acquired at the front end, the reverse amplitude modulation of the potential fault features at the far end is completed. This processing logic can achieve deterministic adaptive exponential amplification of the reflected waves of deep weak faults, compensate for the signal gain, and strengthen the features that were originally too weak to distinguish due to the high-frequency skin effect and dielectric loss. This enables the extremely far-end minute physical damage features to cross the detection sensitivity threshold and become clearly visible, significantly expanding the effective detection boundary of dual-frequency continuous wave detection technology in the scenario of ultra-long network communication coaxial cables, and enhancing the system's ability to perceive and capture deep hidden dangers.

[0051] 6. The dimensionless transient feature probability density obtained by extreme value normalization is calculated, multiplied by its corresponding natural logarithm and negative, and the local instantaneous information entropy sequence is derived. This step captures the essential feature difference at the level of physical order, amplifies the difference between the real structural damage reflection signal and the chaotic environmental background noise, and enables the real fault reflection echo with extremely high order and low entropy value to stand out, separating it from the chaotic and extremely high high-frequency far-end noise region, laying the foundation for subsequent response to the uncontrolled expansion of pure noise background caused by the exponential amplification compensation mechanism.

[0052] 7. By integrating the local instantaneous information entropy over the current time range and performing dual time-axis integration operations throughout the entire observation time window, the transient dynamic ratio of the current cumulative entropy to the total cumulative entropy is calculated. This ratio macroscopically and continuously characterizes the cumulative spread trend of the noise background within the test time series, reflects the nonlinear growth slope of the extremely far-end pure noise, and can accurately capture the critical phase transition node from the signal-dominated region to the pure noise-dominated tail region. It establishes a robust reference benchmark that does not sway violently with a single sudden change in the external environment, and provides conditions for constructing a dynamic penalty and attenuation cutoff mechanism with high environmental adaptability.

[0053] 8. By introducing a directly set system entropy sensitivity coefficient, the extracted transient cumulative entropy ratio is subjected to nonlinear exponential power operation, and a dynamic penalty factor for the extremely far-end noise region is calculated accordingly. This mechanism utilizes the objective evolution trend of the cumulative information entropy increasing absolutely monotonicly and rapidly at the tail end of pure noise, and generates an adaptive attenuation mechanism with forced suppression capability in the pure noise region at the end of the cable. It can cut off false high-amplitude noise peaks, smooth out the far-end clutter that has completely run out of control due to the previous exponential amplification compensation, and complete the pure feature reshaping with extremely high signal-to-noise ratio from the very end of the algorithm architecture, realizing the dynamic balance between weak signal extraction and false alarm suppression.

[0054] 9. By directly applying the dynamic penalty factor derived from accumulated information entropy, and simultaneously applying it to the modulated phase gradient features that have been amplified exponentially at the front end, a cable fault feature index with high purity is formed. Then, the specific time node corresponding to the global maximum value of this feature sequence is extracted. Combined with the nominal high-frequency electromagnetic wave transmission speed in the coaxial cable, the actual spatial physical distance between the weak fault point and the test head end is accurately calculated. This achieves complete multi-dimensional data aggregation from complex frequency domain feature stripping to time domain energy modulation and then to precise spatial domain physical mapping, successfully solving the problem of locating and detecting actual physical hidden dangers in engineering. Attached Figure Description

[0055] Figure 1 This is a schematic diagram of the basic process of the present invention. Detailed Implementation

[0056] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0057] Example 1, refer to Figure 1 A method for detecting coaxial cables used in network communication based on intelligent sensors includes the following steps:

[0058] Initial operation and data acquisition steps: Connect a high-frequency intelligent voltage sensor and a dual-frequency continuous wave transmitter to the test end of the coaxial cable under test; control the transmitter to synchronously inject the first frequency continuous test wave and the second frequency continuous test wave into the cable; control the intelligent voltage sensor to continuously acquire the time domain data of the reflected voltage at the cable end within the set objective observation time window; extract the first frequency reflected voltage signal and the second frequency reflected voltage signal respectively.

[0059] Dual-frequency signal analytical domain conversion steps: The first frequency reflected voltage signal is mapped to the complex analytical domain using a transformation algorithm to obtain the first frequency analytical voltage signal; the second frequency reflected voltage signal is mapped to the complex analytical domain using the same transformation algorithm to obtain the second frequency analytical voltage signal.

[0060] The steps for extracting the cross power spectrum and phase gradient are as follows: Calculate the product of the complex conjugate of the first frequency analytic voltage signal and the second frequency analytic voltage signal to obtain the transient cross power spectrum; calculate the product of the complex conjugate of the transient cross power spectrum and its own first-order time derivative; extract the imaginary part of the above product result to obtain the amplitude-weighted phase gradient.

[0061] The energy-reverse exponential adaptive modulation steps are as follows: Calculate the transient accumulated energy based on the first frequency analytical voltage signal and the second frequency analytical voltage signal; construct an adaptive exponential compensation weight based on the transient accumulated energy; multiply the adaptive exponential compensation weight by the amplitude-weighted phase gradient to obtain the phase gradient characteristics after exponential modulation.

[0062] The steps of the dimensionless entropy-driven dynamic penalty truncation are as follows: extract the dimensionless transient feature probability density based on the exponentially modulated phase gradient features; calculate the local instantaneous information entropy based on the dimensionless transient feature probability density; extract the transient cumulative entropy ratio based on the local instantaneous information entropy; and construct the dynamic penalty factor based on the transient cumulative entropy ratio.

[0063] The steps for locating and outputting weak cable faults are as follows: Multiply the dynamic penalty factor by the phase gradient feature after exponential modulation to obtain the final cable fault characteristic index; combine the nominal transmission speed of the high-frequency electromagnetic wave of the coaxial cable with the final cable fault characteristic index to calculate and output the physical distance of the fault point.

[0064] The dual-frequency signal analytical domain conversion step includes:

[0065] Perform Hilbert principal value integral transform on the first frequency reflected voltage signal to construct the first frequency analytical voltage signal;

[0066] The same analytic domain mapping operation is performed on the second frequency reflected voltage signal, that is, Hilbert principal value integral transform is performed to construct the second frequency analytic voltage signal.

[0067] The steps for calculating the transient accumulated energy include:

[0068] The absolute amplitudes of the first frequency analytical voltage signal and the second frequency analytical voltage signal are obtained respectively;

[0069] Calculate the sum of squares of the two absolute amplitudes to obtain the transient total energy density;

[0070] The transient cumulative energy is obtained by performing time-axis integration on the total transient energy density.

[0071] The steps of constructing the adaptive exponential compensation weights and obtaining the exponentially modulated phase gradient features include:

[0072] The total test energy gained when accumulating points to the end of the time limit;

[0073] Calculate the ratio of transient cumulative energy to the total test energy, and use this ratio as the independent variable of the natural exponential function to obtain the adaptive exponential compensation weight.

[0074] The amplitude-weighted phase gradient is multiplied by the adaptive exponential compensation weight to complete the modulation, thus obtaining the exponentially modulated phase gradient characteristics.

[0075] The steps for extracting the dimensionless transient feature probability density include:

[0076] Within the entire objective observation time window, extract the global maximum value of the absolute value of the exponentially modulated phase gradient feature;

[0077] Extract the absolute value of the instantaneous exponentially modulated phase gradient feature;

[0078] Divide the instantaneous absolute value by the aforementioned global maximum value and perform extreme value normalization to obtain the dimensionless transient characteristic probability density.

[0079] The steps for calculating the local instantaneous information entropy include:

[0080] Calculate the natural logarithm of the dimensionless transient characteristic probability density;

[0081] Multiply the dimensionless transient characteristic probability density by the natural logarithm;

[0082] Taking the negative value of the above product result yields the local instantaneous information entropy.

[0083] The steps for extracting the transient cumulative entropy percentage include:

[0084] Perform time-axis integration on the local instantaneous information entropy within the current time range to obtain the integration result at the current moment;

[0085] The local instantaneous information entropy is integrated over the entire objective observation time window to obtain the total time window integration result.

[0086] Divide the current integral result by the total time window integral result to obtain the transient cumulative entropy ratio.

[0087] The steps for constructing the dynamic penalty factor include:

[0088] Obtain the known system entropy sensitivity coefficient that is set directly;

[0089] Calculate the system entropy sensitivity coefficient raised to the power of the transient cumulative entropy ratio;

[0090] The dynamic penalty factor is generated by subtracting the power value obtained from the above calculation from the number one.

[0091] The steps for obtaining the final cable fault characteristic index and outputting the physical distance to the fault point include:

[0092] The final cable fault characteristic index is obtained by multiplying the exponentially modulated phase gradient feature with the dynamic penalty factor.

[0093] Extract the specific time point that causes the final cable fault characteristic index to reach its maximum value;

[0094] Multiply the nominal high-frequency electromagnetic wave transmission speed of the coaxial cable by the specific time point to obtain the reference distance span value;

[0095] Dividing the baseline distance span by two, the physical distance from the weak fault point in the cable to the test head is output. This technical solution addresses the specific long-cable environment of high-frequency network communication coaxial cables operating under complex temperature gradients and where high-frequency signals are prone to exponential attenuation. It proposes a weak fault detection algorithm based on dual-frequency signal analysis and entropy-driven truncation. The core of the solution lies in extracting the cross-power spectrum phase gradient without singularities, combining it with transient accumulated energy for inverse exponential adaptive modulation, and using information entropy to construct a dynamic penalty factor for output truncation. In actual coaxial cable testing, temperature gradients can cause severe common-mode phase shift interference, and the subtle characteristics of hidden faults deep within the cable can be severely attenuated and masked by dielectric loss. If only the weak signal at the far end is amplified exponentially, the pure noise background at the far end will inevitably explode out of control, causing serious false alarms in the system. In this specific environment, this solution not only effectively cancels the temperature drift and avoids the phase entanglement singularity caused by the traditional arctangent function, but also utilizes the difference in information entropy between the real signal and the noise. While successfully amplifying the deep weak fault characteristics in reverse, it accurately suppresses and cuts off the pure noise background at the far end that is out of control due to exponential amplification. Thus, it greatly improves the sensitivity of the system to detect weak faults at the far end while ensuring an extremely low false alarm rate.

[0096] Example 2: A method for detecting coaxial cables used in network communication based on smart sensors, further comprising:

[0097] Initial operation and data acquisition: Connect a high-frequency intelligent voltage sensor and a dual-frequency continuous wave transmitter to the test end of the coaxial cable of the network under test;

[0098] The transmitter synchronously injects a frequency into the cable. and The continuous test wave, the intelligent voltage sensor within the set objective observation time window Inside, the time-domain data of the reflected voltage at the cable head end is continuously collected, and the first frequency reflected voltage signal is obtained respectively. Second frequency reflected voltage signal ;

[0099] Dual-frequency signal analytical domain conversion: using a transformation algorithm to convert the first frequency reflected voltage signal into an analytical domain. and the second frequency reflected voltage signal Mapping to the complex analytic domain, the analytic voltage signal at the first frequency is obtained respectively. With the analytical voltage signal at the second frequency ;

[0100] Singularity-free cross-power spectrum and phase gradient extraction: The transient cross-power spectrum of the two analytical voltage signals above is calculated to compensate for the common-mode phase shift caused by the cable temperature gradient. The specific calculation is as follows:

[0101] In the formula, This represents the calculated transient cross-power spectrum. This represents the first frequency analytical voltage signal. Indicates the second frequency analytical voltage signal The complex conjugate;

[0102] Using the cross-power spectrum and its first-order time derivative, the transient phase gradient features weighted by the square of the amplitude are extracted to avoid the phase entanglement singularity of the arctangent function. The specific calculation is as follows:

[0103] In the formula, This represents the calculated amplitude-weighted phase gradient. This represents the operation of taking the imaginary part of a complex number. Represents transient cross power spectrum The complex conjugate, This represents the first derivative of the transient cross-power spectrum with respect to time.

[0104] Energy-inverse exponential adaptive modulation: based on the analytical voltage signal and Calculate transient cumulative energy This is used to construct adaptive exponential compensation weights. And the adaptive exponential compensation weights Acting on the amplitude-weighted phase gradient The phase gradient characteristics after exponential modulation are obtained. To amplify the characteristics of weak faults at the remote end in reverse;

[0105] Dimensionless entropy-driven dynamic penalty truncation: based on the aforementioned phase gradient features Extracting dimensionless transient feature probability density And calculate the local instantaneous information entropy. Further extract the proportion of transient cumulative entropy To construct a dynamic penalty factor To cut off the background noise at the far end that has run out of control due to exponential amplification;

[0106] Cable weak fault location and output: The dynamic penalty factor Acting on the phase gradient feature The final cable fault characteristic index is obtained. And in conjunction with the nominal transmission speed of high-frequency electromagnetic waves in the cable Output physical distance of the fault point .

[0107] Among them, for injection frequency of and The continuous test wave, whose value range is set at... to Within the radio frequency band, and the frequency difference between the two Set between 1% and 5% of the base frequency, for example, set... , The higher the frequency setting, the shorter the wavelength of the electromagnetic wave, and the higher the spatial resolution and reflection sensitivity of the system for minor physical damage inside the cable (such as micro-oxidation of the shielding layer and minor deformation of the insulation layer). However, according to the high-frequency transmission line theory, the skin effect and dielectric loss will cause the attenuation coefficient of the high-frequency signal in the cable to increase exponentially, which will seriously shorten the effective detection distance of the system. The lower the frequency setting, the slower the energy attenuation of the signal when it is transmitted inside the cable, and the longer the cable distance can be penetrated and detected. However, low-frequency long-wavelength signals will diffract and are not sensitive to weak fault points in the cable whose size is smaller than the wavelength. This will cause minor hidden dangers to fail to excite enough reflection energy and phase change, resulting in missed detection.

[0108] For the objective observation time window constant Its value is based on the maximum total physical length of the coaxial cable under test. With nominal wave speed It is determined that the theoretical minimum value is To allow for computational redundancy, the value range is set as follows: The magnitude is on the order of microseconds. A larger time window ensures that the reflected signal from the far end of the cable is fully captured without effective signal truncation. However, if the value is too large, it will cause the integration domain to contain too much "pure noise background" beyond the cable end. This will not only unnecessarily increase the computational load of hardware sampling and algorithm processing, but also dilute the total energy throughout the entire time period. The baseline value interferes with the sensitivity of the subsequent entropy cutoff mechanism; the smaller the time window is set, the lower the computational load and the more effectively pure noise interference at the end can be avoided, but if Less than the time required for the electromagnetic wave to reach the farthest fault and return ( The weak fault reflection wave at the farthest point will be physically cut off on the time axis, and the algorithm cannot obtain this segment of data, resulting in a blind zone at the far end. By calculating the transient cross power spectrum of two complex analytic voltage signals, and using the method of multiplying the complex conjugate of the cross power spectrum with its first-order time derivative to obtain the imaginary part, the transient phase gradient feature after amplitude square weighting is extracted. The stable and continuous phase gradient is directly obtained by using the product differentiation rule of complex plane of analytic signals. The arctangent function operation that is extremely dependent on in traditional detection algorithms is abandoned. Thus, the severe phase entanglement phenomenon that is easily triggered under high-frequency noise interference is avoided from the underlying operation logic, as well as the mathematical singularity of derivative operation and false interference pulses caused by it. The feature analysis stability and fidelity of weak fault reflection signal in high-frequency harsh electromagnetic environment are improved.

[0109] In the dual-frequency signal analytical domain conversion step, the Hilbert principal value integral transform of the first frequency reflected voltage signal is performed to construct the first frequency complex signal, specifically calculated as follows:

[0110] In the formula, This represents the analytical voltage signal at the first frequency. This represents the first frequency reflected voltage signal initially acquired. Represents the imaginary unit. Represents the integral variable;

[0111] The same analytic domain mapping operation is performed on the second-frequency reflected voltage signal to construct the second-frequency complex signal, specifically calculated as follows:

[0112] In the formula, This represents the analytical voltage signal at the second frequency. This represents the initially acquired second-frequency reflected voltage signal. By synchronously injecting dual-frequency continuous test waves at the test head of the coaxial cable and acquiring the reflected voltage signal, and then using Hilbert principal value integral transform, the acquired time-domain reflected voltage signal is mapped to the complex analytic domain to obtain the analytic voltage signal of the corresponding frequency. This not only effectively eliminates redundant negative frequency component interference in the real-domain physical signal and accurately and without distortion extracts transient amplitude and transient phase characteristics, but also provides a precisely aligned reference data structure for subsequent use of frequency difference characteristics to remove transmission delay phase shift interference caused by uneven temperature gradients along the cable. This effectively ensures the basic detection accuracy and reference data reliability of the weak cable damage detection system under complex temperature fluctuation environments.

[0113] The specific calculation of the energy distribution state of the quantized high-frequency signal in the energy inverse exponential adaptive modulation step includes:

[0114] The transient total energy density of the two frequency bands at each time step is calculated as follows:

[0115] In the formula, Represents the transient total energy density. and These respectively represent the analytical voltage signals and The absolute amplitude;

[0116] The distribution function is obtained by integrating the transient total energy density over time. The specific calculation is as follows:

[0117] In the formula, This represents the transient accumulated energy. Represents the transient total energy density. This represents the integral variable. By calculating the transient total energy density of the dual-frequency analytical voltage signal at each moment in real time and performing integral calculations along the time axis to obtain the transient cumulative energy, and then combining it with the total test energy over the entire time period to construct an adaptive exponential compensation weight, the objective distribution law of the exponential energy attenuation of the high-frequency test signal in the long-distance cable medium with the passage of time and distance is quantified. This can provide a suitable adaptive gain benchmark for the reverse compensation of subsequent deep weak fault signals, and solve the problem of the detection blind zone where the extremely small amplitude of the high-frequency detection signal caused by dielectric loss at the far end of the cable is compressed twice by the strong signal at the front end, resulting in the masking of the weak echo of the real fault.

[0118] The energy-inverse exponential adaptive modulation step, which involves calculating the gain modulation for deep weak faults, includes:

[0119] An adaptive exponential gain compensation weight is constructed using the ratio of accumulated energy to total energy over the entire period. The specific calculation is as follows:

[0120] In the formula, This indicates the adaptive exponential compensation weight. This represents the transient accumulated energy. Represents the objective observation time window constant. Indicates the integration up to the end of the time. The total test energy obtained at that time Represents the natural exponential function;

[0121] The compensation weight is applied to the phase gradient to complete the modulation, and the specific calculation is as follows:

[0122] In the formula, This represents the phase gradient characteristics after exponential modulation. This represents the amplitude-weighted phase gradient. This represents the adaptive exponential compensation weight. By extracting the adaptive exponential compensation weight, which is dynamically constructed based on the energy decay law, and directly applying it to the amplitude-weighted phase gradient features acquired at the front end, the reverse amplitude modulation of the potential fault features at the far end is completed. This processing logic can achieve deterministic adaptive exponential amplification of the reflected waves of deep weak faults, compensate for the signal gain, and enhance the features that were originally too weak to distinguish due to the high-frequency skin effect and dielectric loss. This enables the subtle physical damage features at the far end to cross the detection sensitivity threshold and become clearly visible, significantly expanding the effective detection boundary of dual-frequency continuous wave detection technology in the scenario of ultra-long network communication coaxial cables, and enhancing the system's ability to perceive and capture deep hidden dangers.

[0123] The dimensionless entropy-driven dynamic penalty truncation step involves calculating the conversion of the modulated signal into probability density form, which includes:

[0124] The global maximum amplitude of the modulated signal is extracted over the entire observation period, and the extreme value normalization of the exponentially modulated phase gradient feature is performed, specifically calculated as follows:

[0125] In the formula, Represents the dimensionless transient characteristic probability density, with its value range constrained by... Inside, This represents the absolute value of the phase gradient characteristic after exponential modulation. Indicates the objective observation time window Inside The global maximum value.

[0126] In the dimensionless entropy-driven dynamic penalty truncation step, the local instantaneous information entropy is calculated based on the feature probability density, specifically as follows:

[0127] In the formula, Represents instantaneous information entropy. This represents the dimensionless transient characteristic probability density. This represents the natural logarithm function. The dimensionless transient characteristic probability density, obtained by normalizing the extreme values, is calculated, multiplied by its corresponding natural logarithm value, and then negative. This process is used to deduce and calculate the local instantaneous information entropy sequence. This step captures the essential differences in physical order, amplifying the distinction between real structural damage reflection signals and chaotic environmental background noise. This allows real fault reflection echoes with extremely high order and low entropy to stand out, separating them from the chaotic and highly entropy-laden high-frequency far-end noise region. This lays the foundation for subsequent responses to the uncontrolled expansion of pure noise background caused by exponential amplification compensation mechanisms.

[0128] In the dimensionless entropy-driven dynamic penalty truncation step, the proportion of transient cumulative entropy is extracted to characterize the noise trend, specifically calculated as follows:

[0129] In the formula, This represents the percentage of transient cumulative entropy. Represents instantaneous information entropy. Represents the objective observation time window constant. This represents the integral variable. By integrating the local instantaneous information entropy over the current time range and simultaneously performing dual time-axis integration operations throughout the entire observation time window, the transient dynamic ratio of the current accumulated entropy to the total accumulated entropy is calculated. This ratio macroscopically and continuously characterizes the cumulative spread trend of the noise background within the test time series, reflects the nonlinear growth slope of the extremely far-end pure noise, and can accurately capture the critical phase transition node from the signal-dominated region to the pure noise-dominated tail region. It establishes a robust reference benchmark that does not sway drastically with a single abrupt change in the external environment, providing conditions for constructing a dynamic penalty and attenuation cutoff mechanism with high environmental adaptability.

[0130] In the dimensionless entropy-driven dynamic penalty truncation step, a dynamic penalty mechanism is generated in the pure noise region, specifically calculated as follows:

[0131] In the formula, Indicates the dynamic penalty factor. This represents the percentage of transient cumulative entropy. This represents the directly set sensitivity coefficient of the known system entropy, with a value range between 1.5 and 3.0.

[0132] Among them, the sensitivity coefficient of system entropy The larger the value, the greater the impact: due to the proportion of transient cumulative entropy. Hengzai Within a closed interval, according to the formula ,when The larger the value, Growth is extremely slow in the initial stages, and the penalty factor... It will remain at a high level close to 1 for a longer period, which delays the intervention of the penalty mechanism and helps to preserve deeper and weaker effective fault reflection signals at the far end of the cable; however, this also means that the violent false bursts caused by exponential amplification in the pure noise region at the far end cannot be cut off in time, which can easily lead to serious false alarms in the system output; the smaller the value, the more... The smaller the value (approaching 1), the more drastically and sensitively the system reacts to the cumulative trend of local information entropy, and the greater the penalty factor. The descent curve will be extremely steep, which can suppress all noise background at the far end and ensure zero false alarm rate; however, the negative impact is that the excessively rapid penalty attenuation will cause "false alarms", which will erase the weak fault signals that actually exist at the end of the cable along with the noise, resulting in a decrease in the system detection sensitivity. By introducing a directly set system entropy sensitivity coefficient, the proportion of the extracted transient cumulative entropy is calculated nonlinearly by exponential power, and a dynamic penalty factor for the extremely far noise region is generated accordingly. This mechanism utilizes the objective evolution trend of the cumulative information entropy increasing absolutely monotonicly and rapidly at the tail end of pure noise, and generates an adaptive attenuation mechanism with forced suppression capability in the pure noise region at the end of the cable. It can cut off false high-amplitude noise peaks, smooth out the far-end clutter that has completely run out of control due to the exponential amplification compensation in the early stage, and complete the pure feature reshaping with extremely high signal-to-noise ratio from the very end of the algorithm architecture, realizing a dynamic balance between weak signal extraction and false alarm suppression.

[0133] The specific calculation process for the cable weak fault location and output step includes:

[0134] Applying the dynamic penalty factor to the modulated signal yields a pure fault characteristic, specifically calculated as follows:

[0135] In the formula, Indicates the final cable fault characteristic index, This represents the phase gradient characteristics after exponential modulation. Indicates a dynamic penalty factor;

[0136] The spatial physical distance is calculated by extracting the time node corresponding to the maximum value of the feature sequence. The specific calculation is as follows:

[0137] In the formula, This indicates the physical distance from the point of cable abnormality or minor fault to the test start point. This indicates the nominal high-frequency electromagnetic wave transmission speed of the coaxial cable. This represents the process of obtaining the final cable fault characteristic index. The variable at a specific time point when it reaches its maximum value; by directly applying the dynamic penalty factor derived from the cumulative information entropy, it is synchronously applied to the modulated phase gradient feature at the front end after inverse exponential amplification, and fused to form a cable fault feature index with high purity. Then, the specific time point corresponding to when the feature sequence reaches the global maximum value is extracted. Combined with the nominal high-frequency electromagnetic wave transmission speed in the coaxial cable, the actual spatial physical distance of the weak fault point from the test head end is accurately calculated. This achieves complete multi-dimensional data aggregation from complex frequency domain feature stripping to time domain energy modulation and then to precise physical mapping in the spatial domain, successfully solving the problem of locating and detecting actual physical hidden dangers in engineering.

[0138] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.

[0139] The above description is only a preferred embodiment of the present invention. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the technical principles of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.

Claims

1. A method for detecting coaxial cables used in network communication based on intelligent sensors, characterized in that, Includes the following steps: S1: Simultaneously inject a first frequency continuous test wave and a second frequency continuous test wave into the cable, and extract the first frequency reflected voltage signal and the second frequency reflected voltage signal respectively; S2: Map the first frequency reflected voltage signal to the complex analytic domain to obtain the first frequency analytic voltage signal; map the second frequency reflected voltage signal to the complex analytic domain to obtain the second frequency analytic voltage signal; S3: Calculate the product of the complex conjugate of the first frequency analytical voltage signal and the second frequency analytical voltage signal to obtain the transient cross power spectrum; calculate the product of the complex conjugate of the transient cross power spectrum and its own first-order time derivative; extract the imaginary part of the above product result to obtain the amplitude-weighted phase gradient; S4: Calculate the transient cumulative energy based on the first frequency analytical voltage signal and the second frequency analytical voltage signal, obtain the total test energy obtained when the integration reaches the end of the time, calculate the ratio of the transient cumulative energy to the total test energy, and use this ratio as the independent variable of the natural exponential function to obtain the adaptive exponential compensation weight; multiply the adaptive exponential compensation weight by the amplitude-weighted phase gradient to obtain the phase gradient characteristics after exponential modulation. S5: Extract the dimensionless transient feature probability density based on the exponentially modulated phase gradient features, calculate the natural logarithm of the dimensionless transient feature probability density, multiply the dimensionless transient feature probability density by the natural logarithm, and take the negative value of the product to obtain the local instantaneous information entropy; extract the transient cumulative entropy ratio based on the local instantaneous information entropy, obtain the directly set known system entropy sensitivity coefficient, calculate the system entropy sensitivity coefficient of the transient cumulative entropy ratio to the power of the power value, and subtract the power value from the number one to generate the dynamic penalty factor; S6: Multiply the dynamic penalty factor with the exponentially modulated phase gradient feature to obtain the final cable fault characteristic index; extract the specific time node corresponding to the final cable fault characteristic index reaching its maximum value, multiply the nominal high-frequency electromagnetic wave transmission speed of the coaxial cable with the specific time node and divide by the number two to output the physical distance of the fault point.

2. The method for detecting coaxial cables in network communication based on intelligent sensors according to claim 1, characterized in that, Step S2 specifically includes: Perform Hilbert principal value integral transform on the first frequency reflected voltage signal to construct the first frequency analytical voltage signal; The same analytic domain mapping operation is performed on the second frequency reflected voltage signal, that is, Hilbert principal value integral transform is performed to construct the second frequency analytic voltage signal.

3. The method for detecting coaxial cables in network communication based on intelligent sensors according to claim 1, characterized in that, The steps for calculating the transient accumulated energy include: The absolute amplitudes of the first frequency analytical voltage signal and the second frequency analytical voltage signal are obtained respectively; Calculate the sum of squares of the two absolute amplitudes to obtain the transient total energy density; The transient cumulative energy is obtained by performing time-axis integration on the total transient energy density.

4. The method for detecting coaxial cables in network communication based on intelligent sensors according to claim 1, characterized in that, The steps for extracting the dimensionless transient feature probability density include: Within the entire objective observation time window, extract the global maximum value of the absolute value of the exponentially modulated phase gradient feature; Extract the absolute value of the instantaneous exponentially modulated phase gradient feature; Divide the absolute value by the global maximum value and perform extreme value normalization to obtain the dimensionless transient characteristic probability density.

5. The method for detecting coaxial cables in network communication based on intelligent sensors according to claim 1, characterized in that, The steps for extracting the transient cumulative entropy percentage include: Perform time-axis integration on the local instantaneous information entropy within the current time range to obtain the integration result at the current moment; The local instantaneous information entropy is integrated over the entire objective observation time window to obtain the total time window integration result. Divide the current integral result by the total time window integral result to obtain the transient cumulative entropy ratio.

Images