Dynamic testing method and system for grounding resistance of communication infrastructure

Abstract

This invention relates to the field of data processing, and more specifically, to a method and system for dynamic testing of grounding resistance in communication infrastructure. The method includes: collecting historical multidimensional time-series data to construct a historical sample set; clustering the historical sample set according to operating conditions to obtain multiple clusters; pre-setting multiple candidate variational mode decomposition (MODED) numbers for each cluster; performing variational mode decomposition under each candidate MODED number; matching the decomposed voltage and current components; evaluating the separation effect based on the matching results; and determining the MODED number with the best separation effect as the optimal variational mode decomposition MODED number for that cluster; collecting real-time multidimensional time-series data, matching it to the corresponding cluster, and using the optimal MODED number corresponding to that cluster for variational mode decomposition and interference separation, outputting a true grounding resistance sequence. This invention can effectively eliminate the influence of electromagnetic interference and environmental drift, achieving high-precision, all-weather dynamic monitoring of grounding resistance.

Description

Technical Field

[0001] This invention relates to the field of data processing. More specifically, this invention relates to a method and system for dynamically testing the grounding resistance of communication infrastructure. Background Technology

[0002] The grounding system of communication infrastructure is a crucial component for ensuring equipment safety, personal safety, and communication stability. Grounding resistance, as a core indicator for measuring the lightning protection and electrical safety performance of facilities such as base stations and equipment rooms, is vital for timely detection of safety hazards and ensuring the reliable operation of communication networks through online dynamic monitoring. The loop method eliminates the need for long-distance auxiliary grounding electrodes and allows for continuous online measurement, making it suitable for long-term automated monitoring of communication base stations and equipment rooms.

[0003] However, existing dynamic grounding resistance testing methods based on the loop method are susceptible to electromagnetic interference in the communication environment and slow soil drift during the testing process, resulting in large fluctuations and low accuracy in the measurement results. They are unable to truly reflect the dynamic changes in grounding resistance and cannot meet the actual needs of accurate online monitoring around the clock. Summary of the Invention

[0004] To address the technical problem that existing dynamic testing methods for grounding resistance based on loop methods cannot accurately reflect the dynamic changes in grounding resistance, this invention provides solutions in the following aspects.

[0005] In the first aspect, the dynamic testing method for the grounding resistance of communication infrastructure includes: Collect historical multidimensional time-series data of communication infrastructure to construct a historical sample set. The multidimensional time-series data includes operating condition time-series data and grounding loop electrical time-series data. The historical sample set is clustered according to the operating conditions to obtain multiple clusters. For each cluster, multiple candidate variational mode decomposition mode numbers are preset. Variational mode decomposition is performed on the electrical timing data of the grounding circuit in the historical samples within each cluster under each candidate mode number. The decomposed voltage components and current components are matched, and the separation effect under each candidate mode number is evaluated based on the matching results. The candidate mode number with the best separation effect is determined as the optimal variational mode decomposition mode number for the corresponding cluster. Collect real-time multidimensional time-series data of communication infrastructure, match the real-time multidimensional time-series data to the corresponding cluster, use the optimal variational mode decomposition mode number corresponding to the cluster to perform variational mode decomposition and interference separation on the real-time grounding loop electrical time-series data, and output the real grounding resistance sequence.

[0006] Optionally, the operating condition time series data includes at least one of soil temperature time series, soil moisture time series, communication facility transmission power time series, test current frequency time series, and test current amplitude time series; the grounding circuit electrical time series data includes grounding voltage time series and grounding current time series.

[0007] Optionally, the historical sample set can be clustered according to working conditions to obtain multiple clusters, including: Normalize all time series data within each sampling window, and calculate the mean and standard deviation of the operating condition time series data for each sampling window. Combine the calculated mean and standard deviation to form an operating condition vector. K-means clustering was performed using the Euclidean distance between working condition vectors as a similarity metric, and the silhouette coefficient method was used to determine the optimal number of clusters, dividing the historical sample set into multiple clusters.

[0008] Optionally, matching the decomposed voltage components with the current components includes: For each historical sample, the multiple voltage components and multiple current components obtained from variational mode decomposition are paired according to the principle of the closest center frequency, so that each order obtains corresponding voltage and current components. For each paired component pair, calculate the absolute value of the difference between the center frequencies of its voltage and current components, and calculate an exponential function with the product of this negative absolute value and a preset adjustment coefficient as the exponent. The result is used as the matching degree of the component pair.

[0009] Optionally, the evaluation of the separation effect based on the matching results for each candidate modality number includes: For each historical sample, the effective confidence level of each order is calculated. The effective confidence level is obtained by fusing the matching degree of the component pairs matched at that order, the regularity of the voltage component, the regularity of the current component, and the correlation of the voltage component with the soil parameters. The order with the lowest effective confidence level is determined as the electromagnetic interference component, and the absolute value of the difference between the lowest effective confidence level and the second lowest effective confidence level is taken as the separation difference of the historical sample. The order with the lowest effective confidence is removed, and the remaining effective components are reconstructed to obtain the reconstructed voltage sequence and the reconstructed current sequence. The absolute value of the Pearson correlation coefficient between the reconstructed voltage sequence and the reconstructed current sequence is calculated as the reconstruction correlation of the historical sample. The separation effect of the historical sample is obtained by multiplying the separation difference of the historical sample by the reconstructed correlation.

[0010] Optionally, determining the number of candidate modes with the best separation effect as the optimal variational mode decomposition mode number for the corresponding cluster includes: For each cluster, the separation effect of all historical samples in the cluster under any candidate mode number is summed to obtain the comprehensive separation effect corresponding to the candidate mode number. The candidate mode number with the largest comprehensive separation effect is taken as the optimal variational mode decomposition mode number of the cluster.

[0011] Optionally, the output true ground resistance sequence includes: Construct a real-time operating condition vector, calculate the Euclidean distance between the real-time operating condition vector and the center vector of each cluster, and select the cluster with the closest Euclidean distance as the matching cluster; The optimal variational mode decomposition mode number corresponding to the matching cluster is retrieved, and variational mode decomposition is performed on the real-time ground voltage time series and ground current time series to obtain multiple voltage components and multiple current components, and the reconstructed voltage series and reconstructed current series are obtained. The ratio of the reconstructed voltage sequence to the reconstructed current sequence is used as the real-time true ground resistance value, and the true ground resistance sequence is output.

[0012] Secondly, a dynamic testing system for the grounding resistance of communication infrastructure includes: a processor and a memory, wherein the memory stores computer program instructions, and when the computer program instructions are executed by the processor, the dynamic testing method for the grounding resistance of the communication infrastructure described in any one of the claims is implemented.

[0013] The present invention has the following beneficial effects: 1. This invention can adaptively determine the optimal number of variational mode decomposition under different operating conditions based on historical operating data of communication infrastructure, and dynamically match the current operating condition to use the corresponding optimal number of modes for interference separation in real-time testing. Thus, it can effectively eliminate electromagnetic interference and environmental drift under different soil environments, transmission power and other operating conditions, and output a true grounding resistance sequence, which significantly improves the accuracy of dynamic grounding resistance testing and the reliability of all-weather online monitoring.

[0014] 2. By matching the center frequencies of voltage and current components and identifying interference based on effective confidence, this invention can accurately distinguish between valid signals and electromagnetic interference, avoid mode aliasing and false deletion of valid signals, and make the reconstructed voltage and current sequences after separation more in line with the laws of electrical physics, thereby further improving the accuracy of the calculation of the actual grounding resistance. Attached Figure Description

[0015] Figure 1 This is a flowchart of steps S1-S3 in the dynamic testing method for grounding resistance of communication infrastructure according to an embodiment of the present invention.

[0016] Figure 2This is a flowchart of the method for clustering historical sample sets and adaptively optimizing the optimal VMD mode number of each cluster in the dynamic testing method for grounding resistance of communication infrastructure according to an embodiment of the present invention.

[0017] Figure 3 This is a structural block diagram of the dynamic testing system for grounding resistance of communication infrastructure according to an embodiment of the present invention. Detailed Implementation

[0018] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are some embodiments of the present invention, but not all embodiments.

[0019] Reference Figure 1 The dynamic testing method for grounding resistance of communication infrastructure includes steps S1-S3, as detailed below: S1: Collect multi-dimensional time-series data of the communication infrastructure in order to construct a historical sample set.

[0020] Electromagnetic interference and environmental drift mixed in the grounding resistance test signal are closely related to the operating conditions of communication facilities, such as soil temperature and humidity, and transmission power. The characteristics of interference vary significantly under different operating conditions. Data from a single operating condition cannot fully reflect these changing patterns. Therefore, it is necessary to collect historical data covering multiple operating conditions to construct a historical sample set that is representative of the operating conditions.

[0021] A loop-based grounding resistance online testing terminal is deployed on the grounding loop of a communication infrastructure, such as an outdoor macro base station. The sampling frequency is set to 1 kHz, and the sampling window duration is 1 second. Under this configuration, a set of 1000 time-series data points collected in each sampling window is defined as a historical sample.

[0022] Specifically, within each sampling window, two types of time-series data are collected simultaneously: Operating condition timing data includes soil temperature and soil moisture timing sequences collected by digital temperature and humidity sensors buried next to the grounding electrode; communication facility transmission power timing sequences read by the base station operation management center; and test current frequency and test current amplitude timing sequences obtained from the signal source module inside the test terminal.

[0023] Grounding circuit electrical timing data: including the grounding voltage timing sequence acquired by the differential voltage sampling unit, and the grounding current timing sequence acquired by the clamp-on current transformer.

[0024] By repeating the above collection process, data from 10,000 sampling windows can be collected, thus obtaining 10,000 historical samples.

[0025] Furthermore, standard Min-Max normalization is performed on all time-series data within each historical sample.

[0026] Then, for each historical sample's operating condition time series—namely, soil temperature, soil moisture, transmission power, test current frequency, and test current amplitude—the mean and standard deviation were calculated. The five calculated means and five standard deviations were combined to form a 10-dimensional operating condition vector. The mean characterizes the overall level of the parameter within the sampling window, while the standard deviation characterizes the degree of parameter fluctuation; together, they describe the operating condition at the sampling time.

[0027] Ultimately, each historical sample contains: a 10-dimensional operating condition vector, normalized operating condition timing data, and normalized grounding loop electrical timing data.

[0028] The final 10,000 historical samples are aggregated to form a historical sample set.

[0029] S2: Perform working condition clustering on the historical sample set to obtain all clusters, and adaptively optimize the optimal number of VMD modes for each cluster.

[0030] In grounding resistance testing, the original voltage and current signals are a mixture of effective components, electromagnetic interference, and environmental drift. VMD (Variational Mode Decomposition) can adaptively decompose the mixed signal into multiple narrowband components with different center frequencies, thus facilitating the identification and separation of interference.

[0031] However, the interference characteristics and the frequency band distribution of the effective components differ significantly under different operating conditions. If a uniform VMD mode number is used for all operating conditions, aliasing of interference and effective signal modes may occur under some conditions, or the effective signal may be mistakenly deleted.

[0032] Therefore, it is necessary to first cluster the historical samples according to the working conditions, and then, for each working condition, analyze the signal separation effect under different decomposition levels to select the VMD mode number with the best separation effect.

[0033] Reference Figure 2 The process of clustering historical sample sets and adaptively optimizing the optimal VMD modality number for each cluster includes steps S20-S23: S20: Perform K-means clustering on the working condition vectors of the historical sample set to obtain all clusters.

[0034] Since the signal characteristics differ significantly under different operating conditions, it is necessary to first divide the historical samples into several categories based on the similarity of operating conditions, so that the number of decomposition layers can be optimized separately for each category of operating conditions in the future.

[0035] Specifically, K-means clustering is performed using the Euclidean distance between the condition vectors of any two historical samples in the historical sample set as a similarity measure. The silhouette coefficient method is used to evaluate the clustering quality and determine the optimal number of clusters K.

[0036] The historical sample set is then divided into K clusters, with each cluster containing historical samples representing a similar type of communication and soil conditions.

[0037] S21: Preset the number of candidate VMD modes, traverse all candidate VMD modes for each cluster, perform decomposition and component matching, and construct a unique sample subset under each candidate mode.

[0038] To find the optimal number of modes for each cluster, all historical samples within that cluster need to be decomposed using VMD at multiple candidate number of modes, and the decomposed voltage and current components need to be matched and aligned according to frequency to form a specific subset for subsequent evaluation.

[0039] Specifically, a candidate VMD modality set is predefined, for example... Taking the first cluster, which can be labeled as cluster 1, as an example: First, select a value, such as 3, from the above set of candidate VMD modal numbers as the target modal number.

[0040] Next, all historical samples within cluster 1 are traversed, and a 3-level VMD decomposition is performed on the ground voltage and ground current time series of each historical sample. After decomposition, each historical sample yields 3 voltage components and 3 current components.

[0041] Since the physical orders of the components decomposed by the VMD may not be aligned, the voltage and current components need to be paired according to the principle of the closest center frequency.

[0042] That is, for any pair of components, i.e., any voltage component and any current component, calculate the difference in their center frequencies. Physically, if the voltage and current components originate from the same physical process, i.e., the change in actual grounding resistance, their center frequencies should be basically the same. Therefore, the difference in center frequencies can measure the degree of their common origin, i.e., satisfying the following relationship: In the formula, Indicates the first The degree of matching of component pairs, Indicates the first The difference in center frequency between the voltage component and the current component in the component pair. This is a preset adjustment coefficient used to control the sensitivity of the matching degree to the center frequency difference.

[0043] when The smaller, The closer the match is to 1, the higher the degree of matching. All component pairs are iterated through, and the pair with the smallest center frequency difference is selected as the matching pair. Matching continues from the remaining components until all voltage and current components are uniquely paired. After matching is complete, each historical sample at each order yields corresponding voltage and current components.

[0044] After performing the above operations on all historical samples in cluster 1, a dedicated sample subset of cluster 1 under the 3-layer VMD decomposition is formed. This dedicated sample subset contains the matched voltage component set, current component set, and normalized operating condition time series data.

[0045] Repeat the above steps for the remaining modalities in the candidate set to generate a dedicated sample subset for cluster 1 corresponding to each candidate modality. Other clusters are processed in the same way to generate their own dedicated sample subsets for different candidate modalities.

[0046] S22: Calculate the separation performance of each historical sample in the dedicated sample subset for each cluster under each candidate modality number.

[0047] After obtaining the specific subset of samples for each candidate modal number, it is necessary to quantify the effectiveness of each historical sample in separating interference signals under that candidate modal number, so as to compare which candidate modal number is better in the future.

[0048] Taking any historical sample from the exclusive sample subset of cluster 1 under the 3-layer VMD decomposition as an example, perform the following operations: First, for any pair of orders that has been paired, such as voltage component 1 and current component 2, construct their Hankel matrices and perform singular value decomposition. Calculate the ratio of the maximum singular value to the sum of all singular values ​​to obtain the regularity of the voltage component and the regularity of the current component.

[0049] Secondly, the mutual information method was used to calculate the correlation between any order voltage component and the soil temperature and soil moisture time series of the historical sample, respectively. These two correlations were normalized and averaged to obtain the soil parameter correlation of the voltage component. This soil parameter correlation reflects the degree to which the voltage component changes with soil temperature and humidity. A larger value indicates a stronger correlation between the voltage component and changes in the soil environment, and it is more likely to be an effective component reflecting changes in the actual grounding resistance; a smaller value indicates a greater likelihood of being unrelated to electromagnetic interference.

[0050] Next, the matching degree of the component pairs calculated above, the regularity of the voltage component and the regularity of the current component in the matching pairs, and the correlation of the voltage component with the soil parameters are fused together to calculate the effective confidence level of the historical sample at any order, which satisfies the following relationship: In the formula, For the first The first historical sample Effective confidence level of order For the first The order corresponds to the matching degree of the matching pair. For the first The first historical sample The regularity of order voltage components, For the first The first historical sample The regularity of order current components, For the first The first historical sample Correlation of soil parameters with order voltage components.

[0051] The above It integrates the regularity of voltage and current, through the analysis of... and The geometric mean is used because the order is more likely to be a valid signal only when both factors are highly regular. If only one factor is regular while the other is disordered, the geometric mean will be significantly reduced.

[0052] Then traverse the first For each historical sample, all orders are used to obtain the corresponding effective confidence scores. The order with the lowest effective confidence score is identified as an electromagnetic interference component and is removed. The remaining orders are retained as effective components.

[0053] Furthermore, the absolute value of the difference between the minimum effective confidence level (i.e., the effective confidence level corresponding to the electromagnetic interference component) and the second minimum effective confidence level is used as the separation difference of the historical sample. This separation difference measures the confidence gap between the electromagnetic interference component and the closest effective component. The larger the confidence gap, the more significantly the confidence level of the interference component is lower than that of the effective component, the easier it is to distinguish between the two, and the weaker the mode mixing. The smaller the difference, the more similar the confidence levels of the interference and effective components are, making them difficult to distinguish.

[0054] Furthermore, the voltage components corresponding to all the retained effective components are linearly superimposed to obtain the reconstructed voltage sequence. Similarly, the current components are linearly superimposed to obtain the reconstructed current sequence. The Pearson correlation coefficient between the reconstructed voltage and current sequences is calculated, and the absolute value is taken as the reconstruction correlation of the historical sample. The reconstruction correlation of this historical sample reflects the strength of the linear correlation between the reconstructed voltage and current sequences. According to Ohm's law, voltage and current should change proportionally and synchronously when the grounding resistance changes. The closer the reconstruction correlation of this historical sample is to 1, the more the reconstructed sequence conforms to the laws of electrical physics, and the more thoroughly the interference is eliminated.

[0055] Finally, the separation difference of the historical sample is multiplied by the reconstructed correlation to obtain the separation effect of the historical sample. Only when the separation difference is large and the reconstruction result conforms to physical laws, i.e., the reconstruction correlation is high, will the separation effect be close to 1. This ensures that the evaluation of the separation effect considers both statistical separability and physical rationality.

[0056] For all historical samples in the dedicated sample subset of cluster 1 under the 3-layer VMD decomposition, repeat the above process to obtain the separation effect of each historical sample. Then, for the dedicated sample subset of each candidate mode number of all clusters, calculate the separation effect of each historical sample according to the above operation.

[0057] S23: Determine the optimal number of VMD modes for each cluster based on the calculated separation effect.

[0058] For cluster 1, the separation performance of all historical samples under any candidate mode number is summed to obtain the comprehensive separation performance corresponding to that candidate mode number. The candidate mode number with the largest comprehensive separation performance is taken as the optimal mode number for cluster 1. Similarly, the optimal mode number for other clusters can be obtained.

[0059] S3: Collect real-time multi-dimensional time-series data of communication infrastructure, match the multi-dimensional time-series data to the corresponding cluster, perform adaptive decomposition and interference separation, and output the real grounding resistance sequence.

[0060] In real-time operation, the currently collected operating condition data can usually be categorized into a similar operating condition from historical samples. Therefore, dynamically selecting the optimal mode number that best matches the current operating condition can more accurately decompose the real-time signal and separate interference, thereby obtaining a grounding resistance value that is close to the true value.

[0061] Using the same hardware configuration and sampling parameters as S1 above, collect operating condition data and grounding loop electrical data for a new real-time sampling window, and construct a 10-dimensional operating condition vector.

[0062] Calculate the Euclidean distance between the real-time operating condition vector and the center vectors of each cluster in S2 above, and select the cluster with the closest Euclidean distance as the matching cluster. For example, if cluster 1 is matched, then retrieve its corresponding optimal VMD mode number.

[0063] Furthermore, the real-time ground voltage time series and ground current time series are decomposed using the corresponding optimal VMD mode number, and the above-mentioned operation S22 is performed to obtain the reconstructed voltage time series and the reconstructed current time series.

[0064] Furthermore, the reconstructed voltage time series and the reconstructed current time series are inversely transformed back to their original physical dimensions.

[0065] Finally, the ratio of voltage to current after the inverse transformation is used as the real-time grounding resistance value. This yields the real grounding resistance value at each moment within the real-time sampling window, resulting in a continuous real-time grounding resistance sequence that eliminates the effects of electromagnetic interference and environmental drift.

[0066] Through the above S1-S3, the embodiments of the present invention obtain real-time dynamic test results of the grounding resistance of communication infrastructure, namely, real-time real grounding resistance sequence, from historical data modeling, working condition clustering, modality number optimization to real-time signal acquisition, matching, and decomposition. This real grounding resistance sequence can truly reflect the dynamic changes in grounding resistance caused by soil environment changes, grounding body aging, etc., effectively eliminating the influence of electromagnetic interference and environmental drift, thereby realizing high-precision, all-weather online dynamic monitoring of the grounding resistance of communication infrastructure.

[0067] This invention also provides a dynamic testing system for the grounding resistance of communication infrastructure. For example... Figure 3 As shown, the system includes a processor and a memory, the memory storing computer program instructions, which, when executed by the processor, implement a dynamic testing method for the grounding resistance of a communication infrastructure according to the first aspect of the present invention.

[0068] The system also includes other components well known to those skilled in the art, such as communication buses and communication interfaces, the settings and functions of which are known in the art and will not be described in detail here.

[0069] It should be noted that those skilled in the art can make various modifications and improvements without departing from the inventive concept, and these all fall within the scope of protection of this invention. Therefore, the scope of protection of this patent should be determined by the appended claims.

Claims

1. A method for dynamically testing the grounding resistance of communication infrastructure, characterized in that, include: Collect historical multidimensional time-series data of communication infrastructure to construct a historical sample set. The multidimensional time-series data includes operating condition time-series data and grounding loop electrical time-series data. The historical sample set is clustered according to the operating conditions to obtain multiple clusters. For each cluster, multiple candidate variational mode decomposition mode numbers are preset. Variational mode decomposition is performed on the electrical timing data of the grounding circuit in the historical samples within each cluster under each candidate mode number. The decomposed voltage components and current components are matched, and the separation effect under each candidate mode number is evaluated based on the matching results. The candidate mode number with the best separation effect is determined as the optimal variational mode decomposition mode number for the corresponding cluster. Collect real-time multidimensional time-series data of communication infrastructure, match the real-time multidimensional time-series data to the corresponding cluster, use the optimal variational mode decomposition mode number corresponding to the cluster to perform variational mode decomposition and interference separation on the real-time grounding loop electrical time-series data, and output the real grounding resistance sequence.

2. The method for dynamic testing of grounding resistance of communication infrastructure according to claim 1, characterized in that, The operating condition timing data includes at least one of soil temperature timing sequence, soil moisture timing sequence, communication facility transmission power timing sequence, test current frequency timing sequence, and test current amplitude timing sequence; the grounding circuit electrical timing data includes grounding voltage timing sequence and grounding current timing sequence.

3. The method for dynamic testing of grounding resistance of communication infrastructure according to claim 1, characterized in that, The historical sample set was clustered according to working conditions to obtain multiple clusters, including: Normalize all time series data within each sampling window, and calculate the mean and standard deviation of the operating condition time series data for each sampling window. Combine the calculated mean and standard deviation to form an operating condition vector. K-means clustering was performed using the Euclidean distance between working condition vectors as a similarity metric, and the silhouette coefficient method was used to determine the optimal number of clusters, dividing the historical sample set into multiple clusters.

4. The method for dynamic testing of grounding resistance of communication infrastructure according to claim 1, characterized in that, The process of matching the decomposed voltage components with the current components includes: For each historical sample, the multiple voltage components and multiple current components obtained from variational mode decomposition are paired according to the principle of the closest center frequency, so that each order obtains corresponding voltage and current components. For each paired component pair, calculate the absolute value of the difference between the center frequencies of its voltage and current components, and calculate an exponential function with the product of this negative absolute value and a preset adjustment coefficient as the exponent. The result is used as the matching degree of the component pair.

5. The method for dynamic testing of grounding resistance of communication infrastructure according to claim 4, characterized in that, The evaluation of the separation performance for each candidate modality number based on the matching results includes: For each historical sample, the effective confidence level of each order is calculated. The effective confidence level is obtained by fusing the matching degree of the component pairs matched at that order, the regularity of the voltage component, the regularity of the current component, and the correlation of the voltage component with the soil parameters. The order with the lowest effective confidence level is determined as the electromagnetic interference component, and the absolute value of the difference between the lowest effective confidence level and the second lowest effective confidence level is taken as the separation difference of the historical sample. The order with the lowest effective confidence is removed, and the remaining effective components are reconstructed to obtain the reconstructed voltage sequence and the reconstructed current sequence. The absolute value of the Pearson correlation coefficient between the reconstructed voltage sequence and the reconstructed current sequence is calculated as the reconstruction correlation of the historical sample. The separation effect of the historical sample is obtained by multiplying the separation difference of the historical sample by the reconstructed correlation.

6. The method for dynamic testing of grounding resistance of communication infrastructure according to claim 1, characterized in that, The step of determining the number of candidate modes with the best separation effect as the optimal variational mode decomposition mode number for the corresponding cluster includes: For each cluster, the separation effect of all historical samples in the cluster under any candidate mode number is summed to obtain the comprehensive separation effect corresponding to the candidate mode number. The candidate mode number with the largest comprehensive separation effect is taken as the optimal variational mode decomposition mode number of the cluster.

7. The method for dynamic testing of grounding resistance of communication infrastructure according to claim 1, characterized in that, The output true ground resistance sequence includes: Construct a real-time operating condition vector, calculate the Euclidean distance between the real-time operating condition vector and the center vector of each cluster, and select the cluster with the closest Euclidean distance as the matching cluster; The optimal variational mode decomposition mode number corresponding to the matching cluster is retrieved, and variational mode decomposition is performed on the real-time ground voltage time series and ground current time series to obtain multiple voltage components and multiple current components, and the reconstructed voltage series and reconstructed current series are obtained. The ratio of the reconstructed voltage sequence to the reconstructed current sequence is used as the real-time true ground resistance value, and the true ground resistance sequence is output.

8. A dynamic testing system for the grounding resistance of communication infrastructure, characterized in that, include: A processor and a memory, the memory storing computer program instructions that, when executed by the processor, implement the dynamic testing method for grounding resistance of a communication infrastructure according to any one of claims 1-7.

Images