Variable-frequency potential measurement system for intelligent identification of alternating-current and direct-current interference of gas pipeline

Abstract

The invention belongs to the technical field of variable-frequency potential measurement, and provides a variable-frequency potential measurement system for intelligent identification of alternating-current and direct-current interference of a gas pipeline, and the system comprises a data measurement module which applies a variable-frequency current excitation signal of a preset frequency spectrum to the gas pipeline and is based on measurement terminals disposed along the gas pipeline. By introducing a purity dynamic evaluation mechanism based on signal-to-noise ratio and spectrum coherence and combining a self-adaptive signal separation strategy, rapid and accurate identification of alternating-current and direct-current interference of the gas pipeline is realized, the availability of an external reference signal is intelligently judged, the signal is directly used as a high-value guide when the signal is pure, and the reliability of the system is improved. And when the signal is impure, a reinforced guide signal is generated through pipeline signal subspace projection, and interference core features are effectively extracted, so that the speed and precision of interference source separation and contribution degree evaluation are improved in a complex multi-source mixed interference environment, and a reliable basis is provided for engineering protection.

Description

Technical Field

[0001] This invention belongs to the field of variable frequency potential measurement technology, specifically a variable frequency potential measurement system for intelligent identification of AC / DC interference in gas pipelines. Background Technology

[0002] With the intensive use of urban underground space, gas pipelines are often laid near or across high-voltage power transmission lines, subway tracks, electrified railways and other high-power facilities. The resulting stray current interference has become one of the main risks threatening the safe operation of pipelines. These interferences are usually divided into DC interference and AC interference, with complex sources and often coexisting, forming a multi-source strong mixed interference scenario. In order to effectively implement protection, it is necessary to accurately identify the type, source and contribution of interference. Variable frequency potential measurement technology injects test signals of different frequencies into the pipeline and measures its response. However, in complex multi-source interference environments, traditional frequency conversion measurement and analysis methods have some shortcomings. First, existing methods lack an effective real-time evaluation mechanism for the quality of field measurement data, especially the purity of key reference signals. If contaminated or insufficiently representative reference signals are used directly, for example, signals collected from subway return networks may have been mixed with power frequency harmonics, which will seriously mislead subsequent signal separation and identification algorithms, leading to errors in source feature extraction. Second, in the face of dynamically changing and superimposed mixed interference, traditional methods are unable to quickly and accurately separate components and quantify contributions, directly resulting in the inability to accurately identify the dominant interference source in engineering, which in turn leads to serious consequences such as weak protective measures, low investment efficiency, or even protection failure. For example, due to misjudgment, mismatched drainage devices may be installed incorrectly, which not only fails to alleviate corrosion but may also exacerbate risks or cause new safety hazards. Therefore, the present invention provides a frequency conversion potential measurement system for intelligent identification of AC / DC interference in gas pipelines. Summary of the Invention

[0003] In order to overcome the shortcomings of the prior art, at least one technical problem raised in the background art is solved.

[0004] The technical solution adopted by this invention to solve its technical problem is: a frequency conversion potential measurement system for intelligent identification of AC / DC interference in gas pipelines, comprising: Data measurement module: Apply a frequency-converted current excitation signal with a preset spectrum to the gas pipeline. Based on the measurement terminals deployed along the gas pipeline, simultaneously collect the potential response signals of at least two measurement points under frequency-converted excitation as measurement data, as well as external reference signals, and evaluate the purity of the external reference signals to obtain the purity assessment results. Dynamic strategy selection module: Based on the purity evaluation results, convolutional blind source separation is introduced to separate the mixed signals of the frequency conversion measurement response data, and obtain the independent interference source component signals decoupled from the mixed measurement data; High purity processing unit: If the purity is high, signal separation is performed directly guided by an external reference signal; Low purity processing unit: If the purity is low, a gas pipeline signal subspace is constructed based on the signals from multiple measurement points, and an external reference signal is projected onto the gas pipeline signal subspace to generate an enhanced guidance signal. The enhanced guidance signal is then used as a guide for signal separation. Interference source component evaluation module: Identifies the interference source type of independent interference source component signals, obtains the interference source identification results, and evaluates the contribution of the interference source identification results based on the frequency conversion potential measurement response characteristics and signal energy distribution, and obtains an interference source contribution ranking table.

[0005] The beneficial effects of this invention are as follows: This invention introduces a purity dynamic evaluation mechanism based on signal-to-noise ratio and spectral coherence, combined with an adaptive signal separation strategy, to achieve rapid and accurate identification of AC / DC interference in gas pipelines. It also intelligently judges the availability of external reference signals, directly using them as high-value guidance when the signals are pure, and generating enhanced guidance signals through pipeline signal subspace projection when the signals are impure. This effectively extracts the core features of the interference, thereby improving the speed and accuracy of interference source separation and contribution assessment in complex multi-source mixed interference environments, providing a reliable basis for engineering protection. This invention replaces the traditional instantaneous mixing model with a convolutional blind source separation model, characterizing the time delay, attenuation, and multipath effects of interference current propagation in the soil-pipe medium. This achieves signal decoupling that is closer to physical reality. Combined with subsequent multi-dimensional feature intelligent identification and quantitative evaluation, it can not only accurately determine the type of interference source, but also calculate the energy contribution and equivalent interference current density, generating an interference source contribution ranking table, providing a scientific basis for precise protection decisions. Attached Figure Description

[0006] The invention will now be further described with reference to the accompanying drawings.

[0007] Figure 1 This is a framework diagram of a frequency conversion potential measurement system for intelligent identification of AC / DC interference in gas pipelines according to the present invention; Figure 2 This is a flowchart of the steps of a variable frequency potential measurement method for intelligent identification of AC / DC interference in gas pipelines according to the present invention; Figure 3 This is a partial flowchart of a variable frequency potential measurement method for intelligent identification of AC / DC interference in gas pipelines according to the present invention. Detailed Implementation

[0008] To make the technical means, creative features, objectives and effects of this invention easier to understand, the invention will be further described below in conjunction with specific embodiments.

[0009] Example 1 One of the core inventive points of this invention is that, in response to the shortcomings of the current technology, a frequency conversion potential measurement system for intelligent identification of AC / DC interference in gas pipelines is proposed, which optimizes the accuracy and reliability of interference signal separation, thereby realizing accurate identification and contribution assessment of AC / DC interference sources in gas pipelines under complex real-world environments. Please see Figure 1 As shown in the embodiment of the present invention, a frequency conversion potential measurement system for intelligent identification of AC / DC interference in gas pipelines is provided, wherein the system specifically includes: Data measurement module: Apply a frequency-converted current excitation signal with a preset spectrum to the gas pipeline. Based on the measurement terminals deployed along the gas pipeline, simultaneously collect the potential response signals of at least two measurement points under frequency-converted excitation as measurement data, as well as external reference signals, and evaluate the purity of the external reference signals to obtain the purity assessment results. Specifically, the measuring terminal synchronously injects a set of frequency conversion current excitation signals with a preset spectrum into the metal body of the gas pipeline connected to it; While the frequency conversion current excitation signal is applied, each measurement terminal begins to acquire the pipeline potential response signal and the external reference signal; Among them, the pipeline potential response signal acquisition is carried out through a high-impedance input voltage acquisition channel, and the pipeline potential response signal relative to the buried long-term reference electrode is measured simultaneously at at least two different test pile locations. The external reference signal is acquired synchronously through an independent acquisition channel, with at least one signal from the suspected interference source system or its key nodes used as the external reference signal. The acquired raw signals are preprocessed, and the purity of the external reference signal is evaluated. The process is as follows: Among them, the purity assessment adopts the method of time-frequency analysis and correlation calculation; Specifically, the process of using time-frequency analysis is as follows: Short-time Fourier transforms are performed on the external reference signal and all pipe potential response signals respectively to convert the time-domain signals into time-frequency matrix, thereby obtaining the energy distribution of each signal in the time-frequency domain; Based on prior knowledge or peak detection of the pipeline potential response signal spectrum, the characteristic frequency bands of one or more target interference sources are determined; For example, if the goal is to identify subway interference, the characteristic frequency band is preset to be a narrow band range that includes the characteristic harmonics of the subway traction system (such as the 16.7Hz fundamental wave and its odd harmonics). Calculate the ratio of signal power to noise power in the characteristic frequency band of the target interference source, and use it as the signal-to-noise ratio value of the target frequency band; The sum of squares of amplitudes at all time-frequency points within the characteristic frequency band of the target interference source, or the average power spectral density of the characteristic frequency band of the target interference source, is taken as the signal power. Select one or more quiet frequency bands that are adjacent to the characteristic frequency band of the target interference source and have no significant spectral peaks, and calculate the average power spectral density as the noise power; Specifically, the correlation calculation process is as follows: For the external reference signal and each pipe potential response signal, calculate the amplitude squared coherence estimate between them within the effective frequency band covering the main interference energy; The amplitude squared coherence estimates obtained from all pipeline measurement points are arithmetically averaged in the frequency domain to obtain the average spectral coherence coefficient. If the target frequency band signal-to-noise ratio is greater than the first preset threshold and the average spectral coherence coefficient is greater than the second preset threshold at the same time, the purity of the external reference signal is determined to be high. Otherwise, the purity of the external reference signal is determined to be low; The first preset threshold is set based on the minimum signal-to-noise ratio required for reliable identification of target features; the second preset threshold is set based on the minimum coherence coefficient required for the signal and interference to have a significant correlation.

[0010] This module performs real-time purity assessment of external reference signals. By calculating the signal-to-noise ratio of the target frequency band and the spectral coherence with the pipeline signal, it can determine whether the reference signal is qualified to be used as a guide tag. This avoids systematic errors in the entire analysis link caused by using low-quality reference signals and is a prerequisite analysis for the reliable operation of this embodiment.

[0011] Dynamic strategy selection module: Based on the purity evaluation results, convolutional blind source separation is introduced to separate the mixed signals of the frequency conversion measurement response data, and obtain the independent interference source component signals decoupled from the mixed measurement data; High purity processing unit: If the purity is high, signal separation is performed directly guided by an external reference signal; Low purity processing unit: If the purity is low, a gas pipeline signal subspace is constructed based on the signals from multiple measurement points, and an external reference signal is projected onto the gas pipeline signal subspace to generate an enhanced guidance signal. The enhanced guidance signal is then used as a guide for signal separation. It should be noted that the traditional instantaneous mixed blind source separation assumes that the signal is instantaneously and linearly mixed. However, in actual use, when the interference current propagates from the source (such as a subway traction substation) to different measurement points in the pipeline, it will experience time delay, attenuation, and convolution effects due to uneven soil resistivity distribution, differences in pipeline corrosion protection layer conditions, and different electrical lengths of different propagation paths. This violates the instantaneous mixing assumption and leads to a decrease in separation performance. The reason for adopting the convolution blind source separation method is that its model matches the physical process of interference current propagation in the actual soil-pipeline complex medium. The convolution model, by introducing a mixing filter, describes the signal delay, attenuation, and convolution mixing phenomena caused by uneven soil resistivity, differences in corrosion protection layer, and different path lengths. This overcomes the problem of decreased separation performance caused by model mismatch and achieves more accurate separation that is closer to physical reality. The reason for generating an enhanced guidance signal when the purity is low is that even if the initially acquired external reference signal is not pure enough due to the mixing of other interferences, it can still filter out noise and local components that are not related to the interference of the pipeline by projecting it onto the pipeline signal subspace, extract the core features that are strongly related to the actual interference of the pipeline, improve the targeting of the guidance signal, so that the subsequent guidance separation algorithm can avoid being misled and more accurately and stably separate the target interference source components from the complex mixed signal. Specifically, if the purity of the external reference signal is determined to be high, the acquired external signal is directly used as the guiding signal to perform convolutional blind source separation. The process is as follows: Employing a convolutional hybrid model: Where t represents a continuous-time variable, This represents the mixed signal vector observed at time t, i.e., the actual measured data, and is an M×1 column vector. Let represent the vector of the unknown source signal at time t, which is an N×1 column vector. This represents the discrete-time delay index, where L represents the maximum memory length or maximum delay of the hybrid filter. Indicates time delay The M×N dimensional hybrid filter matrix at the location, This represents the additive noise vector at time t; It should be noted that the convolutional mixture model indicates that at any time t, the potential signal measured at the i-th pipe measuring point is... It consists of N different interference source signals. In their respective propagation paths (manifesting as different delays) and decay weight The signal components transmitted from above, plus local noise The result obtained by superposition (convolutional mixing) more realistically reflects the propagation behavior of interference current in complex underground environments than the instantaneous mixing model, and is the basis for achieving accurate separation by those skilled in the art; During the solution process, the time-domain signal is transformed into the frequency domain, and the observed signal is... and guidance signal Performing a piecewise windowed short-time Fourier transform yields the frequency domain representation of the observed signal. Frequency domain representation of the guiding signal Where f is the frequency index and n is the time frame index; Within the characteristic frequency band of the target interference source, the frequency domain representation of the guiding signal is used. Based on prior information highly correlated with a desired source signal, constraints are constructed. A fast convolutional blind source separation algorithm, such as a frequency domain method based on joint diagonalization, is then employed. The steps are as follows: Calculate the frequency domain representation of the observed signal Covariance matrix over multiple time frames; Frequency domain representation of the pilot signal For reference, the separation filter can be estimated using generalized eigenvalue decomposition or adaptive filtering techniques. ,in, The separation filter matrix or demixing matrix at the discrete frequency point f is the approximate inverse or compensation of the mixed filter sequence in the frequency domain in the time-domain convolutional mixing model. The separated frequency domain signal The components are as independent as possible, and one of the components and Strong correlation; The separated frequency domain signal The signal is reconstructed into independent interference source components in the time domain through inverse short-time Fourier transform. ,in, This represents the set of independent interference source signals separated from the mixed signal, where N represents the total number of independent interference source components estimated by the signal separation method of this invention.

[0012] Specifically, if the purity of the external reference signal is determined to be low, an enhanced guiding signal is generated through analysis and processing. This enhanced guiding signal is then used as the guiding signal to perform convolutional blind source separation. The process is as follows: Representing the observed signal in the time domain Zero-mean preprocessing is performed to obtain the preprocessed observation signal matrix; Based on the preprocessed observation signal matrix, a signal subspace characterizing the main interference modes of the gas pipeline is constructed. Calculate the time-domain representation of the observed signal The covariance matrix is ​​obtained, and eigenvalue decomposition is performed on the covariance matrix; Based on the preset cumulative variance contribution rate threshold (e.g., 0.95), select the eigenvectors corresponding to the top K largest eigenvalues, where K is the smallest integer that satisfies the condition that the cumulative variance contribution rate is greater than or equal to the cumulative variance contribution rate threshold. The space spanned by the K eigenvectors is the gas pipeline signal subspace. External reference signals that are determined to have low purity are subjected to the same zero-mean processing to obtain preprocessed external reference signals. The preprocessed external reference signals are then projected onto the gas pipeline signal subspace to extract components that are strongly correlated with the interference experienced by the gas pipeline. Enhanced guidance signal Obtained through projection calculations. ,in, This represents the preprocessed external reference signal. Let M×K dimensional matrix represent the column vectors that form a set of orthonormal bases for the gas pipeline signal subspace. Representation matrix Transpose of; It should be noted that the above projection operation realizes the signal purification process, utilizing the main feature patterns learned from the gas pipeline's own response. As a filter, it filters out impurities unrelated to pipeline interference from the impure external reference signal, and outputs a higher quality enhanced guiding signal that is more suitable for guiding subsequent convolution blind source separation. Using the enhanced guiding signal as semi-supervised information, convolutional blind source separation is performed on the preprocessed observation signal matrix; The same convolutional mixing model and convolutional blind source separation algorithm framework based on fast frequency domain approximation are used in the high purity processing flow. It is important to note that, unlike the high-purity process where the pure reference signal is used as a strong constraint, the low-purity process uses the enhanced guide signal as the initialization vector for algorithm iteration or as a soft constraint term in the objective function (for example, adding a term to the objective function of independent component analysis to maximize the correlation coefficient between a certain output component and the enhanced guide signal). The different functions of the above processing are: to use the effective information in the guiding signal to guide the search direction, while tolerating possible residual uncertainties, thereby improving the robustness of the separation process under low-quality guidance; After algorithm iteration, the signal of independent interference source components is output.

[0013] The core function of this module is to transform measurement data into clearly identifiable interference source components, fundamentally optimizing the problem of accurate separation of multi-source interference in complex environments. Specifically, this is reflected in: Firstly, the purity assessment results of the aforementioned modules are used as key decisions to flexibly address inconsistent signal quality. When the reference signal is pure, high-value guiding information is fully utilized; when the reference signal is impure, an intelligent enhancement mechanism is used, thereby ensuring that a relatively optimal separation process can be initiated under any actual working condition, improving engineering applicability and reliability. Secondly, by introducing a convolutional blind source separation model and method, the time delay, attenuation and multipath effects of interference current propagation in the complex medium of soil-pipe are described, thereby reducing the problem of degraded separation performance. The interference source component signal separated in this way is closer to the real interference source emission waveform, which provides a basis for accurate data in the future. Thirdly, in the case of low purity, by constructing a pipeline signal subspace and projecting and enhancing the reference signal, the core features that are truly related to the interference of the pipeline can be extracted from the contaminated reference signal, and an enhanced guidance signal can be generated. Even when the quality of the guidance information is poor, it can effectively guide the algorithm to search the correct solution space, thereby improving the probability and accuracy of successfully separating the target interference source under adverse conditions. Fourthly, by outputting high-quality independent interference source component signals, it serves as a direct input for the subsequent interference source component evaluation module to perform type identification and contribution calculation, thus determining the accuracy and reliability of the system's final output.

[0014] Interference source component evaluation module: Identifies the interference source type of independent interference source component signals, obtains the interference source identification results, and evaluates the contribution of the interference source identification results based on the frequency conversion potential measurement response characteristics and signal energy distribution, and obtains the interference source contribution ranking table; Specifically, multi-dimensional feature extraction is performed on the signal of each independent interference source component; Time-domain feature extraction: Calculate the statistical and waveform features of the signal, including but not limited to: absolute mean, effective value (RMS), peak value, peak-to-peak value, waveform factor, peak factor, impulse factor, as well as the skewness and kurtosis of the signal; Frequency domain feature extraction: Discrete Fourier transform is performed on the independent interference source component signals to extract their spectral structure features. Key features include: dominant fundamental frequency (e.g., 50Hz, 16.7Hz), main harmonic frequency set and its amplitude ratio relative to the fundamental frequency, total harmonic distortion rate, and energy proportion of specific frequency bands (e.g., 0-100Hz, 100-1kHz). By analyzing the fundamental frequency and harmonic structure, the operating characteristics of specific electrical equipment or systems can be directly mapped. Time-frequency feature extraction: Short-time Fourier transform is used to obtain the two-dimensional distribution of signal energy in the time-frequency plane, and dynamic pattern features are extracted from the time spectrum, such as the duration and periodicity of high-energy regions, the time-varying envelope of specific frequency components (such as characteristic harmonics of subways), and the repetition frequency and duty cycle of pulse events, which are used to capture interference patterns associated with dynamic processes such as train operation and load switching. The extracted time-domain, frequency-domain, and time-frequency-domain features are combined into a high-dimensional comprehensive feature vector; The high-dimensional integrated feature vector is input into the pre-trained perturbation pattern classification model; Among them, the interference pattern classification model can use a multi-class machine learning model, which can use support vector machine (SVM) or random forest model. The above models have been trained using a large number of labeled interference signal samples (covering subway, high-speed rail, high-voltage lines, DC systems, etc.) and can learn the discrimination boundary of different interference sources in the feature space. The interference mode classification model analyzes the input high-dimensional integrated feature vector and outputs the probability distribution of each predefined interference source category. The category with the highest probability is determined as the interference source type of the component signal (such as DC subway traction interference), and the highest probability value is used as the identification confidence level. Generate a list of interference source identification results for all N components, including: component index, interference source type label, and identification confidence score; Specifically, the process of evaluating the contribution of interference source identification results and obtaining a ranking table of interference source contributions is as follows: For each identified interference source component, a contribution index is calculated based on a preset analysis time window, including: The analysis time window can be a complete evaluation period, such as 24 hours, to cover the daily variation cycle of the interference source; Calculate the total energy value of the interference source component signal within the analysis time window, and use it as a single energy value; The calculation process for the total energy value is as follows: the square of the instantaneous amplitude of the signal component is continuously integrated within the time window T (or, in actual digital processing, the discrete sampling points are summed). The result of the integration (or the summation) is the total energy value of the signal component within the time period T. Calculate the percentage of a single energy value relative to the total energy of all N interference source components to obtain the energy contribution percentage; Based on the amplitude of the impedance to ground of the pipeline system at a given frequency obtained from frequency-dependent potential measurement, the dominant characteristic frequency of the interference source component signal is determined, the effective voltage value of the interference source component signal is calculated, and the equivalent interference current density is calculated according to Ohm's law. The energy contribution ratio and the equal interference current density are standardized, and the standardized values ​​are summed and averaged to obtain the comprehensive contribution. Based on the calculated overall contribution, all identified interference sources are sorted in descending order to generate an interference source contribution ranking table.

[0015] The core function of this module is to transform the independent interference source component signals decoupled from the number of components into quantitative contribution assessment conclusions that can directly guide engineering protection through intelligent identification and precise calculation. Specifically, it first performs feature extraction and pattern recognition on each component to determine the type of physical interference source. Combining the impedance characteristics obtained from frequency conversion measurements, it calculates key quantitative indicators such as the energy proportion contribution and equivalent interference current density of each interference source, and generates an interference source contribution ranking table to clearly indicate the dominant risk source and its relative degree of harm. Thus, the complex signal analysis results are directly output as a basis for supporting precise governance decisions.

[0016] As one specific implementation process in this embodiment: A gas pipeline in the core area of ​​a city runs parallel to a subway tunnel for about 2 kilometers and passes above a 110kV high-voltage cable corridor. There is also a large commercial complex substation grounding grid in the vicinity. The pipeline section has repeatedly experienced accelerated corrosion due to damage to the anti-corrosion layer. Traditional monitoring methods can only detect violent fluctuations in the overall potential and cannot identify the main source of interference, resulting in poor effectiveness of drainage measures. Three synchronous measurement terminals were set up along the pipeline, and a variable frequency current excitation of 1-100Hz was applied to synchronously collect the pipeline potential response. At the same time, the "external reference signal_metro" was collected at the nearby subway return cabinet, and the "external reference signal_power grid" was collected at the high-voltage cable grounding lead-out line. The purity assessment revealed that the "external reference signal_metro" contained strong 50Hz power frequency harmonics (from power grid coupling), and its target frequency band signal-to-noise ratio (for the 16.7Hz characteristic of metro) was low. Furthermore, its average spectral coherence coefficient with the pipeline signal did not reach the high purity threshold, so it was judged to have low purity. In contrast, the "external reference signal_power grid" had a pure spectrum and was judged to have high purity. For power grid interference (high purity), the high purity processing unit directly uses the power grid reference signal as a guide to quickly separate the power frequency and its harmonic interference components; In response to subway interference (low purity), the low purity processing unit is activated: A pipeline signal subspace is constructed based on the response signals from three pipeline measuring points; The impure "external reference signal_metro" is projected into this subspace, filtering out grid coupling noise unrelated to the pipeline, and generating an enhanced guidance signal (mainly containing the fluctuating components of metro traction characteristics). Guided by this enhanced signal, the DC and harmonic interference components of the subway with time-varying characteristics are separated. The entire process employs a convolutional blind source separation model, which effectively addresses the time delay and attenuation caused by subway interference propagating through the soil to different measuring points, resulting in more realistic waveforms of the separated subway interference components. The separated components were analyzed for characteristics. Component A: 50Hz fundamental frequency, rich in higher harmonics → identified as "high voltage AC transmission interference". Component B: 16.7Hz characteristic harmonic superimposed on fluctuating DC, and its intensity is synchronized with the subway timetable → identified as "subway DC traction interference". Component C: broadband noise → identified as "background noise". Contribution assessment: The energy proportion and equivalent interference current density of each component were calculated over 24 hours. The results showed that the equivalent interference current density of the subway interference component far exceeded the safety standard, and although its energy proportion was slightly lower than that of the power grid interference, it had the highest comprehensive contribution score because it was DC in nature. Output: The system generates a report that clearly identifies "subway DC traction interference" as the dominant risk source for this pipeline section, quantifies its corrosion risk level, and provides data on the contribution of power grid interference. The report includes a ranking table of interference source contributions, providing clear priorities for mitigation: priority should be given to installing polarity drainage devices for stray currents from the subway.

[0017] Example 2 Based on the same inventive concept as the frequency conversion potential measurement system for intelligent identification of AC / DC interference in gas pipelines in the foregoing embodiments, such as Figure 2 and Figure 3 As shown, this application provides a variable frequency potential measurement method for intelligent identification of AC / DC interference in gas pipelines, wherein the method specifically includes the following steps: Step S10: Apply a frequency conversion current excitation signal with a preset spectrum to the gas pipeline. Based on the measurement terminals deployed along the gas pipeline, simultaneously collect the potential response signals of at least two measurement points under frequency conversion excitation as measurement data, as well as external reference signals, and evaluate the purity of the external reference signals to obtain the purity evaluation results. Step S20: Based on the purity assessment results, convolutional blind source separation is introduced to separate the mixed signals of the frequency conversion measurement response data, and the independent interference source component signals decoupled from the mixed measurement data are obtained; Step S201: If the purity is high, then signal separation is performed directly using the external reference signal as a guide; Step S202: If the purity is low, a gas pipeline signal subspace is constructed based on the signals from multiple measurement points, and an external reference signal is projected onto the gas pipeline signal subspace to generate an enhanced guidance signal. The enhanced guidance signal is then used as a guide for signal separation. Step S30: Identify the interference source type of the independent interference source component signal, obtain the interference source identification result, and evaluate the contribution of the interference source identification result based on the frequency conversion potential measurement response characteristics and signal energy distribution to obtain the interference source contribution ranking table.

[0018] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely illustrative of the principles of the invention. Various changes and modifications can be made to the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the present invention as claimed. The scope of protection of the present invention is defined by the appended claims and their equivalents.

Claims

1. A variable frequency potential measurement system for intelligent identification of AC / DC interference in gas pipelines, characterized in that: include: Data measurement module: Apply a frequency-converted current excitation signal with a preset spectrum to the gas pipeline. Based on the measurement terminals deployed along the gas pipeline, simultaneously collect the potential response signals of at least two measurement points under frequency-converted excitation as measurement data, as well as external reference signals, and evaluate the purity of the external reference signals to obtain the purity assessment results. Dynamic strategy selection module: Based on the purity evaluation results, convolutional blind source separation is introduced to separate the mixed signals of the frequency conversion measurement response data, and obtain the independent interference source component signals decoupled from the mixed measurement data; High purity processing unit: If the purity is high, signal separation is performed directly guided by an external reference signal; Low purity processing unit: If the purity is low, a gas pipeline signal subspace is constructed based on the signals from multiple measurement points, and an external reference signal is projected onto the gas pipeline signal subspace to generate an enhanced guidance signal. The enhanced guidance signal is then used as a guide for signal separation. Interference source component evaluation module: Identifies the interference source type of independent interference source component signals, obtains the interference source identification results, and evaluates the contribution of the interference source identification results based on the frequency conversion potential measurement response characteristics and signal energy distribution, and obtains an interference source contribution ranking table.

2. The frequency conversion potential measurement system for intelligent identification of AC / DC interference in gas pipelines according to claim 1, characterized in that: The process of obtaining the purity assessment results: The signal-to-noise ratio and average spectral coherence coefficient of the target frequency band were calculated using time-frequency analysis and correlation calculation methods, respectively. If the target frequency band signal-to-noise ratio is greater than the first preset threshold and the average spectral coherence coefficient is greater than the second preset threshold at the same time, the purity of the external reference signal is determined to be high. Otherwise, the purity of the external reference signal is determined to be low.

3. The frequency conversion potential measurement system for intelligent identification of AC / DC interference in gas pipelines according to claim 2, characterized in that: The process of obtaining the signal-to-noise ratio value of the target frequency band: Calculate the ratio of signal power to noise power in the characteristic frequency band of the target interference source, and use it as the signal-to-noise ratio value of the target frequency band; The sum of squares of amplitudes at all time-frequency points within the characteristic frequency band of the target interference source, or the average power spectral density of the characteristic frequency band of the target interference source, is taken as the signal power. Select one or more quiet frequency bands that are adjacent to the characteristic frequency band of the target interference source and have no significant spectral peaks, and calculate the average power spectral density as the noise power.

4. The frequency conversion potential measurement system for intelligent identification of AC / DC interference in gas pipelines according to claim 2, characterized in that: The process of obtaining the average spectral coherence coefficient is as follows: For the external reference signal and each pipe potential response signal, calculate the amplitude squared coherence estimate between them within the effective frequency band covering the main interference energy; The amplitude squared coherence estimates obtained from all pipeline measurement points are arithmetically averaged in the frequency domain to obtain the average spectral coherence coefficient.

5. A frequency conversion potential measurement system for intelligent identification of AC / DC interference in gas pipelines according to claim 1, characterized in that: Convolutional blind source separation employs a convolutional hybrid model: Where t represents a continuous-time variable, This represents the mixed signal vector observed at time t. This represents the vector of the unknown source signal at time t. This represents the discrete-time delay index, and L represents the maximum memory length of the hybrid filter. Indicates time delay The M×N dimensional hybrid filter matrix at the location, This represents the additive noise vector at time t.

6. A frequency conversion potential measurement system for intelligent identification of AC / DC interference in gas pipelines according to claim 5, characterized in that: In the high-purity processing unit, when solving the convolutional mixture model, the time-domain signal is converted to the frequency domain, and the observed signal and the guiding signal are subjected to a segmented windowed short-time Fourier transform to obtain the frequency domain representation of the observed signal and the frequency domain representation of the guiding signal. Within the characteristic frequency band of the target interference source, prior information highly correlated with a desired source signal is used in the frequency domain representation of the guiding signal to construct constraints. A fast convolutional blind source separation algorithm is then employed, with the following steps: Calculate the covariance matrix of the frequency domain representation of the observed signal over multiple time frames; Using the frequency domain representation of the guiding signal as a reference, the separation filter is estimated through generalized eigenvalue decomposition or adaptive filtering techniques. The separated frequency domain signals are reconstructed into independent interference source components in the time domain through inverse short-time Fourier transform.

7. The frequency conversion potential measurement system for intelligent identification of AC / DC interference in gas pipelines according to claim 1, characterized in that: The process of enhancing the guidance signal is as follows: The time-domain representation of the observed signal is preprocessed with zero mean to construct a signal subspace characterizing the main interference modes of the gas pipeline; Calculate the covariance matrix of the time-domain representation of the observed signal, and perform eigenvalue decomposition on the covariance matrix; Based on the preset cumulative variance contribution rate threshold, the feature vectors corresponding to cumulative variance contribution rates greater than or equal to the cumulative variance contribution rate threshold are selected, which are the gas pipeline signal subspace. External reference signals that are determined to have low purity are subjected to the same zero-mean processing to obtain preprocessed external reference signals. The preprocessed external reference signals are then projected onto the gas pipeline signal subspace to extract components that are strongly correlated with the interference experienced by the gas pipeline. Enhanced guidance signals are obtained through projection operations.

8. A frequency conversion potential measurement system for intelligent identification of AC / DC interference in gas pipelines according to claim 1, characterized in that: When performing convolutional blind source separation in a low-purity processing unit; The enhanced guiding signal is used as the initialization vector for algorithm iteration or as a soft constraint term for the objective function; After algorithm iteration, the signal of independent interference source components is output.

9. A frequency conversion potential measurement system for intelligent identification of AC / DC interference in gas pipelines according to claim 1, characterized in that: The process of obtaining the interference source identification results is as follows: Multi-dimensional feature extraction is performed on each independent interference source component signal, including time-domain feature extraction, frequency-domain feature extraction, and time-frequency feature extraction; The extracted features are then combined into a high-dimensional comprehensive feature vector. The high-dimensional integrated feature vector is input into the pre-trained perturbation pattern classification model; Output the probability distribution of each predefined interference source category, determine the category with the highest probability as the interference source type of the component signal, and use the highest probability value as the identification confidence level to generate a list of interference source identification results for all N components.

10. A frequency conversion potential measurement system for intelligent identification of AC / DC interference in gas pipelines according to claim 1, characterized in that: The process of obtaining the interference source contribution ranking table is as follows: For each identified interference source component, a contribution index is calculated based on a preset analysis time window; Calculate the total energy value of the interference source component signal within the analysis time window, and use it as a single energy value; Calculate the percentage of a single energy value relative to the total energy of all N interference source components to obtain the energy contribution percentage; Based on the amplitude of the impedance to ground of the pipeline system at a given frequency obtained from frequency-dependent potential measurement, the dominant characteristic frequency of the interference source component signal is determined, the effective voltage value of the interference source component signal is calculated, and the equivalent interference current density is calculated according to Ohm's law. The energy contribution ratio and the equal interference current density are standardized, and the standardized values ​​are summed and averaged to obtain the comprehensive contribution. Based on the calculated overall contribution, all identified interference sources are sorted in descending order to generate an interference source contribution ranking table.

Images