Electric hot melting sleeve connection monitoring method for polyurethane directly-buried thermal insulation pipe

Abstract

The invention discloses an electric hot melting sleeve connection monitoring method for a polyurethane directly-buried thermal insulation pipe, and belongs to the technical field of electric hot melting sleeve installation monitoring. The method specifically comprises the steps that distributed electrode sets are embedded into annular contact surfaces on the two sides of an electric hot melting sleeve, and an electric coupling network is formed by the distributed electrode sets and a conductive enhancing structure in a polyurethane thermal insulation layer; during electrifying and heating, applying a broadband alternating current signal which is dynamically modulated and covers a dielectric relaxation frequency band of a melting interface to the electrode group; collecting electrode pair voltage response signals in real time, and resolving a complex impedance spectrum containing real part and imaginary part impedance components; analyzing real part impedance non-linear attenuation characteristics, and if the real part impedance non-linear attenuation characteristics do not accord with a rule, analyzing an imaginary part impedance phase angle fluctuation rule, extracting dielectric anomaly features, matching the dielectric anomaly features with a preset model, and performing early warning if the dielectric anomaly features exceed a tolerance range; and finally, positioning the spatial distribution of the fusion defects based on the impedance response of the electrode group.

Description

Technical Field

[0001] The present invention relates to the technical field of monitoring the installation of electrofusion sleeves, and specifically relates to a method for monitoring the electrofusion sleeve connection of polyurethane directly buried insulation pipes. Background Art

[0002] In the field of energy transportation, polyurethane directly buried insulation pipes are widely used in many systems such as central heating, cooling, and oil transportation due to their good heat insulation performance, corrosion resistance, and long service life. These systems have extremely high requirements for the sealing and reliability of pipe connections. Because once problems occur in the pipe connection parts, not only will a large amount of energy be wasted, but also safety accidents may be triggered, affecting the normal operation of the entire transportation system. Therefore, ensuring the connection quality of polyurethane directly buried insulation pipes is of crucial importance. And as a commonly used connection method, the monitoring of the connection quality of electrofusion sleeves has become a key link to ensure the stable operation of the pipe system.

[0003] Currently, there have been some studies and practices on monitoring the connection quality of electrofusion sleeves of polyurethane directly buried insulation pipes. Some methods evaluate the connection quality by detecting the change in conductivity. Electrodes are set at the connection part to measure the current and voltage, and then the conductivity is calculated. Whether the connection is good is judged according to the change in conductivity. There are also some that use ultrasonic flaw detection methods to detect whether there are defects in the connection part by emitting ultrasonic waves and analyzing the characteristics of the reflected waves. In addition, there are also monitoring methods based on resistance changes, which evaluate the connection state by monitoring the real-time change in the resistance of the connection part.

[0004] However, these existing monitoring methods have many deficiencies. On the one hand, the real-time performance is poor. Most methods are used for detection after the connection is completed, and it is difficult to grasp the connection state in real time during the process of heating and melting the electrofusion sleeve. Once a problem is found, irreversible losses are often caused. On the other hand, the detection accuracy is limited. For some microscopic defects, such as tiny bubbles and impurities, it is difficult to accurately detect them, and these microscopic defects may gradually expand over time, affecting the long-term use performance of the pipeline. And the spatial positioning ability is insufficient, and it is difficult to accurately determine the specific spatial position of the defect in the connection part, which makes the subsequent repair work lack pertinence and low efficiency. Summary of the Invention

[0005] The purpose of the present invention is to provide a method for monitoring the electrofusion sleeve connection of polyurethane directly buried insulation pipes to solve the following technical problems: The existing electrofusion sleeve monitoring methods have poor real-time performance and it is difficult to locate some microscopic defects.

[0006] The purpose of the present invention can be achieved through the following technical solutions: A method for monitoring the electrofusion sleeve connection of polyurethane directly buried insulation pipes includes the following steps: Embed a distributed electrode group into the annular contact surfaces on both sides of the electrothermal fusion sleeve, and the distributed electrode group forms an electrical coupling network with the conductive enhancement structure inside the polyurethane thermal insulation layer; During the process of heating the electrothermal fusion sleeve by energization, apply a dynamically modulated broadband alternating current signal to the electrode group, and the frequency range of the broadband alternating current signal covers the dielectric relaxation frequency band of the melting interface; Real-time collect the voltage response signals of each electrode pair in the distributed electrode group, and calculate the complex impedance spectrum of the electrothermal fusion sleeve based on the voltage response signals. The complex impedance spectrum includes a real part impedance component and an imaginary part impedance component; Analyze the non-linear attenuation characteristics of the real part impedance component. If the non-linear attenuation characteristics do not conform to the expected law, generate a melting defect warning signal; By analyzing the fluctuation law of the phase angle of the imaginary part impedance component changing with frequency, extract the dielectric anomaly characteristics of the microstructure of the melting interface; match the dielectric anomaly characteristics with a preset melting interface uniformity model. If the matching result exceeds the preset tolerance range, generate a melting defect warning signal; Locate the spatial distribution of the melting defect based on the impedance response of the distributed electrode group.

[0007] As a further solution of the present invention: the conductive enhancement structure is a metal wire mesh or a carbon fiber array embedded in the polyurethane thermal insulation layer. The metal wire mesh or the carbon fiber array is evenly distributed along the circumferential direction of the polyurethane thermal insulation layer, and forms a full circumferential electrical connection with the annular electrode mounting surface of the distributed electrode group through welding or conductive adhesive.

[0008] As a further solution of the present invention: the dynamic modulation method of the broadband alternating current signal includes: Mark the time interval of energizing and heating the electrothermal fusion sleeve as [t1, t2]. The broadband alternating current signal includes a low-frequency continuous sweep signal and a high-frequency pulse modulation signal; When in the time interval [t1, t0], t0 is a preset value, apply a low-frequency continuous sweep signal, and the frequency range of the low-frequency band corresponds to the response frequency band of the overall conductivity of the polyurethane material in the molten state; When in the time interval (t0, t2], superimpose a high-frequency pulse modulation signal on the low-frequency continuous sweep signal. The frequency range of the high-frequency band corresponds to the local dielectric relaxation frequency band caused by microscopic defects at the melting interface. The microscopic defects include bubbles or impurities; During the electrothermal sleeve energization heating process, impedance response data of the electrode group is collected in real time, and the interval size of frequency change in the low-frequency continuous sweep signal and the ratio of pulse duration to period in the high-frequency pulse modulation signal are adjusted according to the impedance response data, so that the broadband alternating current signal is adaptively matched with the melting process of the polyurethane material.

[0009] As a further solution of the present invention: the calculation of the complex impedance spectrum specifically includes: Perform synchronous demodulation processing on the voltage response signal, separate and mark the part of the voltage response signal that is in the same frequency and phase as the excitation signal as the real part voltage component, and separate and mark the part that is orthogonal to the phase of the excitation signal as the imaginary part voltage component; Calculate the real part impedance through the real part voltage component, obtain the change rate of the real part impedance in the radial space, and construct a dynamic distribution model of the melting layer thickness; calculate the imaginary part impedance based on the imaginary part voltage component, use the spatial position as the coordinate, represent the value of the imaginary part impedance in the form of a cloud map, generate a dielectric property distribution cloud map, and obtain the local extreme points of the imaginary part impedance in the cloud map.

[0010] As a further solution of the present invention: the process of extracting the dielectric anomaly characteristics of the microstructure of the melting interface includes: Perform multi-scale time-frequency analysis on the imaginary part impedance component, decompose the imaginary part impedance component signal into different frequency components, obtain the fluctuation changes of different frequency components at different time points, and extract the non-stationary fluctuation components of the phase angle in a specific frequency band; According to the extracted non-stationary fluctuation components of the phase angle, construct an energy spectrum, which is used to represent the energy distribution of the phase angle fluctuation at different frequencies, calculate the integral value of the energy spectrum within a preset defect characteristic frequency band, and the integral value represents the phase angle fluctuation energy of all frequencies within this frequency band; when the integral value exceeds the reference energy threshold, it is determined that there is an unfused defect at the melting interface; Obtain the frequency position of the peak value in the energy spectrum. When a low-frequency energy peak appears in the energy spectrum, the unfused defect is a macroscopic crack defect; when a high-frequency energy peak appears in the energy spectrum, the unfused defect is a microscopic bubble / impurity defect.

[0011] As a further solution of the present invention: the arrangement mode of the distributed electrode group is: In the axial direction of the electrothermal fusion sleeve, n groups of annular electrodes are arranged, and there is an axial overlap in the detection areas of adjacent annular electrodes; each group of annular electrodes includes m concentric annular electrodes with different radial directions, and each layer of annular electrodes includes o electrode units, and the electrode units are distributed on the annular electrodes in a circumferentially symmetric manner. n, m, and o are preset values. Each electrode unit is connected to a switching circuit. Through the control of the switching circuit, different electrode units are combined in different ways to form multiple independent detection circuits.

[0012] As a further solution of the present invention: Locating the spatial distribution of the melting defect includes axial positioning, radial positioning, and circumferential positioning. The positioning process is as follows: Obtain the impedance response gradient change in the axially overlapping detection area, and perform axial detection starting from one end of the melting interface. When it is recognized that the impedance response gradient at any position increases by more than the preset value, mark this position as the starting point of the axial defect, and continue to identify along the axial direction until the value at any position is lower than the preset value, then mark this position as the ending point of the axial defect. The area between the starting point and the ending point of the defect is the axially positioned area of the defect; Obtain the impedance responses of different layers of concentric annular electrodes within the axially positioned area, and calculate the concentric annular electrode layer where the defect is located according to the impedance response differences of the concentric annular electrodes at different radial positions, which is the radially positioned range of the defect; Obtain the impedance responses generated by each detection circuit within the radially positioned range of the axially positioned area, and calculate the uniformity of the melting layer in the circumferential direction by analyzing the impedance response differences of the detection circuits. If the impedance difference between any adjacent electrode units exceeds the set ratio, it is determined that there is a melting defect in the corresponding circumferential area.

[0013] As a further solution of the present invention: The construction method of the preset melting interface uniformity model includes: Under standard experimental conditions, monitor and collect data on the completely sealed melting interface, record the change of the real part impedance with time at different heating stages and different temperatures to form a real part impedance decay curve; extract the fluctuation patterns of the imaginary part phase angle at different frequencies to form an imaginary part phase fluctuation template, and generate an impedance reference database; The melting interface uniformity model includes a relationship model between the real part impedance gradient and the melting layer thickness and an imaginary part phase fluctuation pattern classification model; Input the real part impedance data in the impedance reference database into a convolutional neural network, learn the change characteristics of the real part impedance at different positions and different conditions, establish a non-linear mapping relationship between the real part impedance gradient and the melting layer thickness, and obtain the relationship model between the real part impedance gradient and the melting layer thickness; Input the imaginary part phase fluctuation template data in the impedance reference database into the support vector machine algorithm. The template data contains the imaginary part phase fluctuation patterns corresponding to different defect types. The support vector machine algorithm performs classification training on the template data, stores the imaginary part phase fluctuation characteristics corresponding to different defect types, generates several phase fingerprint recognition libraries, and obtains an imaginary part phase fluctuation pattern classification model. Input the real-time monitoring data into the trained molten interface uniformity model, and output the molten mass score of the real part impedance data and the defect confidence parameter of the imaginary part phase data.

[0014] As a further solution of the present invention: it further includes: Based on the axial positioning, radial positioning, circumferential positioning, and heating time parameters, establish a three-dimensional defect evolution model. The three-dimensional defect evolution model includes the growth rate of the defect volume with heating time and the change trajectory of the shape and position.

[0015] The beneficial effects of the present invention: The present invention uses a distributed electrode group and a conductive enhancement structure to form a full circumferential electrical coupling network, combined with the dynamic modulation technology of a broadband alternating current signal, and can synchronously collect impedance response data during the electrothermal sleeve heating process. By real-time analyzing the non-linear attenuation characteristics of the real part impedance and the fluctuation law of the imaginary part phase angle, it can directly identify interface defects during the melting stage, breaking through the lag limitation of traditional off-line detection.

[0016] Through the multi-scale time-frequency analysis algorithm, decompose the imaginary part impedance component into different frequency components, and extract the phase angle fluctuation characteristics of a specific frequency band. This technology can effectively distinguish macroscopic cracks from microscopic bubble / impurity defects, and its detection sensitivity is one order of magnitude higher than that of ultrasonic flaw detection technology, and it can capture interface non-uniformity problems that are difficult to discover by traditional methods.

[0017] Based on the impedance gradient change analysis of the axially overlapping detection area, combined with the response difference of the radially concentric electrode layers and the circumferential electrode unit combination detection, realize the precise positioning of defects in three-dimensional space. By quantifying the impedance response differences at different positions, this method can accurately determine the axial distribution range, radial depth, and circumferential position of defects, providing a reliable basis for targeted repair.

[0018] According to the change of the electrical characteristics of the polyurethane material during the melting process, dynamically adjust the sweep frequency interval and pulse duty cycle to make the excitation signal always match the material state. This technology effectively solves the signal distortion problem in traditional fixed-frequency detection and ensures stable monitoring accuracy during the material phase change process. Brief Description of the Drawings

[0019] The present invention will be further described below with reference to the drawings.

[0020] Figure 1 This is the process schematic diagram of the present invention. Specific embodiments

[0021] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0022] Please refer to Figure 1 As shown, the present invention is a method for monitoring the electrofusion sleeve connection of a polyurethane directly buried insulation pipe, including the following steps: Step 1: Construction of a distributed electrode network: Symmetrically embed multiple layers of distributed electrode groups in the annular contact surfaces on both sides of the electrofusion sleeve. The electrode groups are made of high-temperature resistant conductive materials, and their arrangement density is optimized according to the dielectric properties of the polyurethane material. The electrode groups form an integrated structure with the electrofusion sleeve matrix through a precision injection molding process to ensure stable electrical performance during the high-temperature melting process. The distributed electrode groups are conductively connected in a full circumferential direction with the conductive enhancement structure (such as a metal wire mesh or a carbon fiber array) inside the polyurethane insulation layer to form a three-dimensional electrical coupling network, which covers the entire melting interface area and provides a low-impedance channel for broadband signal transmission.

[0023] Step 2: Dynamic broadband signal excitation: During the power-on heating stage of the electrofusion sleeve, a dynamically modulated broadband alternating current signal is applied to the electrode groups by an intelligent signal generator. The signal includes a continuous sweep frequency signal (10 Hz - 100 kHz) in the low-frequency band and a pulse modulation signal (1 MHz - 10 MHz) in the high-frequency band, and its frequency range covers the entire dielectric relaxation frequency band of the polyurethane material from the solid state to the molten state. The signal modulation strategy adopts a double closed-loop control: the temperature feedback loop adjusts the signal amplitude in real time according to the heating power, and the impedance feedback loop dynamically optimizes the sweep frequency interval and the pulse duty cycle to ensure that the excitation signal is synchronized with the material phase change process.

[0024] Step 3: Multi-dimensional impedance data acquisition: The voltage response signals of each electrode pair in the distributed electrode groups are obtained in real time through a high-precision synchronous acquisition system, and the sampling rate reaches more than 100 kS / s. The lock-in amplification technology is used to orthogonally decompose the voltage signal to separate the real part voltage component in phase with the excitation signal and the imaginary part voltage component orthogonal in phase. Based on Ohm's law, the complex impedance values of the corresponding electrode pairs are calculated, and a three-dimensional impedance database containing more than 2000 measurement points is constructed to provide data support for full-space coverage for subsequent analysis.

[0025] Step 4. Real part impedance non-linear analysis: Use the polynomial fitting algorithm to extract the characteristic parameters of the attenuation curve of the real part impedance, including the attenuation rate, inflection point position, curve symmetry, etc. Establish a non-linear attenuation model based on support vector regression. This model integrates parameters such as the temperature field distribution and material thermal conductivity and can predict the impedance attenuation trajectory under ideal conditions. When the deviation between the measured curve and the prediction model exceeds the set threshold, trigger the molten defect warning mechanism.

[0026] Step 5. Extraction of imaginary part phase fluctuation characteristics: Perform continuous wavelet transform on the imaginary part impedance component to decompose the phase fluctuation signals in 12 different frequency bands. Extract the instantaneous phase of each frequency band signal through Hilbert transform and construct a time-frequency domain joint feature matrix. Use a convolutional neural network to perform pattern recognition on the feature matrix. When the phase fluctuation energy in a specific frequency band (such as 100 kHz - 500 kHz) is detected to exceed 1.5 times the reference value, it is determined that there are microscopic defects.

[0027] Step 6. Three-dimensional spatial positioning of defects: Based on the impedance gradient analysis of the axial overlapping electrode group, use the cubic spline interpolation algorithm to determine the axial distribution range of the defect. Through the impedance response difference of the radial layered electrodes, use the Bayesian estimation method to invert the radial depth of the defect. Use the impedance matrix of the circumferential electrode array and combine the Tikhonov regularization algorithm to reconstruct the thickness distribution cloud map of the molten layer and identify the circumferential non-uniform area. Finally, generate a defect positioning report containing XYZ three-dimensional coordinates to provide accurate guidance for subsequent repair.

[0028] In a preferred embodiment of the present invention, the conductive enhancement structure is a metal wire mesh or carbon fiber array embedded in the polyurethane insulation layer. The metal wire mesh or carbon fiber array is evenly distributed along the circumference of the polyurethane insulation layer and forms a full circumferential electrical connection with the annular electrode mounting surface of the distributed electrode group through welding or conductive adhesive.

[0029] In another preferred embodiment of the present invention, the dynamic modulation method of the broadband alternating current signal includes: Set the time interval for energizing and heating the electrothermal sleeve as [t1, t2]. This broadband alternating current signal is composed of a low-frequency continuous sweep signal and a high-frequency pulse modulation signal.

[0030] In the time interval [t1, t0] (where t0 represents the time point at which the initial heating of the electrothermal fusion sleeve by electricity ends), a continuous swept-frequency signal in the low-frequency band is applied. The frequency range of this low-frequency band, for example, from 10 Hz to 100 kHz, corresponds to the response frequency band of the overall conductivity of the polyurethane material when it is in a molten state. In the initial stage of heating, as the polyurethane material transforms from a solid state to a molten state, its overall conductivity begins to change, and the continuous swept-frequency signal in the low-frequency band can effectively detect this change in overall conductivity. For example, as the material gradually melts, the change in the internal molecular structure makes the movement of free electrons more active, resulting in an increase in overall conductivity. The continuous swept-frequency signal in the low-frequency band can capture this changing trend, providing basic data for subsequent monitoring and analysis.

[0031] When entering the time interval (t0, t2], a high-frequency pulsed modulation signal is superimposed on the continuous swept-frequency signal in the low-frequency band. The frequency range of the high-frequency band, such as from 1 MHz to 10 MHz, corresponds to the local dielectric relaxation frequency band caused by microscopic defects (including bubbles, impurities, etc.) at the molten interface. In the later stage of heating, when the polyurethane material further melts, microscopic defects may occur at the molten interface. These defects affect the local dielectric properties of the material, triggering local dielectric relaxation phenomena. The high-frequency pulsed modulation signal can sensitively detect this change in local dielectric properties, helping to discover potential microscopic defects. For example, when there are bubbles at the molten interface, the electric field distribution around the bubbles changes, resulting in a change in the local dielectric constant, and the high-frequency pulsed modulation signal can detect this change.

[0032] During the entire process of heating the electrothermal fusion sleeve by electricity, the impedance response data of the electrode group are collected in real time. Based on these data, the continuous swept-frequency signal in the low-frequency band and the high-frequency pulsed modulation signal are dynamically adjusted. For the continuous swept-frequency signal in the low-frequency band, if the collected impedance response data show that the conductivity of the material changes slowly, the frequency change interval can be appropriately increased, such as from 1 kHz to 2 kHz; if the conductivity changes rapidly, the frequency change interval is decreased, such as from 1 kHz to 0.5 kHz. For the high-frequency pulsed modulation signal, when it is detected that there may be microscopic defects, the ratio of the pulse duration to the period (duty cycle) is increased, such as from 30% to 50%; when there are no obvious signs of microscopic defects, the duty cycle is appropriately decreased, such as from 30% to 20%. Through this dynamic adjustment, the broadband alternating current signal can be adaptively matched with the melting process of the polyurethane material, significantly improving the accuracy and effectiveness of monitoring.

[0033] In another preferred embodiment of the present invention, the solution of the complex impedance spectrum specifically includes: First, perform synchronous detection processing on the voltage response signal collected in real time. This process is achieved by means of a high-precision synchronous detection circuit or relevant software algorithms. Specifically, through phase-locked loop technology, the voltage response signal is accurately compared with the excitation signal. On this basis, the part of the voltage response signal that is in the same frequency and phase as the excitation signal can be precisely identified and separated, and then this part of the signal is marked as the real part voltage component; at the same time, using quadrature phase detection technology, the part of the voltage response signal that is in quadrature phase with the excitation signal is also separated and marked as the imaginary part voltage component. This precise separation process lays a solid foundation for subsequent impedance calculation and related characteristic analysis.

[0034] Next, calculate the real part impedance based on the separated real part voltage component. According to Ohm's law, combined with the known magnitude of the excitation current, through the corresponding calculation program, the real part impedance value can be accurately obtained. To further understand the molten state in depth, it is also necessary to obtain the change rate of the real part impedance in the radial space. This requires multi-point measurement of the real part impedance at different radial positions. For example, measurement points are set at regular intervals (such as 0.5 mm) along the radial direction, and the real part impedance values at each point are collected. By analyzing these multi-point data and using differential algorithms or methods of curve fitting to find the slope, the change rate of the real part impedance in the radial space is calculated. Based on these change rate data, combined with the physical characteristic parameters of the polyurethane material during the melting process and the heat transfer principle, a model that can accurately reflect the dynamic distribution of the molten layer thickness is constructed. This model can intuitively show the change of the molten layer thickness in the radial direction over time during the electrothermal sleeve energization heating process.

[0035] After completing the processing related to the real part impedance, calculate the imaginary part impedance based on the imaginary part voltage component. Similarly, according to Ohm's law and the excitation current information, the imaginary part impedance is calculated through a specially designed algorithm. To more intuitively and comprehensively present the distribution characteristics of the imaginary part impedance in space, taking the spatial position as the coordinate, with the help of professional drawing software or self-written drawing programs, the values of the imaginary part impedance are visually represented in the form of a cloud map, thereby generating a dielectric property distribution cloud map. In this cloud map, different colors or gray levels represent different magnitudes of the imaginary part impedance values. By carefully observing and analyzing the cloud map and using image recognition algorithms or manual screening methods, the local extreme points of the imaginary part impedance in the cloud map are obtained. These local extreme points often correspond to the regions where the dielectric properties change significantly at the molten interface, and have important indicative significance for judging the existence of microscopic defects and evaluating the melting quality.

[0036] In a preferred case of this embodiment, the process of extracting the dielectric anomaly characteristics of the microscopic structure of the molten interface includes: First, multi-scale time-frequency analysis is carried out on the imaginary impedance component. This process uses advanced mathematical tools, such as wavelet transform algorithms. By selecting appropriate wavelet basis functions, the imaginary impedance component signal is decomposed according to different time scales and frequency scales, thereby decomposing it into a series of different frequency components. During the decomposition process, the time sliding window technology is used to accurately obtain the fluctuation changes of these different frequency components at different time points during the heating process of the electric hot melt sleeve. Then, the phase angle analysis algorithm is used to focus on a specific frequency band, such as the frequency range of 100kHz to 500kHz, which is more sensitive to microstructural changes, and carefully extract the non-stationary fluctuation components of the phase angle in this frequency band. These non-stationary fluctuation components often contain key information on the microstructural changes of the molten interface and are the core data for subsequent analysis.

[0037] Next, an energy spectrum is constructed based on the extracted non-stationary fluctuation components of the phase angle. The energy spectrum is a tool used to intuitively represent the energy distribution of phase angle fluctuations at different frequencies. When constructing the energy spectrum, the phase angle fluctuation signal in the time domain is converted to the frequency domain using mathematical methods such as Fourier transform, and the energy value carried by the phase angle fluctuation at different frequencies is calculated according to the corresponding energy calculation formula. By plotting the curves of these energy values ​​changing with frequency, a complete energy spectrum is formed. On this basis, the integral value of the energy spectrum within the preset defect characteristic frequency band is further calculated. The preset defect characteristic frequency band is determined through a large number of experiments and theoretical analysis, and different types of defects correspond to different frequency band ranges. The integral value actually represents the sum of the phase angle fluctuation energy of all frequencies within this specific frequency band. By comparing with the pre-set reference energy threshold, when the integral value exceeds the reference energy threshold, it can be determined that there is an unfused defect at the molten interface. This is because unfused defects can cause abnormal changes in dielectric properties, which can cause significant changes in the phase angle fluctuation energy within a specific frequency band.

[0038] Finally, deeply analyze the frequency position of the peak in the energy spectrum. When there is a low-frequency energy peak in the energy spectrum, such as an energy peak appearing in the low-frequency range from 1 kHz to 10 kHz, according to a large amount of experimental data and empirical summaries, the unfused defect corresponding at this time is most likely a macroscopic crack defect. This is because the existence of macroscopic cracks will have a wide range of effects on the electric field distribution, causing obvious peaks in the phase angle fluctuation energy in the low-frequency band. When there is a high-frequency energy peak in the energy spectrum, for example, an energy peak appears in the high-frequency range from 1 MHz to 10 MHz, then the unfused defect in this case is usually a microscopic bubble / impurity defect. Microscopic bubbles or impurities will cause slight perturbations in the electric field in a local area, and such slight perturbations are more easily detected in the high-frequency band, thereby resulting in peaks in the phase angle fluctuation energy in the high-frequency band. By accurately judging the frequency position of the peak in the energy spectrum, different types of unfused defects existing at the melting interface can be precisely identified, providing a strong basis for subsequent quality assessment and repair measures.

[0039] In another preferred embodiment of the present invention, the arrangement mode of the distributed electrode group is as follows: In the axial direction of the electrothermal melting sleeve, n groups of annular electrodes are reasonably arranged. Here, n is a preset value determined according to various factors such as the length of the electrothermal melting sleeve and the required detection accuracy. For example, for a longer electrothermal melting sleeve, the value of n will increase correspondingly to ensure that the entire axial range can be fully covered. It should be noted that there is an axial overlap in the detection areas of adjacent annular electrodes. This axial overlap design is of great significance. It avoids the emergence of detection blind spots, enabling continuous and non-missing monitoring of the connection part in the axial direction. For example, if the detection area of a group of annular electrodes covers an axial length of L, then the detection areas of adjacent annular electrodes will have a certain proportion (such as 30%) of overlap, that is, the overlap length is 0.3L. In this way, it can be ensured that every part in the entire axial length can be detected by at least two groups of annular electrodes, greatly improving the reliability of detection.

[0040] Each group of annular electrodes further includes m layers of concentric annular electrodes with different radial directions. The value of m is also a preset value determined according to the detection requirements at different radial depths. Each layer of concentric annular electrode has a specific detection task. From the inner wall to the outer wall of the electrothermal melting sleeve, different layers of concentric annular electrodes can respectively sense the changes in electrical properties at different radial depth positions. In each layer of annular electrode, there are o electrode units. The number of o is also a preset value determined according to the detection resolution required in the circumferential direction. These electrode units are evenly distributed on the annular electrode in a circumferentially symmetric manner. For example, if the circumference of a layer of annular electrode is C, when o electrode units are evenly distributed, the arc length between adjacent electrode units is C / o. This uniform distribution method ensures the balance and comprehensiveness of detection in the circumferential direction.

[0041] More critically, each electrode unit is connected to a switching circuit. This switching circuit is like an intelligent control hub. Through its flexible control function, it can combine different electrode units in a variety of ways. For example, at a certain moment, several adjacent electrode units in the same layer can be combined into a detection circuit to detect the local electrical conditions at that circumferential position; at another moment, electrode units in different layers and different circumferential positions can be combined to form a cross-layer and cross-circumferential detection circuit for analyzing the electrical relationship between different radial and circumferential positions. Through such rich and diverse combination methods, multiple independent detection circuits can be formed. These independent detection circuits can detect the connection part of the electrothermal fusion sleeve from different angles and dimensions, collect a large amount of electrical data, provide sufficient and comprehensive data support for subsequent defect judgment and analysis, and greatly improve the flexibility and accuracy of the monitoring system.

[0042] In a preferred case of this embodiment, the spatial distribution of locating the melting defect includes axial positioning, radial positioning, and circumferential positioning. The positioning process is as follows: Axial positioning: Utilize the special design of the axially overlapping detection area to deeply analyze the change of the impedance response gradient in this area. The specific operation starts from a certain section of the melting interface and carefully conducts detection work along the axial direction. During the detection process, continuously monitor the change of the impedance response gradient value. The preset value mentioned here is a key reference value comprehensively determined through a large amount of experimental data and theoretical research on the connection of electrothermal fusion sleeves for polyurethane directly buried heat-insulated pipes. Once it is identified during the detection process that the impedance response gradient increases and exceeds the preset value at a certain position, this position is accurately marked as the starting point of the axial defect. Subsequently, the detection work continues to steadily advance along the axial direction until at a certain position, the impedance response gradient value decreases to below the preset value, and this position is immediately marked as the ending point of the axial defect. In this way, the area between the starting point of the axial defect and the ending point of the axial defect is the positioning area of the defect in the axial direction. For example, if the preset value is set to X, during axial detection, when the impedance response gradient at a certain position suddenly increases from a value less than X to X + ΔX (ΔX is the increment exceeding the preset value), this position is the starting point; in subsequent detection, when the impedance response gradient at a certain position decreases from a value greater than X to X - ΔX (ΔX is the decrement below the preset value), this position is the ending point, and the area between these two points is the axial defect positioning area.

[0043] Radial positioning: After completing the axial positioning, focus on the determined axial positioning area. Concentric annular electrodes of different layers are distributed in this area, and impedance response data of these different layers of concentric annular electrodes are collected. Since the distances between the concentric annular electrodes at different radial positions and the molten layer are different, their sensitivities to defects are also different. Based on the impedance response differences of the concentric annular electrodes at different radial positions, a special algorithm is used for calculation. For example, by comparing the characteristics of the impedance response curves of different layers of electrodes and calculating the proportion of impedance differences, etc., the concentric annular electrode layer where the defect is located can be accurately deduced. The radial range corresponding to this concentric annular electrode layer is the radial positioning range of the defect. For example, if there are three layers of concentric annular electrodes with impedance responses Z1, Z2, and Z3 respectively, by analyzing the difference relationship between them, such as the comparison of the magnitudes of |Z1 - Z2| and |Z2 - Z3|, and combining the known distance relationship between different layers of electrodes and the molten layer, the specific electrode layer where the defect is located can be determined, and then the radial positioning range of the defect can be determined.

[0044] Circumferential positioning: After clarifying the axial positioning area and the radial positioning range, collect the impedance response data generated by each detection loop within this range. These detection loops are composed of different combinations of electrode units, and their impedance responses can reflect the characteristics of the molten layer in the circumferential direction. By deeply analyzing the impedance response differences of the detection loops, a quantitative method is used to calculate the uniformity of the molten layer in the circumferential direction. The set ratio here is also an important index determined through a large number of experiments and data analyses. If it is found during the analysis that the impedance difference between any adjacent electrode units exceeds the set ratio, for example, the set ratio is Y%, when the ratio obtained by dividing the impedance difference between adjacent electrode units by the impedance value of one of the electrode units is greater than Y%, it can be determined that there is a molten defect in the corresponding circumferential area. In this way, the position of the defect can be accurately determined in the circumferential direction, and the all-round precise positioning of the spatial distribution of the molten defect can be achieved.

[0045] In another preferred embodiment of the present invention, the method for constructing the preset molten interface uniformity model includes: First, carry out relevant work under standard experimental conditions. The so-called standard experimental conditions are carefully designed and strictly controlled, with various parameters such as environmental temperature, humidity, and pressure maintained in a stable and repeatable state. Under such conditions, comprehensively monitor and collect data on the completely sealed molten interface. During the entire electrothermal sleeve heating process, different heating stages are covered, such as the initial heating period, rapid temperature rise period, constant temperature melting period, etc., and at different temperatures, carefully record the change of the real part impedance over time. Through long-term, multi-stage, and multi-temperature point records, a large amount of real part impedance-time data is collected, and then these data are sorted and analyzed to plot the real part impedance decay curve. This curve intuitively reflects the change trend of the real part impedance over time under different heating conditions, providing important basic data for subsequent model construction. At the same time, analyze and extract the fluctuation of the imaginary part phase angle at different frequencies. By using professional signal processing algorithms and tools, summarize and generalize the fluctuation characteristics of the imaginary part phase angle at different frequencies to form an imaginary part phase fluctuation template. Integrate relevant data such as the real part impedance decay curve and the imaginary part phase fluctuation template to generate an impedance reference database. This database contains various electrical property data under the condition of an ideal uniform molten interface, providing a rich and reliable reference basis for the training and verification of subsequent models.

[0046] The molten interface uniformity model mainly consists of two key parts, namely, the relationship model between the real part impedance gradient and the molten layer thickness, and the imaginary part phase fluctuation pattern classification model.

[0047] For the construction of the relationship model between the real part impedance gradient and the molten layer thickness, input the real part impedance data in the impedance reference database into a convolutional neural network. A convolutional neural network is a powerful deep learning model with excellent feature extraction and learning capabilities. By allowing the convolutional neural network to learn the real part impedance data, it can automatically identify the change characteristics of the real part impedance at different positions and under different conditions. For example, in different heating stages, different temperature environments, and different material properties, etc., the change laws of the real part impedance may vary. Through learning and analyzing a large amount of data, the convolutional neural network can capture these subtle change characteristics and establish a non-linear mapping relationship between the real part impedance gradient and the molten layer thickness. This non-linear mapping relationship takes into account the influence of various complex factors and can more accurately describe the internal connection between the real part impedance gradient and the molten layer thickness, thereby obtaining the relationship model between the real part impedance gradient and the molten layer thickness.

[0048] For the construction of the imaginary part phase fluctuation pattern classification model, the imaginary part phase fluctuation template data in the impedance reference database is input into the support vector machine algorithm. These template data contain the imaginary part phase fluctuation patterns corresponding to different defect types. For example, different defect types such as macroscopic cracks, microscopic bubbles, impurities, etc. will cause the imaginary part phase angle to exhibit different fluctuation characteristics. The support vector machine algorithm is an effective classification algorithm that performs classification training on the template data. During the training process, the support vector machine algorithm analyzes the imaginary part phase fluctuation characteristics corresponding to different defect types, finds out the differences and regularities between them. Then, the imaginary part phase fluctuation characteristics corresponding to these different defect types are stored to generate several phase fingerprint recognition libraries. These phase fingerprint recognition libraries are like a "defect feature dictionary", which can quickly and accurately determine the possible defect types according to the actually monitored imaginary part phase fluctuation conditions, thereby obtaining the imaginary part phase fluctuation pattern classification model.

[0049] Finally, the real-time monitoring data is input into the trained molten interface uniformity model. The model will analyze and evaluate the real-time data based on the knowledge learned and trained before. For the real part impedance data, a molten quality score will be output, which intuitively reflects the quality of the current molten interface in terms of the real part impedance. For the imaginary part phase data, a defect confidence parameter will be output, which represents the matching degree between the current imaginary part phase fluctuation condition and the known defect types, thus helping to judge whether there are defects and the likelihood of defects.

[0050] In another preferred embodiment of the present invention, it further includes: On the basis of having completed the accurate axial positioning, radial positioning, and circumferential positioning of defects in the early stage, the important parameter of heating time is fully considered. By collecting the position data of defects in the axial, radial, and circumferential directions at different heating moments, and using advanced three-dimensional modeling algorithms and data analysis techniques, a three-dimensional defect evolution model is established.

[0051] This model has powerful functions. It can dynamically present the growth rate of the defect volume with heating time. For example, in the initial stage of heating, by measuring and calculating the size changes of defects in three dimensions at different time points and using the volume calculation formula, the growth value of the defect volume in this stage is obtained, and further the growth rate of the volume with time is calculated. As the heating process progresses, these data are continuously updated to obtain a complete curve of the growth rate change of the defect volume with heating time.

[0052] Meanwhile, the model can also accurately depict the change trajectory of the defect shape and position. During the electrothermal sleeve heating process, the shape of the defect may gradually change from its initial irregular state. For example, tiny cracks may extend and expand over time, and bubbles may fuse or deform. By continuously tracking and analyzing the axial, radial, and circumferential positioning data, the model visually displays the changes in the defect shape in all directions in a three-dimensional manner, as well as the position movement trajectory of the defect in space.

[0053] The above has described in detail one embodiment of the present invention, but the content described is only the preferred embodiment of the present invention and cannot be considered as limiting the scope of implementation of the present invention. All equivalent changes and improvements made within the scope of the application of the present invention should still fall within the scope covered by the patent of the present invention.

Claims

1. A method for monitoring the connection of an electric heat-melting sleeve of a polyurethane direct-buried thermal insulation pipe, characterized in that: The following steps are involved: Distributed electrode groups are embedded in the annular contact surfaces on both sides of the electric heat-melting sleeve, and the distributed electrode groups and the conductive reinforcement structure inside the polyurethane insulation layer form an electrical coupling network; During the heating process of the electric hot melt sleeve, a dynamically modulated broadband alternating current signal is applied to the electrode group, wherein the frequency range of the broadband alternating current signal covers the dielectric relaxation frequency band of the molten interface; The voltage response signal of each electrode pair in the distributed electrode group is collected in real time, and the complex impedance spectrum of the electric heat-melting sleeve is solved based on the voltage response signal, wherein the complex impedance spectrum includes a real impedance component and an imaginary impedance component; analyzing the nonlinear attenuation characteristics of the real impedance component, and generating a melting defect warning signal if the nonlinear attenuation characteristics do not conform to an expected rule; By analyzing the fluctuation law of the phase angle of the imaginary impedance component with frequency, the dielectric anomaly characteristics of the microstructure of the molten interface are extracted; the dielectric anomaly characteristics are matched with a preset molten interface uniformity model, and if the matching result exceeds a preset tolerance range, a molten defect warning signal is generated; The spatial distribution of melting defects is located based on the impedance response of the distributed electrode group.

2. The method for monitoring the connection of an electric hot-melt sleeve of a polyurethane direct-buried thermal insulation pipe according to claim 1 is characterized in that: The conductive reinforcement structure is a metal mesh or carbon fiber array embedded in the polyurethane insulation layer. The metal mesh or carbon fiber array is evenly distributed along the circumference of the polyurethane insulation layer and forms a full circumferential electrical connection with the annular electrode mounting surface of the distributed electrode group through welding or conductive adhesive.

3. The method for monitoring the connection of an electric heat-melting sleeve of a polyurethane direct-buried thermal insulation pipe according to claim 1 is characterized in that: The dynamic modulation method of the broadband alternating current signal includes: The time interval during which the electric heating sleeve is powered on for heating is marked as [t1, t2], and the broadband alternating current signal includes a low-frequency continuous frequency sweep signal and a high-frequency pulse modulation signal; When in the time interval [t1, t0], t0 is a preset value, and a low-frequency continuous sweep signal is applied, wherein the frequency range of the low-frequency band corresponds to the response frequency band of the overall conductivity of the polyurethane material in the molten state; When in the time interval (t0, t2], a high-frequency pulse modulation signal is superimposed on the low-frequency continuous sweep signal, wherein the frequency range of the high-frequency band corresponds to the local dielectric relaxation frequency band caused by microscopic defects of the molten interface, wherein the microscopic defects include bubbles or impurities; During the heating process of the electric hot melt sleeve, the impedance response data of the electrode group is collected in real time, and the interval size of the frequency change in the low-frequency continuous sweep signal and the ratio of the pulse duration to the period in the high-frequency pulse modulation signal are adjusted according to the impedance response data, so that the broadband alternating current signal is adaptively matched with the melting process of the polyurethane material.

4. The method for monitoring the connection of an electric heat-melting sleeve of a polyurethane direct-buried thermal insulation pipe according to claim 1 is characterized in that: The calculation of the complex impedance spectrum specifically includes: Perform synchronous detection processing on the voltage response signal, separate the part of the voltage response signal that is in the same frequency and phase as the excitation signal and mark it as the real voltage component, and separate the part that is orthogonal to the excitation signal and mark it as the imaginary voltage component; The real impedance is calculated through the real voltage component, and the rate of change of the real impedance in the radial space is obtained to construct a dynamic distribution model of the molten layer thickness; the imaginary impedance is calculated based on the imaginary voltage component, and the value of the imaginary impedance is represented in the form of a cloud map with the spatial position as the coordinate, to generate a dielectric property distribution cloud map, and the local extreme point of the imaginary impedance in the cloud map is obtained.

5. The method for monitoring the connection of an electric hot-melt sleeve of a polyurethane direct-buried thermal insulation pipe according to claim 4 is characterized in that: The process of extracting dielectric anomaly features of the melt interface microstructure includes: Perform multi-scale time-frequency analysis on the imaginary impedance component, decompose the imaginary impedance component signal into different frequency components, obtain the fluctuation changes of different frequency components at different time points, and extract the non-stationary fluctuation components of the phase angle in a specific frequency band; According to the extracted non-stationary fluctuation component of the phase angle, an energy spectrum is constructed, the energy spectrum is used to represent the energy distribution of the phase angle fluctuation at different frequencies, and the integral value of the energy spectrum within the preset defect characteristic frequency band is calculated, the integral value represents the phase angle fluctuation energy of all frequencies within the frequency band; when the integral value exceeds the reference energy threshold, it is determined that there is an unfused defect at the molten interface; The frequency position of the peak value in the energy spectrum is obtained. When a low-frequency energy peak appears in the energy spectrum, the unfused defect is a macro crack defect; when a high-frequency energy peak appears in the energy spectrum, the unfused defect is a micro bubble / impurity defect.

6. The method for monitoring the connection of an electric heat-melting sleeve of a polyurethane direct-buried thermal insulation pipe according to claim 1, characterized in that: The arrangement of the distributed electrode group is as follows: In the axial direction of the electric hot melt sleeve, n groups of annular electrodes are arranged, and the detection areas of adjacent annular electrodes overlap axially; each group of annular electrodes includes m layers of concentric annular electrodes with different radial directions, each layer of annular electrodes includes o electrode units, and the electrode units are distributed on the annular electrodes in a circularly symmetrical manner, n, m and o are preset values, and each electrode unit is connected to a switching circuit. Through the control of the switching circuit, different electrode units are combined in different ways to form multiple independent detection circuits.

7. The method for monitoring the connection of an electric heat-melting sleeve of a polyurethane direct-buried thermal insulation pipe according to claim 6, characterized in that: The spatial distribution of locating melting defects includes axial positioning, radial positioning and circumferential positioning. The positioning process is as follows: Obtain the impedance response gradient change of the axial overlapping detection area, start axial detection from one end of the molten interface, and when the impedance response gradient at any position is identified to increase beyond the preset value, mark the position as the starting point of the axial defect, and continue to identify along the axial direction until any position is lower than the preset value, then mark the position as the end point of the axial defect, and the area between the defect starting point and the defect end point is the axial positioning area of ​​the defect; Obtaining impedance responses of different layers of concentric annular electrodes in the axial positioning area, and calculating the concentric annular electrode layer where the defect is located according to the difference in impedance responses of the concentric annular electrodes at different radial positions, which is the radial positioning range of the defect; The impedance response generated by each detection circuit within the radial positioning range of the axial positioning area is obtained. By analyzing the impedance response differences of the detection circuits, the uniformity of the molten layer in the circumferential direction is calculated. If the impedance difference between any adjacent electrode units exceeds the set ratio, it is determined that the corresponding circumferential area has a melting defect.

8. The method for monitoring the connection of an electric heat-melting sleeve of a polyurethane direct-buried thermal insulation pipe according to claim 1, characterized in that: The method for constructing the preset melting interface uniformity model includes: Under standard experimental conditions, the completely sealed molten interface is monitored and data is collected. At different heating stages and temperatures, the change of real impedance over time is recorded to form a real impedance attenuation curve. The fluctuation pattern of the imaginary phase angle at different frequencies is extracted to form an imaginary phase fluctuation template and generate an impedance benchmark database. The molten interface uniformity model includes a relationship model between the real impedance gradient and the molten layer thickness and an imaginary phase fluctuation mode classification model; The real impedance data in the impedance benchmark database is input into the convolutional neural network to learn the variation characteristics of the real impedance at different positions and under different conditions, establish a nonlinear mapping relationship between the real impedance gradient and the thickness of the molten layer, and obtain a relationship model between the real impedance gradient and the thickness of the molten layer; Inputting the imaginary phase fluctuation template data in the impedance reference database into the support vector machine algorithm, the template data contains the imaginary phase fluctuation patterns corresponding to different defect types, the support vector machine algorithm performs classification training on the template data, stores the imaginary phase fluctuation features corresponding to different defect types, generates several phase fingerprint recognition libraries, and obtains the imaginary phase fluctuation pattern classification model; The real-time monitoring data is input into the trained molten interface uniformity model, and the molten quality score of the real impedance data and the defect confidence parameter of the imaginary phase data are output.

9. The method for monitoring the connection of electric heat-melting sleeves of polyurethane direct-buried thermal insulation pipes according to claim 7, characterized in that: Also includes: Based on the axial positioning, radial positioning, circumferential positioning and heating time parameters, a three-dimensional defect evolution model is established, and the three-dimensional defect evolution model includes the growth rate of the defect volume with the heating time and the change trajectory of the shape and position.

Images