A factory compressed air leakage intelligent detection and plugging control device and method

Abstract

The application discloses a kind of factory compressed air leakage intelligent detection and plugging control device and method, belong to compressed air system intelligent control field.The device includes multidimensional perception network module, for collecting pipe network operating parameter and leakage acoustic wave / vibration waveform data;Leakage diagnosis and feature extraction module, built-in pipe network digital twin model, for leakage positioning and aperture estimation and extract leakage characteristic vector;Medium injection and induction control module, for selecting atomized plugging medium according to leakage aperture and leakage characteristic vector and calculating injection parameter;Medium injection execution module, for spraying atomized plugging medium to leakage point area.The device uses the local low temperature generated by gas throttling expansion at leakage hole, so that the atomized plugging medium entering the leakage hole is phase change frozen to form embolus plugging body in microchannel, realizes phase change self-healing plugging without interrupting downstream gas supply, and evaluates plugging effect and supplementary injection through leakage characteristic vector energy index.

Description

Technical Field

[0001] This invention belongs to the field of intelligent control technology for compressed air systems, specifically relating to an intelligent detection and sealing control device and method for compressed air leakage in factories. Background Technology

[0002] Compressed air, as a crucial power source in modern industry, is widely used in automated production lines in sectors such as automotive manufacturing, electronics, food, and pharmaceuticals. Factory compressed air pipeline networks typically consist of main pipes, branch pipes, and end-user equipment. These networks cover a large area and have numerous nodes. Over long-term operation, leaks are inevitable due to pipe corrosion, loose joints, and aging seals. Leaks waste energy, cause pressure fluctuations in the pipeline network, affect the normal operating pressure of end-user equipment, and consequently reduce production efficiency and product quality.

[0003] In existing technologies, two main methods are used to manage leaks in compressed air pipelines. One is a leak detection and shutdown scheme based on a digital twin model. This scheme uses sensors such as pressure, flow, and sound waves to collect pipeline operating parameters in real time, and combines this data with the pipeline's digital twin model for leak diagnosis and location. When a leak is detected, the shut-off valve of the branch containing the leak is automatically shut off, eliminating the leak through physical isolation. The other method is a live leak sealing scheme, which uses methods such as clamp injection or welding repair to mechanically seal the leak from the outside of the pipeline without interrupting airflow.

[0004] However, the existing technologies mentioned above still have the following drawbacks: whether it is physical isolation by shutting off the valve or mechanical sealing from the outside of the pipe wall, it is necessary to interrupt or partially interrupt the downstream gas supply, or rely on manual on-site operation, and the sealing effect is limited by factors such as the leakage orifice diameter, pipe material and operator experience. It is difficult to achieve accurate and reliable in-situ sealing of leaks without stopping production, reducing pressure and shutting down the pipeline, resulting in the long-term inability to effectively resolve the contradiction between energy conservation and production continuity. Summary of the Invention

[0005] The present invention aims to provide an intelligent detection and sealing control device and method for compressed air leakage in factories, so as to solve the technical problem that it is difficult to achieve in-situ sealing of leakage without stopping production, reducing pressure, or shutting down pipelines in the prior art.

[0006] The objective of this invention can be achieved through the following technical solutions: A smart detection and sealing control device for compressed air leakage in factories includes: The multi-dimensional sensing network module is used to collect pipeline network operating parameters and leakage sound / vibration waveform data in real time; The leak diagnosis and feature extraction module has a built-in digital twin model of the pipeline network, which is used to integrate the operating parameters to judge leaks, locate leak points and estimate leak apertures, and extract leak feature vectors from the leak sound / vibration waveform data in real time. The medium injection and induction control module is connected to the leakage diagnosis and feature extraction module; The medium injection execution module is connected to the medium injection and induction control module and is used to inject atomized sealing medium into the leak point area. The medium injection and induction control module is configured as follows: Based on the estimated leakage orifice diameter and the leakage characteristic vector, a corresponding atomized sealing medium is selected from a preset medium library, and the injection parameters required for the atomized sealing medium to enter the leakage orifice and undergo phase change solidification in the microchannel of the leakage orifice are calculated based on the pipeline digital twin model. The medium injection execution module is controlled to inject the atomized sealing medium into the leak point area according to the injection parameters. The local low temperature generated by the gas throttling and expansion at the leak hole causes the atomized sealing medium entering the leak hole to undergo phase change and solidify in the microchannel of the leak hole, forming a plug-like sealing body. After completing the current injection operation, the sealing effect is evaluated by the change in the energy index in the leakage characteristic vector, and the injection parameters are adjusted for supplementary injection if the sealing is not completely successful.

[0007] As a preferred technical solution of the present invention, it further includes an online aerodynamic parameter identification module, which is connected to the leakage diagnosis and feature extraction module and the medium injection and induction control module respectively; the online aerodynamic parameter identification module uses a pre-trained neural network model, takes a continuous multi-frame leakage feature vector sequence as input, outputs the aerodynamic damping coefficient and effective flow area of ​​the leakage hole, and updates the local impedance of the leakage node in the pipeline network digital twin model in real time accordingly; The media injection and induction control module is configured to calculate the injection parameters based on the updated pipeline digital twin model.

[0008] As a preferred technical solution of the present invention, the leakage feature vector extracted from the leakage acoustic / vibration waveform data by the leakage diagnosis and feature extraction module includes at least three of the following: spectral centroid, energy attenuation rate, short-time energy, spectral kurtosis, energy entropy, and energy ratio of each frequency band; the neural network model is a long short-term memory network or a temporal convolutional network.

[0009] As a preferred technical solution of the present invention, when the media injection and induction control module calculates the injection parameters based on the pipeline network digital twin model, it uses the airflow velocity field and temperature field at the leak hole as a basis to simulate and calculate the motion trajectory of atomized sealing media particles of different sizes and different injection velocities in the leak jet field and the probability of being captured by the leak hole, and selects the parameter combination that makes the capture probability the highest as the injection direction, injection rate and injection dose.

[0010] As a preferred embodiment of the present invention, the medium injection and induction control module is further configured to: The leakage characteristic vector energy index within multiple consecutive time windows after injection is compared with the baseline value before injection. The energy decay rate is obtained based on the comparison results. When the energy decay rate exceeds a preset threshold and remains stable, the sealing is deemed successful. Otherwise, the medium type or injection dose is adjusted for supplementary injection.

[0011] As a preferred technical solution of the present invention, it also includes a data interface module for obtaining the operating status and production task scheduling information of each gas-using device from the factory production management system in real time; the medium injection and induction control module also includes a graded processing unit, which is configured to drive the shut-off valve of the branch to close to implement physical hard shutdown when it is determined that the phase change self-healing sealing is successful and the data interface module learns that all gas-using devices downstream of the branch where the leak point is located have switched to standby or no-task state.

[0012] As a preferred technical solution of the present invention, the atomized blocking medium in the preset medium library includes at least one of nano-sized hygroscopic powder, low-temperature coagulating aerosol precursor, and high-viscosity condensable droplets. The media injection and induction control module is configured to select the corresponding sealing medium type and particle size distribution based on the estimated leakage orifice diameter.

[0013] As a preferred embodiment of the present invention, the media injection execution module includes an inspection robot mounted on a pipeline track and a nozzle mounted on the multi-degree-of-freedom robotic arm of the inspection robot. And / or, the media injection execution module includes an array of miniature nozzles that can be remotely triggered and directionally injected at multiple fixed positions distributed along the pipeline network.

[0014] A method for intelligent detection and sealing of compressed air leaks in factories, the method comprising the following steps: S1: Real-time acquisition of pipeline network operating parameters and leakage sound / vibration waveform data; use the pipeline network digital twin model to detect and locate leaks and estimate leak apertures; and extract leakage feature vectors from the waveform data in real time. S2: Based on the estimated leakage orifice diameter and the leakage characteristic vector, select the atomized sealing medium from the preset medium library, and calculate the injection parameters required for the atomized sealing medium to enter the leakage orifice and undergo phase change solidification in the microchannel of the leakage orifice based on the pipeline digital twin model; S3: The driving medium injection execution module injects the atomized sealing medium into the leak point area according to the injection parameters. Utilizing the local low temperature generated by the gas throttling and expansion at the leak hole, the atomized sealing medium undergoes phase change and solidifies in the microchannel of the leak hole, forming a plug-like sealing body, thereby achieving phase change self-healing sealing without interrupting the downstream gas supply. S4: After the injection is completed, the sealing effect is evaluated by the change in the energy index in the leakage characteristic vector. If the sealing is not completely successful, the injection parameters are adjusted for supplementary injection.

[0015] As a preferred embodiment of the present invention, the following steps are included before step S2: Using a pre-trained neural network model, with a continuous multi-frame leakage feature vector sequence as input, the aerodynamic damping coefficient and effective flow area of ​​the leakage orifice are output, and the local impedance of the leakage node in the pipeline network digital twin model is updated in real time accordingly. In step S2, the injection parameters are calculated based on the updated pipeline digital twin model.

[0016] Compared with the prior art, the beneficial effects of this application are as follows: 1. The device of the present invention utilizes the local low temperature generated by the gas throttling and expansion at the leakage hole to cause the atomized sealing medium injected into the leakage point area to undergo phase change and solidify in the microchannel of the leakage hole to form a plug-like sealing body. This achieves phase change self-healing sealing without interrupting the downstream gas supply, avoiding the production interruption problem caused by shutting off valves or mechanical plugging in the prior art.

[0017] 2. The device of the present invention evaluates the sealing effect by the change of energy index in the leakage characteristic vector, and adjusts the injection parameters for supplementary injection when the sealing is not completely successful, thus ensuring the reliability of the sealing and forming a closed-loop control from medium selection, parameter calculation, injection sealing to effect evaluation.

[0018] 3. The local low temperature generated by the throttling and expansion of gas at the leak hole is actively converted into the phase change solidification driving condition of the sealing medium, so that the sealing medium forms an embolism inside the micron-level microchannel of the leak hole. This achieves in-situ self-healing sealing by "using leak to stop leak" without interrupting the downstream gas supply. The sealing effect is evaluated by the leakage characteristic vector energy index, and supplementary injection is performed if it is not completely successful. This effectively solves the technical contradiction in the existing technology that cannot balance energy saving and production continuity.

[0019] 4. The method of the present invention collects pipeline network operating parameters and leakage sound / vibration waveform data in real time, uses a pipeline network digital twin model to detect and locate leaks and estimate leak orifice diameter, extracts leakage feature vectors from waveform data in real time, selects media and calculates injection parameters based on leak orifice diameter and leakage feature vectors, evaluates the sealing effect through energy index after injection, and supplements injection if not completely successful, thus realizing a complete closed-loop control from leak detection to sealing evaluation. Attached Figure Description

[0020] To facilitate understanding by those skilled in the art, the present invention will be further described below with reference to the accompanying drawings.

[0021] Figure 1 This is a system block diagram of the present invention.

[0022] Figure 2 This is a schematic diagram of the method flow of the present invention.

[0023] Figure 3 This is a schematic diagram illustrating the phase change self-healing sealing principle of the present invention.

[0024] Figure 4 This is a schematic diagram of the deployment of the media injection execution module of the present invention. Detailed Implementation

[0025] To further illustrate the technical means and effects of the present invention in achieving its intended purpose, the following detailed description of the specific implementation methods, structures, features, and effects of the present invention, in conjunction with the accompanying drawings and preferred embodiments, is provided.

[0026] Example 1 In existing technologies, two main methods are used to manage leaks in compressed air pipelines. One is a leak detection and shutdown scheme based on a digital twin model. This scheme uses sensors to collect pipeline operating parameters in real time and combines them with the digital twin model of the pipeline to diagnose and locate leaks. When a leak is detected, the shut-off valve of the branch where the leak point is located is automatically shut off to eliminate the leak through physical isolation. The other is a pressurized leak sealing scheme, which uses methods such as clamp injection or welding repair to mechanically seal the leak from the outside of the pipeline without interrupting air supply. However, both of these methods require interruption or partial interruption of downstream air supply, or rely on manual on-site operation. The sealing effect is also limited by factors such as the leak diameter, pipeline material, and operator experience. It is difficult to achieve accurate and reliable in-situ sealing of leaks without interrupting production, reducing pressure, or shutting down the pipeline, resulting in the long-standing inability to effectively resolve the contradiction between energy conservation and production continuity.

[0027] The core improvement of this embodiment lies in: utilizing the passive physical phenomenon of local low temperature generated by gas throttling and expansion at the leak hole, it is actively transformed into a phase change solidification driving condition for the sealing medium. By spraying atomized sealing medium into the leak point area, the medium enters the microchannel of the leak hole and undergoes phase change solidification under the action of local low temperature, forming a plug-like sealing body. This achieves phase change self-healing sealing without interrupting the downstream gas supply, effectively solving the technical contradiction between energy conservation and production continuity.

[0028] like Figures 1 to 4 As shown, this embodiment provides an intelligent detection and sealing control device for compressed air leaks in a factory, including a multi-dimensional sensing network module, a leak diagnosis and feature extraction module, a media injection and induction control module, and a media injection execution module. The multi-dimensional sensing network module is used to collect real-time pipeline operating parameters and leak sound or vibration waveform data from the factory's compressed air system. This multi-dimensional sensing network module includes intelligent flow sensors and pressure sensors deployed on various branches and key nodes of the compressed air pipeline network, as well as ultrasonic sensors or vibration sensors deployed on key monitoring nodes. The ultrasonic sensors and vibration sensors are used to capture high-frequency characteristic signals generated by leaks and are synchronized with the pressure sensors in time to enable subsequent multi-source information fusion and localization.

[0029] The leak diagnosis and feature extraction module is connected to the multi-dimensional sensing network module and has a built-in digital twin model of the pipeline network. This module is used to integrate the operating parameters collected by the multi-dimensional sensing network module to determine leaks, locate leak points, and estimate leak orifice diameters, and to extract leak feature vectors in real time from leak acoustic or vibration waveform data. The pipeline network digital twin model includes the pipeline network topology, pipe segment impedance characteristics, and aerodynamic models of gas-using equipment, and can simulate the pressure distribution and flow distribution within the pipeline network under different operating conditions. The leak feature vectors extracted by the leak diagnosis and feature extraction module from the leak acoustic or vibration waveform data include at least three of the following: spectral centroid, energy attenuation rate, short-time energy, spectral kurtosis, energy entropy, and energy ratio of each frequency band.

[0030] Specifically: the spectral centroid is used to characterize the concentrated frequency of the energy distribution of the leaking acoustic signal, providing a frequency domain reference for leak aperture estimation; the energy attenuation rate reflects the attenuation characteristics of the leaking acoustic signal propagating along the pipe wall, providing distance estimation parameters for leak location; short-time energy and spectral kurtosis are used to assess the severity of the current leak and trigger the plugging decision threshold; energy entropy and energy ratio of each frequency band are used to distinguish different types of leak modes (such as perforation leaks and crack leaks), providing a classification basis for the medium selection unit to select appropriate plugging medium type and particle size distribution. These leak characteristic vectors collectively constitute the data foundation for the entire process of this device, from leak detection, location, aperture estimation, medium selection to plugging effect evaluation.

[0031] The media injection and induction control module is connected to the leak diagnosis and feature extraction module. This module internally includes a media selection unit, an injection parameter calculation unit, and a sealing effect evaluation unit. The media injection execution module is connected to the media injection and induction control module and is used to inject atomized sealing media into the leak area.

[0032] In some embodiments, the media injection execution module is a multi-degree-of-freedom robotic arm end nozzle mounted on an inspection robot that travels along the pipeline track.

[0033] In other embodiments, it may be a micro-nozzle array that can be remotely triggered and directionally sprayed from multiple fixed locations distributed along the pipeline network.

[0034] The working configuration of the media injection and induction control module is as follows.

[0035] First, the media selection unit selects a suitable atomizing plugging medium from a pre-set media library based on the leak pore size estimate and leak feature vector provided by the leak diagnosis and feature extraction module. The pre-set media library contains at least one of the following: nano-sized hygroscopic powder, low-temperature solidifying aerosol precursor, and high-viscosity condensable droplets. The media selection unit selects the corresponding plugging medium type and particle size distribution based on the leak pore size estimate. When the leak pore size is small (equivalent diameter ≤ 1 mm), nano-sized hygroscopic powder is selected to utilize its high permeability and hygroscopic expansion properties. When the leak pore size is medium (equivalent diameter between 1 mm and 3 mm), a low-temperature solidifying aerosol precursor is selected, utilizing its rapid phase transformation capability within the microchannels of the leak pore. When the leakage orifice diameter is relatively large (equivalent diameter between 3mm and 5mm), a combination of high-viscosity condensable droplets and nano-sized hygroscopic powder is selected as the medium. First, the high-viscosity droplets fill the main space of the leakage orifice to form a matrix, and then the nano-sized powder enhances the density of the sealing body, thereby achieving rapid filling and stable solidification. The above-mentioned criteria for "smaller," "medium," and "larger" are based on the equivalent diameter of the leakage orifice, which is calculated by converting the leakage orifice area into the diameter of a circle with an equal area.

[0036] Then, the injection parameter calculation unit calculates the injection parameters required for the atomized sealing medium to enter the leak orifice and undergo phase change solidification within the microchannel of the leak orifice based on the digital twin model of the pipeline network. The injection parameters include injection direction, injection rate, and injection dosage. When calculating the injection parameters, the airflow velocity field and temperature field at the leak orifice are used as a basis to simulate and calculate the motion trajectory of atomized sealing medium particles of different sizes and injection velocities in the leak jet field and the probability of them being captured by the leak orifice. The parameter combination that maximizes the capture probability is then selected.

[0037] Specifically, a local flow field model near the leak hole is established in the digital twin model of the pipeline network. The velocity field distribution of the leak jet is determined based on the leak hole diameter and pipeline pressure. Simultaneously, the temperature field distribution generated by gas throttling expansion is calculated using the Joule-Thomson effect. Based on this, numerical simulations are performed on the trajectories of media particles of different sizes in the jet field. The effects of gas drag, gravity, inertial force, and thermophoretic force on particle motion are considered, and the capture probability of particles at the microchannel inlet of the leak hole is calculated. The formula for calculating the capture probability is: ,in, For capture probability, The number of media particles entering the microchannel of the leak hole. The total number of media particles ejected is denoted as . By iterating through different combinations of parameters with different ejection directions, ejection rates, and ejection doses, the parameter combination that maximizes the capture probability is selected as the final ejection parameters.

[0038] Furthermore, to ensure reliable phase change solidification sealing under various leakage conditions, this device can also be equipped with an auxiliary cooling unit. This auxiliary cooling unit includes an active pre-cooling nozzle and / or a vortex tube micro-refrigeration component located at the front end of the media injection execution module. When the localized low temperature generated by the throttling expansion at the leak orifice is insufficient to rapidly solidify the atomized sealing medium (e.g., when the leak orifice diameter is extremely small resulting in a weak throttling effect, or when the pipeline pressure is low resulting in insufficient temperature drop), the media injection and induction control module can activate the auxiliary cooling unit to pre-cool the atomized sealing medium before injection and / or provide auxiliary cooling to the pipe wall area surrounding the leak orifice, based on the difference between the temperature at the leak orifice calculated by the pipeline digital twin model and the solidification point temperature of the sealing medium. This ensures that the medium entering the leak orifice can successfully complete phase change solidification within the microchannel. The activation threshold of the auxiliary cooling unit can be preset to automatically activate when the estimated temperature at the leak orifice is more than 10°C higher than the solidification point temperature of the sealing medium.

[0039] Meanwhile, to address the challenge of high-speed jets generated by gas throttling at the leak hole making it difficult for atomized sealing medium particles to reverse their entry into the leak hole, this embodiment also includes a flow velocity reduction guide shroud at the front end of the medium injection execution module. This shroud has a conical or arc-shaped structure and is positioned outside the leak hole when close to it. Its internal cavity effectively blocks and diverts the leak jet, reducing the local airflow velocity near the leak hole outlet to 1 / 5 to 1 / 20 of its original value, forming a relatively low-velocity medium capture zone at the leak hole inlet. Within this capture zone, the gas drag force on the atomized sealing medium particles is significantly reduced, allowing the particles to more easily overcome airflow resistance and enter the leak hole microchannel. The inner wall of the flow velocity reduction guide shroud can also be coated with micro-nano-level rough textures or electrostatic adsorption coatings to further improve the retention and capture probability of particles at the microchannel inlet.

[0040] Then, the medium injection execution module injects atomized sealing medium into the leak area according to the calculated injection parameters. Utilizing the localized low temperature generated by the gas throttling and expansion at the leak orifice, the atomized sealing medium entering the leak orifice undergoes a phase change and solidifies within the microchannel of the leak orifice, forming a plug-like seal. This achieves phase change self-healing sealing without interrupting downstream gas supply. Its working principle is as follows: Figure 3 As shown: When high-pressure gas is injected outward through the leak hole, the gas undergoes throttling and expansion at the leak hole, generating a local low-temperature region according to the Joule-Thomson effect. After the medium injection execution module injects atomized sealing medium into the leak point area, the medium particles enter the leak jet field and move along an optimal trajectory to the inlet of the leak hole microchannel under the guidance of the digital twin model simulation, where they are captured. After entering the leak hole microchannel, the medium undergoes phase change and solidification under the action of local low temperature, changing from a liquid or semi-liquid state to a solid state, forming a dense plug-like sealing body inside the leak hole microchannel, thereby inhibiting further gas leakage and achieving phase change self-healing sealing. The entire process does not require shutting down the pipeline, reducing the pipeline pressure, or interrupting the downstream gas supply.

[0041] After injection, the sealing effect evaluation unit assesses the sealing effect by analyzing changes in the energy index within the leakage characteristic vector. Specifically, the evaluation method involves comparing the energy index of the leakage characteristic vector over multiple consecutive time windows after injection with the baseline value before injection. If the energy attenuation rate exceeds a preset threshold and remains stable, the sealing is considered successful; otherwise, the medium type or injection dosage is adjusted for supplementary injection. The formula for calculating the energy attenuation rate is: ,in, Energy decay rate, The energy baseline value of the leakage characteristic vector before injection is taken as the average value of the energy index within multiple consecutive time windows before injection. This represents the energy index value within the current time window after injection. When the threshold is exceeded and the situation remains stable for multiple consecutive time windows, the blocking is considered successful.

[0042] If the sealing is determined to be incompletely unsuccessful, the medium injection and induction control module adjusts the injection parameters to perform supplementary injection. The adjustment strategy determines the cause of the sealing failure based on the changing trend of the energy decay rate sequence.

[0043] If the energy decay rate is high in the early stage after injection but gradually decreases thereafter, it indicates that the plug is eroded and destroyed by the airflow after formation. In this case, choose a medium with higher viscosity or faster solidification rate, and appropriately increase the injection dosage to form a thicker plug.

[0044] If the energy decay rate remains low throughout the monitoring period, it indicates that the atomized sealing medium particles have failed to effectively enter the leak hole. In this case, adjust the spray direction to make the nozzle more accurately align with the leak hole, and increase the spray rate to increase the momentum of the particles, helping the particles overcome the drag of the airflow and enter the leak hole.

[0045] If the energy decay rate fluctuates significantly, it indicates that the sealing body is unstable. In this case, increase the injection dosage and extend the injection duration to make the sealing body denser and more stable. After adjustment, perform supplementary injection according to the new injection parameters and evaluate the sealing effect again until the sealing is successful or the preset maximum number of supplementary injections is reached.

[0046] It should be noted that the pipeline digital twin model involved in this embodiment is not only used for leak detection and location, but also serves as the core simulation platform for calculating injection parameters. During the initial deployment of the device, the model is calibrated based on actual pipeline parameters, including pipe length, pipe diameter, number and location of elbows, valve type and characteristic curves, air compressor performance curves, and air consumption characteristics of each air-consuming device. During normal operation, the model continuously receives real-time data from the multi-dimensional sensing network module and continuously corrects the model parameters through a data assimilation algorithm to maintain consistency with the actual pipeline network state.

[0047] This embodiment also provides a method for intelligent detection and sealing control of compressed air leakage in factories, including the following steps.

[0048] Step S1: Real-time acquisition of pipeline network operating parameters and leakage acoustic or vibration waveform data; use a pipeline network digital twin model for leak detection, location, and orifice estimation; and extract leakage feature vectors from the waveform data in real time. Leak detection and location employs multi-source information fusion, utilizing the pressure wave attenuation time difference and acoustic wave arrival time difference in the pipeline network digital twin model for matching calculations to pinpoint the specific branch and location of the leak. Orifice estimation analyzes the spectral and energy characteristics of the leakage acoustic or vibration waveform data, combined with pressure distribution and flow data from the pipeline network digital twin model, and uses a pre-trained regression model to output an estimated orifice size.

[0049] Step S2: Based on the estimated leak orifice diameter and leak characteristic vector, select an atomized sealing medium from a preset medium library, and calculate the injection parameters required for the atomized sealing medium to enter the leak orifice and undergo phase change solidification within the microchannel of the leak orifice based on the pipeline network digital twin model. The calculation process for the injection parameters is as follows: In the pipeline network digital twin model, combined with the airflow velocity field and temperature field at the leak orifice, simulate and calculate the motion trajectory of atomized sealing medium particles of different sizes and injection velocities in the leak jet field and the probability of them being captured by the leak orifice. Select the parameter combination that maximizes the capture probability as the injection direction, injection rate, and injection dosage.

[0050] Step S3: The driving medium injection execution module injects atomized sealing medium into the leak point area according to the injection parameters. Utilizing the local low temperature generated by the gas throttling and expansion at the leak hole, the atomized sealing medium undergoes phase change and solidifies in the microchannel of the leak hole, forming a plug-like sealing body, thereby achieving phase change self-healing sealing without interrupting the downstream gas supply.

[0051] Step S4: After injection, the sealing effect is evaluated by the change in energy index in the leakage characteristic vector. If the sealing is not completely successful, the injection parameters are adjusted for supplementary injection. The specific method for evaluating the sealing effect is as follows: the energy index of the leakage characteristic vector within several consecutive time windows after injection is compared with the baseline value before injection. Based on the comparison results, the energy attenuation rate is obtained. When the energy attenuation rate exceeds a preset threshold and remains stable, the sealing is considered successful; otherwise, the medium type or injection dosage is adjusted for supplementary injection.

[0052] Example 2 It is understandable that, in Example 1, the pipeline digital twin model relied upon by the media injection and induction control module to calculate injection parameters was calibrated under normal operating conditions, and its local impedance parameters were set based on the normal state of the pipelines and valves. However, when a leak occurs, the physical parameters of the leak hole itself may deviate from the impedance parameters of the leak node in the model. This model mismatch will lead to a decrease in the calculation accuracy of the injection parameters, thereby affecting the probability of the sealing medium being captured by the leak hole and the success rate of sealing on the first attempt.

[0053] The core improvement of this embodiment lies in the addition of an online aerodynamic parameter identification module based on Embodiment 1. The neural network model is used to identify the aerodynamic damping coefficient and effective flow area of ​​the leaking hole online based on the real-time extracted leakage feature vector. The local impedance of the leaking node in the pipeline network digital twin model is updated in real time accordingly, so that the calculation of injection parameters is based on the updated high-fidelity model, thereby significantly improving the accuracy of injection parameters and the success rate of sealing.

[0054] like Figure 1 As shown, to address the aforementioned problems, this embodiment provides an intelligent detection and sealing control device for compressed air leaks in factories. Based on Embodiment 1, it further includes an online aerodynamic parameter identification module. This module is connected to both the leak diagnosis and feature extraction module and the media injection and induction control module.

[0055] The online aerodynamic parameter identification module utilizes a pre-trained neural network model. Taking a continuous multi-frame sequence of leakage feature vectors as input, it outputs the aerodynamic damping coefficient and effective flow area of ​​the leaking orifice, and updates the local impedance of the leaking node in the pipeline network digital twin model in real time accordingly. The neural network model can be a Long Short-Term Memory (LSTM) network or a temporal convolutional network. Taking the LSM network as an example, its input is a continuous multi-frame sequence of leakage feature vectors. Each frame of the leakage feature vector includes at least three feature parameters from the following: spectral centroid, energy decay rate, short-time energy, spectral kurtosis, energy entropy, and energy ratio of each frequency band. The network captures the dynamic evolution of the leakage feature vectors over time through a forget gate, input gate, and output gate, outputting the aerodynamic damping coefficient and effective flow area of ​​the leaking orifice. The aerodynamic damping coefficient characterizes the damping effect of the leaking orifice on the airflow and is related to the geometry of the leaking orifice, wall roughness, and airflow state; the effective flow area reflects the actual opening degree of the leaking orifice. By substituting the identified aerodynamic damping coefficient and effective flow area into the local impedance equation of the leaking node in the digital twin model of the pipeline network, the local impedance parameters of that node can be updated. The expression for the local impedance equation is: ,in, The local impedance of the leaking node is given. This is the aerodynamic damping coefficient. For effective circulation area, The imaginary unit, The frequency of airflow pulsation. This is the local inertia coefficient.

[0056] The media injection and induction control module calculates injection parameters based on the updated digital twin model of the pipeline network, making the physical parameters of the leak holes on which the simulation calculation is based closer to the real state, thereby significantly improving the accuracy of the injection parameters.

[0057] It should be noted that the neural network model is trained using a combination of offline pre-training and online fine-tuning. In the offline pre-training phase, leakage feature vector sequences and corresponding measured aerodynamic damping coefficients and effective flow areas from a large database of laboratory simulation experiments or historical leakage case databases are used as training samples. The model is trained by minimizing the mean square error loss function between the predicted and measured values. In the online fine-tuning phase, after the actual sealing is completed, the leakage feature vector sequence of this leakage event and the actual leakage orifice parameters inferred from the sealing effect are used as new samples to incrementally update the model parameters. This allows the model to continuously adapt to different operating conditions and leakage types, improving identification accuracy and generalization ability.

[0058] This embodiment also provides a method for intelligent detection and sealing of compressed air leaks in factories. Before step S2 of the method in embodiment one, it further includes: using a pre-trained neural network model, taking a continuous multi-frame leakage feature vector sequence as input, outputting the aerodynamic damping coefficient and effective flow area of ​​the leak hole, and updating the local impedance of the leak node in the pipeline network digital twin model in real time accordingly; in step S2, the injection parameters are calculated based on the updated pipeline network digital twin model.

[0059] Example 3 Understandably, in Example 2, after the sealing was completed, the medium injection and induction control module evaluated the sealing effect and performed supplementary injection if it was not completely successful. However, the plug-like seal formed by phase change self-healing sealing is essentially a temporary seal, and its long-term durability may be affected by factors such as pipeline pressure fluctuations, temperature changes, and medium aging. Maintaining a temporary seal is necessary when the downstream gas-using equipment on the branch where the leak is located is in operation; however, when all downstream equipment is switched to standby or no-task status, if the leak can be physically shut off to completely isolate it, more favorable conditions can be created for further reducing leakage and subsequent permanent repair.

[0060] The core improvement of this embodiment is that, based on embodiment two, a data interface module and a hierarchical processing unit are added. By communicating with the factory production management system, the real-time operating status and production task scheduling information of downstream gas-using equipment are obtained. After the phase change self-healing sealing is successful, it automatically determines whether all downstream equipment has switched to standby or no-task state, and automatically drives the shut-off valve to close when the conditions are met, so as to achieve a seamless transition from temporary sealing to permanent isolation and form a complete sealing strategy combining soft and hard.

[0061] like Figure 1 As shown, to address the aforementioned problems, this embodiment provides an intelligent detection and sealing control device for compressed air leaks in factories, which, based on Embodiment 2, also includes a data interface module. The data interface module is used to obtain real-time operating status and production task scheduling information of each air-consuming device from the factory production management system. The factory production management system can be a Manufacturing Execution System or an Enterprise Resource Planning System. The data interface module interacts with it via a standard industrial communication protocol, obtaining information including the start / stop status of each air-consuming device, the currently executing production task number, task priority, estimated remaining completion time of the task, and production scheduling plans for future periods.

[0062] The medium injection and induction control module also includes a graded processing unit, which is connected to the sealing effect evaluation unit and the data interface module. The working logic of the graded processing unit is as follows: When the sealing effect evaluation unit determines that the phase change self-healing sealing is successful, the graded processing unit first maintains a monitoring state, continuously monitoring the leakage characteristic vector energy index through the multi-dimensional sensing network module to ensure the continued effectiveness of the temporary sealing. Simultaneously, the graded processing unit continuously queries the operating status of each gas-consuming device downstream of the branch where the leak point is located through the data interface module. When all downstream gas-consuming devices are switched to standby or no-task status, and production scheduling information confirms that no new production tasks will be assigned to this branch within a preset future time period, the graded processing unit generates a shutdown command, driving the shut-off valve of this branch to close to implement a physical hard shutdown, thereby eliminating the leakage hazard through physical isolation.

[0063] It should be noted that when the hierarchical processing unit confirms that there are no subsequent tasks for downstream equipment, it can send a confirmation request to the factory production management system through the data interface module. After receiving confirmation or automatic authorization from the production management system, it triggers a shutdown command to avoid accidental shutdown due to delays in scheduling information updates. After the physical hard shutdown is executed, the hierarchical processing unit records the shutdown event to the visual traceability module, including information such as shutdown time, shutdown branch number, duration of blocking before shutdown, and cumulative energy-saving data.

[0064] Furthermore, to ensure the feasibility of this solution in industrial settings, it should be noted that the applicable range for leak hole equivalent diameters is 0.01mm to 5mm, with a preferred range of 0.05mm to 3mm. In compressed air pipelines, common leak holes caused by corrosion perforation and loose joints typically have equivalent diameters between 0.05mm and 3mm, falling within the preferred range for this solution. Common compressed air pipeline leak types and their corresponding equivalent diameters include: microcrack leaks (0.01~0.1mm), gasket aging leaks (0.05~0.5mm), loose joint leaks (0.1~2mm), and corrosion perforation leaks (0.5~5mm). When the equivalent diameter of the leak hole exceeds 5mm, physical auxiliary methods such as temporary diameter reduction using mechanical clamps or pipe section replacement can be used first, followed by the phase change sealing step of this solution. This further expands the applicable operating conditions of this solution while ensuring system safety.

[0065] The basic principles, main features, and advantages of this application have been described above. Those skilled in the art should understand that this application is not limited to the above embodiments. The embodiments and descriptions in the specification are merely the principles of this application. Various changes and modifications can be made to this application without departing from its spirit and scope, and all such changes and modifications fall within the scope of the claimed application.

Claims

1. A smart detection and sealing control device for compressed air leakage in factories, characterized in that, include: The multi-dimensional sensing network module is used to collect pipeline network operating parameters and leakage sound / vibration waveform data in real time; The leak diagnosis and feature extraction module has a built-in digital twin model of the pipeline network, which is used to integrate the operating parameters to judge leaks, locate leak points and estimate leak apertures, and extract leak feature vectors from the leak sound / vibration waveform data in real time. The medium injection and induction control module is connected to the leakage diagnosis and feature extraction module; The medium injection execution module is connected to the medium injection and induction control module and is used to inject atomized sealing medium into the leak point area. The medium injection and induction control module is configured as follows: Based on the estimated leakage orifice diameter and the leakage characteristic vector, a corresponding atomized sealing medium is selected from a preset medium library, and the injection parameters required for the atomized sealing medium to enter the leakage orifice and undergo phase change solidification in the microchannel of the leakage orifice are calculated based on the pipeline digital twin model. The medium injection execution module is controlled to inject the atomized sealing medium into the leak point area according to the injection parameters. The local low temperature generated by the gas throttling and expansion at the leak hole causes the atomized sealing medium entering the leak hole to undergo phase change and solidify in the microchannel of the leak hole, forming a plug-like sealing body. After completing the current injection operation, the sealing effect is evaluated by the change in the energy index in the leakage characteristic vector, and the injection parameters are adjusted for supplementary injection if the sealing is not completely successful.

2. The intelligent detection and sealing control device for compressed air leakage in factories according to claim 1, characterized in that, It also includes an online aerodynamic parameter identification module, which is connected to the leak diagnosis and feature extraction module and the medium injection and induction control module, respectively. The online aerodynamic parameter identification module uses a pre-trained neural network model, takes a series of multiple frames of leak feature vector sequences as input, outputs the aerodynamic damping coefficient and effective flow area of ​​the leak hole, and updates the local impedance of the leak node in the pipeline network digital twin model in real time accordingly. The media injection and induction control module is configured to calculate the injection parameters based on the updated pipeline digital twin model.

3. The intelligent detection and sealing control device for compressed air leakage in factories according to claim 2, characterized in that, The leakage feature vector extracted from the leakage acoustic / vibration waveform data by the leakage diagnosis and feature extraction module includes at least three of the following: spectral centroid, energy attenuation rate, short-time energy, spectral kurtosis, energy entropy, and energy ratio of each frequency band; the neural network model is a long short-term memory network or a temporal convolutional network.

4. The intelligent detection and sealing control device for compressed air leakage in factories according to claim 1, characterized in that, When the medium injection and induction control module calculates the injection parameters based on the digital twin model of the pipeline network, it uses the airflow velocity field and temperature field at the leak hole as a basis to simulate and calculate the motion trajectory of atomized sealing medium particles of different sizes and injection velocities in the leak jet field and the probability of being captured by the leak hole. The combination of parameters that maximizes the capture probability is selected as the injection direction, injection rate and injection dose.

5. The intelligent detection and sealing control device for compressed air leakage in factories according to claim 1, characterized in that, The media injection and induction control module is also configured to: The leakage characteristic vector energy index within multiple consecutive time windows after injection is compared with the baseline value before injection. The energy decay rate is obtained based on the comparison results. When the energy decay rate exceeds a preset threshold and remains stable, the sealing is deemed successful. Otherwise, the medium type or injection dose is adjusted for supplementary injection.

6. The intelligent detection and sealing control device for compressed air leakage in a factory according to claim 1, characterized in that, It also includes a data interface module for obtaining the operating status and production task scheduling information of each gas-using device in real time from the factory production management system; the medium injection and induction control module also includes a graded processing unit, which is configured to drive the shut-off valve of the branch to close to implement physical hard shutdown when it is determined that the phase change self-healing sealing is successful and the data interface module learns that all gas-using devices downstream of the branch where the leak point is located have switched to standby or no-task state.

7. The intelligent detection and sealing control device for compressed air leakage in a factory according to claim 1, characterized in that, The atomized blocking medium in the preset medium library includes at least one of nano-sized hygroscopic powder, low-temperature coagulating aerosol precursor, and high-viscosity condensable droplets. The media injection and induction control module is configured to select the corresponding sealing medium type and particle size distribution based on the estimated leakage orifice diameter.

8. A factory compressed air leakage intelligent detection and sealing control device according to any one of claims 1-7, characterized in that, The media injection execution module includes an inspection robot mounted on a pipeline track and a nozzle mounted on the multi-degree-of-freedom robotic arm of the inspection robot. And / or, the media injection execution module includes an array of miniature nozzles that can be remotely triggered and directionally injected at multiple fixed positions distributed along the pipeline network.

9. A method for intelligent detection and sealing control of compressed air leakage in factories, characterized in that, The method employing the intelligent detection and sealing control device for factory compressed air leakage according to any one of claims 1-8 includes the following steps: S1: Real-time acquisition of pipeline network operating parameters and leakage sound / vibration waveform data; use the pipeline network digital twin model to detect and locate leaks and estimate leak apertures; and extract leakage feature vectors from the waveform data in real time. S2: Based on the estimated leakage orifice diameter and the leakage characteristic vector, select the atomized sealing medium from the preset medium library, and calculate the injection parameters required for the atomized sealing medium to enter the leakage orifice and undergo phase change solidification in the microchannel of the leakage orifice based on the pipeline digital twin model; S3: The driving medium injection execution module injects the atomized sealing medium into the leak point area according to the injection parameters. Utilizing the local low temperature generated by the gas throttling and expansion at the leak hole, the atomized sealing medium undergoes phase change and solidifies in the microchannel of the leak hole, forming a plug-like sealing body, thereby achieving phase change self-healing sealing without interrupting the downstream gas supply. S4: After the injection is completed, the sealing effect is evaluated by the change in the energy index in the leakage characteristic vector. If the sealing is not completely successful, the injection parameters are adjusted for supplementary injection.

10. The intelligent detection and sealing control method for compressed air leakage in a factory according to claim 9, characterized in that, The following steps are included before step S2: Using a pre-trained neural network model, with a continuous multi-frame leakage feature vector sequence as input, the aerodynamic damping coefficient and effective flow area of ​​the leakage orifice are output, and the local impedance of the leakage node in the pipeline network digital twin model is updated in real time accordingly. In step S2, the injection parameters are calculated based on the updated pipeline digital twin model.

Images