Intelligent safety early warning system for pressure vessel

Abstract

The embodiment of the invention provides an intelligent safety early warning system for a pressure vessel, and the system comprises a multi-source data collection module which is used for obtaining the data of a pressure vessel body and safety accessories in real time; the edge calculation module is used for executing anomaly detection; the cloud analysis module is used for predicting the residual life according to the abnormal risk level and generating a maintenance strategy; and the early warning feedback module is used for triggering an alarm or executing a safety control action according to the abnormal risk level. According to the system, through a dynamic threshold correction mechanism and multi-source data fusion analysis, sudden abnormity and long-term deterioration risks are recognized in real time, the safety early warning precision is remarkably improved, the false alarm and missing report rate is reduced, and the intelligent safety management and control requirement under the complex working condition is met.

Description

Technical Field

[0001] This application relates to the field of safety early warning, and in particular to an intelligent safety early warning system for pressure vessels. Background Technology

[0002] The safe operation of pressure vessels is a core guarantee for industrial production. Widely used in petrochemical, energy, and pharmaceutical industries, they directly impact enterprise production efficiency and personnel safety. However, with the expansion of industrial scale and the increasing complexity of operating conditions, pressure vessel accidents, such as explosions and leaks, are becoming more frequent, causing not only huge economic losses but also potential environmental disasters. Existing safety standards largely rely on fixed thresholds and manual inspections, making it difficult to dynamically adapt to long-term deterioration processes such as material fatigue and media corrosion, leading to a significant increase in the risk of false alarms or missed alarms.

[0003] In existing technologies, safety early warning systems for pressure vessels often employ a single threshold or a simple linear correction model, which cannot effectively couple short-term sudden anomalies with long-term cumulative risks. For example, traditional methods trigger alarms by setting a fixed pressure threshold, but ignore strength decay caused by material fatigue or wall thinning caused by media corrosion, causing the threshold setting to deviate from the actual safety boundary. In addition, data acquisition relies on external sensors and manual sampling, resulting in insufficient real-time performance and accuracy, and lacking the ability to fuse and analyze multi-source data, making it difficult to achieve accurate risk location.

[0004] Therefore, there is an urgent need for a technical solution that integrates dynamic threshold correction and multi-source data collaborative analysis to improve the reliability and response speed of pressure vessel safety early warning. Summary of the Invention

[0005] This application provides an intelligent safety early warning system for pressure vessels. This system uses a dynamic threshold correction mechanism and multi-source data fusion analysis to identify sudden anomalies and long-term deterioration risks in real time, significantly improving the accuracy of safety early warning, reducing false alarm and missed alarm rates, and meeting the needs of intelligent safety management under complex working conditions.

[0006] Firstly, a pressure vessel intelligent safety early warning system is provided, the system comprising:

[0007] A multi-source data acquisition module is used to acquire data of the pressure vessel body and its safety accessories in real time, including: pressure, temperature, stress deformation and medium flow rate.

[0008] An edge computing module is used to perform anomaly detection, which includes: generating an initial safety threshold based on the pressure vessel design parameters and material fatigue characteristics; dynamically correcting the safety threshold according to the data; and comparing the corrected safety threshold with real-time data to generate an anomaly risk level.

[0009] A cloud-based analytics module, used to predict remaining lifespan and generate maintenance strategies based on the anomaly risk level;

[0010] The early warning feedback module is used to trigger an alarm or perform a safety control action based on the abnormal risk level.

[0011] It should be understood that a pressure vessel is a closed device that holds gas or liquid and bears a certain pressure. The safety accessories of a pressure vessel are devices installed to ensure that the pressure can be effectively monitored, controlled and released during the operation of the equipment. These mainly include safety valves and pressure gauges.

[0012] It should be understood that this system acquires multi-dimensional data such as pressure, temperature, stress deformation, and medium flow velocity in real time through a multi-source data acquisition module. Combined with the two-stage processing mechanism of the edge computing module (short-term threshold judgment and dynamic threshold correction), it effectively solves the technical problem that traditional fixed threshold models cannot adapt to the long-term degradation of material fatigue and medium corrosion. The cloud analysis module uses a digital twin model to simulate stress distribution and medium flow state, and combines transfer learning algorithms to optimize lifetime prediction, significantly improving prediction accuracy. The early warning feedback module triggers control actions in stages through a multi-level alarm mechanism, avoiding excessive intervention in the production process. The overall solution can effectively reduce the false alarm rate and improve the detection accuracy.

[0013] In conjunction with the first aspect, in some implementations of the first aspect, the multi-source data acquisition module includes:

[0014] A distributed stress monitoring unit is deployed on the inner wall of the pressure vessel;

[0015] A multi-parameter sensor unit is disposed on the safety accessory and is used to acquire pressure, temperature and valve opening / closing status data.

[0016] It should be understood that the distributed stress monitoring unit, based on the principle of grating reflection wavelength variation, is distributed along the container wall to monitor local stress concentration and microcrack propagation in real time. This compensates for the insufficient coverage of traditional point sensors and improves the spatial resolution of stress detection. The multi-parameter sensing unit is integrated into the safety accessory, simultaneously collecting pressure, temperature, and valve status data. Cross-validation of multi-source data eliminates errors from single sensors; for example, combining pressure surges with valve opening / closing status to determine if it's a misoperation, thereby reducing the probability of false alarms. Both work together to provide high-precision input data for dynamic threshold correction and digital twin models.

[0017] In conjunction with the first aspect, in some implementations of the first aspect, the edge computing module employs a two-stage processing mechanism when performing anomaly detection, including:

[0018] The first stage involves calculating the mean and variance of the data using a sliding window statistical method to identify short-term data mutations. The length of the sliding window is associated with the current operating stage of the pressure vessel.

[0019] Second stage: Dynamically adjust the safety threshold using the following formula:

[0020]

[0021] Where C is the real-time medium corrosion rate, Cbaseline is the design corrosion rate, N is the current stress cycle number, Nlimit is the material fatigue limit number, and α and β are weighting coefficients.

[0022] It should be understood that the current operating stages of a pressure vessel include: steady-state operation, start-up / shutdown, and medium phase change. This system automatically identifies the current stage by real-time monitoring of medium flow rate, temperature gradient, and valve status, dynamically adjusting the detection strategy to avoid threshold misjudgments caused by changes in operating conditions.

[0023] In conjunction with the first aspect, in some implementations of the first aspect, the values ​​of the weighting coefficients α and β are related to the material type of the pressure vessel and the chemical properties of the medium, and are generated through training with historical failure data.

[0024] In conjunction with the first aspect, in some implementations of the first aspect, the cloud analysis module simulates the stress distribution inside the pressure vessel through a three-dimensional digital twin model. The construction of the model includes: generating an initial geometric structure based on the pressure vessel design drawings; dynamically updating the model boundary conditions according to real-time collected stress, temperature, and medium flow velocity data; and calculating the thermo-mechanical coupling distribution of potential risk areas through algorithms.

[0025] It should be understood that the boundary conditions of the model include: geometric boundaries (the container structure based on the design drawings), physical boundaries (stress and temperature data collected in real time as load inputs), and medium boundaries (fluid dynamic constraints defined by flow velocity and phase data).

[0026] In conjunction with the first aspect, in some implementations of the first aspect, the cloud analysis module uses a transfer learning algorithm to optimize the remaining life prediction, initializes the model parameters through a pre-trained industry historical database, and fine-tunes them based on the current real-time data of the pressure vessel.

[0027] It should be understood that transfer learning, by reusing pre-trained industry databases, extracts common features of pressure vessels and requires only a small amount of current vessel data to fine-tune the output layer parameters, thus quickly building a high-precision life prediction model. Compared to training from scratch, transfer learning can reduce model convergence time and alleviate the problem of insufficient model generalization ability caused by industry data silos.

[0028] In conjunction with the first aspect, in some implementations of the first aspect, the edge computing module includes an offline mode, and the offline mode is constructed in the following ways:

[0029] Extract key feature parameters from the cloud-based analysis module;

[0030] Generate local anomaly classification rules based on local algorithms;

[0031] The local alert logic is triggered by the statistical results of the data in the sliding window.

[0032] It should be understood that the offline mode enables a local lightweight model when communication is interrupted, and continues to perform risk assessment based on historical data stored in the edge computing module and real-time sliding window statistics, so as to avoid system paralysis caused by cloud disconnection.

[0033] In conjunction with the first aspect, in some implementations of the first aspect, the real-time medium corrosion rate C is obtained by means of an electrochemical monitoring unit embedded in the inner wall of the pressure vessel, which calculates the corrosion rate by measuring the change in polarization resistance, using the following formula:

[0034]

[0035] Among them, R p is the measured value of polarization resistance, and k is a proportionality coefficient related to the dielectric composition.

[0036] It should be understood that the corrosion rate reflects the rate of material loss from the container's inner wall due to chemical or electrochemical reactions. In this system, an embedded electrochemical monitoring unit measures the polarization resistance in real time and calculates the corrosion rate. This data is used to dynamically adjust the safety threshold. Furthermore, corrosion data is coupled with the number of stress cycles and input into the fatigue model to predict the remaining life under the combined effects of corrosion and fatigue, avoiding the limitations of traditional methods that only consider a single factor.

[0037] In conjunction with the first aspect, in certain implementations of the first aspect, the window length of the sliding window statistical method is adaptively adjusted according to the operating stage of the pressure vessel, including:

[0038] A long window is used during steady-state operation to smooth out noise interference;

[0039] Switch to a short window during start-up, shutdown, or medium phase change to improve response speed.

[0040] In conjunction with the first aspect, in some implementations of the first aspect, the multi-level alarm of the early warning feedback module includes:

[0041] Level 1 warning: The Level 1 warning system highlights the risk area through a visual interface.

[0042] Level 2 warning: The Level 2 warning automatically limits the operating parameters of the pressure vessel to a safe range;

[0043] A Level 3 warning triggers the interlocking control of the safety accessory to perform emergency pressure relief.

[0044] It should be understood that multi-level alarms implement a progressive response based on risk level: Level 1 alerts highlight the risk area and record data for manual review to prevent misoperation; Level 2 alerts automatically limit operating parameters to maintain production while ensuring safety; Level 3 alerts immediately interlock and control safety accessories to prevent the chain of events from escalating. This mechanism can shorten the time to resumption of production and reduce economic losses caused by unnecessary production interruptions. Attached Figure Description

[0045] Figure 1 This is a schematic diagram of the structure of an intelligent safety early warning system for pressure vessels provided in an embodiment of this application.

[0046] Figure 2 This is a schematic diagram of a multi-source data acquisition module structure provided in an embodiment of this application.

[0047] Figure 3 A flowchart illustrating an offline mode construction method provided in this application embodiment. Detailed Implementation

[0048] The terminology used in the following embodiments is for the purpose of describing particular embodiments only and is not intended to be limiting of this application. As used in the specification and appended claims of this application, the singular expressions “a,” “an,” “the,” “the,” and “this” are intended to also include expressions such as “one or more,” unless the context clearly indicates otherwise. It should also be understood that in the following embodiments of this application, “at least one” and “one or more” refer to one, two, or more than two. The term “and / or” is used to describe the relationship between related objects, indicating that three relationships can exist; for example, A and / or B can indicate: A alone, A and B simultaneously, or B alone, where A and B can be singular or plural. The character “ / ” generally indicates that the preceding and following related objects are in an “or” relationship.

[0049] References to "one embodiment" or "some embodiments" as described in this specification mean that one or more embodiments of this application include a specific feature, structure, or characteristic described in connection with that embodiment. Therefore, the phrases "in one embodiment," "in some embodiments," "in other embodiments," "in still other embodiments," etc., appearing in different parts of this specification do not necessarily refer to the same embodiment, but rather mean "one or more, but not all, embodiments," unless otherwise specifically emphasized. The terms "comprising," "including," "having," and variations thereof mean "including but not limited to," unless otherwise specifically emphasized.

[0050] The safe operation of pressure vessels is a critical guarantee for industrial production, and they are widely used in petrochemical, energy, and pharmaceutical industries, directly impacting enterprise production safety and personnel lives. However, with the expansion of industrial scale and complex operating conditions, accidents such as explosions and leaks occur frequently, causing significant economic losses and environmental risks.

[0051] This application provides an intelligent safety early warning system for pressure vessels. This system uses a dynamic threshold correction mechanism and multi-source data fusion analysis to identify sudden anomalies and long-term deterioration risks in real time, significantly improving the accuracy of safety early warning, reducing false alarm and missed alarm rates, and meeting the needs of intelligent safety management under complex working conditions.

[0052] The technical solutions of the embodiments of this application will now be described in conjunction with the accompanying drawings.

[0053] Figure 1 A schematic diagram of a pressure vessel intelligent safety early warning system provided in this application embodiment is shown. In some examples, the system includes:

[0054] A multi-source data acquisition module is used to acquire data of the pressure vessel body and its safety accessories in real time, including: pressure, temperature, stress deformation and medium flow rate.

[0055] An edge computing module is used to perform anomaly detection, which includes: generating an initial safety threshold based on the pressure vessel design parameters and material fatigue characteristics; dynamically correcting the safety threshold according to the data; and comparing the corrected safety threshold with real-time data to generate an anomaly risk level.

[0056] A cloud-based analytics module, used to predict remaining lifespan and generate maintenance strategies based on the anomaly risk level;

[0057] The early warning feedback module is used to trigger an alarm or perform a safety control action based on the abnormal risk level.

[0058] Figure 2 This application provides a schematic diagram of a multi-source data acquisition module structure. In some examples, the multi-source data acquisition module includes:

[0059] A distributed stress monitoring unit is deployed on the inner wall of the pressure vessel;

[0060] A multi-parameter sensor unit is disposed on the safety accessory and is used to acquire pressure, temperature and valve opening / closing status data.

[0061] In some examples, the edge computing module employs a two-stage processing mechanism when performing anomaly detection, including:

[0062] The first stage involves calculating the mean and variance of the data using a sliding window statistical method to identify short-term data mutations. The length of the sliding window is associated with the current operating stage of the pressure vessel.

[0063] Second stage: Dynamically adjust the safety threshold using the following formula:

[0064]

[0065] Where C is the real-time medium corrosion rate, Cbaseline is the design corrosion rate, N is the current stress cycle number, Nlimit is the material fatigue limit number, and α and β are weighting coefficients.

[0066] In one possible implementation, the abnormal risk level generation strategy includes: first, calculating the mean and variance of real-time data using a sliding window statistical method to identify short-term data mutations (such as sudden pressure increases or temperature jumps); then, combining the medium corrosion rate and the number of stress cycles, using a dynamic attenuation factor model to correct the safety threshold; and finally, classifying the risk level into low, medium, and high levels based on the degree of deviation between the real-time data and the corrected threshold, with the greater the deviation, the higher the risk level.

[0067] In one possible implementation, when the corrosion rate exceeds the design baseline, the allowable pressure limit is reduced proportionally to prevent the risk of cracking due to wall thinning. Simultaneously, corrosion data and stress cycle counts are coupled into the fatigue model to predict the remaining life under the combined effects of corrosion and fatigue, avoiding the limitations of considering only a single factor.

[0068] In some examples, the values ​​of the weighting coefficients α and β are associated with the material type of the pressure vessel and the chemical properties of the medium, and are generated through training with historical failure data.

[0069] In one possible implementation, a logistic regression model is trained using historical failure data, with container material type and media corrosivity as input features, and actual corrosion-fatigue failure cases as labels, outputting the weight values ​​of α and β.

[0070] In some examples, the cloud-based analysis module simulates the stress distribution inside the pressure vessel using a three-dimensional digital twin model. The model is constructed by: generating an initial geometric structure based on the pressure vessel design drawings; dynamically updating the model boundary conditions based on real-time collected stress, temperature, and medium flow velocity data; and calculating the thermo-mechanical coupling distribution of potential risk areas using algorithms.

[0071] In one possible implementation, the three-dimensional model of the pressure vessel is first discretized into tetrahedral or hexahedral meshes, and the mesh is refined in stress concentration areas based on real-time stress data. Then, boundary conditions (such as surface forces corresponding to internal pressure, temperature field, and medium flow velocity) are applied. The ANSYS Mechanical solver is called to calculate nodal displacements and stress distributions. Finally, the maximum equivalent stress value is extracted, and if it exceeds the dynamically corrected safety threshold, it is marked as a red warning area in the digital twin model.

[0072] It should be understood that the ANSYS Mechanical solver is a structural mechanics simulation module integrated into the ANSYS software, supporting linear / nonlinear statics, dynamics, and thermo-mechanical coupling analysis. In this system, the solver loads real-time acquired stress and temperature data as boundary conditions, calculates the stress distribution of the container under complex loads, and automatically outputs the maximum equivalent stress value and the coordinates of the hazardous area, providing a quantitative basis for risk labeling of the digital twin model.

[0073] In some examples, the cloud analytics module uses a transfer learning algorithm to optimize remaining life prediction, initializes model parameters using a pre-trained industry historical database, and fine-tunes them based on the current real-time data of the pressure vessel.

[0074] Figure 3 This application provides a flowchart of an offline mode construction method implementation. In some examples, the edge computing module includes an offline mode, and the offline mode construction method includes:

[0075] Extract key feature parameters from the cloud-based analysis module;

[0076] Generate local anomaly classification rules based on local algorithms;

[0077] The local alert logic is triggered by the statistical results of the data in the sliding window.

[0078] In one possible implementation, the local anomaly classification rule generation strategy includes: using the CART decision tree algorithm, setting the classification features as pressure fluctuation amplitude (the difference between the current value and the mean of the sliding window), temperature change rate (e.g., the rate of temperature rise within 5 minutes), and stress gradient (stress difference between adjacent monitoring units). If the pressure fluctuation exceeds the mean ± 2 times the variance and the temperature change rate is > 3℃ / min, it is judged as high risk; if only the pressure fluctuation exceeds the threshold and the stress gradient is < 10MPa / m, it is judged as medium risk, and the rest are low risk.

[0079] In some examples, the real-time media corrosion rate C is obtained by means of an electrochemical monitoring unit embedded in the inner wall of the pressure vessel, which calculates the corrosion rate by measuring the change in polarization resistance, using the following formula:

[0080]

[0081] Among them, R p is the measured value of polarization resistance, and k is a proportionality coefficient related to the dielectric composition.

[0082] In some examples, the window length of the sliding window statistical method is adaptively adjusted according to the operating stage of the pressure vessel, including:

[0083] A long window is used during steady-state operation to smooth out noise interference;

[0084] Switch to a short window during start-up, shutdown, or medium phase change to improve response speed.

[0085] In some examples, the multi-level alarm of the early warning feedback module includes:

[0086] Level 1 warning: The Level 1 warning system highlights the risk area through a visual interface.

[0087] Level 2 warning: The Level 2 warning automatically limits the operating parameters of the pressure vessel to a safe range;

[0088] A Level 3 warning triggers the interlocking control of the safety accessory to perform emergency pressure relief.

[0089] The above are merely preferred embodiments of the present invention. The scope of protection of the present invention is not limited to the above embodiments. Any equivalent modifications or variations made by those skilled in the art based on the content disclosed in the present invention should be included within the scope of protection set forth in the claims.

Claims

1. A pressure vessel intelligent safety early warning system, characterized in that, The system includes: A multi-source data acquisition module is used to acquire data of the pressure vessel body and its safety accessories in real time, including: pressure, temperature, stress deformation and medium flow rate. An edge computing module is used to perform anomaly detection, which includes: generating an initial safety threshold based on the pressure vessel design parameters and material fatigue characteristics; dynamically correcting the safety threshold according to the data; and comparing the corrected safety threshold with real-time data to generate an anomaly risk level. A cloud-based analytics module, used to predict remaining lifespan and generate maintenance strategies based on the anomaly risk level; The early warning feedback module is used to trigger an alarm or perform a safety control action based on the abnormal risk level.

2. The system according to claim 1, characterized in that, The multi-source data acquisition module includes: A distributed stress monitoring unit is deployed on the inner wall of the pressure vessel; A multi-parameter sensor unit is disposed on the safety accessory and is used to acquire pressure, temperature and valve opening / closing status data.

3. The system according to claim 1, characterized in that, When the edge computing module performs anomaly detection, it adopts a two-stage processing mechanism, including: The first stage involves calculating the mean and variance of the data using a sliding window statistical method to identify short-term data mutations. The length of the sliding window is associated with the current operating stage of the pressure vessel. Second stage: Dynamically adjust the safety threshold using the following formula: Where C is the real-time medium corrosion rate, C 基准 To design the corrosion rate, N is the current stress cycle number. 极限 α represents the material's fatigue limit number of cycles, and β represents the weighting coefficients.

4. The system according to claim 3, characterized in that, The values ​​of the weighting coefficients α and β are related to the material type of the pressure vessel and the chemical properties of the medium, and are generated through training with historical failure data.

5. The system according to claim 1, characterized in that: The cloud-based analysis module simulates the internal stress distribution of the pressure vessel using a three-dimensional digital twin model. The model construction includes: generating an initial geometric structure based on the pressure vessel design drawings; dynamically updating the model boundary conditions based on real-time collected stress, temperature, and medium flow velocity data; and calculating the thermo-mechanical coupling distribution of potential risk areas using algorithms.

6. The system according to claim 1, characterized in that: The cloud-based analytics module employs a transfer learning algorithm to optimize remaining life prediction. It initializes model parameters using a pre-trained industry historical database and fine-tunes them based on the current real-time data of the pressure vessel.

7. The system according to claim 1, characterized in that, The edge computing module includes an offline mode, and the offline mode is constructed in the following ways: Extract key feature parameters from the cloud-based analysis module; Generate local anomaly classification rules based on local algorithms; The local alert logic is triggered by the statistical results of the data in the sliding window.

8. The system according to claim 1, characterized in that, The real-time corrosion rate C is obtained by an electrochemical monitoring unit embedded in the inner wall of the pressure vessel, which calculates the corrosion rate by measuring the change in polarization resistance. The calculation formula is as follows: Among them, R p is the measured value of polarization resistance, and k is a proportionality coefficient related to the dielectric composition.

9. The system according to claim 3, characterized in that, The window length of the sliding window statistical method is adaptively adjusted according to the operating stage of the pressure vessel, including: A long window is used during steady-state operation to smooth out noise interference; Switch to a short window during start-up, shutdown, or medium phase change to improve response speed.

10. The system according to claim 1, characterized in that, The multi-level alarm of the early warning feedback module includes: Level 1 warning: The Level 1 warning system highlights the risk area through a visual interface. Level 2 warning: The Level 2 warning automatically limits the operating parameters of the pressure vessel to a safe range; A Level 3 warning triggers the interlocking control of the safety accessory to perform emergency pressure relief.

Images