Twin-engine-driven comprehensive pipe gallery natural gas leakage diffusion prediction analysis method and system

Abstract

The application relates to the technical field of digital twinning, and provides a comprehensive pipe gallery natural gas leakage diffusion prediction analysis method and system driven by a twinning engine. In the digital twinning platform for comprehensive pipe gallery natural gas leakage diffusion deduction, a pipe gallery geometric digital model consistent with the structure and layout of the comprehensive pipe gallery and a leakage diffusion deduction model integrating a space-time convolution operation and a long short-term memory network mechanism are constructed; the pipe gallery geometric digital model is used for numerical simulation of natural gas leakage diffusion; simulation data obtained is transmitted to an Internet of Things server in real time; the leakage diffusion deduction model is trained; the running state data of different sensors with different protocols is transmitted to the Internet of Things server in real time; the leakage diffusion deduction model is used for deduction of comprehensive pipe gallery natural gas leakage diffusion; a concentration grid of natural gas leakage diffusion is generated; a dynamic process of natural gas concentration change with time and space is simulated; and time-space evolution analysis of natural gas leakage diffusion is realized.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of digital twinning, in particular to a method and system for predicting and analyzing natural gas leakage and diffusion in a utility tunnel based on a twinning engine. BACKGROUND

[0002] As a city underground lifeline project, the utility tunnel has a complex internal environment, multiple pipelines, ventilation systems and other facilities, and a natural gas leakage and diffusion accident can easily cause major safety hazards.

[0003] In the simulation of natural gas leakage and diffusion in a utility tunnel, traditional methods such as the Gaussian diffusion model and the empirical formula are based on simplified mathematical equations and assume that the leaked gas diffuses in a single direction. Although these methods are fast, they have low accuracy and can only reflect one-dimensional concentration field changes, making it difficult to simulate three-dimensional scenarios and adapt to different environments.

[0004] CFD numerical simulation methods based on computational fluid dynamics can accurately simulate the gas diffusion process by establishing the Navier-Stokes equation in a three-dimensional space, combining a turbulence model (such as k-ε, LES) and a component transport equation. Although these methods can accurately simulate real diffusion scenarios, consider factors such as turbulence, obstacles, ventilation and temperature, and provide detailed analysis of complex utility tunnel structures, they have high computational costs and cannot meet the system's response speed requirements, especially in complex environments such as utility tunnels. The computational load is too high, making it difficult to provide the required calculation speed, and the practical application of the digital twinning framework is further difficult to achieve.

[0005] Digital twinning technology that dynamically drives virtual model updates by collecting physical tunnel data in real time through Internet of Things (IoT) sensors is currently widely used. However, this technology only replicates the physical entity by collecting and analyzing sensor data, and the current digital twinning model lacks core components such as complex state changes and simulations in different environments. SUMMARY

[0006] The present application aims to provide a method and system for predicting and analyzing natural gas leakage and diffusion in a utility tunnel based on a twinning engine to solve or alleviate the problems in the prior art.

[0007] To achieve the above-mentioned purpose, the present application provides the following technical solutions:

[0008] The present application provides a method for predicting and analyzing natural gas leakage and diffusion in a utility tunnel based on a twinning engine, comprising:

[0009] In the built digital twin platform for natural gas leakage diffusion deduction of comprehensive pipe gallery, a pipe gallery geometric digital model consistent with the structure and layout of the comprehensive pipe gallery is built through multi-level modeling, and a leakage diffusion deduction model integrating space-time convolution operation and long short-term memory network mechanism is built;

[0010] The natural gas leakage diffusion in the comprehensive pipe gallery is numerically simulated through the pipe gallery geometric digital model, and the simulation data obtained is transmitted to the Internet of Things server in real time to train the leakage diffusion deduction model.

[0011] The running state data of different sensors with different protocols in the comprehensive pipe gallery are transmitted to the Internet of Things server in real time to drive the trained leakage diffusion deduction model to deduce the natural gas leakage diffusion in the comprehensive pipe gallery, and generate the concentration grid of the natural gas leakage diffusion in the comprehensive pipe gallery; and the concentration grid data of the natural gas leakage diffusion in the comprehensive pipe gallery is corrected through the built pipe gallery physical model and leakage constraint model of the comprehensive pipe gallery, to obtain the corrected natural gas concentration grid data of the comprehensive pipe gallery.

[0012] Preferably, a plurality of working conditions of different leakage diffusion factors are set in the pipe gallery geometric digital model to simulate natural gas leakage diffusion, to obtain concentration simulation monitoring point data of different sensors at monitoring points in the pipe gallery geometric digital model, and natural gas concentration simulation data of spatial grids in the pipe gallery geometric digital model, and transmit them to the Internet of Things server; wherein the leakage diffusion factors at least include leakage rate, leakage position, leakage port direction, leakage port size and wind speed.

[0013] Preferably, the obtained natural gas concentration simulation data is spatially interpolated along the length direction, height direction and width direction of the comprehensive pipe gallery respectively to generate ordered spatial grid concentration data.

[0014] The concentration simulation monitoring point data per unit time and the corresponding monitoring point coordinates, and the ordered spatial grid concentration data and the corresponding spatial grid coordinates are classified and stored to generate a standardized data set for training the leakage diffusion deduction model.

[0015] Preferably, a one-dimensional convolution layer is used to locally perceive the concentration simulation monitoring point data per unit time in the standardized data set along the time dimension in a sliding window manner, and the local time sequence features of the concentration simulation monitoring point data obtained by local perception are taken as the input of the leakage diffusion deduction model, and the ordered spatial grid concentration data is taken as the prediction target to train the leakage diffusion deduction model.

[0016] Preferably, the operation state data of different devices in the comprehensive pipe gallery and different protocols is converted into structured information with unified data specifications through the mapping rule of the standardized information model of the device private protocol; the edge node adds three-dimensional grid coordinates and time labels to the structured information according to the requirements of the digital twin model, and actively pushes the structured information to the Internet of Things server in real time.

[0017] Preferably, the mapping rule is:

[0018]

[0019] In the formula, is the operation state data of the first device in the comprehensive pipe gallery at time ; the structured information obtained by converting the operation state data through standardization mapping; is the protocol analysis function of the first device.

[0020] Preferably, the operation state data transmitted to the Internet of Things server in real time is used to deduce the natural gas leakage diffusion of the comprehensive pipe gallery, at the same time, the fireproof door in the comprehensive pipe gallery is closed and the fan is adjusted to the accident ventilation mode, to obtain the natural gas concentration grid data of the comprehensive pipe gallery, and the natural gas concentration grid data is rendered in the pipe gallery geometric digital model.

[0021] Preferably, the leakage constraint correction coefficient of the natural gas concentration grid data of the comprehensive pipe gallery under the leakage constraint is determined through the constructed leakage constraint model; wherein the leakage constraint model is:

[0022]

[0023] In the formula, is the leakage constraint correction coefficient of the natural gas concentration grid data of the comprehensive pipe gallery under the leakage constraint; is the distance from the leakage source to the sensor monitoring point in the comprehensive pipe gallery during the natural gas leakage diffusion; is the real-time wind speed in the comprehensive pipe gallery during the natural gas leakage diffusion; is the state of the fireproof door in the comprehensive pipe gallery during the natural gas leakage diffusion, , indicating that the fireproof door is closed, , indicating that the fireproof door is opened; is the total length of the comprehensive pipe gallery, is the limiting wind speed in the comprehensive pipe gallery, which is a constant, , is the initial state of the fireproof door, ;

[0024] ​As the distance constraint weight for the spread of natural gas leaks, The wind speed constraint weight during the diffusion of a natural gas leak. Weights for fire door status constraints during natural gas leak propagation;

[0025] The evolution of historical natural gas leak and diffusion accidents in the integrated utility tunnel was analyzed using a physical model of the tunnel, and the leakage error correction coefficient for the natural gas concentration grid data of the integrated utility tunnel was determined; wherein, according to the formula:

[0026]

[0027] Calculate the leakage error correction factor for the natural gas concentration grid data of the integrated utility tunnel. In the formula, When studying the evolution of historical natural gas leak and diffusion accidents in integrated utility tunnels, the first Leakage diffusion experimental data at each sampling time point, To provide real leakage diffusion data corresponding to the experimental data on leakage diffusion, This represents the total number of data points from the leakage diffusion experiment.

[0028] According to the formula:

[0029]

[0030] The natural gas concentration grid data of the integrated utility tunnel is corrected to obtain the corrected natural gas concentration grid data of the integrated utility tunnel. In the formula, This is grid data of natural gas concentration obtained by using a leakage diffusion simulation model to simulate the natural gas leakage diffusion in a utility tunnel.

[0031] This application also provides a twin-engine-driven integrated pipeline gallery natural gas leakage diffusion prediction and analysis system, which uses any of the above-described twin-engine-driven integrated pipeline gallery natural gas leakage diffusion prediction and analysis methods to extrapolate the diffusion of natural gas leaks in integrated pipeline galleries. The system includes:

[0032] The twin platform unit is configured to construct a geometric digital model of the integrated pipeline corridor with consistent structure and layout through multi-level modeling in the digital twin platform for natural gas leakage and diffusion simulation of the integrated pipeline corridor, as well as to construct a leakage and diffusion simulation model that integrates spatiotemporal convolution operation and long short-term memory network mechanism.

[0033] The numerical simulation and model training unit is configured to perform numerical simulation of natural gas leakage and diffusion in the integrated pipeline corridor using a geometric digital model of the pipeline corridor, and transmit the obtained simulation data to the Internet of Things server in real time to train the leakage and diffusion inference model.

[0034] The leakage diffusion simulation unit is configured to transmit the operational status data of different sensors and protocols within the integrated utility tunnel to the IoT server in real time. This drives the trained leakage diffusion simulation model to simulate the natural gas leakage diffusion within the integrated utility tunnel, generating a concentration grid for the natural gas leakage diffusion within the integrated utility tunnel. Furthermore, the concentration grid data for the natural gas leakage diffusion within the integrated utility tunnel is corrected using the constructed physical model and leakage constraint model of the integrated utility tunnel, resulting in corrected natural gas concentration grid data for the integrated utility tunnel.

[0035] Beneficial effects:

[0036] The method and system for predicting and analyzing natural gas leakage and diffusion in integrated utility tunnels based on a twin engine provided in this application embodiment involves constructing a multi-level model of the utility tunnel's geometric digital model, consistent with its structure and layout, and a leakage diffusion simulation model integrating spatiotemporal convolution operations and long short-term memory network mechanisms. The geometric digital model is used to numerically simulate the natural gas leakage and diffusion in the integrated utility tunnel, and the resulting simulation data is transmitted in real-time to an IoT server to train the leakage diffusion simulation model. Then, the operational status data of different sensors and protocols within the integrated utility tunnel are transmitted in real-time to the IoT server to determine the trained leakage diffusion simulation model for further simulation of natural gas leakage and diffusion in the integrated utility tunnel, generating a concentration grid for natural gas leakage and diffusion in the integrated utility tunnel. Finally, the concentration grid data for natural gas leakage and diffusion in the integrated utility tunnel is corrected using the constructed physical model and leakage constraint model of the integrated utility tunnel, resulting in corrected natural gas concentration grid data for the integrated utility tunnel. In this way, a geometric digital model of the integrated utility tunnel and a leakage diffusion simulation model with digital twin relationship are established. By using high-quality, high-frequency multi-source real-time data streams from different platforms and different protocols, the spatiotemporal evolution of natural gas leakage diffusion in the integrated utility tunnel can be analyzed to meet the needs of rapid emergency response. Attached Figure Description

[0037] The accompanying drawings, which form part of this application, are used to provide a further understanding of this application. The illustrative embodiments and descriptions of this application are used to explain this application and do not constitute an undue limitation of this application. Wherein:

[0038] Figure 1 This is a flowchart illustrating a twin engine-driven method for predicting and analyzing the diffusion of natural gas leaks in integrated utility tunnels, according to some embodiments of this application.

[0039] Figure 2 This is a schematic diagram of a twin engine-driven method for predicting and analyzing the diffusion of natural gas leaks in integrated utility tunnels, provided according to some embodiments of this application.

[0040] Figure 3 This is a schematic diagram illustrating the principle of classifying and storing data according to some embodiments of this application;

[0041] Figure 4 This is a rendered schematic diagram of a utility tunnel model provided according to some embodiments of this application;

[0042] Figure 5 This is a schematic diagram of a physical model of an integrated utility tunnel provided according to some embodiments of this application;

[0043] Figure 6 This is a schematic diagram of another integrated utility tunnel physical model provided according to some embodiments of this application;

[0044] Figure 7 This is a schematic diagram illustrating spatial interpolation of natural gas concentration simulation data provided according to some embodiments of this application;

[0045] Figure 8 This is a schematic diagram of a digital twin model for extrapolating the diffusion of natural gas leaks in an integrated utility tunnel, provided according to some embodiments of this application.

[0046] Figure 9 This is a schematic diagram of the structure of a twin engine-driven integrated utility tunnel natural gas leakage diffusion prediction and analysis system according to some embodiments of this application. Detailed Implementation

[0047] The present application will now be described in detail with reference to the accompanying drawings and embodiments. Various examples are provided by way of explanation and not by way of limitation. In fact, those skilled in the art will understand that modifications and variations can be made to the present application without departing from the scope or spirit of the present application. For example, a feature shown or described as part of one embodiment may be used in another embodiment to produce yet another embodiment. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention should fall within the scope of protection of the embodiments of the present invention.

[0048] In the simulation of natural gas leak diffusion in pipeline corridors, traditional mathematical models, such as Gaussian diffusion models and empirical formulas, are usually based on simplified assumptions, such as assuming that the leaking gas diffuses in a single direction. While these methods are computationally fast and easy to implement, they have limitations in accuracy. Traditional methods can only reflect one-dimensional concentration field changes and cannot simulate complex diffusion processes in three-dimensional space. This results in poor adaptability to complex real-world scenarios, especially in complex pipeline corridor environments, where they struggle to accurately reflect the dynamic processes of gas leaks and diffusion.

[0049] In contrast, numerical simulation methods based on computational fluid dynamics (CFD) offer a more accurate solution. CFD simulations, by establishing the Navier-Stokes equations in three-dimensional space and combining them with turbulence models (such as k-ε and LES) and component transport equations, can describe in detail the diffusion process of gases under different environmental conditions. This method can not only accurately recreate actual diffusion scenarios but also comprehensively consider the influence of multiple physical factors, such as turbulence effects, obstacles, ventilation, and temperature changes, on the diffusion process, thus providing more accurate data support for the safety assessment of complex utility tunnel structures. However, while CFD methods are accurate, their computational cost is high, often making them difficult to use in real-time systems, especially in complex environments such as integrated utility tunnels. The computational demands of CFD are enormous and may not meet the response speed requirements of real-time monitoring systems. Excessive computational load makes it difficult for the system to provide the necessary efficient computation within a limited time, especially in critical application scenarios such as emergency response, where computational delays can affect decision-making effectiveness. Therefore, although CFD can provide accurate simulation results, its practical application within a digital twin framework still faces significant challenges.

[0050] Digital twin technology uses Internet of Things (IoT) sensors to collect environmental data in real time within physical utility tunnels, driving dynamic updates to a virtual model to create a digital copy corresponding to the real system. The core advantage of digital twin technology lies in its ability to achieve real-time monitoring and state prediction of the physical entity, providing a visual representation of changes in various variables within the utility tunnel. Through the collection and analysis of sensor data, digital twins can provide utility tunnel managers with real-time dynamic monitoring and support decision-making. However, simply replicating the physical entity through sensor data collection and analysis does not fully realize the potential of digital twins. Existing digital twin models often lack sufficient simulation capabilities and cannot accurately reflect complex state changes, especially gas diffusion behavior under different environmental conditions.

[0051] Based on this, this application proposes a method for predicting and analyzing the diffusion of natural gas leaks in integrated utility tunnels based on a twin engine-driven approach, such as... Figures 1 to 8 As shown, the method includes:

[0052] Step S101: In the digital twin platform for natural gas leakage and diffusion simulation of the integrated pipeline corridor, a geometric digital model of the pipeline corridor with the same structure and layout as the integrated pipeline corridor is constructed through multi-level modeling, and a leakage and diffusion simulation model integrating spatiotemporal convolution operation and long short-term memory network mechanism is constructed.

[0053] In this embodiment, a digital twin platform for the simulation of natural gas leakage and diffusion in a utility tunnel is built using Unity3d. A rendered geometric digital model of the utility tunnel, which is consistent with the structure and layout of the utility tunnel and is constructed through multi-level modeling, and a leakage diffusion simulation model that integrates spatiotemporal convolution operations and long short-term memory network mechanisms are imported into the digital twin platform. This establishes a digital twin relationship between the geometric digital model of the utility tunnel and the leakage diffusion simulation model in the digital twin platform. Then, by using high-quality, high-frequency multi-source real-time data streams from different platforms and with different protocols, the spatiotemporal evolution of natural gas leakage and diffusion in the utility tunnel can be analyzed.

[0054] In this embodiment, based on the specific as-built drawings of the integrated utility tunnel, a multi-level model is constructed for the main structure, ancillary facilities, and pipelines entering the tunnel, resulting in a geometric digital model of the integrated utility tunnel that is consistent with the actual structure and layout. Specifically, the coordinates of each pile are obtained based on the longitudinal and plan views in the specific as-built drawings of the integrated utility tunnel. Positioning lines for the integrated utility tunnel are generated from these coordinates and imported into the model building software. Standard sections, ventilation sections, personnel access sections, and intersection sections of the integrated utility tunnel are then modeled and arranged according to the positioning lines.

[0055] A geometric model of the pipeline entering the corridor is constructed by organizing the relevant attribute information of the pipeline. Specifically, the latitude and longitude and connection status of each node are recorded along the pipeline from start to finish, and the attribute parameters (including type, size, and material) of the pipe segments are collected. Then, the `Create(Document, Elementld, Elementld, Elementld, XYZ, XYZ)` function is called to construct the pipeline model. Material attributes are assigned to the pipeline model by calling the `(PGMATERIALS, ParameterType.Material, true)` method, and its color representation is set accordingly.

[0056] By using pre-defined key models in the model building software, when overlapping with pipe segments, the system automatically adjusts and aligns the relevant pipes to achieve a smooth connection. During pipe fitting connections, the system ensures accurate configuration of primary and secondary pipes to prevent connection failures. Pipe connections are achieved using connectors. By calculating the spacing between pipe segments, appropriate connecting components and fittings are selected to complete the pipe connection and the construction of the access corridor.

[0057] When constructing the geometric model of the ancillary facilities, the integrated utility tunnel is systematically divided according to specific requirements and in conjunction with the as-built drawings, including fire protection systems, ventilation systems, etc., and the components contained in each divided system are analyzed. Subsequently, the ancillary facilities are classified or assigned to specific compartments. Based on the relative coordinates of various ancillary facilities, the geometric positions of the ancillary facility layout points are accurately determined, and the NewCurveByPoints(ReferencePointArray points) function is applied to generate spline curves for the ancillary facility layout based on specified reference points.

[0058] Next, by calling the MoveElement method, namely MoveElement(Document document, ElementId elementToMove, XYZ translation), the displacement operation of the auxiliary facility in three-dimensional space to a specific coordinate position is realized. Finally, by applying two Rotate(Lineaxis, doubleangle) operations, the precise positioning and arrangement of the auxiliary facility are achieved.

[0059] When constructing the equipment data transmission and control model, the OPC UA-IoT architecture is used to communicate with different IoT sensors, industrial sensors and PLC devices in the integrated utility tunnel using a unified communication protocol. Data interaction is performed with natural gas sensors, fire doors and fans in the natural gas compartment of the integrated utility tunnel. In the event of a natural gas leak in the natural gas compartment of the integrated utility tunnel, the natural gas sensors transmit data to the IoT server through the OPC UA-IoT architecture, and the IoT server sends control commands to the fire doors, fans, etc., so that the fire doors are closed and the fans are adjusted to emergency ventilation.

[0060] Step S102: Numerical simulation of natural gas leakage and diffusion in the integrated pipeline corridor is performed using the geometric digital model of the pipeline corridor. The obtained simulation data is transmitted to the Internet of Things server in real time to train the leakage and diffusion simulation model.

[0061] In this embodiment, the leakage is simulated by using a geometric digital model of the integrated utility tunnel that is consistent with the structure and layout of the integrated utility tunnel. This simulation obtains three-dimensional grid data (natural gas concentration simulation data) and monitoring point data (concentration simulation monitoring point data) of natural gas leakage under multiple working conditions for different leakage diffusion factors (different leakage locations, leakage rates, different leakage outlet directions, leakage outlet sizes, and wind speeds, etc.) within the integrated utility tunnel, and transmits them to the Internet of Things server.

[0062] In a specific example, a geometric digital model of the natural gas compartment within the integrated utility tunnel is constructed, and corresponding boundary conditions are set. A standard section of the natural gas compartment's fire compartment model is selected as a template. The fire compartment model has geometric dimensions of 3.5m × 1.8m × 200m, with standard DN300 pipelines, supports measuring 0.6m × 0.5m × 0.4m, and a top support measuring 0.4m × 0.08m × 200m. Fire extinguishing devices are installed within the fire compartment. Air inlets and exhaust outlets are located on both sides of the fire compartment. During this process, the utility tunnel's geometric digital model needs to be simplified accordingly. The model's geometric dimensions remain unchanged, but the ventilation and exhaust outlets are simplified to 1m × 1m.

[0063] Then, the parameters for the numerical simulation were set. Natural gas mainly consists of methane (…). Natural gas is composed of methane, which typically accounts for over 85% of the total natural gas content. Other gaseous components (such as ethane, propane, butane, and carbon dioxide) generally constitute a very small proportion and can be ignored; therefore, the composition is defined as methane-air. The operating temperature of the natural gas pipeline is set to 288K, and the natural gas satisfies the ideal gas law. A Simple solver based on a pressure-coupled algorithm is used. The computational framework is mainly constructed based on the energy conservation equation, the k-ε turbulent closed-loop model, and the component transport equation. Unsteady-state analysis methods are employed to investigate the time-varying characteristics of the natural gas diffusion mode within the underground integrated utility tunnel.

[0064] Next, the boundary conditions for natural gas pipeline leakage and diffusion in the integrated utility tunnel were determined, including various types such as inlet, outlet, and wall surface. The natural gas leak point and air inlet were defined as inlet conditions, while the air outlet was used as the outlet boundary condition. By establishing a natural gas leak model and calculating the actual leakage rate, the leak point was set as the mass flow inlet, and relevant parameters were configured according to the compressible fluid characteristics. Local total pressure and static pressure data, as well as the temperature of the natural gas at the leak point, were input into the mass flow inlet interface. The air inlet area was set as the velocity inlet, and the following formula was used:

[0065]

[0066] Calculate the ventilation velocity under leakage conditions; where, For the air inlet velocity, These are the length, width, and height of the natural gas compartment, respectively. These are the length and width of the air inlet, respectively.

[0067] The natural gas concentration monitoring points in the integrated utility tunnel are arranged according to standards. Similarly, in the numerical simulation of natural gas pipeline leakage and diffusion in the integrated utility tunnel, the monitoring points are also arranged according to the same standards in the geometric digital model of the tunnel. In specific leakage conditions, the leakage rate is set as follows: Leakage outlets are installed at 10-meter intervals, and the diameters of the leakage holes are set as follows: Meanwhile, considering wind speed, the ventilation conditions are divided into normal ventilation (wind speed...). Emergency ventilation (wind speed) Two categories were set up, with a total of 1710 different working conditions. The concentration data at the monitoring point was recorded every second within 240 seconds of leakage time for each working condition, as well as the grid data of the numerical simulation geometric model (i.e., the natural gas concentration simulation data of the spatial grid in the pipeline geometric digital model).

[0068] Simulation data is obtained through simulations of different operating conditions, and different simulation results will be obtained due to different causes of leakage. Under pipeline damage conditions, the simulation data shows that the concentration rises rapidly in the early stage of leakage, the concentration fluctuates drastically at the monitoring points, and is accompanied by local peaks or pulse changes; for leaks caused by material aging, the simulation data shows that the concentration change curve is smooth and the concentration distribution at the monitoring points is uniform; for leaks caused by human damage, the concentration at the monitoring points facing the leak outlet rises rapidly during the simulation process, or there are stage characteristics of a sharp increase in concentration and a significant increase in diffusion range in a short period of time. Based on the differences in the concentration change patterns of the simulation data, the cause of the natural gas leak in the pipeline can be preliminarily determined.

[0069] The differences in natural gas leakage under integrated utility tunnel scenarios compared to other scenarios are mainly due to the complex three-dimensional enclosed structure of the integrated utility tunnel (including obstacles such as pipes and supports) and the forced ventilation system, which restricts gas diffusion and significantly increases its directionality, resulting in a highly non-uniform and multi-gradient concentration distribution. To address this, mesh data obtained from numerical simulations under different leakage conditions are spatially interpolated along the length, height, and width directions of the integrated utility tunnel to transform them into ordered meshes. Specifically, the simulated natural gas concentration data is spatially interpolated along the length, height, and width directions of the integrated utility tunnel to convert it into ordered spatial mesh concentration data. This adapts to the concentration field characteristics under the structural constraints and ventilation effects of the integrated utility tunnel, accurately fitting its complex three-dimensional concentration gradient changes.

[0070] Specifically, firstly, in the longitudinal direction of the utility tunnel ( Interpolation is performed in the direction of the axis: for fixed The lateral monitoring section of the coordinates is taken from the upstream and downstream nodes. ) and( ) measured concentration value , According to the target point Calculating the longitudinal median concentration value using distance weighting of coordinates Similarly, for adjacent elevation surfaces ( ) same cross-section node , The calculation yields... This achieves longitudinal concentration gradient fitting. Then, in the transverse ( Perform secondary interpolation in the axis direction: , Based on this, extract the same elevation ( Lower section left and right nodes , The concentration value, according to Coordinate distance weight calculation The same process was applied to the other elevation surface data. This completes the two-dimensional fitting of the concentration field in the transverse section. Finally, in the vertical direction ( (Axis) Perform cubic interpolation: based on intermediate concentration values ​​at different elevations , (or , ),according to Target point is calculated using a weighted average of vertical distance coordinates. concentration value A continuous concentration field is constructed by coupling the concentration of three-dimensional grid nodes, providing refined data support for leakage diffusion simulation.

[0071] In a specific application scenario, firstly, along Interpolation of axes: For fixed axes and Value, in Linear interpolation is performed along the axis, resulting in two interpolation results. There are eight points in space, and their spatial coordinates are... , This is the concentration value at that point. Specifically, according to the formula:

[0072]

[0073] Calculate the two interpolation results and .

[0074] Then, the two interpolation results were calculated. and According to the formula:

[0075]

[0076] Along Interpolate the axis to obtain the interpolation result. and .

[0077] Finally, the interpolation results are... and According to the formula:

[0078]

[0079] Alternatively, according to the formula:

[0080]

[0081] along Interpolate the axis to obtain the target point. function value , i.e., target point concentration value .

[0082] After spatial interpolation processing of the natural gas concentration simulation data of the spatial grid in the obtained pipeline geometric digital model, in order to facilitate the training and evaluation of the leakage diffusion inference model, the grid data of numerical simulation and the concentration simulation monitoring point data per unit time at the monitoring point are classified and stored as a row. That is, the concentration simulation monitoring point data per second and the corresponding monitoring point coordinates, as well as the ordered spatial grid concentration data and the corresponding spatial grid coordinates, can be classified and stored to generate a standardized dataset for training the leakage diffusion model.

[0083] By optimizing the storage structure of different data sources, multi-source heterogeneous data is transformed into a spatiotemporally aligned format. By reshaping the arrangement of the data matrix, time-series information and spatial coordinates are aligned in an orderly manner along the channel dimension. The concentration simulation monitoring point data and the ordered spatial grid concentration data corresponding to each working condition after spatial interpolation are arranged according to the same time, and the spatial coordinates and time coordinates are stored separately, thus saving two different datasets.

[0084] Here, the concentration simulation monitoring point data from the sensor are arranged in sequential rows, and the concentration data from the ordered spatial grid are arranged according to... The axes are arranged in ascending order; simultaneously, for each operating condition, the concentration simulation monitoring point data of the sensor and the ordered spatial grid concentration data are arranged in rows according to time steps. ( The first column (where a positive integer represents the number of sensors) contains the concentration simulation monitoring point data for the sensors, and the remaining columns contain the concentration data for the ordered spatial grid. The sensor concentration simulation monitoring point data is arranged in the order of the monitoring point coordinates, and the ordered spatial grid concentration data is arranged in the order of the spatial grid coordinates.

[0085] Next, the generated standardized dataset was divided using a hierarchical random partitioning method, with 80% of the data used as the training set, 10% as the validation set, and 10% as the test set. A one-dimensional convolutional layer was used to locally sense the concentration simulation monitoring point data per unit time in the standardized dataset along the time dimension using a sliding window approach. The local temporal features of the concentration simulation monitoring point data obtained from the local sensing were used as the input to the leakage diffusion inference model. The leakage diffusion inference model was trained using the ordered spatial grid concentration data of the spatial grid as the prediction target.

[0086] In the fusion of spatiotemporal convolutional operations and long short-term memory network mechanisms, convolutional neural networks are used to extract local temporal features from the temporal data of sensors, with the input receiving dimension being... The sensor time-series data matrix is ​​used, where the time window length of 10 corresponds to 10 consecutive sampling times, and each time point contains multi-dimensional monitoring data from 13 sensors. In the time-series feature extraction stage, a one-dimensional convolutional layer (Conv1D) is first used for local feature mining. This layer is configured with 64 convolutional kernels of size 3, using a sliding window approach to perform local sensing along the time dimension. Each convolutional kernel generates a non-linear feature map through the ReLU activation function. Each convolutional kernel extracts features from a small window in the input sequence. The output of the convolutional layer is a feature map after convolution, preserving the temporal features of the input data and transforming them into a more abstract representation, as shown in the following equation:

[0087]

[0088] In the formula, This represents the output of the convolutional layer. This represents the output of the pooling layer. As weight, For deviation, This represents the convolution operation.

[0089] In this way, by integrating spatiotemporal convolution operations with long short-term memory network mechanisms, the number of parameters is effectively reduced and the computation speed is accelerated. Furthermore, by sharing convolution kernels, the most representative local features are automatically discovered without the need for manual feature engineering intervention. These features are then passed as input to the long short-term memory network to form a deep learning of temporal features.

[0090] The high-order features output by the convolutional layers are then fed into a two-layer long short-term memory network for temporal dependency modeling; the first layer of the long short-term memory network contains 128 hidden units, and its output dimension is [missing information]. The temporal state matrix. Let the input sequence be... The time-recurrent neural network unit at time step The calculation process is as follows:

[0091]

[0092] In the formula, Indicates the output of the forget gate. Indicates the input gate output. Indicates the state of candidate cells. , These represent the weight matrices of the forget gate and the input gate, respectively. , This is a deviation.

[0093] By passing the output of the first-layer Long Short-Term Memory (LSTM) network to the second-layer LSM network, which deeply aggregates the temporal state matrix through 256 hidden units, and finally outputs a 256-dimensional feature vector representing the dynamic evolution of the system, this cascaded structure achieves progressive learning from local features to short-term dependencies and then to long-term patterns. The second-layer LSM network receives... The calculation process is similar as input, but the parameter dimension changes. Finally, the hidden state at the last time step is taken, and the calculation process is as follows:

[0094]

[0095] In the formula, This represents the hidden state output of the second-layer Long Short-Term Memory network at the final time step. This represents the computational unit of the second-layer Long Short-Term Memory network. This represents the time-series state matrix output by the first layer of the Long Short-Term Memory network after processing the time-series data.

[0096] In the spatiotemporal feature fusion stage, the 256-dimensional temporal feature vector output by the Long Short-Term Memory network is expanded through a RepeatVector layer to... The spatiotemporal feature matrix was obtained and then stitched together with the three-dimensional coordinate information of 1000 grid points and 13 sensors to form a... The fusion feature space is constructed. This process uses a tensor broadcasting mechanism to align coordinate information with temporal features point by point, enabling each spatial node to perceive both the dynamic evolution of the system and its own prior spatial location. The feature decoding stage employs a TimeDistributed fully connected layer for distributed processing, with a hidden layer consisting of 128 neurons to transform spatial dimension features. Finally, the output layer generates predicted values ​​for 1000 grid nodes through linear regression.

[0097] Finally, a 3D model combining the pipe gallery geometric digital model, equipment data transmission and control model, and natural gas leak diffusion simulation model was implemented in Unity3D. The rendered geometric model was imported into Unity3D for rendering. In terms of material processing, the traditional shader parameters in the Revit material library were converted to a metal-roughness workflow parameter system conforming to the PBR standard. The roughness calculation adopted the formula... Nonlinear mapping was performed, and the original BIM texture was retopologically processed using Substance Designer to generate an Albedo map with a resolution of 2048×2048. Multi-quadrant UV unwrapping technology was used to bake the normals of complex nodes such as pipe flanges, effectively preserving surface details at the 0.1mm level.

[0098] Appropriate materials and textures were applied to each element. The ground used a green material to simulate the real-world environment; pipes used red and black materials to distinguish different types of pipes; and the support structure used a metal material to reflect its metallic texture. During the rendering process, suitable light sources were set to simulate realistic lighting effects. The spatial lighting system was constructed using the IES standard lighting model, and a 16-bit dynamic range industrial scene HDRI environment map was generated using HDR Light Studio. Considering the longitudinal extension characteristics of the pipe gallery, a uniform lighting scheme based on array light calculation was designed. Specifically, rectangular surface lights with an intensity of 1500 lumens were placed every 5 meters along the pipe gallery axis, and noise was eliminated using the Cycles renderer's optical path noise reduction algorithm. The metal support material used the GGX micro-surface reflection model, with a parameter combination of metallicity of 0.95 and roughness of 0.3 to achieve anisotropic reflection effects on the carbon steel surface.

[0099] On the Unity platform, the OpcUaHelper library is used to implement OPC UA client functionality, thereby establishing a communication connection with the OPC UA server. By sending requests to the OPC UA server, data resources can be retrieved, and the retrieved information can then be processed and analyzed accordingly. Simultaneously, when the Unity platform needs to update the machine learning model, it proactively sends variable update requests to the OPC UA server. Upon receiving the variable change information, the server immediately executes the corresponding operations.

[0100] Step S103: Transmit the operational status data of different sensors and protocols within the integrated utility tunnel to the IoT server in real time, and use the trained leakage diffusion simulation model to simulate the natural gas leakage diffusion in the integrated utility tunnel, generating a concentration grid for the natural gas leakage diffusion in the integrated utility tunnel; and correct the concentration grid data for the natural gas leakage diffusion in the integrated utility tunnel using the constructed physical model of the integrated utility tunnel and the leakage constraint model, to obtain the corrected natural gas concentration grid data for the integrated utility tunnel.

[0101] The trained integrated utility tunnel natural gas leakage and diffusion simulation model is deployed in the Unity engine resource directory. An OPC UA client with secure communication capabilities is built using the .NET Standard library to establish a protocol transmission channel and implement a ValueChange event subscription mechanism for sensor nodes within the natural gas compartment. This allows for real-time transmission of operational status data from different sensors and protocols within the integrated utility tunnel to the IoT server via the OPC UA-IoT data transmission architecture. Specifically, the operational status data of different devices and protocols within the integrated utility tunnel is converted into structured information with unified data specifications using the mapping rules of a standardized information model based on the device's private protocols. Then, edge nodes add 3D mesh coordinates and time tags to the structured information according to the requirements of the digital twin model and actively push it to the IoT server in real time. The mapping rules are as follows:

[0102]

[0103] In the formula, The first set up within the integrated utility tunnel Device time running status data Through standardized mapping The structured information obtained through transformation; For the first Protocol parsing function for each device.

[0104] In other words, the multi-source heterogeneous data collected within the integrated utility tunnel is mapped using a standardized information model based on the equipment's proprietary protocols. This transforms the operational status data generated by heterogeneous devices into structured information conforming to unified data specifications. Then, edge nodes, according to the requirements of the digital twin model, standardize the protocol-converted data (adding the tunnel's 3D grid coordinates and time tags). Finally, they proactively push the standardized data packets to the IoT server, which transmits them to the cloud in real-time via an encrypted channel, providing a real-time and accurate data source for leak propagation simulations. Simultaneously, in the event of a leak, commands are sent via the physical server to close fire doors and switch fans to emergency ventilation mode, obtaining natural gas concentration data from the integrated utility tunnel.

[0105] In a specific example, operational status data includes, but is not limited to, the device status, operating mode, control commands, and alarm information of devices of different functions, attributes, and types (fans, fire doors, and valves). During the operational status data mapping process, a unified namespace is used to organize the object types of each device hierarchically, thereby standardizing the data for easier management and access. For example, a fan object type might include parameters such as speed, temperature, and vibration; a fire door object type might include open and locked states; and a valve object type might include parameters such as opening degree and flow rate. These object types are managed through a unified namespace.

[0106] In this embodiment, the leakage constraint correction coefficient of the natural gas concentration grid data of the integrated utility tunnel under leakage constraints is determined by constructing a leakage constraint model; wherein, the leakage constraint model is:

[0107]

[0108] In the formula, The leakage constraint correction factor is used for the natural gas concentration grid data of the integrated utility tunnel under leakage constraints. This refers to the distance from the leak source to the sensor monitoring point inside the integrated utility tunnel during the spread of a natural gas leak. Real-time wind speed within the integrated utility tunnel during the spread of a natural gas leak; This shows the status of fire doors inside the utility tunnel during the spread of a natural gas leak. This indicates that the fire door is closed. This indicates that the fire door is open; The total length of the integrated utility tunnel. The limiting wind speed within the utility tunnel is a constant value. , This is the initial state of the fire door. .

[0109] As the distance constraint weight for the spread of natural gas leaks, The wind speed constraint weight during the diffusion of a natural gas leak. This refers to the constraint weights for the fire door status during natural gas leak propagation. In a specific application scenario, the distance constraint weights... Wind speed constraint weight Fire door status constraint weights .

[0110] Simultaneously, a physical model of the integrated utility tunnel was constructed. This model, based on the actual three-compartment layout of the project, encompasses key structural units such as standard sections, ventilation openings, hoisting openings, and personnel entrances / exits, and integrates internal pipeline facilities such as gas pipelines and cable trays. By constructing this physical model, historical natural gas leak and diffusion accidents within the integrated utility tunnel were analyzed to obtain experimental data on these accidents. Based on this experimental data and the collected real-world data from historical natural gas leaks and diffusion accidents, the leakage error correction coefficient for the natural gas concentration grid data of the integrated utility tunnel was calculated. Specifically, according to the formula:

[0111]

[0112] Calculate the leakage error correction factor for the natural gas concentration grid data of the integrated utility tunnel. In the formula, When studying the evolution of historical natural gas leak and diffusion accidents in integrated utility tunnels, the first Leakage diffusion experimental data at each sampling time point, To provide real leakage diffusion data corresponding to the experimental data on leakage diffusion, This represents the total number of data points from the leakage diffusion experiment.

[0113] Therefore, according to the formula:

[0114]

[0115] The natural gas concentration grid data of the integrated utility tunnel is corrected to obtain the corrected natural gas concentration grid data of the integrated utility tunnel. In the formula, This is grid data of natural gas concentration obtained by using a leakage diffusion simulation model to simulate the natural gas leakage diffusion in a utility tunnel.

[0116] The predicted grid data is transmitted to the system via OPC UA. The revised natural gas concentration grid data is then rendered in the pipeline's geometric digital model, constructing a digital twin model for the comprehensive pipeline natural gas leak diffusion simulation. This model is then visualized in the three-dimensional space of the pipeline. Here, a heat map is used to present the spatiotemporal distribution characteristics of natural gas concentration, and multi-level risk prevention and control strategies are automatically triggered based on a set natural gas leak concentration threshold. Specifically, in the pipeline physical model, a series of coordinated control measures are implemented, including closing fire doors, switching to emergency ventilation mode, simultaneously closing gas valves within the pipeline, cutting off power to the leak area, activating emergency lighting systems, and triggering audible and visual alarms. This constructs an integrated emergency response system from monitoring and perception to risk warning and risk prevention and control, enabling accurate simulation and emergency response to the leak diffusion process, and improving the situational awareness and intelligent coordinated handling of the pipeline's safety status.

[0117] In the 3D spatial visualization process, the predicted natural gas concentration values ​​on the pipe gallery walls are displayed using a heatmap in the digital twin system. The concentration data of the corresponding grid points are then assigned values ​​in the RGB range (blue to red) according to the proportion of concentration from 0-10%vol and given a base color in Blender. For the rendering of the internal 3D mesh data, a VBD-formatted volumetric gas cloud is added to each preset grid point in Blender. This volumetric gas cloud is colored using a similar mode to the wall surface, except that a value of 0 makes the volumetric gas cloud invisible, while only values ​​from 1-10%vol are set with corresponding colors. The surface with the base color and concentration point data is then baked and imported into Unity 3D for further rendering.

[0118] In Unity, a GasCell component is attached to each mesh cell of the prefabricated utility tunnel model, and MaterialPropertyBlock technology is used to achieve efficient updates of mesh concentration values. A hybrid rendering architecture is adopted, using a physically based volumetric lighting model to simulate the global gas diffusion process, and a GPU Instancing particle system to generate dynamic warning areas. A dual-channel adaptive rendering strategy is used, employing a real-time voxel tracking algorithm in the near field and switching to a pre-baked concentration field isosurface map in the far field. Color difference perturbations and motion blur effects are superimposed on the pipelines to construct an interactive leakage simulation environment with a concentration leakage field. Based on this, a digital twin model of natural gas leakage diffusion in a comprehensive utility tunnel is established. High-quality, high-frequency multi-source real-time data streams from different platforms and protocols are used to analyze the spatiotemporal evolution of natural gas leakage diffusion in the comprehensive utility tunnel, meeting the needs of rapid emergency response.

[0119] like Figure 9 As shown, this application also provides a twin-engine-driven integrated pipeline gallery natural gas leakage diffusion prediction and analysis system. The system uses the twin-engine-driven integrated pipeline gallery natural gas leakage diffusion prediction and analysis method of any of the above embodiments to extrapolate the natural gas leakage diffusion in integrated pipeline galleries. The system includes:

[0120] The twin platform unit 901 is configured to construct a geometric digital model of the integrated pipeline corridor with consistent structure and layout through multi-level modeling in the digital twin platform for natural gas leakage and diffusion simulation of the integrated pipeline corridor, and to construct a leakage and diffusion simulation model that integrates spatiotemporal convolution operation and long short-term memory network mechanism.

[0121] The numerical simulation and model training unit 902 is configured to perform numerical simulation of natural gas leakage and diffusion in the integrated pipeline corridor using a geometric digital model of the pipeline corridor, and transmit the obtained simulation data to the Internet of Things server in real time to train the leakage and diffusion inference model.

[0122] The leakage diffusion simulation unit 903 is configured to transmit the operational status data of different sensors and protocols within the integrated utility tunnel to the Internet of Things server in real time. This drives the trained leakage diffusion simulation model to simulate the natural gas leakage diffusion within the integrated utility tunnel, generating a concentration grid for the natural gas leakage diffusion within the integrated utility tunnel. Furthermore, the concentration grid data for the natural gas leakage diffusion within the integrated utility tunnel is corrected using the constructed physical model and leakage constraint model of the integrated utility tunnel, resulting in corrected natural gas concentration grid data for the integrated utility tunnel.

[0123] The twin-engine-driven integrated pipeline natural gas leakage diffusion prediction and analysis system provided in this application can realize the steps and processes of the twin-engine-driven integrated pipeline natural gas leakage diffusion prediction and analysis method in any of the above embodiments, and achieve the same technical effect, which will not be described in detail here.

[0124] In the description of this invention, it should be understood that the terms "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., refer to specific features, structures, materials, or characteristics described in connection with that embodiment or example, which are included in at least one embodiment or example of the present invention. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples.

[0125] The above description is merely a preferred embodiment of this application and is not intended to limit this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the protection scope of this application.

Claims

1. A method for predicting and analyzing the diffusion of natural gas leaks in integrated utility tunnels based on a twin engine-driven approach, characterized in that, include: In the digital twin platform for natural gas leakage and diffusion simulation of the integrated pipeline corridor, a geometric digital model of the pipeline corridor with the same structure and layout as the integrated pipeline corridor is constructed through multi-level modeling, and a leakage and diffusion simulation model integrating spatiotemporal convolution operation and long short-term memory network mechanism is constructed. Numerical simulation of natural gas leakage and diffusion in integrated utility tunnels is conducted using a geometric digital model of the tunnel. The simulation data is then transmitted in real time to an IoT server to train the leakage and diffusion simulation model. The operational status data of different sensors and protocols within the integrated utility tunnel are transmitted in real time to the IoT server to drive the trained leakage diffusion simulation model to simulate the natural gas leakage diffusion in the integrated utility tunnel, generating a concentration grid for the natural gas leakage diffusion in the integrated utility tunnel; and the concentration grid data for the natural gas leakage diffusion in the integrated utility tunnel is corrected by the constructed physical model of the integrated utility tunnel and the leakage constraint model, resulting in corrected natural gas concentration grid data for the integrated utility tunnel. Among them, by constructing a leakage constraint model, the leakage constraint correction coefficient of the natural gas concentration grid data of the integrated pipeline corridor under leakage constraints is determined; The leakage constraint model is as follows: ； In the formula, The leakage constraint correction factor is used for the natural gas concentration grid data of the integrated utility tunnel under leakage constraints. This refers to the distance from the leak source to the sensor monitoring point inside the integrated utility tunnel during the spread of a natural gas leak. Real-time wind speed within the integrated utility tunnel during the spread of a natural gas leak; This shows the status of fire doors inside the utility tunnel during the spread of a natural gas leak. This indicates that the fire door is closed. This indicates that the fire door is open; The total length of the integrated utility tunnel. The limiting wind speed within the utility tunnel is a constant value. , This is the initial state of the fire door. ; As the distance constraint weight for the spread of natural gas leaks, The wind speed constraint weight during the diffusion of a natural gas leak. Weights for fire door status constraints during natural gas leak propagation; The evolution of historical natural gas leak and diffusion accidents in the integrated utility tunnel was analyzed using a physical model of the tunnel, and the leakage error correction coefficient for the natural gas concentration grid data of the integrated utility tunnel was determined; wherein, according to the formula: ； Calculate the leakage error correction factor for the natural gas concentration grid data of the integrated utility tunnel. In the formula, To obtain leakage diffusion experimental data at sampling time points when studying the evolution of historical natural gas leakage diffusion accidents in integrated utility tunnels, To provide real leakage diffusion data corresponding to the experimental data on leakage diffusion, , This represents the total number of data points from the leakage diffusion experiment. According to the formula: ； The natural gas concentration grid data of the integrated utility tunnel is corrected to obtain the corrected natural gas concentration grid data of the integrated utility tunnel. In the formula, This is grid data of natural gas concentration obtained by using a leakage diffusion simulation model to simulate the natural gas leakage diffusion in a utility tunnel.

2. The method for predicting and analyzing natural gas leakage and diffusion in integrated utility tunnels based on twin engine drive as described in claim 1, characterized in that, In the geometric digital model of the pipeline corridor, multiple sets of operating conditions with different leakage diffusion factors are set to simulate the leakage of natural gas. The concentration simulation monitoring point data at different sensor monitoring points in the geometric digital model of the pipeline corridor, as well as the natural gas concentration simulation data of the spatial grid in the geometric digital model of the pipeline corridor, are obtained and transmitted to the Internet of Things server. Among them, the leakage diffusion factors include at least the leakage rate, leakage location, leakage outlet direction, leakage outlet size, and wind speed.

3. The method for predicting and analyzing natural gas leakage and diffusion in integrated utility tunnels based on twin engine drive as described in claim 2, characterized in that, Spatial interpolation was performed on the obtained natural gas concentration simulation data along the length, height, and width directions of the integrated utility tunnel to generate ordered spatial grid concentration data. The concentration simulation monitoring point data per unit time and the corresponding monitoring point coordinates, as well as the concentration data of the ordered spatial grid and the corresponding spatial grid coordinates, are classified and stored to generate a standardized dataset for training the leakage diffusion simulation model.

4. The method for predicting and analyzing natural gas leakage and diffusion in integrated utility tunnels based on twin engine drive as described in claim 3, characterized in that, A one-dimensional convolutional layer is used to locally sense the concentration of simulated monitoring point data per unit time in a standardized dataset along the time dimension using a sliding window. The local temporal features of the concentration simulated monitoring point data obtained from the local sensing are used as the input of the leakage diffusion inference model. The leakage diffusion inference model is trained using ordered spatial grid concentration data as the prediction target.

5. The method for predicting and analyzing natural gas leakage and diffusion in integrated utility tunnels based on twin engine drive according to claim 1, characterized in that, The operational status data of different equipment and protocols installed in the integrated utility tunnel are converted into structured information with unified data specifications through the mapping rules of the established equipment private protocol standardized information model; According to the requirements of the digital twin model, the edge nodes add three-dimensional grid coordinates and time tags to the structured information and actively push them to the IoT server in real time.

6. The method for predicting and analyzing natural gas leakage and diffusion in integrated utility tunnels based on twin engine drive according to claim 5, characterized in that, The mapping rule is: ； In the formula, The first set up within the integrated utility tunnel Device time running status data Through standardized mapping The structured information obtained through transformation; For the first Protocol parsing function for each device.

7. The method for predicting and analyzing natural gas leakage and diffusion in integrated utility tunnels based on twin engine drive according to claim 1, characterized in that, The spread of natural gas leaks in the integrated utility tunnel is simulated by transmitting real-time operational status data to the IoT server. At the same time, the fire doors in the integrated utility tunnel are closed and the fans are adjusted to emergency ventilation mode to obtain the natural gas concentration grid data of the integrated utility tunnel. The natural gas concentration grid data is then rendered in the geometric digital model of the utility tunnel.

8. A predictive analysis system for natural gas leakage and diffusion in integrated utility tunnels based on a twin engine, characterized in that, The method for predicting and analyzing the diffusion of natural gas leaks in integrated utility tunnels based on twin engine-driven integrated utility tunnels, as described in any one of claims 1-7, is used to simulate the diffusion of natural gas leaks in integrated utility tunnels. The system includes: The twin platform unit is configured to construct a geometric digital model of the integrated pipeline corridor with consistent structure and layout through multi-level modeling in the digital twin platform for natural gas leakage and diffusion simulation of the integrated pipeline corridor, as well as to construct a leakage and diffusion simulation model that integrates spatiotemporal convolution operation and long short-term memory network mechanism. The numerical simulation and model training unit is configured to perform numerical simulation of natural gas leakage and diffusion in the integrated pipeline corridor using a geometric digital model of the pipeline corridor, and transmit the obtained simulation data to the Internet of Things server in real time to train the leakage diffusion inference model. The leakage diffusion simulation unit is configured to transmit the operational status data of different sensors and protocols within the integrated utility tunnel to the IoT server in real time. This drives the trained leakage diffusion simulation model to simulate the natural gas leakage diffusion within the integrated utility tunnel, generating a concentration grid for the natural gas leakage diffusion within the integrated utility tunnel. Furthermore, the concentration grid data for the natural gas leakage diffusion within the integrated utility tunnel is corrected using the constructed physical model and leakage constraint model of the integrated utility tunnel, resulting in corrected natural gas concentration grid data for the integrated utility tunnel.

Images