LNG film type storage tank body gas pressure monitoring system and method

Abstract

The application discloses a kind of LNG film type storage tank body gas pressure monitoring system and method, belong to the field of tight detection.The system includes gas pressure sampling module, agent model simulation module and prediction monitoring module;The gas pressure acquisition module is equipped with gas flowmeter and gas pressure sampling sensor;The agent model simulation module is connected to the gas pressure sampling module to obtain the gas pressure difference range value of the main shielding space, the secondary shielding space;The prediction monitoring module is according to the gas pressure difference range value of the main shielding space, the secondary shielding space is through neural network model in real time prediction the gas pressure difference range value of different parts of the LNG film type storage tank.The present application can analyze, monitor and predict in real time LNG film type storage tank when carrying out pressure operation, and the maximum pressure difference information between the two spaces of tank body under different pressure and pumping speed, provide intuitive and reliable tank pressure state information.

Description

Technical Field

[0001] This invention belongs to the field of airtightness detection and relates to an LNG membrane-type storage tank pressure monitoring system and method. Background Technology

[0002] Membrane-type liquefied natural gas (LNG) storage tanks, also known as LNG membrane-type storage tanks, use an enclosure system to ensure the overall airtightness of the tank. For example... Figure 1a , Figure 1b As shown, the enclosure system of a membrane-type liquefied natural gas (LNG) storage tank is divided into a main shielding space, a secondary shielding space, and an internal tank space for storing LNG. The enclosure system includes a main shielding layer, a secondary shielding layer, and an outer tank. The space between the main and secondary shielding layers is the main shielding space, and the space between the secondary shielding layer and the outer tank is the secondary shielding space. The main shielding layer is made of 1.2mm thick 304L stainless steel corrugated plate with longitudinal and transverse grooves. The secondary shielding layer is made of a three-in-one sheet material consisting of two layers of fiberglass cloth and one layer of aluminum foil. Both the main and secondary shielding spaces are filled with insulation material, which is reinforced polyurethane foam sandwiched between upper and lower pressure plates. The main shielding layer is constructed by lap welding of corrugated plates, with a weld length of nearly 10,000 meters, making welding defects difficult to avoid. However, due to plate thickness and structural limitations, the application of conventional non-destructive testing techniques for detecting weld defects in the tank wall is limited. Therefore, tank wall sealing testing is often used to directly replace structural defect detection. When performing a sealing test on a tank, it is necessary to frequently change the air pressure between different structures of the tank to create a pressure difference for testing. However, since different structures have different pressure bearing capacities, it is necessary to precisely control the pressure difference between different structural spaces of the tank. However, existing testing systems only have the ability to sample pressure at a single point, and their ability to perceive and monitor the overall air pressure of the tank is relatively lacking. Summary of the Invention

[0003] This invention provides a gas pressure monitoring system and method for LNG membrane storage tanks, which can analyze, monitor and predict the maximum pressure difference between the two spaces of the LNG membrane storage tank and the tank pressure status information under different pressures and pumping speeds during pressure operation in real time.

[0004] To achieve the above objectives, the present invention adopts the following technical solution:

[0005] This application provides an LNG membrane-type storage tank pressure monitoring system for monitoring the pressure of the LNG membrane-type storage tank; the LNG membrane-type storage tank is provided with a main shielding layer, a secondary shielding layer and an outer tank, the space between the main shielding layer and the secondary shielding layer is the main shielding space, and the space between the secondary shielding layer and the outer tank is the secondary shielding space;

[0006] The system includes a barometric pressure sampling module, a proxy model simulation module, and a prediction and monitoring module.

[0007] The gas pressure acquisition module is equipped with a gas flow meter and a gas pressure sampling sensor. The gas flow meter is connected to the main shielding space and the secondary shielding space. The gas pressure sampling sensor is located on the main shielding layer and at the position of the outer tank corresponding to the secondary shielding layer.

[0008] The proxy model simulation module is connected to the air pressure sampling module to obtain the air pressure difference range value of the main shielding space and the secondary shielding space;

[0009] The prediction and monitoring module uses a neural network model to predict the pressure difference range of different parts of the LNG membrane storage tank in real time based on the pressure difference range of the main shielding space and the secondary shielding space.

[0010] Furthermore, the gas pressure sampling module also includes a vacuum pump, which is connected to the gas flow meter and evacuates the gas flow meter.

[0011] Furthermore, the agent model simulation module and the prediction monitoring module are located in a single device terminal.

[0012] Furthermore, the proxy model simulation module uses simulation software to simulate the gas pressure distribution data at any location of the LNG membrane storage tank during pressure operation, and simulates the maximum and minimum values ​​of the gas pressure difference between the gas pressure data at the location of the gas pressure sampling sensor and the main shielding space and the secondary shielding space.

[0013] Furthermore, the proxy model simulation module uses simulation software to simulate the gas pressure distribution data at any location of the LNG membrane storage tank under different pumping speeds, and simulates the maximum and minimum values ​​of the gas pressure difference between the gas pressure data at the location of the gas pressure sampling sensor and the main shielding space and the secondary shielding space.

[0014] Furthermore, the prediction and monitoring module includes two neural network models: a simulation network model and a prediction network model. The simulation network model learns the pressure difference range values ​​of the main shielding space and the secondary shielding space, and transmits the learning results to the prediction network model. The prediction network model predicts the pressure difference range values ​​of different parts of the LNG membrane storage tank.

[0015] Furthermore, during the learning process of the simulation network model on the pressure difference range values ​​of the main shielding space and the secondary shielding space, the simulation network model simulates the pressure difference range values ​​of different parts of the LNG membrane storage tank based on the pressure difference range values ​​of the main shielding space and the secondary shielding space to obtain simulated pressure difference range values, and optimizes the learning results by combining the simulated pressure difference range values; the prediction network model optimizes the predicted pressure difference range values ​​of different parts of the LNG membrane storage tank by combining the simulated pressure difference range values.

[0016] This application also provides a method for monitoring the gas pressure of an LNG membrane-type storage tank, using the LNG membrane-type storage tank gas pressure monitoring system described above. The method includes the following steps:

[0017] Obtain the pressure difference range values ​​of the main shielding space and the secondary shielding space of the LNG membrane storage tank;

[0018] Based on the pressure difference range between the main shielding space and the secondary shielding space, a neural network model is used to predict the pressure difference range at different locations of the LNG membrane storage tank in real time.

[0019] Furthermore, the step of predicting the pressure difference range of different parts of the LNG membrane storage tank in real time using a neural network model based on the pressure difference range of the main shielding space and the secondary shielding space includes:

[0020] The simulated pressure difference range is obtained by simulating the pressure difference range of different parts of the LNG membrane storage tank based on the pressure difference range of the main shielding space and the secondary shielding space.

[0021] The pressure difference range values ​​of the main shielding space and the secondary shielding space are learned, and the learning results are optimized by combining the simulated pressure difference range values.

[0022] Based on the learning results, the range of pressure difference values ​​at different parts of the LNG membrane storage tank is predicted.

[0023] Furthermore, the step of predicting the pressure difference range of different parts of the LNG membrane storage tank in real time using a neural network model based on the pressure difference range of the main shielding space and the secondary shielding space also includes:

[0024] The predicted pressure difference range for different parts of the LNG membrane storage tank is optimized by combining the simulated pressure difference range.

[0025] In summary, the advantages and positive effects of this invention, based on all the above technical solutions, can be summarized as follows:

[0026] This invention is based on a proxy model and neural network, which can realize real-time analysis, monitoring and prediction of the maximum pressure difference between the two spaces of the LNG membrane storage tank under different pressures and pumping speeds during pressure operation, and provide intuitive and reliable tank pressure status information. Attached Figure Description

[0027] Figure 1a This is a schematic diagram of the cross-sectional structure of an existing LNG membrane storage tank.

[0028] Figure 1b A schematic diagram of the longitudinal section structure of an existing LNG membrane storage tank;

[0029] Figure 2 This is a schematic diagram of the LNG membrane-type storage tank pressure monitoring method of the present invention;

[0030] Figure 3 This is a flowchart of a method for monitoring the gas pressure of an LNG membrane storage tank according to the present invention;

[0031] Figure 4 This is a schematic diagram of the connection structure of the air pressure sampling module of the present invention;

[0032] Figure 5a This is a schematic diagram of the installation position of the air pressure sampling sensor in the air pressure sampling module of the present invention;

[0033] Figure 5b This is a schematic diagram of the installation position of the gas flow meter in the gas pressure sampling module of the present invention;

[0034] Figure 6 This is a schematic diagram of the architecture of the agent model simulation module of the present invention;

[0035] Figure 7 This is a schematic diagram of the neural network structure of the prediction and monitoring module of the present invention;

[0036] Figure 8 This is a flowchart of another LNG membrane-type storage tank pressure monitoring method according to the present invention;

[0037] Figure 9 This is a flowchart illustrating the steps of the present invention to predict in real time the pressure difference range of different parts of the LNG membrane storage tank based on the pressure difference range of the main shielding space and the secondary shielding space using a neural network model. Detailed Implementation

[0038] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0039] Example 1

[0040] Embodiment 1 of this application provides an LNG membrane-type storage tank pressure monitoring system and method, which is an intelligent pressure monitoring system assisted in the tightness detection process. It consists of a pressure sampling module, a proxy model simulation module, and a prediction monitoring module. The pressure acquisition module is a hardware system for real-time acquisition and sensing of tank pressure data. The proxy model simulation module simulates the current tank pressure distribution to obtain the maximum and minimum pressure difference between the two spaces in the tank. The prediction monitoring module is deployed through a neural network model to achieve real-time prediction of the maximum and minimum pressure difference between the two spaces in the tank.

[0041] like Figure 1a , Figure 1b The diagram shown is a schematic of the LNG membrane storage tank structure, which mainly includes a main shielding layer 1, a secondary shielding layer 2, and an outer wall 3. Figure 1a The diagram shows a cross-sectional view of an LNG membrane storage tank. Figure 1b The diagram shows a longitudinal cross-sectional view of an LNG membrane-type storage tank. The sealed space within the main shielding layer 1 is the tank interior space 4; the sealed space between the main shielding layer 1 and the secondary shielding layer 2 is the main shielding space 5; and the enclosed space between the secondary shielding layer 2 and the outer wall 3 is the secondary shielding space 6.

[0042] like Figure 2 The diagram illustrates the principle of the detection method of this invention. Pumping speed information is input into a simulation network. After obtaining the predicted value of the pressure data at the sampling points, the deviation between the simulated pressure data and the predicted result is obtained. This deviation is used to determine the gradient descent direction and backpropagation is performed to improve robustness. During this process, real sampling point data is used to supplement and correct any deficiencies in the model simulation. After obtaining a high-performance simulation network, its predicted result is used as input, and the simulated pressure difference maximum / minimum value data is used as output to train the prediction network. After obtaining both a high-performance simulation network and a prediction network, the entire system is deployed.

[0043] like Figure 3 The flowchart shown is a detection method of the present invention, and its steps are as follows:

[0044] First, by establishing a tank pressure proxy model for LNG membrane storage tanks, the overall distribution of tank pressure under different pumping speeds of vacuum pumps is obtained, thereby simulating a large amount of pressure data at sampling points and the maximum and minimum pressure difference data between the two spaces of the tank, providing a basic training dataset for neural network training.

[0045] Secondly, through real-time field tests, real-time air pressure data at different positions in the two spaces of the tank under different pumping speeds were collected to provide a small amount of real data for the neural network model to supplement and correct its accuracy.

[0046] Finally, the two neural networks of the established prediction module were trained. First, both networks were trained together using simulated model data, enabling the simulated network to predict the pressure at the sampling point based on pumping speed, and the prediction network to predict the maximum pressure difference between the two spaces in the tank based on the pressure distribution at the sampling point. Then, real data was used to supplement and correct the simulated network, improving its prediction accuracy and robustness, thereby enhancing the overall predictive capability of the network.

[0047] like Figure 4 The diagram shows the connection structure of the air pressure sampling module of this invention. The core of the monitoring system is a device terminal, whose main function is to collect the sensing signals from the sensor system and deploy the trained predictive monitoring module. This terminal can directly display the maximum pressure difference information, providing a reference for operators to perform pressure operations.

[0048] Figure 5a This is a schematic diagram of the installation position of the air pressure sampling sensor 8 in the air pressure sampling module of the present invention; Figure 5b This is a schematic diagram showing the installation position of the gas flow meter 7 in the gas pressure sampling module of the present invention. Wherein, Figure 5a Multiple pressure sampling points were set up at different typical locations inside the tank to record the air pressure data at different locations at different times under the actual pumping speed. Figure 5b It is a two-space pumping speed sensor, which mainly consists of two gas flow meters 7, used to record instantaneous gas flow.

[0049] like Figure 6 The diagram shows the architecture of the proxy model simulation module of this invention. Its main function is to use simulation software to create a simulation model of the entire tank, simulate the pumping process from atmospheric pressure to the maximum pressure required by the process on the model, and generate the air pressure distribution of various parts of the tank at different times under different pumping speeds. The simulated data is then organized into a dataset of different pumping speeds, sampling points, and the location of the maximum pressure difference, providing data support for neural network training.

[0050] like Figure 7 The diagram shows the neural network structure of the predictive monitoring module of this invention, which mainly consists of two fully connected neural networks connected end-to-end. In this network, the prediction results calculated by the simulation network are directly input into the prediction network. The initial network design sets one intermediate layer for the simulation network and two intermediate layers for the prediction network, but these can be appropriately increased later depending on the size of the tank and the complexity of its internal structure to improve robustness. The activation functions for the network input and intermediate layers are ReLU activation functions, the output layer uses a Sigmoid activation function, and the loss function is the cross-entropy loss function. After training with a large amount of data, the network can be directly deployed on a terminal and connected to a monitoring and sensing system.

[0051] Example 2

[0052] Embodiment 2 of this application provides an LNG membrane-type storage tank pressure monitoring system for monitoring the pressure of the LNG membrane-type storage tank; such as Figure 1a , Figure 1b As shown, the LNG membrane storage tank is provided with a main shielding layer 1, a secondary shielding layer 2 and an outer tank. The space between the main shielding layer 1 and the secondary shielding layer 2 is the main shielding space 5, and the space between the secondary shielding layer 2 and the outer tank is the secondary shielding space 6.

[0053] Combination Figure 4 , Figure 5a , Figure 5b As shown, the system includes a pressure sampling module, a proxy model simulation module, and a prediction and monitoring module. The pressure sampling module is equipped with a gas flow meter 7 and a pressure sampling sensor 8. The gas flow meter 7 is connected to the main shielding space 5 and the secondary shielding space 6. The pressure sampling sensor 8 is located on the main shielding layer 1 and at the position on the outer tank corresponding to the secondary shielding layer 2. The proxy model simulation module is connected to the pressure sampling module to obtain the pressure difference range value of the main shielding space 5 and the secondary shielding space 6. The prediction and monitoring module predicts the pressure difference range value of different parts of the LNG membrane storage tank in real time based on the pressure difference range value of the main shielding space 5 and the secondary shielding space 6 using a neural network model.

[0054] Furthermore, the gas pressure sampling module also includes a vacuum pump, which is connected to the gas flow meter 7 and evacuates the gas flow meter 7.

[0055] Furthermore, the agent model simulation module and the prediction monitoring module are located in a single device terminal.

[0056] Furthermore, the proxy model simulation module uses simulation software to simulate the gas pressure distribution data at any location of the LNG membrane storage tank during pressure operation, and simulates the maximum and minimum values ​​of the gas pressure difference between the gas pressure data at the location of the gas pressure sampling sensor 8 and the main shielding space 5 and the secondary shielding space 6.

[0057] Furthermore, the proxy model simulation module simulates the gas pressure distribution data at any location of the LNG membrane storage tank under different pumping speeds using simulation software, and simulates the maximum and minimum values ​​of the gas pressure difference between the gas pressure data at the location of the gas pressure sampling sensor 8 and the main shielding space 5 and the secondary shielding space 6.

[0058] like Figure 7As shown, the prediction and monitoring module includes two neural network models: a simulation network model and a prediction network model. The simulation network model learns the pressure difference range of the main shielding space 5 and the secondary shielding space 6, and transmits the learning results to the prediction network model. The prediction network model predicts the pressure difference range of different parts of the LNG membrane storage tank.

[0059] Furthermore, during the learning process of the pressure difference range values ​​of the main shielding space 5 and the secondary shielding space 6, the simulation network model simulates the pressure difference range values ​​of different parts of the LNG membrane storage tank based on the pressure difference range values ​​of the main shielding space 5 and the secondary shielding space 6 to obtain simulated pressure difference range values, and optimizes the learning results by combining the simulated pressure difference range values; the prediction network model optimizes the predicted pressure difference range values ​​of different parts of the LNG membrane storage tank by combining the simulated pressure difference range values.

[0060] like Figure 8 As shown, this application also provides a method for monitoring the gas pressure of an LNG membrane storage tank, using the LNG membrane storage tank gas pressure monitoring system described above. The method includes the following steps:

[0061] S1. Obtain the pressure difference range of the main shielding space 5 and the secondary shielding space 6 of the LNG membrane storage tank;

[0062] S2. Based on the pressure difference range of the main shielding space 5 and the secondary shielding space 6, the pressure difference range of different parts of the LNG membrane storage tank is predicted in real time using a neural network model.

[0063] like Figure 9 As shown, the step of predicting the pressure difference range of different parts of the LNG membrane storage tank in real time using a neural network model based on the pressure difference range of the main shielding space 5 and the secondary shielding space 6 includes:

[0064] S21. Based on the pressure difference range values ​​of the main shielding space 5 and the secondary shielding space 6, the pressure difference range values ​​of different parts of the LNG membrane storage tank are simulated to obtain the simulated pressure difference range values.

[0065] S22. Learn the air pressure difference range values ​​of the main shielding space 5 and the secondary shielding space 6, and optimize the learning results by combining the simulated air pressure difference range values;

[0066] S23. Based on the learning results, predict the range of gas pressure difference values ​​at different parts of the LNG membrane storage tank.

[0067] like Figure 9As shown, the step of predicting the pressure difference range of different parts of the LNG membrane storage tank in real time using a neural network model based on the pressure difference range of the main shielding space 5 and the secondary shielding space 6 further includes:

[0068] S24. Optimize the predicted pressure difference range values ​​for different parts of the LNG membrane storage tank by combining the simulated pressure difference range values.

[0069] In summary, the advantages and positive effects of this invention, based on all the above technical solutions, can be summarized as follows:

[0070] This invention is based on a proxy model and neural network, which can realize real-time analysis, monitoring and prediction of the maximum pressure difference between the two spaces of the LNG membrane storage tank under different pressures and pumping speeds during pressure operation, and provide intuitive and reliable tank pressure status information.

Claims

1. A gas pressure monitoring system for an LNG membrane-type storage tank, characterized in that, Used for monitoring the gas pressure of LNG membrane storage tanks; the LNG membrane storage tank is provided with a main shielding layer, a secondary shielding layer and an outer tank, the space between the main shielding layer and the secondary shielding layer is the main shielding space, and the space between the secondary shielding layer and the outer tank is the secondary shielding space; The system includes a barometric pressure sampling module, a proxy model simulation module, and a prediction and monitoring module. The gas pressure sampling module is equipped with a gas flow meter and a gas pressure sampling sensor. The gas flow meter is connected to the main shielding space and the secondary shielding space. The gas pressure sampling sensor is located on the main shielding layer and at the position on the outer tank corresponding to the secondary shielding layer. The proxy model simulation module is connected to the air pressure sampling module to obtain the air pressure difference range value of the main shielding space and the secondary shielding space; The prediction and monitoring module uses a neural network model to predict the pressure difference range of different parts of the LNG membrane storage tank in real time based on the pressure difference range of the main shielding space and the secondary shielding space.

2. The LNG membrane-type storage tank pressure monitoring system according to claim 1, characterized in that, The gas pressure sampling module also includes a vacuum pump, which is connected to the gas flow meter and evacuates the gas flow meter.

3. The LNG membrane-type storage tank pressure monitoring system according to claim 1, characterized in that, The agent model simulation module and the prediction monitoring module are located in a single device terminal.

4. The LNG membrane-type storage tank pressure monitoring system according to claim 1, characterized in that, The proxy model simulation module uses simulation software to simulate the gas pressure distribution data at any location of the LNG membrane storage tank during pressure operation, and simulates the maximum and minimum values ​​of the gas pressure difference between the gas pressure data at the location of the gas pressure sampling sensor and the main shielding space and the secondary shielding space.

5. The LNG membrane-type storage tank pressure monitoring system according to claim 1, characterized in that, The proxy model simulation module uses simulation software to simulate the gas pressure distribution data at any location of the LNG membrane storage tank under different pumping speeds, and simulates the maximum and minimum values ​​of the gas pressure difference between the gas pressure data at the location of the gas pressure sampling sensor and the main shielding space and the secondary shielding space.

6. The LNG membrane-type storage tank pressure monitoring system according to claim 1, characterized in that, The prediction and monitoring module includes two neural network models: a simulation network model and a prediction network model. The simulation network model learns the pressure difference range of the main shielding space and the secondary shielding space, and transmits the learning results to the prediction network model. The prediction network model predicts the pressure difference range of different parts of the LNG membrane storage tank.

7. The LNG membrane-type storage tank pressure monitoring system according to claim 6, characterized in that, During the learning process of the simulation network model on the pressure difference range of the main shielding space and the secondary shielding space, the simulation network model simulates the pressure difference range of different parts of the LNG membrane storage tank based on the pressure difference range of the main shielding space and the secondary shielding space to obtain simulated pressure difference range values, and optimizes the learning results by combining the simulated pressure difference range values; the prediction network model optimizes the predicted pressure difference range values ​​of different parts of the LNG membrane storage tank by combining the simulated pressure difference range values.

8. A method for monitoring the gas pressure of an LNG membrane-type storage tank, characterized in that, The method of using the LNG membrane-type storage tank pressure monitoring system according to any one of claims 1 to 7 includes the following steps: Obtain the pressure difference range values ​​of the main shielding space and the secondary shielding space of the LNG membrane storage tank; Based on the pressure difference range between the main shielding space and the secondary shielding space, a neural network model is used to predict the pressure difference range at different locations of the LNG membrane storage tank in real time.

9. The method for monitoring the gas pressure of an LNG membrane-type storage tank according to claim 8, characterized in that, The step of predicting the pressure difference range of different parts of the LNG membrane storage tank in real time using a neural network model based on the pressure difference range of the main shielding space and the secondary shielding space includes: The simulated pressure difference range is obtained by simulating the pressure difference range of different parts of the LNG membrane storage tank based on the pressure difference range of the main shielding space and the secondary shielding space. The pressure difference range values ​​of the main shielding space and the secondary shielding space are learned, and the learning results are optimized by combining the simulated pressure difference range values. Based on the learning results, the range of pressure difference values ​​at different parts of the LNG membrane storage tank is predicted.

10. The method for monitoring the gas pressure of an LNG membrane-type storage tank according to claim 9, characterized in that, The step of predicting the pressure difference range of different parts of the LNG membrane storage tank in real time using a neural network model based on the pressure difference range of the main shielding space and the secondary shielding space further includes: The predicted pressure difference range for different parts of the LNG membrane storage tank is optimized by combining the simulated pressure difference range.

Images