Inflation control method and device for pressure storage tank and medium

Abstract

The invention belongs to the technical field of pressure storage tank control methods, and provides a pressure storage tank inflation control method and device and a medium. According to the pressure storage tank inflation control method, a physical model of the dynamic inflation process is built; acquiring unknown parameters of the physical model according to the calibration data and the physical model; obtaining a trained machine learning model through calibration data training; obtaining a soft measurement model through the calibrated physical model and the trained machine learning model; acquiring real-time data, and acquiring a real-time mass flow through the real-time data and the soft measurement model; obtaining a current valve opening instruction according to the real-time data and the real-time mass flow; according to the method, by fusing multi-source sensing information such as strain and temperature and combining a soft measurement model established through early-stage calibration, non-intrusive pressure and flow high-precision sensing is achieved, closed-loop inflation control is conducted based on the non-intrusive pressure and flow high-precision sensing, and therefore on the premise that safety is guaranteed, the inflation efficiency is remarkably improved, and the system cost is reduced.

Description

Technical Field

[0001] This application belongs to the technical field of pressure storage tank control methods, and specifically relates to a pressure storage tank inflation control method, device and medium. Background Technology

[0002] Compressed air energy storage (CASS) is a crucial technology for addressing the intermittency of renewable energy and achieving peak shaving and valley filling in the power grid. One method of storing high-pressure air in CASS systems is by integrating multiple pressure tanks into a cluster. This approach offers flexibility in deployment, cluster distribution, and decentralized pressure control, making it suitable for sites where underground gas resources are difficult to access. During periods of low grid load, electricity drives a compressor to obtain compressed air, which is then sealed in the pressure tank. During periods of high load, the compressed air is released, passing through an expander to drive a generator. Real-time monitoring of the CASS storage pressure is essential for the control and safety of the equipment. Ensuring safety, improving storage efficiency, and reducing construction costs during the filling and discharging of these pressure tanks present significant challenges in practical engineering projects.

[0003] The pressure monitoring and control methods for pressure storage tank clusters in existing compressed air energy storage systems are as follows: (1) In terms of safety monitoring, the common practice is to measure pressure by opening holes in the pressure header, and pressure sensors are generally not installed by opening holes in the gas storage tank body. Each additional measuring point introduces a potential risk of high-pressure gas leakage, which poses a serious safety hazard to large-scale clusters; at the same time, there is a lack of direct monitoring of the stress and strain state of the container body, and it is impossible to assess the structural health and fatigue damage of the container body in real time.

[0004] (2) The control is crude and inefficient: the inflation control is mostly based on a single pressure feedback to perform simple switching or proportional control, which cannot sense the temperature field changes and the actual mass flow rate during the inflation process. This causes the inflation process to deviate from the ideal isothermal process, resulting in significant compression heat loss. After cooling, the pressure drops significantly, reducing the energy storage density and system efficiency.

[0005] (3) High cost and poor economic efficiency: Equipping each gas storage tank or branch with a high-precision flow meter will greatly increase the initial investment cost of large gas storage clusters; at the same time, the lack of fine perception of the status of individual containers makes it difficult to achieve coordinated optimization control at the cluster level (such as flow distribution and gas filling sequence scheduling).

[0006] (4) Lack of life management basis: It is impossible to obtain stress and strain data reflecting the fatigue state of the container in real time, making it difficult to implement predictive maintenance based on actual damage accumulation, which may lead to over-maintenance or under-maintenance.

[0007] In summary, existing technologies suffer from three major drawbacks: high safety risks, high economic costs, and low operating efficiency, making it difficult to meet the growing demands of modern large-scale compressed air energy storage systems for safe, economical, and intelligent operation.

[0008] Therefore, there is an urgent need for a pressure tank filling control method that is highly safe, economical, and efficient. Summary of the Invention

[0009] To address the above problems, this invention provides a method for controlling the inflation of a pressure storage tank, comprising the following steps: Obtain N sets of calibration data during the inflation process of the pressure storage tank; Construct a physical model of the dynamic inflation process of a pressure storage tank; Based on the calibration data and the physical model, obtain the unknown parameters of the physical model; By substituting the unknown parameters into the physical model, a calibrated physical model can be obtained. The calibrated physical model is introduced into the machine learning model, and the trained machine learning model is obtained by training with calibrated data. A soft measurement model is obtained by using a calibrated physical model and a trained machine learning model; Acquire real-time data, and obtain real-time quality flow rate through real-time data and soft measurement models; The current valve opening command is obtained based on real-time data and real-time mass flow rate, and the pressure tank valve is controlled according to the valve opening command.

[0010] Furthermore, obtain N sets of calibration data during the pressure tank filling process, including the following steps: In the environmental test chamber, under the simulated extreme environmental temperatures of winter and summer, multiple sets of inflation experiments were conducted at different initial pressures in the pressure tank and at three constant flow rates: low, medium, and high. The measurement data obtained for each group of experiments are used as calibration data.

[0011] Furthermore, the calibration data includes wall strain data, outer wall temperature data, inlet air temperature data, and mass flow rate data. The wall strain data is acquired through a strain gauge array.

[0012] Furthermore, the physical model is as follows:

[0013] in, Specific heat capacity at constant volume For the mass of the gas inside the tank, The gas volume average temperature, For quality flow, Specific heat capacity at constant pressure Intake air temperature, The convective heat transfer coefficient is... For heat exchange area, This refers to the wall temperature.

[0014] Furthermore, the soft measurement model is as follows:

[0015] in, The output of the calibrated physical model, The output of the trained machine learning model is α, which is the fusion weight determined based on the model's performance on the validation set.

[0016] Furthermore, the current valve opening command is obtained based on real-time data and real-time mass flow rate, and the pressure storage tank valve is controlled according to the valve opening command, including the following steps: The real-time pressure inside the pressure tank is obtained based on the real-time strain data of the wall surface. Based on the real-time pressure, real-time mass flow rate, external wall temperature data, and inlet air temperature data inside the pressure tank, the optimal valve opening command is calculated in real time using a multi-objective optimization control algorithm. Control the pressure tank valves according to the valve opening command.

[0017] A pressure tank inflation control device, comprising, The sensing device acquires N sets of calibration data during the inflation process of the pressure storage tank; A computing device is used to construct a physical model of the dynamic inflation process; based on calibration data and the physical model, unknown parameters of the physical model are obtained; the unknown parameters are input into the physical model to obtain a calibrated physical model; the calibrated physical model is introduced into a machine learning model, and a trained machine learning model is obtained by training with calibration data; a soft measurement model is obtained through the calibrated physical model and the trained machine learning model. The control device acquires real-time data and obtains real-time mass flow rate through real-time data and a soft measurement model. The actuator obtains the current valve opening command based on real-time data and real-time mass flow rate, and controls the pressure tank valve according to the valve opening command.

[0018] Furthermore, the sensing device includes strain gauges, an outer wall temperature sensor, an inlet temperature sensor, and a temporary flow meter. Several circumferential arrays of strain gauges and an outer wall temperature sensor are fixedly installed on the outer surface of the pressure tank, forming a distributed strain gauge array. A regulating valve, an inlet temperature sensor, and a temporary flow meter are fixedly installed on the inlet pipeline of the pressure tank.

[0019] A pressure tank inflation control device includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements the steps of the method described above.

[0020] A computer-readable storage medium storing a computer program that, when executed by a processor, implements the steps of the method described above.

[0021] Beneficial effects: 1. The pressure tank inflation control method of the present invention constructs a physical model of the dynamic inflation process; obtains unknown parameters of the physical model based on calibration data and the physical model; introduces the calibrated physical model into a machine learning model, and obtains a trained machine learning model through training with calibration data; obtains a soft sensor model through the calibrated physical model and the trained machine learning model; acquires real-time data, and obtains real-time mass flow through the real-time data and the soft sensor model; obtains the current valve opening command based on the real-time data and real-time mass flow, and controls the pressure tank valve according to the valve opening command. This method achieves non-invasive, high-precision pressure and flow sensing by fusing multi-source sensor information such as strain and temperature, combined with the soft sensor model established through prior calibration, and performs closed-loop inflation control based on this, thereby significantly improving inflation efficiency and reducing system costs while ensuring safety.

[0022] 2. The calibration data of the pressure tank inflation control method of the present invention includes wall strain data, outer wall temperature data, inlet air temperature data and mass flow rate data; non-contact strain measurement is used to invert pressure, avoiding the leakage risk introduced by installing pressure sensors through openings on high-pressure vessels.

[0023] 3. The pressure tank filling control method of the present invention uses an externally installed distributed strain gauge array as the core sensing element without the need for an intrusive flow meter. This array not only obtains the internal pressure of the container with high precision and non-invasively through mechanical inversion, completely eliminating the risk of high-pressure leakage, but also simultaneously provides the stress distribution and strain history of the container body, providing direct data for structural health monitoring and fatigue life assessment. Combined with an external wall temperature sensor array, it achieves comprehensive and safe sensing of the filling thermodynamic process.

[0024] Other features and advantages of this application will be set forth in the following description and will be apparent in part from the description or may be learned by practicing the application. The objectives and other advantages of this application may be realized and obtained by means of the structures pointed out in the description and the accompanying drawings. Attached Figure Description

[0025] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0026] Figure 1 A schematic flowchart of the pressure tank inflation control method in an embodiment of this application is shown.

[0027] Figure 2 A flowchart illustrating the calibration and modeling stage in an embodiment of this application is shown.

[0028] Figure 3 A flowchart illustrating the online operation and control phase in an embodiment of this application is shown.

[0029] Figure 4 A schematic diagram of the process for inferring mass flow rate in an embodiment of this application is shown.

[0030] Figure 5 A schematic diagram of the pressure tank inflation control device in an embodiment of this application is shown.

[0031] Figure 6 A schematic diagram of the sensor unit arrangement of the pressure tank inflation control device in an embodiment of this application is shown.

[0032] Figure 7 A schematic diagram of the strain gauge arrangement in an embodiment of this application is shown.

[0033] Figure 8 A schematic diagram of the process for evaluating the sealing of strike-slip faults in Embodiment 5 of the present invention is shown.

[0034] Explanation of reference numerals in the attached drawings: 201, strain gauge; 202, external wall temperature sensor; 203, inlet temperature sensor; 204, PLC; 205, DCS; 206, regulating valve; 207, inlet pipeline; 208, temporary flow meter. Detailed Implementation

[0035] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0036] Example 1 refer to Figure 1 A method for controlling the inflation of a pressure storage tank, comprising: Obtain N sets of calibration data during the inflation process of the pressure storage tank; Construct a physical model of the dynamic inflation process; Based on the calibration data and the physical model, obtain the unknown parameters of the physical model; By substituting the unknown parameters into the physical model, a calibrated physical model can be obtained. The calibrated physical model is introduced into the machine learning model, and the trained machine learning model is obtained by training with calibrated data. A soft measurement model is obtained by using a calibrated physical model and a trained machine learning model; Acquire real-time data, and obtain real-time quality flow rate through real-time data and soft measurement models; This invention provides a method and system for controlling the inflation of a pressure tank in a compressed air energy storage system. The method acquires the current valve opening command based on real-time data and real-time mass flow rate, and controls the pressure tank valve accordingly. The method integrates multi-source sensing information (stress, strain, wall temperature, and inlet temperature) with a soft-sensor model (derivation of easily measurable variables and data calibration). By fusing multi-source sensor information such as strain and temperature, and combining it with a pre-calibrated soft-sensor model, the method achieves non-invasive, high-precision sensing of pressure and flow rate. Based on this, closed-loop inflation control is implemented, thereby significantly improving inflation efficiency and reducing system costs while ensuring safety.

[0037] Furthermore, obtain N sets of calibration data during the pressure tank filling process, including the following steps: In the environmental test chamber, under simulated extreme winter and summer temperatures, multiple sets of inflation experiments were conducted at different initial pressures in the pressure tank, using three constant flow rates: low, medium, and high. The measurement data from each set of experiments were used as calibration data.

[0038] Calibration data includes wall strain data, external wall temperature data, inlet air temperature data, and mass flow rate data. Wall strain data is acquired through a strain gauge array. Non-contact strain measurement is used to invert pressure, avoiding the leakage risks introduced by installing pressure sensors through openings in high-pressure vessels. Simultaneously, distributed strain monitoring can directly assess structural health and fatigue status, enabling safety early warning.

[0039] Furthermore, the physical model is as follows: ; in, Specific heat capacity at constant volume For the mass of the gas inside the tank, The gas volume average temperature, For quality flow, Specific heat capacity at constant pressure Intake air temperature, The convective heat transfer coefficient is... For heat exchange area, This refers to the wall temperature.

[0040] Furthermore, the soft measurement model is as follows:

[0041] in, The output of the calibrated physical model, The output of the trained machine learning model is α, which is the fusion weight determined based on the model's performance on the validation set.

[0042] Furthermore, the current valve opening command is obtained based on real-time data and real-time mass flow rate, and the pressure storage tank valve is controlled according to the valve opening command, including the following steps: The real-time pressure inside the pressure tank is obtained based on the real-time strain data of the wall surface. Based on the real-time pressure, real-time mass flow rate, external wall temperature data, and inlet air temperature data inside the pressure tank, the optimal valve opening command is calculated in real time using a multi-objective optimization control algorithm. Control the pressure tank valves according to the valve opening command.

[0043] Example 2 This invention provides a pressure tank inflation control method, which is a compressed air energy storage system pressure tank inflation control method based on multi-source sensing and soft measurement, including a calibration modeling stage and an online operation and control stage.

[0044] refer to Figure 2 During the calibration and modeling phase, the following steps are performed: Step 1: Before the pressure storage tank (such as a gas storage tank) leaves the factory, temporarily install a temporary flow meter 208 on the inlet pipe 207. Conduct a series of inflation (deflation) tests on the pressure storage tank under environmental test chamber conditions or simulating different seasonal operating conditions. The tests cover different initial pressures, initial temperatures, and various constant inflation (deflation) flow rates. Throughout the testing process, synchronously collect the following data at high frequency: wall strain data obtained through an array of strain gauges 201 attached to the outer wall of the container. The outer wall surface temperature data obtained by the outer wall temperature sensor 202 attached to the outer wall of the container. The inlet air temperature data is obtained by the inlet temperature sensor 203, which is installed on the inlet pipe 207 and located on the side of the regulating valve 206 away from the pressure tank. And the actual mass flow rate data obtained through the temporary flow meter 208. Temporary pressure data is obtained through temporary pressure gauges. Simultaneously, distributed strain gauge 201 monitoring can directly assess structural health and fatigue status, enabling safety early warning.

[0045] Step 2: Based on the detailed calibration data described above, a soft measurement model is constructed. The preferred approach for constructing this model is a fusion of physical mechanisms and data-driven methods: a thermodynamic-heat transfer coupled physical model describing the dynamic inflation process is established based on the principles of mass conservation, energy conservation, and heat transfer. (1) in, Specific heat capacity at constant volume For the mass of the gas inside the tank, The gas volume average temperature, For quality flow, Specific heat capacity at constant pressure Intake air temperature, The convective heat transfer coefficient is... For heat exchange area, This refers to the wall temperature.

[0046] Step 3: Use calibration data to test the unknown parameters (such as the equivalent heat transfer coefficient) in the above physical model. Specific heat capacity at constant volume and specific heat capacity at constant pressure The physical model is then identified and calibrated. Next, using the calibrated physical model output as a foundation, a machine learning model (such as an ANN) is introduced and trained with calibration data to learn nonlinear features and errors not covered by the physical model, resulting in a trained machine learning model. Finally, the two (the trained machine learning model and the calibrated physical model) are fused through weighted or concatenated methods to form a high-precision soft measurement model. This model can use real-time acquired strain (which can be inverted into pressure) to form a high-precision soft measurement model. ), outer wall temperature and intake air temperature As input, accurately calculate the real-time mass flow rate. .

[0047] refer to Figure 3 During the online operation and control phase, the following steps are performed: Step 4: Remove the temporary flow meter. In actual operation, data is collected in real time only through the permanently installed strain gauge (201), external wall temperature sensor 202, and inlet temperature sensor 203.

[0048] Step 5: Input the real-time data into the trained soft sensing model to calculate the real-time mass flow rate online. Simultaneously, the real-time pressure inside the container was retrieved through distributed strain gauge data fusion. .

[0049] Specifically, before the pressure storage tank (such as a gas storage tank) leaves the factory, the wall strain data is obtained through an array of strain gauges 201 attached to the outer wall of the container. The outer wall surface temperature data is obtained through the outer wall temperature sensor 202 attached to the outer wall of the container. And constructing pressure and strain data by acquiring temporary pressure data through temporary pressure gauges. and external wall surface temperature data The relational model is optimized by acquiring multiple sets of data. In actual use, strain data is acquired. External wall surface temperature data Real-time pressure can be inferred from the model. Step 6, based on the calculation , , and A multi-objective optimization control algorithm is adopted (e.g., taking approximating the isothermal process, maximizing efficiency, and minimizing the fatigue damage increment at the current inflation rate as comprehensive objectives) to calculate the current optimal valve opening command in real time.

[0050] Step 7: According to the instruction, adjust the electric regulating valve on the inlet pipeline of the pressure tank to achieve refined closed-loop control of the gas flow rate of a single pressure tank or a cluster of pressure tanks.

[0051] Furthermore, regarding the construction of soft measurement models (see reference...) Figure 4 During the calibration stage before the gas storage tank leaves the factory, a temporary flow meter 208 is temporarily installed. The temporary flow meter 208 adopts a high-precision Coriolis mass flow meter. In the environmental test chamber, two extreme environmental temperatures, winter and summer, are simulated. Multiple sets of gas filling experiments are carried out at low, medium and high constant flow rates under different initial pressures of the storage tank. During the experiment, PLC204 synchronously records all sensor data and real flow data. After obtaining a large amount of data, offline model training is carried out. First, a physical model is established according to formula (1), and the effective average convective heat transfer coefficient is identified using data. Then, with , , Its rate of change is used as an input feature, with actual flow rate. A three-layer feedforward neural network is trained using the output labels. Finally, the output of the soft measurement model... Determined by the following formula: ; in, The output of the calibrated physical model, The output of the trained machine learning model is represented by α, which is the fusion weight (e.g., 0.3) determined based on the model's performance on the validation set. This fusion model combines physical interpretability with strong nonlinear fitting capabilities.

[0052] Example 3 refer to Figure 6 A pressure tank inflation control device, comprising: Sensing units: including eight units arranged in a specific array (e.g., arranged in a 45-degree horizontal and vertical ring in the middle of the cylinder, for reference). Figure 7 The strain gauge 201 is attached to the outer wall of the pressure tank for comprehensive monitoring of the wall strain distribution; it includes multiple outer wall temperature sensors 202 (e.g., 4) arranged axially evenly; and an inlet temperature sensor 203 installed before the regulating valve 206.

[0053] Data processing and model calculation unit: Typically integrated into the local PLC204, responsible for acquiring sensor data, running soft sensing models, and outputting estimated flow rates. and inversion pressure .

[0054] Control Unit: Integrated into the DCS205 of the compressed air energy storage system, it receives local data, executes optimized control algorithms, and generates valve control commands.

[0055] Execution unit: an electrically operated regulating valve installed on the inlet branch pipe of each pressure tank. The core inventive point of this invention lies in its creative proposal of an integrated solution of "non-intrusive sensing - soft measurement modeling - intelligent closed-loop control" for the inflation process of large-scale pressure tank clusters in scenarios such as compressed air energy storage.

[0056] like Figure 6 As shown, the system of the present invention is applied to a large gas storage tank. In the middle of the cylinder of the gas storage tank, according to... Figure 7 The strain gauges 201 are arranged as shown: eight strain gauges (circular array) are arranged perpendicularly to each other at 45-degree intervals along the circumference, forming a strain sensor array. Simultaneously, four external wall temperature sensors 202 are evenly arranged axially along the tank body. An inlet temperature sensor 203 is installed on the inlet pipe 207 before the regulating valve 206.

[0057] All sensor signals are connected to the local data acquisition module (integrated in PLC204). PLC204 has two pre-installed core algorithms: one is an algorithm for fusing strain data to invert the pressure inside the tank; the other is an algorithm that runs the aforementioned fusion soft sensor model to calculate the real-time mass flow rate based on strain (pressure), external wall temperature, and inlet temperature. PLC204 will then display the calculated real-time pressure... Real-time quality flow Average external wall temperature and inlet temperature The data is uploaded to the DCS205 in the central control center via industrial Ethernet.

[0058] In this embodiment, the control objective is set as follows: to make the inflation process as close to an isothermal process as possible while meeting the safety constraint of the maximum inflation pressure rise rate. The control algorithm is based on the real-time uploaded data. , , , and ambient temperature By combining the heat capacity parameters of the gas storage tank, the average temperature of the gas inside the tank is estimated in real time. (For example, through a simplified lumped-parameter energy equation). Then, with With the goal of minimizing the absolute value of the control signal, a PID or model predictive control algorithm is used to calculate the control signal (such as an opening command) to be applied to the regulating valve 206. This command is issued by the DCS205 to the PLC204, which ultimately drives the regulating valve 206 to operate, thus forming a complete closed-loop control circuit.

[0059] For a cluster consisting of hundreds of such gas storage tanks, each tank is equipped with this independent system. The DCS205 can aggregate the status of all tanks and, under the constraint of total air intake flow, prioritize the rapid inflation of gas storage tanks with lower pressure and temperatures closer to ambient temperature, or dynamically adjust the inflation strategy based on the cumulative fatigue damage of each tank, thereby achieving optimized operation at the cluster level.

[0060] The technical improvements of this invention are mainly reflected in the following three aspects: (1) Improved Sensing Level: From Invasive Single-Point to Non-Invasive Array. The traditional practice of installing contact pressure sensors by drilling holes in high-pressure vessels and installing permanent flow meters on each branch has been abandoned. A distributed strain gauge 201 array is adopted as the core sensing element. This array can not only obtain the internal pressure of the vessel with high precision and non-invasively through mechanical inversion, completely eliminating the risk of high-pressure leakage, but also simultaneously provide the stress distribution and strain history of the vessel body, providing direct data for structural health monitoring and fatigue life assessment. Combined with an external wall temperature sensor array, comprehensive and safe sensing of the gas filling thermodynamic process is achieved. The expensive high-precision permanent flow meters on each gas storage tank branch are eliminated, and "permanent soft measurement capabilities" are obtained through "one-time calibration," greatly reducing hardware costs and long-term maintenance expenses.

[0061] (2) Modeling Integration: From a single model to a fusion of thermodynamic mechanisms and data-driven approaches. To solve the problem of flow measurement without flowmeters, a modeling method of "pre-calibration and soft measurement replacement" was invented. Before the container leaves the factory, calibration tests are conducted under various operating conditions using temporary flowmeters to obtain massive amounts of data. Based on this, a composite soft measurement model integrating physical mechanism models and data-driven models is constructed. First, a physical model is established based on thermodynamics and heat transfer principles to ensure interpretability, and then the calibration data is used to calibrate its parameters; subsequently, a machine learning model (such as an ANN neural network) is trained to capture complex nonlinear relationships; finally, the two are intelligently integrated. This model only requires real-time strain (pressure) and temperature data to accurately calculate the mass flow rate online, realizing the permanent replacement of "hardware flowmeters" with "software algorithms," significantly reducing costs while ensuring accuracy. By acquiring mass flow rate and internal temperature field information in real time through soft measurement, which are traditionally difficult to measure online, it provides the possibility of achieving near-isothermal inflation and optimal closed-loop control, thereby improving the overall efficiency of the energy storage system.

[0062] (3) Intelligent Control: From Open-Loop Coarse-Fit to Multi-Objective Closed-Loop Optimization. Based on the rich and high-precision real-time information (pressure, estimated flow rate, temperature field) provided by the above-mentioned non-intrusive sensing and soft measurement, the inflation control is upgraded from the traditional single pressure threshold switch control to real-time closed-loop feedback control based on multi-objective optimization. The control algorithm can dynamically optimize and adjust the opening of the inlet valve of each container with the goals of approximating isothermal inflation (improving efficiency), balancing pressure distribution, and minimizing fatigue damage. This transforms the inflation process from a passive and inefficient safety operation to an active and efficient energy management process, while laying the decision-making foundation for cluster-level collaborative scheduling and life management. Rich local status information (pressure, flow rate, temperature) is provided for each gas storage tank. After this information is uploaded to the central control center, it can provide data support for cluster-level collaborative optimization scheduling. Calibration can be completed before the container leaves the factory, simplifying on-site debugging and reducing engineering complexity. Once a gas storage tank model is determined and the model is calibrated, it can be directly deployed in other projects.

[0063] Example 4 refer to Figure 5 A pressure tank inflation control device, comprising, The sensing device acquires N sets of calibration data during the inflation process of the pressure storage tank; A computing device is used to construct a physical model of the dynamic inflation process; based on calibration data and the physical model, unknown parameters of the physical model are obtained; the unknown parameters are input into the physical model to obtain a calibrated physical model; the calibrated physical model is introduced into a machine learning model, and a trained machine learning model is obtained by training with calibration data; a soft measurement model is obtained through the calibrated physical model and the trained machine learning model. The control device acquires real-time data and obtains real-time mass flow rate through real-time data and a soft measurement model. The actuator obtains the current valve opening command based on real-time data and real-time mass flow rate, and controls the pressure tank valve according to the valve opening command.

[0064] Furthermore, the sensing device includes strain gauges 201, an outer wall temperature sensor 202, an inlet temperature sensor 203, and a temporary flow meter 208. Several circumferential arrays of strain gauges 201 and outer wall temperature sensors 202 are fixedly installed on the outer surface of the pressure tank, forming a distributed strain gauge array. A regulating valve 206, an inlet temperature sensor 203, and a temporary flow meter 208 are fixedly installed on the inlet pipe 207 of the pressure tank.

[0065] Example 5 refer to Figure 8 A pressure tank filling control device includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it performs the steps of the method as described in Embodiment 1 or 2. Specifically, the pressure tank filling control device can be a computing device such as a desktop computer, laptop, handheld computer, or cloud server. The pressure tank filling control device may include, but is not limited to, a processor and a memory.

[0066] Example 6 A computer-readable storage medium stores a computer program that, when executed by a processor, implements the steps of the method as described in Embodiment 1 or 2. Specifically, the computer program may be divided into one or more modules / units, one or more of which are stored in a memory and executed by a processor to perform the present invention. The one or more modules / units may be a series of computer program instruction segments capable of performing a specific function, these instruction segments describing the execution process of the computer program in a pressure tank inflation control device.

[0067] Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application.

Claims

1. A method for controlling the inflation of a pressure storage tank, characterized in that, Includes the following steps: Obtain N sets of calibration data during the inflation process of the pressure storage tank; Construct a physical model of the dynamic inflation process of a pressure storage tank; Based on the calibration data and the physical model, obtain the unknown parameters of the physical model; By substituting the unknown parameters into the physical model, a calibrated physical model can be obtained. The calibrated physical model is introduced into the machine learning model, and the trained machine learning model is obtained by training with calibration data. A soft measurement model is obtained by using a calibrated physical model and a trained machine learning model; Acquire real-time data, and obtain real-time quality flow rate through real-time data and soft measurement models; The current valve opening command is obtained based on real-time data and real-time mass flow rate, and the pressure tank valve is controlled according to the valve opening command.

2. The method for controlling the inflation of a pressure storage tank according to claim 1, characterized in that, The process of acquiring N sets of calibration data during the pressure tank filling process includes the following steps: In the environmental test chamber, under the simulated extreme environmental temperatures of winter and summer, multiple sets of inflation experiments were conducted at different initial pressures in the pressure tank and at three constant flow rates: low, medium, and high. The measurement data obtained for each group of experiments are used as calibration data.

3. A method for controlling the inflation of a pressure storage tank according to claim 1 or 2, characterized in that, The calibration data includes wall strain data, outer wall temperature data, inlet air temperature data, and mass flow rate data. The wall strain data is acquired through a strain gauge array.

4. The method for controlling the inflation of a pressure storage tank according to claim 1, characterized in that, The physical model is as follows: in, Specific heat capacity at constant volume For the mass of the gas inside the tank, The gas volume average temperature, For quality flow, Specific heat capacity at constant pressure Intake air temperature, The convective heat transfer coefficient is... For heat exchange area, This refers to the wall temperature.

5. The method for controlling the inflation of a pressure storage tank according to claim 1, characterized in that, The soft measurement model is as follows: in, The output of the calibrated physical model, The output of the trained machine learning model is α, which is the fusion weight determined based on the model's performance on the validation set.

6. The method for controlling the inflation of a pressure storage tank according to claim 1, characterized in that, The step of obtaining the current valve opening command based on real-time data and real-time mass flow rate, and controlling the pressure storage tank valve according to the valve opening command, includes the following steps: The real-time pressure inside the pressure tank is obtained based on the real-time strain data of the wall surface. Based on the real-time pressure, real-time mass flow rate, external wall temperature data, and inlet air temperature data inside the pressure tank, the optimal valve opening command is calculated in real time using a multi-objective optimization control algorithm. Control the pressure tank valves according to the valve opening command.

7. A pressure tank inflation control device, characterized in that, include, The sensing device acquires N sets of calibration data during the inflation process of the pressure storage tank; A computing device is used to construct a physical model of the dynamic inflation process; based on calibration data and the physical model, unknown parameters of the physical model are obtained; the unknown parameters are input into the physical model to obtain a calibrated physical model; the calibrated physical model is introduced into a machine learning model, and a trained machine learning model is obtained by training with calibration data; a soft measurement model is obtained through the calibrated physical model and the trained machine learning model. The control device acquires real-time data and obtains real-time mass flow rate through real-time data and a soft measurement model. The actuator obtains the current valve opening command based on real-time data and real-time mass flow rate, and controls the pressure tank valve according to the valve opening command.

8. The pressure tank inflation control device according to claim 7, characterized in that, The sensing device includes strain gauges (201), an outer wall temperature sensor (202), an inlet temperature sensor (203), and a temporary flow meter (208). Several circumferential arrays of strain gauges (201) and outer wall temperature sensors (202) are fixedly installed on the outer surface of the pressure tank. The several circumferential arrays of strain gauges (201) form a distributed strain gauge array. A regulating valve (206), an inlet temperature sensor (203), and a temporary flow meter (208) are fixedly installed on the inlet pipe (207) of the pressure tank.

9. A pressure tank inflation control device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the steps of the method as described in any one of claims 1-6.

10. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by a processor, it implements the steps of the method as described in any one of claims 1-6.

Images