Steam pipe network system optimization scheduling method and device based on deep learning and digital twin coupling

Abstract

The invention belongs to the related technical field of energy system optimization, and discloses a steam pipe network system optimization scheduling method and equipment based on deep learning and digital twin coupling. The method comprises the following steps: (1) inputting obtained historical information of each measuring point of an entity steam pipe network into a machine learning model to train a set of single-step and multi-step machine learning prediction model suitable for the steam pipe network system, wherein the machine learning model is a long-short-term memory network model; meanwhile, establishing an equation-based digital twinborn model according to the obtained steam pipe network topological graph, the satellite graph, the connection relation between the pipelines and the historical data of the measuring points; and (2) inputting a user side steam flow prediction value predicted by the machine learning model into the digital twinborn model to obtain a valve opening degree of each node in the steam pipe network and a boiler output parameter suggested adjustment value so as to realize optimal scheduling of the steam pipe network system. The heat efficiency of the system is improved, and energy is saved.

Description

Technical Field

[0001] The present invention belongs to the technical field related to the optimization of energy systems, and more specifically, relates to an optimized scheduling method and device for a steam pipe network system based on the coupling of deep learning and digital twins. Background Art

[0002] Steam is one of the essential elements in modern industrial production, and a large amount of steam is used in many scenarios involving heating, humidification, and providing energy power. The steam pipe network is a key link connecting steam production and steam consumption, and the thermal efficiency of the steam pipe network directly determines the relationship between the user's use of steam and the steam production of the steam supplier.

[0003] Steam pipe networks applied to modern industrial parks usually have the characteristics of large fluctuations in steam consumption, a large number of steam users, a large spatial distribution of the pipe network, a complex pipeline structure, and variable gas consumption demands among different steam users. In most of the currently maturely applied technologies, pipeline management systems fail to effectively integrate flow prediction and simulation models, resulting in low accuracy in predicting and monitoring the operating state of pipelines, which affects the scheduling and optimization of the system. Traditional pipeline management systems rely on manual or simple numerical simulation models. Although they can reflect the overall usage of the steam pipe network to a certain extent, they cannot dynamically adjust the system operating parameters in real time and are difficult to intuitively display the system state and regulation results. In addition, most of the existing pipeline system optimization methods are static simulations, lacking the ability to interact with real-time data and unable to provide immediate feedback during actual operation, resulting in poor timeliness of system optimization.

[0004] The current situation where accurate prediction and simulation of individual users and the overall flow system cannot be achieved may lead to the following problems: The demand for steam by steam users changes continuously over time. If steam production and the steam transmission pipeline network cannot respond to this demand change in a timely manner, it will result in excessive steam production, thus causing waste of steam production. Users' usage patterns and preferences for steam vary to a certain extent at different times, in different seasons, and under different weather conditions. If the steam system cannot predict the future demand of users, steam regulation can only be carried out according to empirical formulas or existing regulation modes, and cannot make early adjustments for special gas usage situations. Then, the demand for steam by some users in the steam pipe network cannot be fully met.

[0005] Based on the different gas consumption demand characteristics of steam users in different industrial parks, personalized regulation methods need to be implemented based on their characteristics. Most of the existing mature industrial pipeline system optimization methods cannot achieve a more convenient local adaptation design for pipe networks with different gas usage characteristics, which will result in the inability to guarantee the thermal efficiency and economic efficiency of the regulation system when facing different pipe network systems. Summary of the Invention

[0006] In view of the above defects or improvement requirements of the prior art, the present invention provides an optimized scheduling method and device for a steam pipe network system based on the coupling of deep learning and digital twin, aiming to solve the problem of low steam utilization rate in various industrial parks.

[0007] To achieve the above object, according to one aspect of the present invention, there is provided an optimized scheduling method for a steam pipe network system based on the coupling of deep learning and digital twin, the method comprising the following steps:

[0008] (1) Input the historical information of each measuring point of the entity steam pipe network obtained into a machine learning model to train a set of single-step and multi-step machine learning prediction models suitable for the steam pipe network system, and the machine learning model is a long short-term memory network model;

[0009] (2) Establish an equation-based digital twin model according to the obtained steam pipe network topology map, satellite map, connection relationship between pipelines, and historical data of each measuring point;

[0010] (3) Input the predicted value of the steam flow at the user end predicted by the machine learning model into the digital twin model to obtain the flow rate, temperature, and pressure values of each node in the steam pipe network, and design the adjustment mechanisms of the boiler fuel regulating valve, air intake volume, main gas valve, and each branch valve based on the flow rate, temperature, and pressure values calculated by the digital twin model for each node and the actual measured values at the next moment, so as to realize the optimized scheduling of the steam pipe network system.

[0011] Further, after obtaining the recommended adjustment values of the valve opening degrees and boiler output parameters of each node, the input quantity and the output of the recommended adjustment value are updated in real time with time, and at the same time, the parameters of the machine learning model and the digital twin model are updated.

[0012] Further, construct the control strategy and control system of the industrial park heating system, and then optimize the heat pipe network by the control system according to the obtained recommended adjustment values, and summarize the output in the form of a visualization interface, and refresh it in real time according to the operation of the system.

[0013] Further, the visualization module extracts the information of the sensor measuring points in the system, and displays the real-time flow rate, pressure distribution, and prediction results of the pipeline system through 3D graphics, and the user can adjust the flow rate and pressure of the pipeline in real time through the interaction interface.

[0014] Further, the system refresh time is adjusted according to the sampling frequency of different pipe network sensors and the steam load change characteristics.

[0015] Furthermore, the digital twin model includes a boiler end boundary condition module, a user end boundary condition module, a heat transfer module, a pressure loss module, a physical pipeline module, and a steam physical property module. The boiler end boundary condition module sets the working state of the boiler at the parameter setting end of the model; different steam user end requirements are set through the user end boundary condition module, and different steam pressures and instantaneous steam mass flows at the outlets of different user ends are defined to adapt to the changes during the actual use of the simulation system.

[0016] Furthermore, according to the heat transfer theory, the heat transfer module calculates the heat transfer by coupling two parts: the convective heat transfer between the steam inside the pipe wall and the inner wall of the pipe, and the heat transfer from the steam pipeline pipe wall to the insulation layer.

[0017] Furthermore, the pressure loss module calculates the pressure loss in the pipeline using the frictional loss and local resistance model; the physical pipeline module defines the geometric and physical properties of the pipeline; for the fluid inside the pipe, it is mainly controlled by the momentum and energy conservation equations:

[0018] Δp geo [i] + Δp fric [i] + (p[i - 1] - p[i + 1]) = 0

[0019]

[0020] where, Δp geo [i] is the geometric pressure loss of the i-th unit, Δp fric [i] is the frictional pressure loss of the i-th unit, p[i - 1] and p[i + 1] are the pressures upstream and downstream of the i-th unit respectively, is the change rate of the average specific enthalpy of the i-th unit with respect to time, t is time, h i and h i-1 are the specific enthalpies of the i-th unit and the (i - 1)-th unit respectively, and are the mass flow rates of the i-th unit and the (i - 1)-th unit respectively, is the heat flow of the i-th unit, V i is the volume of the i-th unit, is the change rate of the pressure of the i-th unit with respect to time, is the average specific enthalpy of the i-th unit, is the change rate of the average density of the i-th unit with respect to time, M i is the mass of the i-th unit.

[0021] The present invention also provides an optimized scheduling system for a steam pipe network system based on the coupling of deep learning and digital twin. The system includes a memory and a processor. The memory stores a computer program, and when the processor executes the computer program, it executes the optimized scheduling method for the steam pipe network system based on the coupling of deep learning and digital twin as described above.

[0022] The present invention also provides a computer-readable storage medium. The computer-readable storage medium stores machine-executable instructions. When the machine-executable instructions are called and executed by a processor, the machine-executable instructions cause the processor to implement the optimized scheduling method for the steam pipe network system based on the coupling of deep learning and digital twin as described above.

[0023] Generally speaking, compared with the prior art by the above technical solutions conceived by the present invention, the optimized scheduling method and device for the steam pipe network system based on the coupling of deep learning and digital twin provided by the present invention mainly have the following beneficial effects:

[0024] 1. The digital twin model established by the present invention can simulate the dynamic changes of the steam pipe network, predict the pressure, temperature and flow rate at each position of the pipeline, and calculate the heat loss, pressure loss and flow loss. It can realize real-time optimization and control of the thermal system based on the changes in user needs and the thermal properties of the pipe network, improve the thermal efficiency of the system and save energy.

[0025] 2. This digital twin model can not only support optimized scheduling and energy-saving operations, but also be used for daily working condition monitoring and abnormal warning, providing a scientific basis for the stable operation and energy consumption management of the pipe network system.

[0026] 3. By adopting the optimized scheduling method for the steam pipe network system based on the coupling of deep learning and digital twin proposed by the present invention, the thermal efficiency of the steam pipe network system can be improved by referring to the recommended adjustment values, and the abnormal state in the pipeline can be monitored in real time using the visualization interface, providing a scientific basis and a reliable solution for the stable operation and energy consumption management of the pipe network system. BRIEF DESCRIPTION OF THE DRAWINGS

[0027] Figure 1 is a flowchart of an optimized scheduling method for a steam pipe network system based on the coupling of deep learning and digital twin provided by the present invention;

[0028] Figure 2 is a schematic diagram of the physical structure modeling of a single pipeline of the steam pipe network in an embodiment of the present invention;

[0029] Figure 3 is a schematic diagram of the heat exchange mode of the pipe wall of the steam pipe network in an embodiment of the present invention;

[0030] Figure 4It is a flowchart for predicting the steam demand of a machine learning model prediction system in an embodiment of the present invention;

[0031] Figure 5 It is a schematic structural diagram of the establishment of a deep learning model provided by the present invention;

[0032] Figure 6 It is a flowchart for building a deep learning model system provided by the present invention. Detailed implementation manners

[0033] In order to make the objectives, technical solutions and advantages of the present invention clearer, the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention. In addition, the technical features involved in the various embodiments of the present invention described below can be combined with each other as long as they do not conflict with each other.

[0034] The present invention provides an optimized scheduling method for a steam pipe network system based on the coupling of deep learning and digital twin, which can improve the thermal efficiency of the steam pipe network system by referring to the recommended adjustment value, and use a visual interface to monitor the abnormal state in the pipeline in real time, providing a scientific basis and a reliable solution for the stable operation and energy consumption management of the pipe network system.

[0035] Please refer to Figures 1 to 5 , the method mainly includes the following steps:

[0036] S1. Obtain the historical information and current dynamic data of each measuring point of the physical steam pipe network, and obtain the topological map, measured satellite map and local weather data of the steam pipe network.

[0037] Collect data of each measuring point for more than one year, and use data processing methods to remove outliers from the data.

[0038] S2. Input the historical information of each obtained measuring point into the machine learning model to train a set of single-step and multi-step machine learning prediction models suitable for the steam pipe network system; specifically, for the gas consumption characteristics in the industrial park, use the long short-term memory network (LSTM) to train the dynamic data of each dimension. Among them, a bidirectional LSTM model is selected according to the gas consumption characteristics of the industrial park, and an early stopping mechanism is introduced during the training process.

[0039] The specific sub-steps are:

[0040] S201. Preprocess the dynamic data.

[0041] S202. Retain the variables in the dynamic data that have an impact on the prediction, and then convert the time information into a numerical "year-month-day column" as a time feature.

[0042] S203. Detect whether there are outliers in the dynamic data. If so, remove the outliers.

[0043] S204. Load the data processed in step S203 from the specified database file, and use the data loading function to divide the dynamic data into a training set and a test set.

[0044] S205. Normalize the dynamic data and perform normalization processing on the training set and the test set to improve the model training efficiency.

[0045] S206. Create a data window to convert the data into the sequence format required by the LSTM model, and create a sequence data window.

[0046] S207. Define the LSTM model structure.

[0047] Build the LSTM model structure using the Python language, and determine the number of model layers and neurons according to the data characteristics.

[0048] S208. Compile the model.

[0049] Compile the LSTM model and set the loss function and optimizer.

[0050] S209. Model training and evaluation.

[0051] S210. Train the model.

[0052] Use the training set data to train the LSTM model with the goal of the lowest loss value. Prevent overfitting through the early stopping mechanism and save the best model during the training process.

[0053] S211. Evaluate the model.

[0054] Use the test set data to evaluate the model and calculate evaluation metrics such as mean absolute error (MAE), root mean square error (RMSE), and the proportion of cumulative error.

[0055] S212. Visualize the prediction results.

[0056] Use visualization means to compare the prediction results with the actual data, generate comparison charts with different time window lengths, and the display range can be modified.

[0057] S213. Save the prediction results.

[0058] Save the prediction results and the actual measurement point data to the database file for later use.

[0059] S214. Save the evaluation report.

[0060] Output the evaluation metrics on the terminal after the model training is completed and save them in the computer storage medium.

[0061] S3. Establish an equation-based digital twin model according to the obtained steam pipe network topology map, satellite map, connection relationships between pipelines, and historical data of each measurement point;

[0062] For the consideration of system scalability, select Modelica language for modeling. The digital twin model is developed based on an open physical modeling computer language, including the following module settings:

[0063] S301. Set the boiler end boundary condition module

[0064] This boiler end boundary condition module sets the working state of the boiler at the parameter setting end of the model, including parameters such as steam temperature, steam pressure, instantaneous steam mass flow rate, boiler water supply, and boiler fuel supply, and uses these data as boundary conditions to drive the digital twin model simulation.

[0065] S302. Set the user end boundary condition module

[0066] Set the requirements of different steam user ends through this user end boundary condition module, and define different steam pressures and instantaneous steam mass flows at the outlets of different user ends to adapt to the changes in the actual use of the simulation system.

[0067] S303. Set the heat transfer module

[0068] According to the heat transfer theory, this heat transfer module couples and calculates the steam flow in the pipe network into two parts: the convective heat transfer between the steam inside the pipe wall and the inner wall of the pipe and the heat transfer from the pipe wall of the steam pipeline to the insulation layer. For the former, the steam flow state and various physical property data of the steam pipeline are mainly considered. When considering controlling the heat flux density, its heat transfer amount is controlled by the following formula.

[0069] When the laminar heat transfer mainly occurs inside the pipe:

[0070]

[0071] Among them, Nu m,q,1 = 4.364

[0072]

[0073] When the turbulent heat transfer mainly occurs inside the pipe:

[0074]

[0075] (Gnielinski method)

[0076]

[0077] (Dittus / Boelter method)

[0078] ξ = (1.8 log 10 Re - 1.5) -2

[0079] Q = αA(T w - T)

[0080] where Nu m,q is the Nusselt number in the laminar flow state, Nu m,q,1 is a constant term, Nu m,q,2 is the Nusselt number related to the Reynolds number and Prandtl number of the fluid, Nu m,q,3 is another Nusselt number related to the Reynolds number and Prandtl number of the fluid, ξ is the wall friction factor, α is the convective heat transfer coefficient and is calculated separately for the laminar or turbulent flow states of the steam flow, Re is the Reynolds number, Pr is the Prandtl number, d i is the pipe diameter or the characteristic of the fluid flow, l is the characteristic length in the flow direction, λ is the thermal conductivity of the fluid, x is a parameter related to the pipe material and structure, Q is the heat flux, A is the heat transfer area, T w is the wall temperature, T is the average temperature of the fluid, and at the same time, this digital twin model also considers the smooth transition of the convective heat transfer coefficient between the laminar and turbulent flows.

[0081] For the latter, the changes in the thermal conductivity brought about by the pipe wall material, roughness, insulation material and thickness are mainly considered, and its heat transfer amount is controlled by the following formula.

[0082]

[0083] T outer = T

[0084]

[0085] where cp is the unit heat capacity of the pipe wall, U is the internal energy stored in the pipe wall material, T is the temperature of the fluid, d is the inner diameter of the pipe, d o is the outer diameter of the pipe, d i is the inner diameter of the pipe, Δx is the length of the pipe, is the heat flux from the inner wall surface to the fluid, is the heat flux from the outer wall surface to the environment, and are the heat fluxes at adjacent positions, T outer is the outer wall surface temperature, T is the fluid temperature, T inner is the inner wall surface temperature, λ is the thermal conductivity of the fluid, CF λis the correction factor for heat conduction, which varies with the thickness and properties of the pipe wall insulation material. For actual pipelines, by calibrating the heat loss per unit area of the steam pipeline under standard conditions, the heat transfer heat flux density or wall temperature on the pipe wall can be defined in the digital twin model to form boundary conditions, so that the heat transfer heat loss of the steam in the pipeline can be simulated and calculated in the digital twin model based on the above control equations using an open physical modeling computer language.

[0086] Among them, the heat transfer of the pipeline to the outside is set using the heat flux density to improve the accuracy of the heat dissipation simulation.

[0087] S304. Set the pressure loss module

[0088] The pressure loss in the pipeline is calculated using the friction loss along the way and the local resistance model. The former considers the Reynolds number, steam flow rate, and pipeline geometric characteristics, and the latter uses the calibrated local pressure loss to define the pressure loss borne by the steam passing through valves and bends in the digital twin model. For single-phase flow, its control equations are as follows.

[0089]

[0090]

[0091] For two-phase flow, its control equations are as follows:

[0092]

[0093] Among them, Δp is the pressure loss per unit length of the pipeline, Δp is the friction factor, L is the pipeline length, d is the pipeline diameter, m is the mass flow rate, ρ is the fluid density, A is the pipeline cross-sectional area, Re is the Reynolds number, K is the absolute roughness of the pipeline, β is the correction coefficient of the friction factor for two-phase flow, Δx FM is a small section after the pipeline is discretized, η liq and η vap are the dynamic viscosities of the liquid and steam respectively, ρ liq and ρ vap are the densities of the liquid and steam respectively, is the mass fraction, is a constant related to the pipeline and fluid characteristics.

[0094] S305. Set the physical pipeline module

[0095] This physical pipeline module defines the geometric and physical properties of the pipeline, constructs pipe segments with different diameters according to the actual pipe network, and determines the elevation and geometric shape of the pipeline. Secondly, the operation accuracy of the digital twin model is determined by discretizing the modeling pipe ends. For the fluid in the pipe, it is mainly controlled by the momentum and energy conservation equations:

[0096] Δp geo[i] + Δp fric [i] + (p[i - 1] - p[i + 1]) = 0

[0097]

[0098] where Δp geo [i] is the geometric pressure loss of the i-th unit, and Δp fric [i] is the frictional pressure loss of the i-th unit, p[i - 1] and p[i + 1] are the pressures upstream and downstream of the i-th unit respectively, is the rate of change of the average specific enthalpy of the i-th unit with respect to time, t is time, h i and h i-1 are the specific enthalpies of the i-th unit and the (i - 1)-th unit respectively, and are the mass flow rates of the i-th unit and the (i - 1)-th unit respectively, is the heat flow of the i-th unit, V i is the volume of the i-th unit, is the rate of change of the pressure of the i-th unit with respect to time, is the average specific enthalpy of the i-th unit, is the rate of change of the average density of the i-th unit with respect to time, M i is the mass of the i-th unit.

[0099] The physical pipeline module is mainly based on the actual pipe network structure and is built according to the pipe network design drawing. Therefore, the optimization method of the present invention has wide applicability to various steam pipe networks and extremely high pertinence to a certain pipe network after specific implementation.

[0100] S306. Set up the steam physical property module

[0101] Calculate the steam physical properties in the pipe network in the digital twin model according to the operating conditions of the pipe network, including steam specific enthalpy, steam temperature, steam pressure, steam flow velocity and steam flow rate.

[0102] S4. Input the results predicted by the machine learning model as in S213 into the digital twin model to obtain the recommended adjustment values of the valve openings and boiler output parameters at each node in the pipe network, and update the input and output of the recommended adjustment values in real time with time. At the same time, update the parameters of the machine learning model and the digital twin model. The specific sub-steps are as follows:

[0103] S401. Transfer the prediction results.

[0104] Take the predicted value of the steam flow rate at the user end obtained in S213 as the boundary condition and input it into the digital twin model established in S3.

[0105] S402. Run the digital twin simulation optimization.

[0106] Run the digital twin model and obtain the optimized recommendation adjustment values for the current working conditions.

[0107] S403. Save the results.

[0108] Save the results of the simulation optimization in a temporary digital storage medium. The new files generated after the next simulation will overwrite the results of this simulation. The results will not enter the real measurement point database and will only be temporarily saved in the temporary storage medium.

[0109] S5. Construct the control strategy and control system for the heating system of this industrial park. The specific sub-steps are as follows:

[0110] S501. Determine the process structure of the thermal system in this industrial park;

[0111] S502. Obtain the main parameters of the thermal system from historical data and on-site detection;

[0112] S503. Determine the control objectives of the control system based on the gas consumption in the park and the characteristics of the gas supply end;

[0113] S504. Design the control loop of this thermal system; adopt a pid controller to achieve the control of each index, and adopt a closed-loop control system structure.

[0114] Based on the system control objectives determined in S503, refer to the general system control strategy and the change relationship between various parameters in the thermal system to build a control loop suitable for this system. The system refresh time can be appropriately adjusted according to the sampling frequency of different pipeline sensors and the change characteristics of the steam load.

[0115] S505. Tune the controller parameters

[0116] Use common parameter tuning methods in engineering to tune the controller parameters, and verify the dynamic characteristics of the control loop after parameter tuning through experiments.

[0117] S6. Optimize the system's thermal pipeline network according to the recommended values obtained in S4 with the control loop built in S5, and summarize the output in the form of a visual interface, and refresh it in real time according to the operation of the system; specifically include:

[0118] S601. Set up the visual display module

[0119] The visualization module extracts the information of the sensor measurement points in the system, and displays the real-time flow rate, pressure distribution and prediction results of the pipeline system through 3D graphics. Users can adjust the flow rate and pressure of the pipeline in real time through the interactive interface to achieve the best optimization effect.

[0120] Among them, the considered visualization module can be implemented through the Vue.js framework and other frameworks that can achieve similar functions, enabling users to control the simulation process through input boxes and buttons and view real-time data updates.

[0121] S602. Provide an interactive optimization function

[0122] Users can input flow data through the interface or obtain measurement point data from sensors in real time through the data acquisition function provided by the system. The system adjusts the system optimization parameters according to the data and provides real-time feedback on the optimization results through visual charts.

[0123] The present invention also provides an optimized scheduling system for a steam pipe network system based on the coupling of deep learning and digital twins. The system includes a memory and a processor. The memory stores a computer program, and when the processor executes the computer program, it executes the optimized scheduling method for the steam pipe network system based on the coupling of deep learning and digital twins as described above.

[0124] The present invention also provides a computer-readable storage medium. The computer-readable storage medium stores machine-executable instructions. When the machine-executable instructions are called and executed by the processor, the machine-executable instructions cause the processor to implement the optimized scheduling method for the steam pipe network system based on the coupling of deep learning and digital twins as described above.

[0125] It is easy for those skilled in the art to understand that the above are only preferred embodiments of the present invention and are not intended to limit the present invention. Any modifications, equivalent replacements, and improvements made within the spirit and principles of the present invention shall be included within the protection scope of the present invention.

Claims

1. A steam pipe network system optimization scheduling method based on deep learning and digital twin coupling, characterized in that: The method comprises the following steps: (1) inputting the historical information of each measuring point of the physical steam pipe network into a machine learning model to train a set of single-step and multi-step machine learning prediction models suitable for the steam pipe network system, wherein the machine learning model is a long short-term memory network model; (2) Establish an equation-based digital twin model based on the obtained steam pipe network topology map, satellite map, connection relationship between pipelines, and historical data of each measuring point; (3) The user-side steam flow prediction value predicted by the machine learning model is input into the digital twin model to obtain the flow, temperature and pressure values ​​of each node in the steam network. Based on the flow, temperature and pressure values ​​calculated by the digital twin model at each node and the actual measured values ​​at the next moment, the adjustment mechanism of the boiler fuel regulating valve, air intake, main air valve and each branch valve is designed to achieve optimal scheduling of the steam network system.

2. The steam pipe network system optimization scheduling method based on deep learning and digital twin coupling according to claim 1, characterized in that: After obtaining the valve opening of each node and the recommended adjustment value of the boiler output parameter, the input quantity and the output of the recommended adjustment value are updated in real time over time, and the parameters of the machine learning model and the digital twin model are updated at the same time.

3. The steam pipe network system optimization scheduling method based on deep learning and digital twin coupling according to claim 2, characterized in that: The control strategy and control system of the industrial park heating system are constructed, and the thermal network is then optimized by the control system according to the obtained recommended adjustment values, and the output is summarized in the form of a visual interface and refreshed in real time according to the operation of the system.

4. The method for optimizing and scheduling a steam pipe network system based on deep learning and digital twin coupling according to claim 3, characterized in that: The visualization module extracts the sensor measurement point information in the system and displays the real-time flow, pressure distribution and prediction results of the pipeline system through 3D graphics. Users can adjust the flow and pressure of the pipeline in real time through the interactive interface.

5. The method for optimizing and scheduling a steam pipe network system based on deep learning and digital twin coupling according to claim 3, characterized in that: The system refresh time is adjusted according to the sampling frequency of different pipe network sensors and the changing characteristics of steam load.

6. The method for optimizing and scheduling a steam pipe network system based on deep learning and digital twin coupling according to claim 1, characterized in that: The digital twin model includes a boiler-end boundary condition module, a user-end boundary condition module, a heat transfer module, a pressure loss module, a physical pipeline module and a steam property module. The boiler-end boundary strip module sets the working state of the boiler at the parameter setting end of the model; the needs of different steam user ends are set through the user-end boundary condition module, and different steam pressures and instantaneous steam mass flow rates at different user-end outlets are defined to adapt to changes in the simulation system during actual use.

7. The method for optimizing and scheduling a steam pipe network system based on deep learning and digital twin coupling according to claim 6, characterized in that: According to the heat transfer theory, this heat transfer module divides the flow of steam in the pipe network into two parts: the convective heat transfer between the steam and the inner wall of the pipe, and the heat transfer between the steam pipe wall and the insulation layer to the outside.

8. The method for optimizing and scheduling a steam pipe network system based on deep learning and digital twin coupling according to claim 6, characterized in that: The pressure loss module calculates the pressure loss in the pipeline using the along-the-line loss and local resistance models. The physical pipeline module defines the geometric and physical properties of the pipeline. For the fluid in the pipe, it is mainly controlled by the momentum and energy conservation equations: Δp geo [i]+Δp fric [i]+(p[i-1]-p[i+1])=0 Among them, Δp geo [i] is the geometric pressure loss of the i-th unit, Δp fric [i] is the friction pressure loss of the i-th unit, p[i-1] and p[i+1] are the pressures upstream and downstream of the i-th unit, respectively. is the rate of change of the average specific enthalpy of the ith unit with time, t is the time, h i and h i-1 are the specific enthalpies of the ith unit and the i-1th unit, respectively, and are the mass flow rates of the i-th unit and the i-1-th unit, is the heat flow of the ith unit, V i is the volume of the ith unit, is the rate of change of the pressure of the ith unit over time, is the average specific enthalpy of the ith unit, is the rate of change of the average density of the i-th unit over time, M i is the mass of the ith unit.

9. A steam pipe network system optimization scheduling system based on deep learning and digital twin coupling, characterized by: The system includes a memory and a processor, the memory stores a computer program, and the processor executes the steam pipe network system optimization scheduling method based on deep learning and digital twin coupling as described in any one of claims 1 to 8 when executing the computer program.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores machine-executable instructions. When the machine-executable instructions are called and executed by the processor, the machine-executable instructions prompt the processor to implement the steam pipe network system optimization scheduling method based on deep learning and digital twin coupling as described in any one of claims 1 to 8.

Images