A gas pressure regulating self-adjusting intelligent monitoring system based on digital twinning

Abstract

The present application relates to the technical field of industrial process intelligent monitoring, and provides a gas pressure regulating self-adjusting intelligent monitoring system based on digital twinning, which comprises a data acquisition unit, a digital twinning unit and an execution and feedback unit. The data acquisition unit acquires the operation data of the gas pressure regulating equipment in real time, and transmits the preprocessed data to the digital twinning unit. The digital twinning unit constructs a physical mechanism submodel based on the geometric topology and heat flow coupling characteristics of the gas pressure regulating equipment, and loads a data-driven submodel to form a hybrid digital twinning model; in combination with real-time data mapping and disturbance-safety boundary condition set, the predictive iterative simulation is performed to generate multiple sets of pressure regulating schemes and select the optimal scheme. The execution and feedback unit converts the scheme into standardized control instructions to drive the pressure regulating equipment and returns the operation data to realize closed-loop optimization, and realizes real-time monitoring, high-precision prediction and self-adaptive regulation and control of the gas pressure regulating process under complex working conditions, and improves the safety, stability and energy utilization efficiency.

Description

Technical Field

[0001] This invention relates to the field of intelligent monitoring technology for industrial processes, specifically to an intelligent monitoring system for gas pressure regulation and self-adjustment based on digital twins. Background Technology

[0002] Gas pressure regulating systems are a key component of urban gas transmission and distribution networks and industrial centralized gas supply systems. Their primary task is to maintain a stable gas pressure supplied downstream within a set range, regardless of fluctuations in source gas pressure and user load. With the expansion of urbanization and industrial gas consumption, the structure of gas transmission and distribution networks is becoming increasingly complex, and gas demand exhibits significant dynamic fluctuations. During peak hours, in the event of sudden load changes, or abnormal upstream gas source pressure, pressure regulating equipment must respond rapidly; otherwise, it may lead to pipeline pressure imbalances, user-end equipment shutdowns, or even safety accidents.

[0003] Existing gas pressure regulation controls largely rely on manual operation, remote commands, or preset control logic based on fixed parameters. While some systems possess a degree of automatic adjustment capability, feedback regulation exhibits significant lag when facing rapidly changing operating conditions. Furthermore, they often operate through single commands or periodic adjustments, making it difficult to establish continuous closed-loop adaptive control. Especially under conditions of multiple input sources, complex pipeline networks, and variable loads, existing methods rely on historical experience or static models for setting pressure regulation parameters, failing to balance response speed, pressure stability, and operational safety margins. This not only limits the dynamic regulation capabilities of the pressure regulation system but also increases operational energy consumption and maintenance risks.

[0004] To address this, a gas pressure regulation self-regulating intelligent monitoring system based on digital twins is proposed. Summary of the Invention

[0005] The purpose of this invention is to provide a gas pressure regulation self-adjustment intelligent monitoring system based on digital twins, including a data acquisition unit, a digital twin unit, and an execution and feedback unit. The data acquisition unit acquires real-time operating data of the gas pressure regulating equipment, preprocesses it, and transmits it to the digital twin unit. The digital twin unit constructs a physical mechanism sub-model based on the geometric topology and thermal-fluid coupling characteristics of the gas pressure regulating equipment, and loads a data-driven sub-model to form a hybrid digital twin model. It then performs predictive iterative simulation by combining real-time data mapping and a disturbance-safety boundary condition set to generate multiple pressure regulation schemes and select the optimal scheme. The execution and feedback unit transforms the scheme into standardized control commands to drive the pressure regulating equipment and transmits back operating data to achieve closed-loop optimization. This enables real-time monitoring, high-precision prediction, and adaptive control of the gas pressure regulation process under complex operating conditions, improving safety, stability, and energy utilization efficiency.

[0006] To achieve the above objectives, the present invention provides the following technical solution:

[0007] A gas pressure regulating and self-regulating intelligent monitoring system based on digital twins includes:

[0008] The data acquisition unit acquires gas operation data, preprocesses it, and transmits it to the digital twin unit;

[0009] The digital twin unit combines the physical topology and thermal-fluid coupling operation characteristics of the gas pressure regulating equipment to construct a physical mechanism sub-model, and integrates the data-driven sub-model trained based on the gas operation data to generate a hybrid digital twin model;

[0010] The gas operation data is mapped to virtual device nodes; a gas disturbance-safety boundary condition set is introduced into the hybrid digital twin model, disturbance factors are extracted from it, and matched with the current state of the virtual device nodes to perform disturbance simulation of the pressure regulation condition, generate gas pressure regulation response curves, and calculate multiple gas pressure regulation schemes in combination with valve control parameters.

[0011] The operation index score results are obtained by weighted summation of the multiple gas pressure regulation schemes; candidate gas pressure regulation schemes are screened according to safety threshold and stability requirements based on the weighted operation index score results, and the scheme with the best comprehensive performance is selected through multi-objective optimization, generating an encrypted standardized command data stream.

[0012] Preferably, the data acquisition unit acquires multi-dimensional operating parameters of the gas operation process and performs noise filtering, time series alignment and deviation correction on the multi-source data to generate standardized data that meets the calculation requirements of the physical mechanism sub-model.

[0013] The data acquisition unit includes a data packaging and transmission module, which encapsulates the standardized data into data frames with timestamps and device identifiers, and transmits them to the digital twin unit via wired and / or wireless industrial communication links.

[0014] Preferably, the process of constructing the physical mechanism sub-model by the digital twin unit includes: performing position analysis and size recognition on the physical components of the gas pressure regulating device, digitally encoding the analysis results according to the actual spatial coordinates and connection order, and establishing three-dimensional geometry and connection topology in the virtual environment;

[0015] Based on the aforementioned heat-fluid coupling operation characteristics modeling, the gas flow path is divided into computational units, each unit is assigned gas physical property parameters and pressure regulation boundary conditions, and the fluid flow equation, heat transfer equation and pressure regulation dynamic control equation are solved by multi-physics coupling to generate a physical mechanism sub-model that reflects the actual operation law and can be called by the hybrid digital twin model.

[0016] Preferably, the process of generating the hybrid digital twin model by the digital twin unit includes: loading the data-driven sub-model trained with gas operation data into the computing environment of the hybrid digital twin model, and establishing a data interaction channel with the constructed physical mechanism sub-model according to a preset interface protocol; during the operation phase, preprocessed data from the data acquisition unit is synchronously input into the physical mechanism sub-model and the data-driven sub-model, and physical calculation and data-driven prediction are executed simultaneously through multi-threaded calculation and / or multi-core distributed computing; wherein, the data-driven sub-model is responsible for real-time prediction, and the physical mechanism sub-model is used for periodic calibration of the data-driven model;

[0017] After completing the dual-model calculation, the two sets of prediction results are fused. The fusion includes assigning weights based on the accuracy evaluation results, performing weighted summation on different prediction values, and performing consistency correction on the output sequence to generate a unified prediction result of the hybrid digital twin model. The prediction result is cached after generation as a core input data stream that can be called for perturbation injection and simulation calculation.

[0018] Preferably, the process of the digital twin unit performing real-time data mapping includes: calling the synchronization mechanism to add a timestamp to the preprocessed data transmitted by the data acquisition unit, and comparing and correcting the deviation between the timestamp and the clock reference of the virtual device node; after completing time synchronization, parsing the real-time data into node status parameters according to the device identifier and data type, and injecting the parsing result into the corresponding virtual device node object;

[0019] During the injection process, the operating parameters associated with the virtual device nodes, including pressure, flow rate, and temperature, are updated according to the connection relationship of the nodes in the virtual topology, so that the device state in the virtual environment remains dynamically consistent with the physical device state. After the node state update is completed, the update result is written into the digital twin runtime cache as the input data stream for introducing disturbance conditions and predictive iterative simulation calculations.

[0020] Preferably, the process of the digital twin unit performing disturbance injection and simulation calculation includes: reading disturbance factors from the gas disturbance-safe boundary condition set, parsing the pressure fluctuation range, flow change amplitude and temperature anomaly threshold involved in the disturbance factors into input parameters that conform to the simulation calculation format, and matching and binding them with the current virtual device node's operating status data;

[0021] After binding is completed, a multi-condition predictive iterative simulation calculation process is started to generate a corresponding gas pressure regulation response curve for each combination of disturbance factor and operating state. The response curve is then correlated with valve control parameters to obtain multiple gas pressure regulation schemes with valve openings. The gas pressure regulation schemes are written into the scheme cache area after generation as input data streams for performance evaluation and optimal scheme selection.

[0022] Preferably, when performing performance evaluation and scheme selection, the digital twin unit receives a set of valve opening gas pressure regulation schemes introduced by the disturbance conditions and output by the predictive iterative simulation stage. For each scheme, it calls the performance evaluation module to read the corresponding three operating indicators: response time, pressure fluctuation amplitude, and safety margin. Then, it performs a weighted calculation on the operating indicators according to the preset weight parameters to generate the corresponding comprehensive performance score.

[0023] After completing the comprehensive scoring calculation of all schemes, the scoring results are sorted by numerical value, and a set of candidate schemes is selected based on the operational safety threshold and control stability requirements. In the set of candidate schemes, a multi-objective control parameter optimization algorithm is further invoked to evaluate the three operational indicators of response time, pressure stability and safety margin based on simulation results. Within the set threshold range, the candidate scheme with the best comprehensive performance is selected, and the finally retained optimal gas pressure regulation scheme is converted into a standardized command data stream.

[0024] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0025] 1. This invention combines the physical topology of a gas pressure regulating device with a thermal-fluid coupling operation characteristic model to construct a physical mechanism sub-model, which works in conjunction with a data-driven sub-model trained based on gas operation data to form a hybrid digital twin model. This hybrid modeling approach ensures the accuracy of physical constraints while possessing adaptive update capabilities, enabling more accurate prediction of gas system operation trends and output of regulation reference data.

[0026] 2. This invention introduces a gas disturbance-safety boundary condition set in a virtual environment. Through multi-condition predictive iterative simulation, it verifies the equipment response under different operating scenarios in advance. It can generate feasible gas pressure regulation schemes in advance when sudden disturbances or extreme operating conditions occur, significantly reducing the risk of equipment instability or overpressure and improving the safety margin of system operation.

[0027] 3. This invention uses a weighted comprehensive evaluation method to comprehensively score multiple gas pressure regulation schemes in terms of steady-state performance, dynamic response time, pressure fluctuation amplitude, and energy consumption, and selects the scheme with the best response delay, fluctuation amplitude, and safety margin, so as to achieve a balance between regulation speed, system stability, and energy utilization efficiency, and reduce the operating cost of gas transmission and distribution. Attached Figure Description

[0028] Figure 1 This is a schematic diagram of the architecture of the gas pressure regulating and self-regulating intelligent monitoring system based on digital twin, as an example of the present invention.

[0029] Figure 2 This is a flowchart illustrating the hybrid digital twin model of an example of the present invention.

[0030] Figure 3 This is a schematic diagram of the gas pressure regulating device and digital twin mapping of an example of the present invention. Detailed Implementation

[0031] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0032] Please see Figures 1 to 3 This invention provides a secure communication system in a traffic control network, the technical solution of which is as follows:

[0033] Example 1

[0034] Reference Figure 1 This embodiment provides a gas pressure regulation self-adjusting intelligent monitoring system based on digital twins, including a data acquisition unit, a digital twin unit, and an execution and feedback unit. The data acquisition unit acquires real-time operating data of the gas pressure regulating equipment, preprocesses it, and transmits it to the digital twin unit. The digital twin unit constructs a physical mechanism sub-model based on the geometric topology and thermal-fluid coupling characteristics of the gas pressure regulating equipment, and loads a data-driven sub-model to form a hybrid digital twin model. It then performs predictive iterative simulation by combining real-time data mapping and a disturbance-safety boundary condition set to generate multiple pressure regulation schemes and select the optimal scheme. The execution and feedback unit transforms the scheme into standardized control commands to drive the pressure regulating equipment and transmits back operating data to achieve closed-loop optimization. This enables real-time monitoring, high-precision prediction, and adaptive control of the gas pressure regulation process under complex operating conditions, improving safety, stability, and energy utilization efficiency.

[0035] Furthermore, the data acquisition unit acquires multi-dimensional operating parameters during the gas operation process, including core parameters characterizing the dynamic features of gas transmission and distribution, such as pressure, temperature, flow rate, valve opening degree, and equipment vibration status. The data acquisition unit achieves multi-source parallel acquisition by configuring multiple types of sensors and generates timestamps based on a unified sampling clock to ensure the comparability of parameters in the time domain.

[0036] The preprocessing operation for the collected multi-source data includes: employing a noise suppression strategy, which includes low-pass filtering of high-frequency equipment vibration noise, with a cutoff frequency preferably 1 / 10 of the sensor sampling frequency, and adaptive Kalman filtering of random measurement noise. The filtering can dynamically adjust the measurement noise covariance according to the measurement residual; aligning the time series of data from different sampling nodes by triggering a synchronization mechanism such as a unified clock based on GPS or NTP. When hardware synchronization is not feasible, timestamp interpolation algorithms such as linear interpolation can be used to supplement a small amount of missing data to avoid data phase deviation caused by asynchronous sampling; and correcting outliers for data points that exceed the equipment range or deviate significantly from historical trends. The correction strategy is based on a combination of sliding window mean and trend prediction model to ensure data continuity and physical rationality.

[0037] The data acquisition unit is equipped with a data packaging and transmission module. Standardized data is encapsulated into data frames with timestamps and unique device identifiers, and a CRC check field is added to the data frames to ensure data integrity during transmission. The data frames are transmitted to the digital twin unit via a wired industrial Ethernet and / or wireless industrial communication link. The link layer employs a retransmission acknowledgment mechanism and a sequential buffer queue to ensure the sequential and real-time arrival of data at the digital twin unit, meeting the requirements of the hybrid digital twin model for input data accuracy and synchronization.

[0038] Through the aforementioned preprocessing and encapsulation mechanisms, the data acquisition unit can effectively eliminate noise and time drift in multi-source sensor signals in complex gas transmission and distribution environments, improving the spatiotemporal consistency of cross-node data. Simultaneously, by encapsulating data frames with timestamps and device identifiers, the unit achieves traceability and anomaly localization capabilities for operational data, providing the digital twin unit with a continuous, accurate, and synchronous input data stream, significantly improving the stability and reliability of prediction and control calculations.

[0039] Further, see attached document. Figure 3The digital twin unit's process for constructing the physical mechanism sub-model includes: firstly, analyzing the position and identifying the dimensions of each physical component of the gas pressure regulating equipment. This analysis process employs a multi-sensor data fusion algorithm. Firstly, an iterative nearest-point algorithm is used to register the local geometric point cloud obtained from structured light measurement with the global point cloud obtained from laser scanning, forming high-density three-dimensional point cloud data. Subsequently, point cloud segmentation and feature extraction algorithms are used to identify the boundaries of components such as valve bodies and pipelines. Combined with industrial-grade image recognition algorithms, surface features of the components are obtained from two-dimensional images. Finally, all data is mapped to a unified coordinate system to generate a three-dimensional geometric model. The dimension identification step precisely measures parameters such as the length, width, height, wall thickness, and key mounting hole spacing of the components during the identification process, and generates corresponding physical property labels based on the material properties of the components. The identification results are then digitally encoded according to the actual spatial coordinates and the physical connection order between components. This encoding includes information such as node identifiers, connection directions, and interface types to ensure the accuracy of the virtual topology construction and physical property matching.

[0040] In the virtual environment, a three-dimensional geometric model corresponding to the physical object is established based on the digital encoding results, and a connection topology reflecting the actual structure of the gas pressure regulating equipment is constructed. The connection topology construction process supports hierarchical decomposition and reorganization of complex connection structures, enabling component visualization, structural editing, and traceable connection relationships within the virtual space. After completing the virtualization of the physical topology, the thermal-fluid coupling operation characteristic modeling process is invoked, dividing the gas flow path into several computational units based on the geometric structure and process function. Initial physical property parameters are assigned to each computational unit, and in the iterative calculation of the thermal-fluid coupling solution, the gas density, viscosity, and other physical property parameters are dynamically updated based on the real-time temperature and pressure fields. Boundary conditions such as inlet pressure, outlet pressure, flow limit, and heat source intensity are also set.

[0041] The fluid flow equation, heat transfer equation, and pressure regulation dynamic control equation are integrated and calculated in a unified manner through a heat-fluid coupling solution step. This process uses the finite volume method to discretize and iteratively solve the physical fields of each computational unit, and uses a specific coupled solver, such as the PISO algorithm, to alternately iterate the flow field and temperature field until convergence conditions are met. The pressure regulation dynamic control equation serves as the boundary condition; within each time step, the opening of the virtual valve is dynamically adjusted based on the real-time parameters of the virtual node, simulating the dynamic response behavior of the pressure regulating valve in actual operation. The physical mechanism sub-model generated through the above steps not only fully reflects the spatial structural characteristics and operational physical properties of the gas pressure regulating equipment, but also supports multi-condition switching, boundary disturbance analysis, and operational performance prediction in the simulation of the digital twin unit, providing an interpretable basic model support for the intelligent control of gas transmission and distribution systems.

[0042] Through the above construction process, the digital twin unit can accurately reproduce the spatial structure and physical connection relationships of the gas pressure regulating equipment. Based on this, a high-fidelity physical mechanism model is established using real physical property parameters. This model can not only reflect the flow field distribution and heat transfer characteristics during the gas pressure regulating process, but also dynamically respond to changes in control strategies. This provides a solid physical foundation for predictive calculations and operational optimization, effectively improving the system's reliability and control accuracy under complex operating conditions.

[0043] Further, see attached document. Figure 2 When generating the hybrid digital twin model, the data-driven sub-model trained based on gas operation data is first loaded into the computing environment. This computing environment can be a high-performance industrial server, a cloud computing platform, or an edge computing node. Subsequently, a bidirectional data interaction channel is established with the constructed physical mechanism sub-model according to a preset interface protocol. The interface protocol supports unified data format, precise timestamp synchronization, automatic filtering of abnormal data, and packet loss compensation, thereby ensuring that the data transmission between the two models meets operational requirements in terms of both real-time performance and consistency.

[0044] During operation, the standardized dataset from the data acquisition unit, after noise filtering, time alignment, and bias correction, is synchronously input into the physical mechanism sub-model and the data-driven sub-model. This synchronization process is achieved through a distributed cache queue and a global clock management mechanism, ensuring the temporal consistency and integrity of the data even under high concurrency and multi-source input conditions. The physical mechanism sub-model performs precise numerical calculations based on device topology and thermal-fluid coupling characteristics, generating prediction results under physical constraints. The data-driven sub-model uses deep learning, temporal regression, or other data-driven algorithms to perform real-time predictions on the same input data, enabling rapid output of high-frequency prediction values ​​to respond to system control requirements.

[0045] Furthermore, to improve computational efficiency and response speed, the dual-model execution process preferably employs multi-threaded computation and / or multi-core distributed computing. The physical mechanism computation task is decomposed into multiple parallel computing units, while data-driven prediction is completed using batch inference and GPU acceleration. The output of the physical mechanism sub-model is used for offline supervised training or online parameter fine-tuning of the data-driven sub-model, periodically correcting the drift error of its prediction results to ensure the long-term stability of prediction accuracy.

[0046] After completing the dual-model calculations, the two sets of prediction results are fused. This fusion includes dynamically allocating the weights of the physical mechanism sub-model and the data-driven sub-model based on historical operating data and current accuracy assessment results. Weighted summation of different predicted values ​​yields a preliminary fused output. Consistency correction is then performed on the output sequence, including trend smoothing, outlier removal, and physical constraint re-verification, ensuring that the prediction results numerically meet the boundary conditions of physical feasibility and the control strategy. During simulation, the physical mechanism sub-model serves as the accuracy benchmark and is solved over a longer time step. The data-driven sub-model predicts the system response more rapidly at a higher frequency. The outputs of the two models are fused using a weighted fusion algorithm, where the data-driven model has a higher weight initially, gradually decreasing as the prediction time progresses, prioritizing the more stable physical mechanism model prediction results.

[0047] The generated hybrid digital twin model's unified prediction output is cached in a high-performance data buffer, serving as the core input data stream for disturbance injection and multi-condition simulation calculations. This data stream can be read in real time and used for predictive control, operational optimization, and emergency strategy generation in the gas pressure regulation process. It also supports high-speed transmission between different computing nodes, enabling collaborative optimization of multi-level control systems.

[0048] The implementation of the process enables the hybrid digital twin model to possess both the rapid response characteristics of a data-driven model and the interpretability of a physical mechanism model. In the dynamic control of gas pressure regulation, it can balance real-time performance, accuracy, and stability, significantly improving the system's operational reliability and optimization capabilities under complex and variable operating conditions.

[0049] Furthermore, the process of real-time data mapping performed by the digital twin unit includes: First, invoking a synchronization mechanism to add a high-resolution timestamp of a unified format to the preprocessed multi-dimensional operational data transmitted by the data acquisition unit. The timestamp accuracy is preferably at the millisecond level or higher to ensure timing consistency across devices and networks. Subsequently, the timestamp is compared with the global clock reference of the virtual device node, and a deviation estimation algorithm is used to correct any clock deviations to ensure timing synchronization of data across all nodes in the virtual environment.

[0050] After time synchronization is complete, the system parses real-time data into node status parameters based on the device identifier, data type, and numerical validity verification results contained in the data frame. These parameters may include, but are not limited to, pressure, flow rate, temperature, valve position angle, and device operating mode. The parsing process preferably employs a multi-threaded data decoding mechanism to maintain real-time processing performance under high-frequency data input. The parsed node status parameters are accurately mapped and injected into the corresponding virtual device node objects. These node objects have a unique identifier in the virtual topology and retain their mapping relationship with the actual physical nodes.

[0051] Furthermore, when performing real-time data mapping, the digital twin unit calls the synchronous mapping mechanism to add timestamps to the preprocessed data transmitted by the data acquisition unit and compares it with the global clock reference of the virtual device node. Clock differences are eliminated through a dual algorithm of constant deviation correction and dynamic drift compensation. The device identifier and data type in the data frame are parsed, the parameters are mapped to the corresponding virtual node objects, and the operating status parameters of adjacent nodes are updated in conjunction with the virtual topology relationship to achieve low-latency dynamic consistency between the virtual and physical systems.

[0052] Compared with the traditional single-node independent synchronization method, the synchronization mechanism improves the time synchronization accuracy and adapts to complex operating conditions such as high-frequency disturbances and large-scale gas transmission and distribution across nodes.

[0053] During the injection process, the system automatically triggers the state update logic of associated nodes based on the connection relationships of nodes in the virtual topology, dynamically adjusting the operating parameters related to the nodes. This includes, but is not limited to, pressure changes in upstream and downstream pipe sections, flow distribution corrections, and temperature gradient adjustments affected by heat or cold sources, ensuring that the operating state of the equipment in the virtual environment remains dynamically consistent with the actual state of the physical equipment. This dynamic update is not limited to directly connected device nodes but can also be extended to multi-level associated devices through topology traversal, forming a globally consistent operating state mapping.

[0054] After the node state update is completed, the updated virtual node state and its associated operating parameters are written to the digital twin runtime cache. The cache adopts a high-performance in-memory database or a distributed caching architecture, supporting low latency and multi-threaded concurrent access. The cached data will serve as the core input data stream for perturbation condition injection and predictive iterative simulation calculations, ensuring that the simulation calculations are based on the latest and time-consistent system operating state, thereby improving the reliability and real-time performance of the prediction results.

[0055] Through the above mapping process, the digital twin unit can achieve low-latency synchronization between the virtual and physical systems under complex and variable gas operating conditions, providing a stable and reliable data foundation for predictive control, emergency strategy optimization, and multi-condition simulation of the system.

[0056] Furthermore, the process of perturbation injection and simulation calculation performed by the digital twin unit includes: First, reading the perturbation factors to be injected from a pre-built set of gas perturbation-safety boundary conditions. These perturbation factors include, but are not limited to, parameter sets such as pressure fluctuation range, flow rate change amplitude, temperature anomaly threshold, sudden load change mode, and valve action delay characteristics. The system parses the read perturbation factors, converts them into a unified data format required by the simulation calculation module, and standardizes the numerical units, dimensions, and data precision involved to ensure the comparability and stability of the calculations.

[0057] Furthermore, when performing disturbance injection and simulation calculations, the digital twin unit adopts a predictive iterative simulation strategy. It reads disturbance factors from the gas disturbance-safety boundary condition set and matches and binds them with the real-time mapped node state parameters to generate multiple sets of disturbance-operating state combinations. During the simulation process, the physical mechanism sub-model selects a longer time step than the data-driven sub-model for numerical calculation, while the data-driven sub-model selects a shorter time step than the physical mechanism sub-model for high-frequency iterative calculation during the prediction phase. After receiving the updated results from the physical mechanism sub-model, it optimizes the prediction accuracy through parameter fine-tuning and deviation correction. The outputs of the two are merged and updated in each iteration cycle to form multi-condition response curves, which are then associated with the valve control parameter library to generate corresponding multiple sets of gas pressure regulation schemes.

[0058] The predictive iterative simulation strategy simulates various combinations of disturbances and operating states in a virtual environment in advance, enabling dynamic response prediction of the gas pressure regulating system under different operating conditions. This retains the high accuracy of the physical mechanism sub-model while leveraging the high-frequency prediction of the data-driven sub-model to accelerate computation. Compared to traditional single-model simulation, it can quickly generate multiple executable valve gas pressure regulating schemes in complex scenarios such as sudden pressure fluctuations and flow surges, improving system safety and control flexibility.

[0059] After parsing, the disturbance factor is matched and bound to the current virtual device node's operating status data. This matching process is based on the node's unique identifier and topological location relationship, and a consistency verification mechanism is used to verify the physical applicability of the disturbance factor and the bound node. For example, it checks whether pressure disturbances are within the node's design pressure range and whether flow rate changes conform to the pipe diameter and flow resistance characteristics. After binding, the system generates multiple disturbance-operating status combinations based on the type and magnitude of the disturbance factor, and uses these as the initial input for predictive iterative simulation calculations.

[0060] Subsequently, a multi-condition predictive iterative simulation calculation process is initiated. This process is preferably based on a multi-threaded or multi-core distributed computing architecture, allocating each perturbation-operating state combination to independent computing units for parallel solution. During the simulation, the physical mechanism sub-model serves as the accuracy benchmark, solved over a longer time step. The data-driven sub-model, on the other hand, makes rapid predictions of the system response at a higher frequency. The outputs of the two models are fused using a weighted fusion algorithm, with the data-driven model having a higher weight in the early stages of prediction, gradually decreasing as the prediction time progresses, prioritizing the prediction results from the more stable physical mechanism model.

[0061] For each disturbance-operating state combination, the system generates a corresponding gas pressure regulation response curve, which describes the dynamic changes in pressure, flow rate, and temperature over time under the influence of the disturbance. The response curve is correlated with a valve control parameter library, and a genetic algorithm under constraints is used to generate multiple gas pressure regulation schemes for valve openings. Each scheme is accompanied by corresponding dynamic response characteristic values ​​and steady-state performance evaluation indicators.

[0062] After generation, the gas pressure regulation scheme is written into a high-performance scheme cache. This cache supports parallel access and historical scheme retrieval, facilitating rapid retrieval during performance evaluation and optimal scheme selection. Through the aforementioned disturbance injection and simulation calculation process, the system can verify the operational stability and control strategy effectiveness of the gas pressure regulation system under different disturbance conditions in a virtual environment. This provides directly applicable adjustment references for actual operation, significantly improving the safety and adaptive control capabilities of the gas transmission and distribution system under complex and extreme conditions.

[0063] Furthermore, when performing performance evaluation and scheme selection, the digital twin unit first receives a set of gas pressure regulation schemes for valve opening, introduced by disturbance conditions and output by the predictive iterative simulation stage. This set includes different control strategies generated under various operating conditions and their associated operating parameter curves. The system then calls the performance evaluation module to sequentially read the three core operating indicators corresponding to each gas pressure regulation scheme in the set: response time, pressure fluctuation amplitude, and pressure safety margin based on the maximum pressure limit. The response time reflects the time required for the system to reach steady state from the occurrence of a disturbance; the pressure fluctuation amplitude measures the stability of the system operation; and the pressure safety margin assesses the scheme's risk control capability under extreme operating conditions.

[0064] Furthermore, when performing performance evaluation and scheme selection, the digital twin unit adopts a weighted adaptive adjustment and safety margin judgment algorithm to perform weighted calculations on three indicators for each gas pressure regulation scheme: response time, pressure fluctuation amplitude, and safety margin. The weight values ​​are dynamically adjusted according to real-time operating status, historical data statistics, and external control strategy requirements. Based on the comprehensive score, a safety margin judgment is performed to eliminate schemes below the safety threshold. In conjunction with control stability requirements, a set of candidate schemes is selected, and finally, the gas pressure regulation scheme with the best comprehensive performance is determined and converted into a standardized control command data stream.

[0065] The algorithm dynamically adjusts the weights of three core operational indicators to achieve adaptive optimization under different operational objectives. It also incorporates a safety margin assessment to eliminate potentially risky solutions, ensuring the selection results achieve an optimal balance between safety, response speed, and pressure stability. Compared to fixed-weight and single-indicator selection methods, this significantly improves the reliable operation of gas pressure regulating systems in complex environments, including under sudden load changes, pressure fluctuations, or abnormal operating conditions.

[0066] During the index reading process, the system performs data cleaning and consistency verification on the original simulation results, including removing outliers, interpolating and completing missing data, and ensuring that the calculation results of different schemes can be directly compared under the same time resolution and data unit. Subsequently, a weighted calculation is performed on the three indices according to the pre-set weight parameters, which can be obtained through historical operation data analysis, expert experience setting, or online adaptive adjustment, to generate the corresponding comprehensive performance score.

[0067] After calculating the comprehensive score for all schemes, the scores are sorted in descending order of numerical value. The sorted results are then preliminarily screened based on operational safety thresholds and control stability requirements, forming a candidate scheme set. The operational safety thresholds include the maximum permissible pressure fluctuation range, minimum safety margin, and maximum response delay. Control stability requirements include steady-state maintenance time and overshoot limitation. Only schemes meeting these conditions are included in the candidate set, thus eliminating strategies with potential safety risks or insufficient control performance at the source.

[0068] Within the candidate solution set, the system further invokes a multi-objective control parameter optimization algorithm. This algorithm preferably employs a multi-objective evolutionary algorithm, Pareto optimal solution set screening, or constrained optimization methods to jointly analyze and weigh the three operational indicators: response time, pressure stability, and safety margin. By searching within the set indicator threshold ranges, the system can identify solutions that achieve the optimal balance among different operational indicators.

[0069] Once the candidate scheme with optimal overall performance is determined, it is converted into a standardized instruction data stream that meets the requirements of the execution unit interface. This data stream includes specific valve opening control commands, execution timing information, control mode identifiers, and fault-tolerant parameter settings. To ensure data security and reliability during transmission, the instruction data stream undergoes AES-based encryption, asymmetric key-based digital signature, and SHA-256-based integrity verification before output to prevent tampering or loss in the communication link. Finally, the securely reinforced instruction data stream is output to the execution and feedback unit to achieve precise control of the physical gas pressure regulating system. After execution, the results are fed back to the digital twin unit via a feedback channel for continuous optimization and model correction.

[0070] This invention collects and preprocesses gas operating parameters through a data acquisition unit, ensuring the accuracy and temporal consistency of the input data. The digital twin unit employs a hybrid modeling approach, combining a physical mechanism sub-model and a data-driven sub-model, balancing prediction accuracy and real-time performance. Based on real-time data mapping and disturbance-safety boundary condition injection, multi-condition predictive simulations are conducted in a virtual environment, rapidly generating multiple feasible gas pressure regulation schemes. A weighted comprehensive evaluation based on multiple indicators selects the scheme with the optimal overall performance in terms of response speed, pressure stability, and safety margin. The execution and feedback unit achieves closed-loop control and real-time calibration, significantly improving the adaptive control capability and operational safety of the gas pressure regulation system.

[0071] Example 2

[0072] In this embodiment, the digital twin-based intelligent monitoring system for gas pressure regulation is applied to a centralized gas pressure regulating station in an industrial park. This station, built and operated by a private gas company, is required to provide a stable industrial gas supply to several high-gas-consuming enterprises within the park, such as chemical, metal processing, and glass manufacturing companies. In this scenario, gas consumption exhibits significant periodic fluctuations, and some users experience intermittent high-load shocks during production, placing high demands on the real-time performance, stability, and safety of the pressure regulating system.

[0073] In this scenario, the data acquisition unit deploys multi-type industrial sensor arrays at the main pipeline, branch pipelines, valve inlets and outlets, heat exchange units, and key branch ends of the pressure regulating station. These arrays include monitoring units for pressure, temperature, flow rate, valve opening, and equipment vibration, enabling synchronous acquisition of multi-dimensional operating parameters. To ensure data consistency across nodes, the system employs a unified sampling clock combined with a high-precision time synchronization mechanism to achieve global time alignment. The acquired data undergoes low-pass filtering, adaptive Kalman filtering, time interpolation completion, and outlier correction, significantly reducing the impact of on-site interference, hardware drift, and measurement errors on data quality. This provides a high-precision, highly continuous input data stream for modeling and simulation.

[0074] Based on the actual structure of the pressure regulating station, the digital twin unit acquires the three-dimensional geometric information of the equipment using structured light scanning and laser mapping technologies. It then identifies the structural dimensions and connection methods of key components through point cloud segmentation, feature extraction, and image recognition algorithms. After digital encoding, a high-precision three-dimensional model and connection topology are established in a virtual environment. Combining a heat-fluid coupling modeling method, the gas flow path is divided into multiple computational units, assigned physical property parameters and boundary conditions, and integrated with the fluid flow, heat transfer, and pressure regulation dynamic control equations for iterative solution, forming a high-fidelity physical mechanism sub-model that reflects actual operating conditions.

[0075] During operation, the digital twin unit loads a data-driven sub-model trained and optimized using historical operational data, and establishes real-time interaction with the physical mechanism sub-model through a bidirectional data interface. Acquired data processed by a synchronous mapping mechanism is simultaneously input into both models. The physical mechanism sub-model provides high-precision calculations, while the data-driven sub-model provides rapid predictions. The outputs of both are adaptively fused using weights, with the weight allocation dynamically adjusted based on accuracy assessments. This ensures both rapid response to load changes and maintains the stability and safety margin of pressure output.

[0076] When the park is operating under high load, the system invokes a predictive iterative simulation strategy to extract disturbance factors matching the current operating state from the disturbance-safety boundary condition set, generate multiple sets of disturbance-operating state combinations, and perform parallel simulations to obtain multi-condition response curves. Subsequently, multiple gas pressure regulation schemes are generated by combining the valve control parameter library, and comprehensively evaluated through weight adaptive adjustment and safety margin judgment algorithms. Schemes that do not meet the safety margin are eliminated, and only the scheme with the best balance in terms of response speed, pressure stability, and energy efficiency is retained.

[0077] Ultimately, the optimal solution is converted into an encrypted, standardized control command data stream, which drives the pressure regulating valve to perform adjustments via the execution and feedback unit. The actual operating results are then fed back to the digital twin unit for model correction and continuous optimization. In the described application scenario, the system can achieve stable regulation and continuous gas supply when facing load fluctuations and sudden disturbances, significantly improving the operational safety, control flexibility, and long-term stability of the pressure regulating system.

[0078] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A gas pressure regulating and self-regulating intelligent monitoring system based on digital twins, characterized in that, include: The data acquisition unit acquires gas operation data, preprocesses it, and transmits it to the digital twin unit; The digital twin unit combines the physical topology and thermal-fluid coupling operation characteristics of the gas pressure regulating equipment to construct a physical mechanism sub-model, and integrates the data-driven sub-model trained based on the gas operation data to generate a hybrid digital twin model; The process by which the digital twin unit generates a hybrid digital twin model includes: The data-driven sub-model trained from gas operation data is loaded into the computing environment of the hybrid digital twin model, and a data interaction channel is established with the constructed physical mechanism sub-model according to a preset interface protocol. During the operation phase, preprocessed data from the data acquisition unit is synchronously input into the physical mechanism sub-model and the data-driven sub-model, and physical calculation and data-driven prediction are executed simultaneously through multi-threaded calculation and / or multi-core distributed computing. Among them, the data-driven sub-model is responsible for real-time prediction, and the physical mechanism sub-model is used for periodic calibration of the data-driven model. After completing the dual-model calculation, the two sets of prediction results are fused. The fusion includes assigning weights based on the accuracy evaluation results, performing weighted summation on different prediction values, and performing consistency correction on the output sequence to generate a unified prediction result of the hybrid digital twin model. The prediction result is cached after generation as a core input data stream that can be called for perturbation injection and simulation calculation. The gas operation data is mapped to virtual device nodes; a gas disturbance-safety boundary condition set is introduced into the hybrid digital twin model, disturbance factors are extracted from it, and matched with the current state of the virtual device nodes to perform disturbance simulation of the pressure regulation condition, generate gas pressure regulation response curves, and calculate multiple gas pressure regulation schemes in combination with valve control parameters. The digital twin unit performs the disturbance injection and simulation calculation process as follows: it reads disturbance factors from the gas disturbance-safety boundary condition set, parses the pressure fluctuation range, flow rate change amplitude, and temperature anomaly threshold involved in the disturbance factors into input parameters that conform to the simulation calculation format, and matches and binds them with the current virtual device node's operating status data; after binding, it starts a multi-condition predictive iterative simulation calculation process to generate a corresponding gas pressure regulation response curve for each combination of disturbance factors and operating status, and performs correlation calculations with valve control parameters to obtain multiple sets of gas pressure regulation schemes for valve opening; after generation, the gas pressure regulation scheme is written into the scheme cache area as an input data stream for performance evaluation and optimal scheme selection; The operation index score results are obtained by weighted summation of the multiple gas pressure regulation schemes; candidate gas pressure regulation schemes are screened according to safety threshold and stability requirements based on the weighted operation index score results, and the scheme with the best comprehensive performance is selected through multi-objective optimization, generating an encrypted standardized command data stream.

2. The gas pressure regulating and self-regulating intelligent monitoring system based on digital twin as described in claim 1, characterized in that, The data acquisition unit acquires multi-dimensional operating parameters of the gas operation process and performs noise filtering, time series alignment and deviation correction on the multi-source data to generate standardized data that meets the calculation requirements of the physical mechanism sub-model. The data acquisition unit includes a data packaging and transmission module, which encapsulates the standardized data into data frames with timestamps and device identifiers, and transmits them to the digital twin unit via wired and / or wireless industrial communication links.

3. The gas pressure regulating and self-regulating intelligent monitoring system based on digital twin as described in claim 1, characterized in that, The process of constructing the physical mechanism sub-model by the digital twin unit includes: By performing position analysis and size recognition on the physical components of the gas pressure regulating equipment, the analysis results are digitally encoded according to the actual spatial coordinates and connection sequence, and a three-dimensional geometry and connection topology are established in the virtual environment. Based on the aforementioned heat-fluid coupling operation characteristics modeling, the gas flow path is divided into computational units, each unit is assigned gas physical property parameters and pressure regulation boundary conditions, and the fluid flow equation, heat transfer equation and pressure regulation dynamic control equation are solved by multi-physics coupling to generate a physical mechanism sub-model that reflects the actual operation law and can be called by the hybrid digital twin model.

4. The intelligent monitoring system for gas pressure regulation and self-adjustment based on digital twin as described in claim 1, characterized in that, The process of the digital twin unit performing real-time data mapping includes: The synchronization mechanism is invoked to add timestamps to the preprocessed data transmitted by the data acquisition unit, and the timestamps are compared and the deviation is corrected with the clock reference of the virtual device node. After time synchronization is completed, the real-time data is parsed into node status parameters according to the device identifier and data type, and the parsing results are injected into the corresponding virtual device node object. During the injection process, the operating parameters associated with the virtual device nodes, including pressure, flow rate, and temperature, are updated according to the connection relationship of the nodes in the virtual topology, so that the device state in the virtual environment remains dynamically consistent with the physical device state. After the node state update is completed, the update result is written into the digital twin runtime cache as the input data stream for introducing disturbance conditions and predictive iterative simulation calculations.

5. The intelligent monitoring system for gas pressure regulation and self-adjustment based on digital twin as described in claim 1, characterized in that, When performing performance evaluation and scheme selection, the digital twin unit receives a set of valve opening gas pressure regulation schemes introduced by disturbance conditions and output by predictive iterative simulation. For each scheme, it calls the performance evaluation module to read the corresponding three operating indicators: response time, pressure fluctuation amplitude, and safety margin. It then performs weighted calculations on the operating indicators according to preset weight parameters to generate a corresponding comprehensive performance score. After completing the comprehensive scoring calculation of all schemes, the scoring results are sorted by numerical value, and a set of candidate schemes is selected based on the operational safety threshold and control stability requirements. In the set of candidate schemes, a multi-objective control parameter optimization algorithm is further invoked to evaluate the three operational indicators of response time, pressure stability and safety margin based on simulation results. Within the set threshold range, the candidate scheme with the best comprehensive performance is selected, and the finally retained optimal gas pressure regulation scheme is converted into a standardized command data stream.

Images