Digital twinning system for high-temperature header of power plant

Abstract

The invention discloses a power plant high-temperature header digital twin system, which belongs to the technical field of power plant simulation control systems, and comprises a physical layer which is deployed on a high-temperature header entity, comprises a temperature sensor, a stress sensor, a flow sensor and a data acquisition unit, and is used for acquiring temperature distribution, stress state and medium flow data of a header in real time; the temperature sensor is a distributed optical fiber temperature sensor, the stress sensor is an acoustic emission stress sensor, the sampling frequency is not lower than 1kHz, and data is transmitted to the edge computing node through the industrial Ethernet to be preprocessed; and the data acquisition unit is set to execute a filtering noise reduction algorithm through the edge computing node. High-temperature components are optimized through combination of distributed optical fiber temperature sensors and acoustic emission stress sensors, the problem of data drift of a traditional sensor caused by heat radiation interference of a high-temperature header is solved, and data reliability is improved.

Description

Technical Field

[0001] The present invention belongs to the technical field of power plant simulation control systems, and in particular relates to a digital twin system of a high-temperature header of a power plant. Background Art

[0002] Power plants are industrial facilities that convert raw energy sources, such as fossil fuels, nuclear power, hydropower, and wind power, into electricity. Their core functions include: Energy conversion—converting primary energy sources such as coal, natural gas, uranium, and water into electricity through physical or chemical processes. Power supply transmits this electricity to the grid, providing stable power to households, industries, and public facilities. Environmental regulation reduces pollution emissions through technologies such as desulfurization, denitrification, and wastewater treatment, achieving green power generation. High-temperature headers are core pressure-bearing components in power plant boiler systems. They operate under high temperatures (typically >500°C), high pressures, and alternating stresses for extended periods, making them susceptible to failure risks such as creep damage and fatigue cracking. Digital twin systems create virtual mirrors of these components, enabling real-time interaction between physical entities and digital models, addressing the pain points of traditional monitoring, such as data lag, inaccurate predictions, and reactive maintenance. Existing general-purpose power plant digital twin systems often rely on standard temperature and pressure sensors that are not optimized for high-temperature components. These sensors generate high data noise and require cloud-based processing, which can lead to latency.

[0003] Publication (Announcement) No.: CN118605240A discloses a digital twin system of a power plant system, including a safety alarm system module, a three-dimensional image acquisition module, an equipment operation twin system module, a communication and network module, a modeling module, a data storage and processing module, a three-dimensional display terminal, and a control operation and management module; the safety alarm system module is used to obtain safety monitoring data and transmit it to the data storage and processing module, and then issue an alarm based on the processing status of the data storage and processing module. The digital twin system of the power plant system, through the setting of the safety alarm system module, can obtain safety monitoring data and issue alarms in real time, which helps to improve the safety and emergency response capabilities of the power plant and effectively prevent accidents. Through the setting of the control operation and management module, it can perform operational control and personnel authorization management on three-dimensional acquisition, modeling, display, and safety monitoring data.

[0004] The patent focuses on 3D visual monitoring and safety warnings, encompassing modules such as safety alarms (sensor networks), 3D image acquisition (laser scanning / multi-cameras), equipment operation twins (modeling of combustion / steam-water and other subsystems), 5G communications, and cloud storage. Its innovations include personnel safety monitoring (smart tooling / hardhat sensors), dynamic 3D scene reconstruction (laser scanning + structured light projection), and 5G local area network coverage for power plants. Data preprocessing at edge computing nodes is not mentioned. Common power plant digital twin systems often rely on standard temperature and pressure sensors, lacking optimization for high-temperature components. This is done to address the issues of high data noise and latency associated with cloud-based processing.

[0005] Publication (Announcement) No.: CN115373358A discloses a digital twin control system for a gas power plant, which includes a simulation modeling unit, an on-site operation unit, a virtual DCS unit, and a platform control unit. The digital twin control system of the present invention can map the actual power plant into a virtual system and perform equivalent operation through a mathematical model; it can truly simulate the on-site environment, and the impact of relevant environmental parameters on the operation of the unit is also truly and correctly reflected; it can achieve the purpose of training operators. The software system used by the virtual DCS unit of the present invention is the same as the actual DCS control system, which reduces the system translation and editing time, helps to reduce maintenance costs, is more stable in later use, and reduces the probability of errors.

[0006] The aforementioned patent focuses on a virtual DCS training system and simulation accuracy verification. Modules include a simulation modeling unit, a virtual DCS (a complete replica of the actual DCS software), operator training management, and a stability verification algorithm. Its innovations include consistency between the virtual DCS system and the actual control software (no translation required) and a unique stability verification formula (parameter deviation assessment based on matrix rank calculation). Data preprocessing at the edge computing node is not mentioned. General power plant digital twin systems often rely on standard temperature / pressure sensors and are not optimized for high-temperature components. This is done to address the issues of high data noise and latency associated with cloud-based processing. Summary of the Invention

[0007] The purpose of the present invention is to overcome the shortcomings of the existing technology and provide a digital twin system for high-temperature headers of power plants, which solves the problems in the above-mentioned background technology.

[0008] The purpose of the present invention is achieved as follows: a digital twin system for a high-temperature header in a power plant, comprising: a physical layer, deployed on the high-temperature header entity, including a temperature sensor, a stress sensor, a flow sensor, and a data acquisition unit, for real-time acquisition of the temperature distribution, stress state, and medium flow data of the header; the temperature sensor is a distributed optical fiber temperature sensor, the stress sensor is an acoustic emission stress sensor, and the sampling frequency is not less than 1kHz, and the data is transmitted to the edge computing node via industrial Ethernet for preprocessing; the data acquisition unit is configured to execute a filtering and noise reduction algorithm via the edge computing node. By using a combination of distributed optical fiber temperature sensors + acoustic emission stress sensors to optimize high-temperature components, the problem of traditional sensor data drift caused by thermal radiation interference in high-temperature headers is solved, thereby improving data reliability. By using industrial Ethernet to transmit to the edge computing node for preprocessing, data noise is reduced, and the edge computing node performs real-time filtering and noise reduction, with a noise suppression rate of ≥90%, solving the problem of cloud processing delay.

[0009] Furthermore, the system also includes: a 3D model layer, which constructs a lightweight 3D geometric model based on the header design parameters and integrates material thermodynamic properties, weld structural properties, and historical operating data; a twin mapping layer, which dynamically maps real-time data collected from the physical layer to corresponding components in the 3D model layer through a coding engine, establishing unique ID associations and generating a real-time, updated digital twin; an analysis and warning layer, which integrates life prediction algorithms, stress analysis models, and anomaly detection modules to output header remaining life assessments, crack risk warnings, and optimized operating parameters based on the digital twin; and a visualization layer, which uses a pixel stream rendering engine to output the digital twin and analysis results in a 3D visualization interface, supporting cross-sectional views, thermal map overlays, and historical data backtracking. From the physical layer to the visualization layer, a complete digital twin closed loop is constructed, with unique ID associations and real-time updates, ensuring dynamic data-model synchronization. This system enables full lifecycle management, covering the entire process from data acquisition to decision output, providing a unique digital image of the high-temperature header, and covering system variants with a five-layer architecture, maximizing the scope of basic protection.

[0010] Furthermore, the lightweight processing of the 3D model layer includes: loading the container pipeline model in a hierarchical manner according to its level of sophistication, optimizing rendering efficiency through occlusion culling, and reconstructing regular geometry to reduce the computational load. This hierarchical loading, occlusion culling, and geometry reconstruction reduce the rendering load of models with hundreds of millions of meshes. It also supports loading on low-profile terminals (such as web and mobile devices) in seconds, breaking the limitation of high-precision models relying on GPU clusters, broadening system deployment scenarios, and reducing computing costs by 80%.

[0011] Furthermore, the twin mapping layer's encoding engine is configured to import model encoding files from the power plant identification system, bind the encoding to 3D model components and physical layer sensor IDs, and synchronize real-time data to the corresponding 3D elements via an API. Binding the power plant's KKS code to sensor IDs enables unified mapping of equipment, model, and business data. API data synchronization eliminates information silos, maintains compatibility with industry standards, and reduces manual coding and maintenance costs.

[0012] Furthermore, the life prediction algorithm of the analysis and warning layer is based on a creep-fatigue coupled damage model, with input parameters including real-time temperature gradients, pressure fluctuation spectra, and material microstructure data. The anomaly detection module of the analysis and warning layer utilizes an LSTM time series prediction model, with input parameters including temperature change rate and stress fluctuation spectra. The creep-fatigue coupled model predicts life with an error of ≤10%, addressing the challenge of hidden high-temperature damage. The LSTM model captures time series anomalies, such as sudden temperature changes, enabling millisecond-level risk response. This dual protection core algorithm, combining a coupled damage model and LSTM time series analysis, significantly improves upon traditional linear models (with an error >30%) and reduces unplanned downtime by 70%.

[0013] Furthermore, the visualization layer supports multi-terminal access, including dynamic rendering of digital twins on web browsers, interactive operation interfaces for VR devices, and real-time alert push notifications on mobile devices. The visualization layer also provides customizable heat map threshold controls, enabling dynamic adjustment of warning thresholds. Customizable heat map thresholds support dynamic adjustment of warning strategies, and multi-terminal adaptation (web / VR / mobile) meets full-scenario O&M requirements. VR interactive operations accelerate fault location, while mobile alert push shortens response times.

[0014] Furthermore, it includes a data interface with the power plant's DCS system, configured to receive unit load commands and provide feedback on optimized operating parameters. This implements a "monitor → analyze → execute" closed-loop control system, enabling two-way communication with the power plant's DCS system, receiving load commands and providing feedback on optimized parameters. Deeply integrated with the power plant's core control system, it supports automatic flow control and shutdown protection.

[0015] The power plant high-temperature header operation method based on the above system includes the following steps: continuously collecting header operation data through physical layer sensors; mapping this data to a three-dimensional model layer to generate a real-time digital twin; performing stress hotspot analysis and life prediction based on the twin; automatically adjusting the medium flow or triggering a shutdown protection when local stress exceeds a threshold; and outputting risk reports and maintenance recommendations at the visualization layer. The output of maintenance recommendations drives predictive operations and maintenance by performing stress analysis, life prediction, and automatic control based on the digital twin.

[0016] The beneficial effects of the present invention are as follows: By using a combination of distributed fiber optic temperature sensors and acoustic emission stress sensors to optimize high-temperature components, the problem of traditional sensor data drift caused by thermal radiation interference in high-temperature headers is solved, thereby improving data reliability. By using industrial Ethernet to transmit to edge computing nodes for preprocessing, data noise is reduced, and edge computing nodes perform real-time filtering and noise reduction, with a noise suppression rate of ≥90%, solving the problem of cloud processing delay. By building a complete digital twin closed loop from the physical layer to the visualization layer, unique ID association, and real-time updates, dynamic synchronization of data and models is ensured. Full lifecycle management is achieved: covering the entire process from data acquisition to decision output, providing a unique digital image for high-temperature headers, covering system variants with five-layer architectures, and maximizing the basic protection range. Through hierarchical loading, occlusion culling, and geometry reconstruction, the rendering load of models with hundreds of millions of facets is reduced; low-configuration terminals (such as web / mobile terminals) are supported to load in seconds, breaking through the limitation of high-precision models relying on GPU clusters, broadening system deployment scenarios, and reducing computing power costs by 80%. By binding the power plant KKS code and sensor ID, unified mapping of equipment-model-business data is achieved, API interface synchronizes data, eliminates information silos, achieves compatibility with industrial standards, and reduces manual coding maintenance costs. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1It is a system framework diagram of the present invention; DETAILED DESCRIPTION

[0018] The present invention will be further described in detail below with reference to the accompanying drawings. It should be noted that the drawings are only for the purpose of more clearly illustrating and explaining the present invention. Example 1

[0019] like Figure 1 As shown, this embodiment discloses: A digital twin system for a high-temperature header in a power plant comprises: a physical layer, deployed on the high-temperature header entity, comprising a temperature sensor, a stress sensor, a flow sensor, and a data acquisition unit, for real-time collection of the header's temperature distribution, stress state, and medium flow data; the temperature sensor is a distributed optical fiber temperature sensor, the stress sensor is an acoustic emission stress sensor, and the sampling frequency is not less than 1kHz, and the data is transmitted to an edge computing node via industrial Ethernet for preprocessing; the data acquisition unit is configured to execute a filtering and noise reduction algorithm via the edge computing node. By using a combination of distributed optical fiber temperature sensors and acoustic emission stress sensors to optimize high-temperature components, the problem of traditional sensor data drift caused by thermal radiation interference in high-temperature headers is solved, thereby improving data reliability. By using industrial Ethernet to transmit to an edge computing node for preprocessing, data noise is reduced, and the edge computing node performs real-time filtering and noise reduction, with a noise suppression rate of ≥90%, solving the problem of cloud processing delay.

[0020] A method for operating a high-temperature header in a power plant based on the aforementioned system includes the following steps: continuously collecting header operating data through physical layer sensors; mapping this data to a three-dimensional model layer to generate a real-time digital twin; performing stress hotspot analysis and life prediction based on the twin; automatically adjusting the medium flow rate or triggering a shutdown protection when local stress exceeds a threshold; and outputting risk reports and maintenance recommendations at the visualization layer. The output maintenance recommendations drive predictive operations and maintenance by performing stress analysis, life prediction, and automatic control based on the digital twin. Example 2

[0021] like Figure 1 As shown, this embodiment discloses: A digital twin system for a high-temperature header in a power plant comprises: a physical layer, deployed on the high-temperature header entity, comprising a temperature sensor, a stress sensor, a flow sensor, and a data acquisition unit, for real-time collection of the header's temperature distribution, stress state, and medium flow data; the temperature sensor is a distributed optical fiber temperature sensor, the stress sensor is an acoustic emission stress sensor, and the sampling frequency is not less than 1kHz, and the data is transmitted to an edge computing node via industrial Ethernet for preprocessing; the data acquisition unit is configured to execute a filtering and noise reduction algorithm via the edge computing node. By using a combination of distributed optical fiber temperature sensors and acoustic emission stress sensors to optimize high-temperature components, the problem of traditional sensor data drift caused by thermal radiation interference in high-temperature headers is solved, thereby improving data reliability. By using industrial Ethernet to transmit to an edge computing node for preprocessing, data noise is reduced, and the edge computing node performs real-time filtering and noise reduction, with a noise suppression rate of ≥90%, solving the problem of cloud processing delay.

[0022] To achieve optimal results, the system also includes: a 3D model layer, which constructs a lightweight 3D geometric model based on the header design parameters and integrates material thermodynamic properties, weld structural properties, and historical operating data; a twin mapping layer, which dynamically maps real-time data collected from the physical layer to corresponding components in the 3D model layer through a coding engine, establishing unique ID associations and generating a real-time, updated digital twin; an analysis and warning layer, which integrates life prediction algorithms, stress analysis models, and anomaly detection modules to output header remaining life assessments, crack risk warnings, and optimized operating parameters based on the digital twin; and a visualization layer, which uses a pixel stream rendering engine to output the digital twin and analysis results in a 3D visualization interface, supporting cross-sectional views, thermal map overlays, and historical data backtracking. From the physical layer to the visualization layer, a complete digital twin closed loop is constructed, with unique ID associations and real-time updates, ensuring dynamic data-model synchronization. This system enables full lifecycle management, covering the entire process from data acquisition to decision output, providing a unique digital image of the high-temperature header, and covering system variants including the five-layer architecture, maximizing the scope of basic protection.

[0023] To achieve optimal results, the lightweight processing of the 3D model layer includes: loading the container and pipeline models in different levels of detail, optimizing rendering efficiency through occlusion culling, and reconstructing regular geometry to reduce the computational load. This hierarchical loading, occlusion culling, and geometry reconstruction reduce the rendering load of models with hundreds of millions of meshes. It also supports loading on low-profile terminals (such as web and mobile devices) in seconds, breaking the limitation of high-precision models relying on GPU clusters, broadening system deployment scenarios, and reducing computing costs by 80%.

[0024] To achieve optimal results, the twin mapping layer's encoding engine is configured to import model encoding files from the power plant identification system, bind the encoding to 3D model components and physical layer sensor IDs, and synchronize real-time data to the corresponding 3D elements via an API. Binding the power plant's KKS code to the sensor ID enables unified mapping of equipment, model, and business data. Data synchronization via the API eliminates information silos, complies with industry standards, and reduces manual coding and maintenance costs.

[0025] To achieve optimal results, the life prediction algorithm in the analysis and early warning layer is based on a creep-fatigue coupled damage model. Input parameters include real-time temperature gradients, pressure fluctuation spectra, and material microstructure data. Furthermore, the anomaly detection module in the analysis and early warning layer utilizes an LSTM time series prediction model, with input parameters including temperature change rate and stress fluctuation spectra. The creep-fatigue coupled model predicts life with an error of ≤10%, addressing the challenge of hidden high-temperature damage. The LSTM model captures time series anomalies, such as sudden temperature changes, enabling millisecond-level risk response. This dual protection core algorithm, combining a coupled damage model and LSTM time series analysis, significantly improves upon traditional linear models (with an error >30%) and reduces unplanned downtime by 70%.

[0026] To achieve optimal results, the visualization layer supports multi-terminal access, including dynamic rendering of digital twins via web browsers, interactive user interfaces for VR devices, and real-time alert push notifications on mobile devices. Furthermore, the visualization layer provides customizable heatmap threshold controls, enabling dynamic adjustment of warning thresholds. Customizable heatmap thresholds support dynamic adjustment of warning strategies, and multi-terminal adaptation (web / VR / mobile) supports full-scenario O&M. VR interactive operations accelerate fault location, while mobile alert push shortens response times.

[0027] To achieve optimal results, the system also includes a data interface with the power plant's DCS system, configured to receive unit load commands and provide feedback on optimized operating parameters. This implements a "monitor → analyze → execute" closed-loop control system, enabling two-way communication with the power plant's DCS system, receiving load commands and providing feedback on optimized parameters. Deeply integrated with the power plant's core control system, it supports automatic flow control and shutdown protection.

[0028] A method for operating a high-temperature header in a power plant based on the aforementioned system includes the following steps: continuously collecting header operating data through physical layer sensors; mapping this data to a three-dimensional model layer to generate a real-time digital twin; performing stress hotspot analysis and life prediction based on the twin; automatically adjusting the medium flow rate or triggering a shutdown protection when local stress exceeds a threshold; and outputting risk reports and maintenance recommendations at the visualization layer. The output maintenance recommendations drive predictive operations and maintenance by performing stress analysis, life prediction, and automatic control based on the digital twin.

[0029] The above are only preferred specific embodiments of the present invention, but the scope of protection of the present invention is not limited thereto. Any technician familiar with the technical field can make equivalent replacements or changes based on the technical solutions and concepts of the present invention within the technical scope disclosed by the present invention, and they should be covered by the scope of protection of the present invention.

Claims

1. A digital twin system for high-temperature headers in a power plant, characterized in that: include: The physical layer is deployed on the high-temperature header entity and includes a temperature sensor, a stress sensor, a flow sensor and a data acquisition unit, which are used to collect the temperature distribution, stress state and medium flow data of the header in real time; the temperature sensor is a distributed optical fiber temperature sensor, and the stress sensor is an acoustic emission stress sensor with a sampling frequency of not less than 1kHz. The data is transmitted to the edge computing node via industrial Ethernet for preprocessing; the data acquisition unit is configured to execute a filtering and noise reduction algorithm through the edge computing node.

2. The power plant high temperature header digital twin system according to claim 1, characterized in that: Also includes: The 3D model layer builds a lightweight 3D geometric model based on the header design parameters and couples the material thermodynamic properties, weld structural properties, and historical operation data. The twin mapping layer dynamically maps the real-time data collected by the physical layer with the corresponding components in the 3D model layer through the coding engine, establishes a unique ID association relationship, and generates a real-time updated digital twin; The analysis and early warning layer integrates life prediction algorithms, stress analysis models, and anomaly detection modules to output remaining life assessment of the header, crack risk warning, and optimized operating parameters based on the digital twin; The visualization layer outputs the digital twin and analysis results in a three-dimensional visualization interface through the pixel flow rendering engine, supporting cross-sectional views, heat map overlays, and historical data backtracking.

3. The digital twin system for high-temperature headers of a power plant according to claim 2, characterized in that: The lightweight processing of the three-dimensional model layer includes: loading the header and pipeline model in a graded manner according to the degree of fineness, optimizing rendering efficiency through occlusion culling technology, and reconstructing regular geometric bodies to reduce computing load.

4. The digital twin system for high-temperature headers of a power plant according to claim 2, characterized in that: The encoding engine of the twin mapping layer is configured to import the model encoding file of the power plant identification system, bind the encoding to the three-dimensional model component and the physical layer sensor ID, and synchronize the real-time data to the corresponding three-dimensional element through the API interface.

5. The digital twin system for high-temperature headers of a power plant according to claim 2, characterized in that: The life prediction algorithm of the analysis and warning layer is based on the creep-fatigue coupled damage model, and the input parameters include real-time temperature gradient, pressure fluctuation spectrum and material microstructure data. The anomaly detection module of the analysis and warning layer adopts the LSTM time series prediction model, and the input parameters include temperature change rate and stress fluctuation spectrum.

6. The power plant high-temperature header digital twin system according to claim 2, characterized in that: The visualization layer supports multi-terminal access, including: dynamic rendering of digital twins on the Web browser side, interactive operation interface of VR devices, and real-time alarm push on the mobile side. The visualization layer also provides custom controls for heat map thresholds and supports dynamic adjustment of warning critical values.

7. The power plant high-temperature header digital twin system according to claim 2, characterized in that: It also includes a data interface for the power plant DCS system, which is configured to receive unit load instructions and feed back optimized operating parameters.

8. A method for operating a high-temperature header in a power plant based on the system according to any one of claims 1 to 7, characterized in that: The following steps are involved: Continuously collect header operation data through physical layer sensors; Map data to the 3D model layer to generate a real-time digital twin; Perform stress hotspot analysis and life prediction based on twins; When the local stress is detected to exceed the threshold, the medium flow rate is automatically adjusted or shutdown protection is triggered; Output risk reports and maintenance recommendations at the visualization layer.

Images