Real-time boiler data prediction system

Abstract

The invention relates to the field of boiler prediction, and discloses a boiler real-time data prediction system which is used for improving the accuracy of boiler fault prediction. According to the boiler real-time data prediction system, intrinsic characteristics of combustion fluctuation, convective heat transfer, heat conduction and the like of a flow field in a boiler are extracted through computational fluid mechanics characteristic modal decomposition, a dynamic state evolution model is built by combining mass, energy and momentum conservation laws, and coupling prediction of a temperature field, a pressure field and a flow velocity field is achieved; and a non-equilibrium thermodynamic entropy generation principle is introduced to carry out physical consistency correction on a prediction result, so that high-precision prediction under the constraint of a thermodynamic law is ensured. And finally, mapping the risk indexes to a three-dimensional boiler model through a digital twinning technology, and generating an interactive early warning interface with spatial positioning and risk level visualization. According to the invention, a complete technical chain of data acquisition, feature analysis, state prediction and risk early warning is formed, and early accurate early warning and space positioning capabilities are provided for safe operation of the boiler of the power plant.

Description

Technical Field

[0001] This invention relates to the field of boiler prediction, and more particularly to a real-time boiler data prediction system. Background Technology

[0002] In industries such as power and chemical engineering, boilers, as core heat energy conversion equipment, directly affect the safety and economy of production. With the growth of energy demand and the expansion of equipment scale, the boiler operating environment is becoming increasingly complex. The coupling effect of multiple physical fields such as high temperature, high pressure, and high flow rate leads to frequent equipment failures, especially problems such as tube rupture, oxide scale shedding, and abnormal expansion. These not only cause unplanned downtime losses but may also lead to serious safety accidents.

[0003] Traditional boiler operation and maintenance relies on periodic inspections and experience-based judgment, which has the following limitations: Existing monitoring systems mostly collect single physical quantities and lack spatiotemporal correlation analysis of multi-physics field data, making it difficult to capture dynamic characteristics such as combustion fluctuations and convective heat transfer. Statistical or empirical methods cannot accurately describe the coupled evolution of the complex flow field and thermodynamic field inside the boiler, resulting in a large deviation between the predicted results and the actual results. Predicted data may violate the second law of thermodynamics, leading to non-physical results and affecting the reliability of fault diagnosis; It can only provide qualitative risk warnings, lacks spatial positioning and quantitative risk indicators, and is difficult to guide precise maintenance.

[0004] Therefore, we propose a real-time boiler data prediction system to solve the above problems. Summary of the Invention

[0005] This invention provides a real-time data prediction system for boilers to improve the accuracy of boiler fault prediction.

[0006] The first aspect of this invention provides a real-time boiler data prediction system, comprising: an acquisition module for collecting temperature, pressure, and velocity field data of the boiler during operation, generating a boiler state vector using a dynamic reference scaling method, and establishing a boiler operating state dataset; a decomposition module for extracting combustion fluctuation characteristics, convective heat transfer characteristics, and heat conduction characteristics from the boiler operating state dataset, and constructing a boiler flow field characteristic parameter set; a construction module for constructing a boiler dynamic evolution model using a state evolution model based on physical conservation laws, combined with the boiler flow field characteristic parameter set, and generating multi-physics field prediction data; a correction module for performing thermodynamic consistency correction based on the multi-physics field prediction data, calculating tube rupture risk indicators, oxide scale risk indicators, and expansion risk indicators, and generating a boiler fault early warning indicator set; and a prediction module for mapping the boiler fault early warning indicator set to a three-dimensional model of the boiler heating surface using a digital twin three-dimensional visualization engine, and generating a three-dimensional visualization early warning interface.

[0007] Optionally, in a first implementation of the first aspect of the present invention, the method includes: collecting wall temperature data, steam pressure data, and flue gas velocity data of each heating surface of the boiler to generate an original multiphysics monitoring dataset; establishing a dynamic reference scale benchmark based on boiler design parameters, including a rated temperature benchmark value, a design pressure benchmark value, and a standard flow velocity benchmark value, to generate a dynamic reference scale parameter set; performing continuous proportional calculations on the physical quantity data in the original multiphysics monitoring dataset with the corresponding dynamic reference scale parameters to generate dimensionless temperature field data, dimensionless pressure field data, and dimensionless flow velocity field data; integrating the dimensionless temperature field data, dimensionless pressure field data, and dimensionless flow velocity field data according to spatial location and acquisition time to construct a boiler state vector; and mapping and associating the boiler state vector with the coordinates of the boiler three-dimensional model based on the topological relationship of the boiler heating surface structure to establish a boiler operating state dataset.

[0008] Optionally, in the second implementation of the first aspect of the present invention, the method includes: establishing a computational fluid dynamics model of the internal flow field of the boiler based on the three-dimensional geometry and operating conditions of the boiler, and generating a basic framework for flow field feature analysis; performing modal decomposition on the basic framework of flow field feature analysis, extracting the dominant spatial modes of the internal flow field of the boiler, and generating a set of basis functions for flow field feature modes; projecting the dimensionless velocity field data in the boiler operating state dataset onto the set of basis functions for flow field feature modes, calculating the time coefficients corresponding to each mode, and generating a spatiotemporal coefficient matrix of flow field modes; based on the energy transport mechanism inside the boiler, separating the combustion fluctuation component, the convective heat transfer component, and the heat conduction component from the spatiotemporal coefficient matrix of flow field modes, and constructing a multi-scale flow field feature parameter set; and establishing the correlation between the multi-scale flow field feature parameter set and the heat load distribution of the boiler heating surface through a correlation analysis method, thereby forming a boiler flow field feature parameter set.

[0009] Optionally, in a third implementation of the first aspect of the present invention, the method includes: establishing a boiler system control equation system to generate a state evolution basic framework; embedding combustion fluctuation characteristics, convective heat transfer characteristics, and heat conduction characteristics from the boiler flow field characteristic parameter set into the state evolution basic framework to construct a boiler dynamic evolution model; calculating the time derivative of state variables based on the current state parameters in the boiler operating state dataset using the boiler dynamic evolution model to generate a state evolution rate field; performing time-progression calculations on the state evolution rate field to generate a multiphysics state evolution sequence for future time periods; verifying the multiphysics state evolution sequence using the flow field characteristic parameter set, correcting state data that does not conform to the laws of fluid mechanics, and generating multiphysics prediction data.

[0010] Optionally, in the fourth implementation of the first aspect of the present invention, the method includes: calculating the product of heat flux density and temperature gradient at each location in the multiphysics prediction data to generate entropy generation rate distribution data; performing thermodynamic constraint correction on the multiphysics prediction data based on the entropy generation rate distribution data to eliminate prediction results that violate the second law of thermodynamics and generating thermodynamically consistent state data; calculating the equivalent thermal stress and creep damage accumulation of the heated surface pipe based on the temperature field and stress field data in the thermodynamically consistent state data to generate material damage evolution data; calculating the oxide scale growth rate and peeling risk probability of each heated surface region to generate oxide scale risk distribution data; and combining the material damage evolution data and oxide scale risk distribution data to calculate the tube rupture risk index, oxide scale risk index, and expansion risk index through a risk superposition model to generate a boiler fault early warning index set.

[0011] Optionally, in the fifth implementation of the first aspect of the present invention, the method includes: establishing a three-dimensional geometric model of the boiler heating surface, including structural models of the water-cooled wall, superheater, reheater, and economizer, and generating a three-dimensional digital twin basic model of the boiler; establishing a risk level and color mapping relationship based on the risk level data in the boiler fault early warning indicator set, and generating a risk visualization coding scheme; performing coordinate registration between the spatial location data in the boiler fault early warning indicator set and the three-dimensional digital twin basic model of the boiler, and generating risk distribution mapping data; using real-time rendering technology to combine the risk distribution mapping data with the risk visualization coding scheme, and generating a dynamically updated risk heat map on the three-dimensional digital twin basic model of the boiler; and integrating early warning time, risk type, and handling suggestion information into a three-dimensional visualization interface to generate a three-dimensional visualization early warning interface.

[0012] Optionally, in a sixth implementation of the first aspect of the present invention, a color mapping relationship is established when the risk value is in the medium-risk range of 0.3 to 0.7: R=255; ; B=0; Where r is the risk index value of boiler failure.

[0013] Optionally, in the seventh implementation of the first aspect of the present invention, an operation and maintenance module is further included: a knowledge base containing maintenance schemes corresponding to different fault types is established based on historical maintenance records and boiler equipment operating parameters, generating a boiler maintenance strategy library; according to the risk type and risk level in the boiler fault early warning indicator set, a corresponding maintenance scheme is matched from the boiler maintenance strategy library to generate a maintenance work order; combined with the power plant overhaul plan and equipment operating status, the maintenance work order is optimized and allocated resources to generate an overhaul implementation plan; based on the overhaul implementation plan and real-time operating data, the expected overhaul effect and operating parameter improvement target are calculated, generating an expected effect evaluation report; the maintenance work order, overhaul implementation plan, and expected effect evaluation report are displayed through a three-dimensional visualization early warning interface, generating a boiler operation and maintenance management interface.

[0014] Optionally, in an eighth implementation of the first aspect of the present invention, the expected effect evaluation report includes a comparison formula to quantify the effect: .

[0015] The mechanism of this invention is as follows: by deeply integrating flow field analysis, thermodynamic constraints and state evolution prediction, a complete technology chain from data acquisition, feature extraction, state prediction to risk warning is formed, which realizes early and accurate prediction and spatial location of boiler heating surface failures; Beneficial effects: By using the intrinsic orthogonal decomposition method, the complex flow field is decomposed into independent modes, solving the problem that traditional methods cannot capture dynamic fluctuation characteristics. Correlation analysis is used to establish the relationship between flow field characteristics and heat load distribution, making the parameter set have clear engineering interpretability. Considering the interaction of multiphysics fields, this paper addresses the prediction distortion problem caused by neglecting coupling effects in traditional models. It generates multiphysics field state sequences for future time periods through numerical integration methods, supporting real-time prediction and dynamic adjustment. The prediction data is thermodynamically corrected based on the principle of entropy generation, and risk indicators of pipe burst, oxide scale, and expansion are calculated. By correcting the entropy generation rate distribution, prediction data that violates the second law of thermodynamics is eliminated, thus improving the reliability of the results. By mapping fault early warning indicators to the boiler's three-dimensional digital twin model, a dynamic risk heat map and interactive interface are generated. Through coordinate registration of the three-dimensional model and risk data, the high-risk area can be accurately located, solving the problem of the lack of spatial information in traditional two-dimensional charts. Maintenance strategies are automatically matched based on risk type and level, solving the problem of reliance on manual experience. By comparing the status before and after maintenance, expected effect reports are generated, providing data support for the iteration of operation and maintenance strategies. Attached Figure Description

[0016] Figure 1 This is a schematic diagram of one embodiment of the boiler real-time data prediction system in this invention. Figure 2 This is a schematic diagram of another embodiment of the boiler real-time data prediction system in this invention; Figure 3 This is a schematic diagram of one embodiment of the boiler real-time data prediction device in this invention. Detailed Implementation

[0017] This invention provides a real-time data prediction system for boilers to improve the accuracy of boiler fault prediction. The terms "first," "second," "third," "fourth," etc. (if applicable) in the specification, claims, and accompanying drawings are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" or "having" and any variations thereof are intended to cover a non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0018] For ease of understanding, the specific process of the embodiments of the present invention is described below. Please refer to [link / reference]. Figure 1 One embodiment of the boiler real-time data prediction system in this invention includes: 101. Acquisition module, used to collect temperature field, pressure field and flow velocity field data of boiler operation, generate boiler state vector through dynamic reference scaling method, and establish boiler operation state dataset containing thermodynamic parameters and flow field parameters; It is understood that the executing entity of this invention can be a boiler real-time data prediction device, a terminal, or a server; no specific limitation is made here. This embodiment of the invention will be described using a server as an example.

[0019] It should be noted that, taking a pulverized coal boiler as an example, the boiler is equipped with multiple sensors for real-time monitoring of key parameters.

[0020] In real-time boiler data prediction, it is necessary to collect temperature, pressure, and velocity field data during boiler operation. In this example, six sensor points are arranged in the boiler's heating surface area: temperature sensors are installed at the furnace outlet, superheater inlet, and economizer outlet; pressure sensors are installed at the steam drum and superheater outlet; and velocity sensors are installed in the flue and air duct. Specific collected data are as follows (taking a specific moment as an example): Temperature field data: Furnace outlet temperature 620°C, superheater inlet temperature 540°C, economizer outlet temperature 320°C. Pressure field data: Steam drum pressure 12 MPa, superheater outlet pressure 11.5 MPa. Velocity field data: Flue gas velocity 15 m / s, air duct velocity 8 m / s.

[0021] A dimensionless boiler state vector is generated using the dynamic reference scaling method. The core of this method is selecting dynamic reference values, which are based on boiler design parameters or historical normal operating ranges and adjusted according to the operating conditions. In this example, the reference values ​​selected are the boiler's rated operating values: reference temperature 500 degrees Celsius (based on rated steam temperature), reference pressure 10 MPa (based on design operating pressure), and reference velocity 10 m / s (based on typical flue gas velocity). Dividing the collected actual values ​​by the corresponding reference values ​​yields the dimensionless parameters: Dimensionless temperature: furnace outlet 620 / 500 = 1.24, superheater inlet 540 / 500 = 1.08, economizer outlet 320 / 500 = 0.64. Dimensionless pressure: steam drum pressure 12 / 10 = 1.20, superheater outlet 11.5 / 10 = 1.15. Dimensionless velocity: flue gas velocity 15 / 10 = 1.50, air duct velocity 8 / 10 = 0.80.

[0022] This generates a dimensionless boiler state vector in the form of [Temperature 1, Temperature 2, Temperature 3, Pressure 1, Pressure 2, Velocity 1, Velocity 2], i.e., [1.24, 1.08, 0.64, 1.20, 1.15, 1.50, 0.80]. This vector integrates thermodynamic parameters (temperature, pressure) and flow field parameters (velocity), eliminating the influence of units and facilitating subsequent analysis.

[0023] Time-series data was continuously collected, with sensor readings recorded every 5 seconds, and processed using the same dynamic reference scaling method. The dataset is stored in tabular form, with each row representing a dimensionless state vector at a given time point, labeled with a timestamp. The initial time vector is [1.24, 1.08, 0.64, 1.20, 1.15, 1.50, 0.80], and the next time vector might be [1.22, 1.10, 0.62, 1.18, 1.16, 1.48, 0.82]. The dataset contains hundreds of such vectors, covering different load conditions, forming the historical basis of the boiler's operating status.

[0024] 102. Decomposition module, used to extract combustion fluctuation features, convective heat transfer features and heat conduction features from boiler operating status dataset based on computational fluid dynamics characteristic mode decomposition technology, and to construct boiler flow field characteristic parameter set; It should be noted that we now need to analyze the time-series dataset generated in step 101. This dataset contains dimensionless temperature, pressure, and flow velocity data from multiple measuring points of the boiler over a period of time. Based on computational fluid dynamics characteristic mode decomposition (EMD), the core idea is to decompose these seemingly complex data streams into a series of characteristic patterns (or modes) with clear physical meanings and their weighting coefficients that change over time.

[0025] The dataset is treated as a whole, containing 360 dimensionless state vectors recorded every 5 seconds over the past 30 minutes. The technology decomposes all the data from these 360 ​​time points (including 7 parameters: 3 temperatures, 2 pressures, and 2 flow rates) to obtain a set of feature patterns. Each feature pattern represents a specific, spatially relatively fixed flow or heat transfer structure inside the boiler. The decomposition may yield: Mode 1: This mode exhibits the strongest positive weighting in the furnace outlet and superheater inlet areas, while the economizer area shows a weaker weighting. This mode primarily reflects the thermal fluctuations in the core combustion region.

[0026] Mode 2: It has high weighting in flue gas flow velocity and economizer temperature parameters, and the changes are synchronous, reflecting the intensity of convective heat transfer by flue gas scouring the heated surface.

[0027] Mode 3: It exhibits a specific phase difference relationship in the pressure and temperature parameters at the inlet and outlet of the superheater, reflecting the heat conduction process along the metal tube wall.

[0028] Analyze the time coefficient corresponding to each characteristic mode (i.e., the activity level of the mode at each time point) and extract dynamic characteristic parameters that can quantitatively describe the boiler state. Extract combustion fluctuation characteristics: Analyze the time coefficient of "Mode 1" above. Calculate the standard deviation of this time coefficient to obtain a quantitative combustion stability index. Under stable load, this index may be 0.08; while when the coal quality fluctuates, this index may increase to 0.15. This index is the core parameter of the combustion fluctuation characteristics. Extract convective heat transfer characteristics: Analyze the ratio of the time coefficient of "Mode 2" to its own energy and correlate it with the instantaneous value of the flue gas velocity to calculate a convective heat transfer intensity coefficient. When the flow velocity increases from 1.50 to 1.55, this coefficient may increase from 2.1 to 2.3, indicating enhanced convective heat transfer. Extracting heat transfer characteristics: Analyzing the correlation between the time coefficient of "Mode 3" and the change in temperature difference between the inlet and outlet of the superheater, a heat transfer efficiency factor is calculated. Under normal operating conditions, this factor may be stable at around 0.95; if there is ash accumulation on the pipe wall, leading to deterioration of heat transfer, this factor may drop to 0.88.

[0029] These three key characteristic parameters—combustion stability index (0.08), convective heat transfer intensity coefficient (2.1), and heat transfer efficiency factor (0.95)—along with their corresponding characteristic mode numbers, are packaged together into a new, highly condensed set of boiler flow field characteristic parameters. This parameter set is no longer just raw sensor readings, but reveals the dynamic characteristics of the core physical processes inside the boiler.

[0030] 103. Construction module, used to construct a dynamic evolution model of boiler by adopting a state evolution model based on the law of physical conservation and combining it with the boiler flow field characteristic parameter set, and generate multi-physics field prediction data including temperature field evolution, pressure field evolution and flow velocity field evolution; It should be noted that these deep-seated characteristic parameters are used to predict specific changes in boiler temperature, pressure, and flow rate over a future period. This is achieved using a state evolution model based on physical conservation laws. The core of this model is pre-embedded physical rules of the boiler system, mainly including the laws of mass conservation, energy conservation, and momentum conservation. These laws determine the fundamental laws that the flow, heat transfer, and pressure changes of the working fluid (water, steam, flue gas) must follow.

[0031] The boiler flow field characteristic parameter set obtained in step 102 at the current moment is used as the key input and injected into this physical rule model, thereby constructing a personalized boiler dynamic evolution model. This model is no longer a simple data extrapolation, but a digital simulator constrained by physical laws that reflects the specific operating state of the current boiler.

[0032] The model needs to predict the state over the next 5 minutes, dividing the time into continuous intervals (each 30 seconds). Based on the current state and physical laws, the model progressively extrapolates the changes in each physical quantity: Pressure field evolution: According to the law of conservation of mass and the current combustion stability index (0.08, indicating stable combustion), the model extrapolates the working fluid flow rate and pressure distribution. The model predicts that the steam drum pressure will slowly rise from the current 12.0 MPa to 12.1 MPa due to stable steam production, and the superheater outlet pressure will correspondingly evolve from 11.5 MPa to 11.6 MPa. Velocity field evolution: According to the law of conservation of momentum and the convective heat transfer intensity coefficient (2.1), the model calculates the change in flue gas velocity, predicting that the flue gas velocity will slightly increase from the current 15.0 m / s to 15.2 m / s due to changes in system resistance. Temperature field evolution: This is the most crucial part, dominated by the law of conservation of energy. The model integrates all characteristics for prediction: Combustion fluctuation characteristics (index 0.08) determine the stability of heat input at the furnace outlet, so the temperature at this point changes steadily from 620 degrees Celsius.

[0033] The convective heat transfer characteristics (coefficient 2.1) and the heat conduction characteristics (factor 0.95) together determine the heat exchange efficiency in the heat exchange surface. The model predicts that, since the current heat exchange efficiency is normal, the superheater inlet temperature will rise from 540 degrees Celsius to 543 degrees Celsius, while the economizer outlet temperature will rise from 320 degrees Celsius to 322 degrees Celsius due to the combined effects of the inlet water temperature and flue gas scouring.

[0034] Through this step-by-step deduction based on physical laws, multiphysics prediction data for the next 10 time steps (a total of 5 minutes) was finally generated. This data is a table or data stream containing specific values, clearly showing the predicted evolution trajectory of temperature, pressure, and flow rate at various key points of the boiler over a future period.

[0035] 104. Correction module, used to perform thermodynamic consistency correction on multi-physics field prediction data based on the principle of non-equilibrium thermodynamic entropy generation, calculate tube rupture risk index, oxide scale risk index and expansion risk index, and generate boiler fault early warning index set. It should be noted that the thermodynamic consistency correction for multiphysics prediction data is based on the principle of non-equilibrium thermodynamic entropy generation. The core of this principle is that a real and reasonable physical process will always be accompanied by energy dissipation (i.e., entropy generation). The model built into the server checks whether the prediction data in step 103 satisfies this principle.

[0036] The model discovered that the increase in heat and pressure in the superheater region did not perfectly match in some local details, resulting in a calculated "entropy generation rate" lower than the theoretically reasonable threshold. This indicated that the predicted state was too "ideal" and might underestimate energy losses and risks in actual operation. Therefore, the server fine-tuned the predicted data to ensure thermodynamic consistency, correcting the temperature of a local hot spot on the superheater tube wall from the predicted 543 degrees Celsius to 545 degrees Celsius, more realistically reflecting the potential for actual heat accumulation. This correction process ensured that subsequent risk assessments were based on a more physically realistic scenario.

[0037] Using corrected multiphysics data, three key risk indicators were calculated: First, a tube rupture risk indicator was calculated. This indicator primarily assesses the likelihood of boiler pipes rupturing due to excessive internal pressure, temperature, and pipe wall stress. The server analyzed the corrected data, particularly the spatiotemporal distribution of the pressure and temperature fields. It identified a section of pipe in the superheater with a predicted internal pressure of 11.6 MPa and a localized pipe wall temperature of 545 degrees Celsius. Combining this with the pipe's material and wall thickness design parameters, the equivalent stress at this location was calculated to be 125 MPa. Comparing this stress with the material's allowable stress, a tube rupture risk index of 0.75 was obtained (range 0-1, higher values ​​indicate greater danger), with a risk level of "medium warning." Second, an oxide scale risk indicator was calculated. This indicator assesses the tendency for oxide film growth and detachment on the surface of high-temperature metal pipes. Detached material can clog pipes and cause malfunctions. The server focuses on the peak temperature and temperature fluctuations of the metal pipe wall in the corrected temperature field. Based on the corrected peak temperature of 545 degrees Celsius and referring to the oxide scale growth kinetic model of this type of steel, the oxide scale accumulation rate per unit area is calculated to be 0.08 mm per hour. Comparing this rate with a safety threshold, the oxide scale risk index is 0.60, with a risk level of "low warning". The expansion risk index is calculated: this index assesses the risk of significant thermal stress caused by inconsistent thermal expansion in different boiler components or different parts of the same component due to uneven temperature distribution. The server analyzes the distribution gradient of the corrected temperature field on large components such as the boiler drum and headers, and calculates that the predicted temperature difference between the upper and lower surfaces of the drum reaches 35 degrees Celsius, which may lead to uneven expansion due to the resulting thermal stress. Quantifying this temperature difference and structural constraints, the expansion risk index is 0.40, with a risk level of "attention".

[0038] These three specific risk indicators (tube rupture risk index 0.75, oxide scale risk index 0.60, and expansion risk index 0.40) and their corresponding risk levels and spatial location information ("superheater section A" and "steam drum") are packaged to generate a structured boiler fault early warning indicator set. This indicator set provides accurate data support for the final visualized early warning.

[0039] 105. Prediction module, which uses a digital twin 3D visualization engine to map the boiler fault early warning indicator set to the 3D model of the boiler heating surface, and generates a 3D visualization early warning interface with spatial positioning and risk level indication.

[0040] It should be noted that abstract numerical indicators are transformed into an intuitive 3D visualization interface with spatial location information for operators to monitor. This is achieved by using a digital twin 3D visualization engine, which pre-stores a 1:1 scale 3D model of the boiler's heating surfaces. This model includes detailed 3D graphic models of components such as the furnace, water walls, superheater, reheater, economizer, and steam drum.

[0041] The core operation is mapping, which involves associating risk indicators with corresponding components on the 3D model based on the location information of the early warning indicator set, and setting risk level prompts through graphical elements. The specific process is as follows: Spatial Positioning and Model Coloring: The engine first describes the location of "Superheater Section A" based on the indicator set, precisely locating that specific pipe section in the 3D model. The engine assigns a specific color to the component based on the highest risk level at that location (in this case, a "medium warning" for pipe rupture risk). According to preset rules, "medium warning" corresponds to an orange alert. Therefore, the "Superheater Section A" pipe in the 3D model changes from its normal metallic gray to a bright orange. For the "Steam Drum" area, whose expansion risk is "Caution" (usually lower than a warning level), the engine marks it with a light yellow.

[0042] Risk Level Warning: When the operator hovers the mouse cursor over the orange-colored "Superheater Section A," a detailed information prompt box will immediately pop up. The box clearly lists: Risk Type: Tube Rupture Risk; Risk Index: 0.75; Risk Level: Medium Warning; Recommended Action: Strengthen wall temperature monitoring in this area and consider adjusting the burner tilt angle. Additionally, this component may exhibit a slight pulsed light effect to further attract attention.

[0043] Generating a 3D Visual Early Warning Interface: The engine renders all processing results into a complete, real-time interactive 3D visual early warning interface. In this interface, operators can see a complete, realistic 3D model of the boiler. "Superheater Section A" is displayed in orange, "Steam Drum" in light yellow, and the remaining areas are green or gray, representing safety. A legend bar may also be present on one side of the interface, clearly explaining the risk level corresponding to each color (red, orange, yellow, green). Operators can freely rotate and zoom the 3D model to view the specific location of risk points from different angles. Through this process, step 105 successfully transforms abstract numerical risk indicators into an intuitive, precisely located, and information-rich 3D visualization scene, greatly improving operators' perception of the boiler's health status and their risk response speed.

[0044] In this embodiment of the invention, the collected boiler operating temperature, pressure, and velocity field data are transformed into dimensionless boiler state vectors using a dynamic reference scaling method, eliminating the influence of units and integrating thermodynamic and flow field parameters. Time-series data is continuously collected, processed, and stored to form a historical basic dataset covering different load conditions, providing comprehensive and accurate data support for subsequent analysis and overcoming the limitations of traditional data acquisition and processing methods. Based on computational fluid dynamics characteristic mode decomposition technology, combustion fluctuations, convective heat transfer, and heat conduction characteristics are extracted from complex datasets to construct a boiler flow field characteristic parameter set. This transforms raw sensor readings into dynamic characteristic parameters revealing the core physical processes inside the boiler, greatly enhancing the depth of understanding of the boiler's operating state. Finally, a state evolution model based on physical conservation laws is employed, combined with the boiler flow field characteristic parameter set, to construct a dynamic boiler evolution model. This model, based on the laws of conservation of mass, energy, and momentum, and constrained by physical laws, can accurately predict future changes in boiler temperature, pressure, and flow rate, generating multiphysics prediction data. This overcomes the limitations of simple data extrapolation, achieving more scientific and accurate predictions. Based on the principle of non-equilibrium thermodynamic entropy generation, thermodynamic consistency corrections are applied to the multiphysics prediction data, making the predictions more consistent with physical reality. On this basis, risk indicators for tube rupture, scale buildup, and expansion are calculated, generating a set of boiler fault warning indicators. This effectively avoids underestimating risks due to overly "ideal" predicted conditions, improving the accuracy and reliability of risk assessment. Through a digital twin 3D visualization engine, the boiler fault warning indicator set is mapped to a 3D model of the boiler's heating surface, generating a 3D visualized warning interface with spatial positioning and risk level indications. This transforms abstract numerical indicators into intuitive visual scenarios, enabling operators to quickly perceive the boiler's health status, accurately locate risk points, obtain detailed risk information and suggested actions, greatly improving risk response speed and monitoring efficiency.

[0045] Please see Figure 2 Another embodiment of the boiler real-time data prediction system in this invention includes: 201. Acquisition module, used to collect temperature field, pressure field and flow velocity field data of boiler operation, generate boiler state vector through dynamic reference scale method, and establish boiler operation state dataset containing thermodynamic parameters and flow field parameters; Specifically, a distributed sensor network is used to collect boiler tube wall temperature data, steam pressure data, and flue gas velocity data to generate a raw multiphysics monitoring dataset. A dynamic reference scale benchmark based on boiler design parameters is established, including rated temperature benchmark, design pressure benchmark, and standard flow velocity benchmark, generating a dynamic reference scale parameter set. The physical quantities in the raw multiphysics monitoring dataset are continuously scaled with the corresponding dynamic reference scale parameters to generate dimensionless temperature field data, dimensionless pressure field data, and dimensionless flow velocity field data. The dimensionless temperature field data, dimensionless pressure field data, and dimensionless flow velocity field data are integrated according to spatial location and acquisition time to construct a boiler state vector containing spatiotemporal correlation characteristics. Based on the topological relationship of the boiler heating surface structure, the boiler state vector is mapped and associated with the coordinates of the boiler three-dimensional model to establish a boiler operating status dataset with spatial location identification.

[0046] It should be noted that this is based on a typical power plant boiler system, where the boiler is a supercritical coal-fired boiler and the heating surfaces include water-cooled walls, superheaters, reheaters, and economizers.

[0047] A distributed sensor network is deployed on the boiler's heating surfaces. Fifty thermocouples are arranged on the water-cooled walls to measure tube wall temperature, ten pressure sensors are placed at the superheater outlet to monitor steam pressure, and twenty velocity probes are deployed in the flue to collect flue gas velocity. The sensors continuously collect data at 1-second intervals. An example of raw data at a given moment is provided: the temperature at a point on the water-cooled wall is 450 degrees Celsius, the superheater outlet pressure is 16 MPa, and the flow velocity at a point in the flue is 8 m / s. This data constitutes the original multiphysics monitoring dataset, containing tens of thousands of data points.

[0048] Based on the boiler design parameters, the rated temperature baseline is set to 540 degrees Celsius (corresponding to the rated steam temperature), the design pressure baseline is set to 17.5 MPa (corresponding to the design working pressure), and the standard flow velocity baseline is set to 10 m / s (corresponding to the flue gas velocity under rated load). These baseline values ​​constitute a dynamic reference scale parameter set for dimensionless processing.

[0049] Divide each physical quantity by its corresponding baseline value: temperature 450 degrees Celsius divided by 540 degrees Celsius yields a dimensionless temperature of approximately 0.833; pressure 16 MPa divided by 17.5 MPa yields a dimensionless pressure of approximately 0.914; flow velocity 8 m / s divided by 10 m / s yields a dimensionless flow velocity of 0.8. Similar calculations are performed on all data points to generate dimensionless temperature, pressure, and flow velocity field data.

[0050] These dimensionless data points are integrated according to the sensor's spatial location (3D coordinates x=5.2 m, y=3.1 m, z=10.5 m) and the acquisition time (10:30:00 AM, October 1, 2023). The data from each time point are combined into a dimensionless boiler state vector, in the form of [temperature, pressure, flow rate], corresponding to the vector of the example point above as [0.833, 0.914, 0.8]. The vector contains spatiotemporal correlation characteristics, reflecting the overall state of the boiler.

[0051] Based on the topological relationships of the boiler's heating surface structure (pipe connection sequence and spatial layout), dimensionless state vectors are mapped to corresponding coordinates in the boiler's 3D model, and water-cooled wall region vectors are associated with the coordinates of specific pipe segments in the model. Through this mapping, a boiler operating state dataset with spatial location identifiers is established. This dataset can be used for subsequent analysis and is stored as a time-series table, where each row represents the dimensionless state of a sensor point at a certain time.

[0052] 202. Decomposition module, used to extract combustion fluctuation features, convective heat transfer features and heat conduction features from boiler operating status dataset based on computational fluid dynamics characteristic mode decomposition technology, and to construct boiler flow field characteristic parameter set; Specifically, based on the boiler's three-dimensional geometry and operating conditions, a computational fluid dynamics model of the boiler's internal flow field is established, generating a basic framework for flow field characteristic analysis. The intrinsic orthogonal decomposition method is used to perform modal decomposition on this framework, extracting the dominant spatial modes of the boiler's internal flow field and generating a set of characteristic modal basis functions. The dimensionless velocity field data from the boiler operating state dataset is projected onto the set of characteristic modal basis functions, and the time coefficients corresponding to each mode are calculated, generating a spatiotemporal coefficient matrix of the flow field modes. Based on the boiler's internal energy transport mechanism, combustion fluctuation components, convective heat transfer components, and heat conduction components are separated from the spatiotemporal coefficient matrix of the flow field modes, constructing a multi-scale flow field characteristic parameter set. Correlation analysis is used to establish the relationship between the multi-scale flow field characteristic parameter set and the heat load distribution of the boiler's heating surface, forming a boiler flow field characteristic parameter set with clear physical meaning.

[0053] It should be noted that, based on the boiler's precise three-dimensional geometric model (including the furnace, superheaters at each stage, reheaters, etc.) and the current operating conditions (load rate of 90%), we establish a simplified computational fluid dynamics model. This model is not used for real-time full-scale simulation, but rather serves as a physical background framework. It uses preset boundary conditions, setting a standardized inlet velocity at the burner inlet and a reference static pressure at the furnace outlet, thereby generating a framework representing the typical flow field structure of the boiler under current operating conditions.

[0054] We employed the intrinsic orthogonal decomposition method to analyze the flow field (velocity distribution) calculated by the basic CFM model under steady-state conditions. After decomposition, 10 primary spatial modes (Dominant modes) were obtained. These modes are arranged from largest to smallest energy (contribution to the overall flow field variation). The first mode likely represents the mainstream vortex structure in the central region of the furnace, accounting for 60% of the energy; the second mode likely reflects the recirculation characteristics near the water-cooled wall, accounting for 15% of the energy; subsequent modes capture more subtle flow field fluctuations. The graphical data of these 10 spatial modes constitute the flow field characteristic mode basis function set, which describes various typical flow patterns that may exist inside the boiler.

[0055] The dimensionless velocity field data (from sensors distributed throughout the boiler) obtained in step 201 for the latest time period (the past 5 minutes) are projected onto the aforementioned 10 modal basis functions one by one. This process calculates a time coefficient for each mode. At time t1, the time coefficient for the first mode (main vortex) is 1.05, indicating that the intensity of the main vortex in the actual flow field is slightly higher than that of the basic model; the time coefficient for the second mode (recirculation zone) is -0.2, indicating that the recirculation near the water-cooled wall is weaker at this time. Performing this operation for all time points generates a flow field modal spatiotemporal coefficient matrix of 10 (number of modes) multiplied by N (number of time points).

[0056] Based on the energy transport mechanism inside the boiler, we performed a physical decomposition of the spatiotemporal coefficient matrix. We determined that modes with high frequency of time coefficient changes (fluctuation period on the order of seconds) and strong correlation with the burner fuel quantity change signal (modes 3 and 4) mainly reflect combustion fluctuation characteristics. Modes with stable changes (period on the order of minutes) and whose spatial distribution is highly correlated with the position of the heat-receiving surface tube bundle (modes 5 and 6) mainly reflect convective heat transfer characteristics. Modes with extremely slow changes and whose spatial distribution corresponds to the furnace wall structure (modes 9 and 10) are classified as heat conduction characteristics. Accordingly, we separated the subsets corresponding to these three sets of characteristic components from the matrix.

[0057] Correlation analysis revealed a strong positive correlation between the time coefficients representing convective heat transfer characteristics and the dimensionless temperature data (which indirectly reflects heat load) at corresponding locations in the boiler operating status dataset, with correlation coefficients exceeding 0.9. This establishes a clear physical relationship between flow field characteristics and heat load distribution. Ultimately, we integrated these time-series data representing combustion fluctuations, convective heat transfer, and heat conduction characteristics with clear physical significance to construct a boiler flow field characteristic parameter set.

[0058] 203. Construction module, used to construct a dynamic evolution model of boiler by adopting a state evolution model based on the law of physical conservation and combining it with the boiler flow field characteristic parameter set, and generate multi-physics field prediction data including temperature field evolution, pressure field evolution and flow velocity field evolution; Specifically, based on the laws of conservation of mass, energy, and momentum, a system of control equations for the boiler system is established, generating a basic framework for state evolution under physical constraints. Combustion fluctuation characteristics, convective heat transfer characteristics, and heat conduction characteristics from the boiler flow field feature parameter set are embedded into this basic framework to construct a dynamic evolution model of the boiler considering multi-physics coupling effects. Based on the current state parameters in the boiler operating state dataset, the time derivatives of state variables are calculated using the boiler dynamic evolution model to generate a state evolution rate field. A numerical integration method based on physical constraints is used to perform time-progression calculations on the state evolution rate field, generating a multi-physics state evolution sequence for future time periods. The physical consistency of the multi-physics state evolution sequence is verified using the flow field feature parameter set, correcting state data that does not conform to fluid mechanics laws, and generating the final multi-physics prediction data.

[0059] It should be noted that, based on the laws of conservation of mass, energy, and momentum, a simplified system of governing equations is established for the boiler system. In the furnace region, the mass conservation of a control volume unit is reflected in the balance between flue gas inflow and outflow; energy conservation considers combustion heat release, radiative heat transfer, and convective heat transfer; and momentum conservation considers pressure difference, gravity, and viscous forces. This set of equations constitutes the physical framework for state evolution.

[0060] The boiler flow field characteristic parameter set obtained in step 202 is embedded into the above framework. Specifically, the combustion fluctuation characteristic (a time series representing high-frequency fluctuations in combustion intensity) is used as the source term perturbation in the energy conservation equation; the convective heat transfer characteristic (a parameter representing the heat transfer efficiency of flue gas flowing through the superheater tube bundle) is embedded into the coefficients of energy and momentum exchange; and the heat conduction characteristic (a parameter reflecting the steady-state level of heat dissipation from the furnace wall) is used as the boundary condition of the system. In this way, we construct a dynamic evolution model that is both constrained by physical laws and incorporates the specific flow characteristics of the actual boiler.

[0061] The current boiler operating status dataset (T0) shows a dimensionless temperature of 0.92 at a point on the water-cooled wall, a dimensionless pressure of 0.89 at a point on the superheater, and a dimensionless flow velocity of 1.05 at a point on the furnace. These current state parameters are input into a dynamic evolution model. Based on embedded physical laws and characteristics, the model calculates the trend (time derivative) of each state variable at the current instant. It might calculate that the temperature at that water-cooled wall point is rising at a rate of 0.001 dimensionless units per second, while the pressure at that point is decreasing at a rate of 0.0005 units per second. All these instantaneous rate-of-change data together constitute the "state evolution rate field".

[0062] A numerical integration method (Runge-Kutta method) that ensures physical conservation is employed. Starting with the state at time T0, the calculated evolution rate field is used to advance the time step forward, predicting the state within the next 30 seconds, with each second as a time step. At T0+1 seconds, the model predicts that the temperature of the water-cooled wall at that point may become 0.921 and the pressure 0.8895. Through progressive calculation, a multiphysics state evolution sequence is finally generated, containing multiple future time points such as T0+1s, T0+2s, ..., T0+30s.

[0063] The preliminary prediction sequence is validated using the flow field characteristic parameter set (the reasonable range of convective heat transfer characteristics) obtained in step 202. The check reveals that at T0+15 seconds, the flow velocity at a certain prediction point suddenly increases abnormally, violating the flow characteristics of that region. The model will correct this outlier to conform to the laws of fluid mechanics. After this validation and correction, reliable multiphysics prediction data is finally generated.

[0064] 204. Correction module, used to perform thermodynamic consistency correction on multi-physics field prediction data based on the principle of non-equilibrium thermodynamic entropy generation, calculate tube rupture risk index, oxide scale risk index and expansion risk index, and generate boiler fault early warning index set; Specifically, based on non-equilibrium thermodynamics theory, the product of heat flux density and temperature gradient at each location in the multiphysics prediction data is calculated to generate entropy generation rate distribution data. Thermodynamic constraints are corrected on the multiphysics prediction data based on the entropy generation rate distribution data to eliminate predictions that violate the second law of thermodynamics, generating thermodynamically consistent state data. Based on the temperature and stress field data in the thermodynamically consistent state data, the equivalent thermal stress and creep damage accumulation of the heated surface pipe are calculated to generate material damage evolution data. Combining the material damage evolution data and the oxidation kinetics model, the oxide scale growth rate and peeling risk probability of each heated surface region are calculated to generate oxide scale risk distribution data. Integrating the material damage evolution data and the oxide scale risk distribution data, a risk superposition model is used to calculate tube rupture risk indicators, oxide scale risk indicators, and expansion risk indicators, generating a boiler fault early warning index set that includes risk level and spatial location.

[0065] It should be noted that the predicted state at a future moment (5 minutes after the current moment) is extracted from the multiphysics prediction data obtained in step 203. Based on the theory of nonequilibrium thermodynamics, the entropy production rate (σ) is a measure of the irreversibility of a system. For a micro-element, its calculation can be approximated as the heat flux density (q) and temperature gradient. The dot product divided by the square of the temperature, i.e. At a certain predicted point in the high-temperature superheater, the predicted heat flux density q is 500 kW / m², and the temperature gradient is... The predicted entropy production rate is 50 Kelvin / m, with a local temperature T of 810 Kelvin (approximately 537 degrees Celsius). The calculated entropy production rate at this point is approximately 0.038 W / m³ Kelvin³. We perform this calculation on the predicted data for the entire boiler heating surface, generating entropy production rate distribution data. Any region predicted to have a negative entropy production rate (i.e., spontaneous heat transfer from a low temperature to a high temperature, violating the second law of thermodynamics) is identified and corrected by adjusting its temperature gradient to make the entropy production rate non-negative, thus generating thermodynamically consistent state data.

[0066] Based on the corrected thermodynamic consistency data, we performed mechanical calculations. Taking the aforementioned high-temperature superheater prediction point as an example, its predicted temperature is 537 degrees Celsius, and the pipe wall stress is calculated based on the pressure data. Using the material's creep performance data (P91 steel data provided by the ASME standard), the cumulative creep damage generated during continuous operation at this temperature and stress level for a certain period (the predicted risk period is 1000 hours) is calculated using the Larsen-Miller parameter formula. The calculated damage amount is set to 0.015 (i.e., 1.5% of the material's creep life is consumed). Simultaneously, the equivalent thermal stress caused by uneven temperature distribution is calculated to be 85 MPa. These data together constitute the material damage evolution data for this point.

[0067] Combining the temperature field and material damage data described above, an oxidation kinetics model was used. For P91 steel, at 537°C, the oxide scale growth rate is likely approximately 0.1 μm / hour. After 1000 hours of operation, the oxide scale thickness is predicted to increase by approximately 100 μm. According to empirical models, the risk of spalling increases significantly when the oxide scale thickness exceeds a certain threshold (150 μm) or its growth rate becomes too rapid. We calculated the spalling risk probability at this point to be 30%.

[0068] Based on the above data, the final indicators are calculated using a risk superposition model. The tube rupture risk indicator is a comprehensive function of creep damage, equivalent thermal stress level, and oxide scale peeling risk, and may be quantified as 0.65 (high risk). The oxide scale risk indicator is directly related to oxide scale thickness and peeling probability, and is quantified as 0.70 (high risk). The expansion risk indicator focuses on the uneven thermal expansion caused by the predicted temperature difference between or within different heated surface components, and may be quantified as 0.25 (medium risk) at this point. We correlate these risk indicators (values ​​0-1, corresponding to low, medium, and high risks) at all heated surface locations with their spatial coordinates to ultimately generate a boiler fault early warning indicator set, providing direct input for visualized early warning.

[0069] 205. Prediction module, which uses a digital twin 3D visualization engine to map the boiler fault early warning indicator set to the 3D model of the boiler heating surface, and generates a 3D visualization early warning interface with spatial positioning and risk level prompts. Specifically, a three-dimensional geometric model of the boiler's heating surfaces is established, including precise structural models of the water-cooled walls, superheaters, reheaters, and economizers, generating a three-dimensional digital twin basic model of the boiler with spatial topological relationships. Based on the risk level data in the boiler fault early warning indicator set, a risk level-color mapping relationship is established, generating a risk visualization coding scheme with gradient color warnings. The spatial location data in the boiler fault early warning indicator set is coordinate-registered with the boiler's three-dimensional digital twin basic model to generate risk distribution mapping data with precise spatial positioning. Real-time rendering technology is used to combine the risk distribution mapping data with the risk visualization coding scheme to generate a dynamically updated risk heat map on the boiler's three-dimensional digital twin basic model. Early warning time, risk type, and handling suggestions are integrated into the three-dimensional visualization interface to generate an interactive three-dimensional visualization early warning interface, which supports selecting and viewing risk areas and displaying early warning details.

[0070] It should be noted that, using the precise design drawings of the boiler, a detailed model of the heating surfaces, including water-cooled walls, superheaters, reheaters, and economizers, is constructed in the 3D modeling engine. The high-temperature superheater is modeled as a complex structure composed of hundreds of U-shaped tube bundles, each tube having its accurate 3D coordinates and spatial topological relationships, forming a digital base that can be used for visual interaction.

[0071] Based on the boiler fault early warning index set generated in step 204, where risk index values ​​range from 0 to 1, we establish a color mapping relationship: risk values ​​of 0 to 0.3 are mapped to green (low risk), 0.3 to 0.7 to yellow (medium risk), and 0.7 to 1.0 to red (high risk). To achieve a smooth color transition, a linear interpolation formula can be used to calculate the specific RGB color values. For a risk value r, its red component R = 255 in the yellow (255, 255, 0) to red (255, 0, 0) interval, and the green component... The blue component B=0. This generates a color scheme with a gradient warning effect.

[0072] The spatial location data of the early warning indicator set is precisely aligned with the coordinates of the 3D model. The early warning indicator set records the coordinates (x=5.6 m, y=12.3 m, z=3.1 m) of the 8th tube in the 15th row of the high-temperature superheater, with a tube rupture risk index of 0.85. This coordinate and risk value will be accurately associated with the corresponding tube geometry in the 3D model.

[0073] The visualization engine utilizes the capabilities of the graphics processing unit (GPU) to combine the risk distribution mapping data generated in the previous step with a color coding scheme, rendering the colors onto the surface of the boiler's 3D model in real time. Areas with a risk level of 0.85 are displayed in bright red, areas with a risk level of 0.45 are displayed in yellow, and areas with a risk level of 0.1 are displayed in green. The entire boiler model thus presents a dynamically updated risk heat map, making high-risk areas immediately apparent.

[0074] Based on the aforementioned 3D model, interactive elements are integrated. When a user clicks on the bright red tube segment of the high-temperature superheater, a details panel pops up, displaying "Warning Time: 2023-10-26 10:30:00; Risk Type: Tube Rupture Risk (Advanced); Handling Recommendation: Strengthen soot blowing, and it is recommended to shut down for inspection within 72 hours." The interface supports rotation and zooming, facilitating viewing of the risk distribution from different angles, creating a 3D visual early warning interface that integrates early warning, location, and diagnosis.

[0075] 206. The Operation and Maintenance Module is used to establish a knowledge base containing maintenance solutions corresponding to different fault types based on historical maintenance records and boiler equipment operating parameters, and to generate a boiler maintenance strategy library; according to the risk type and risk level in the boiler fault early warning indicator set, it matches the corresponding maintenance solution from the boiler maintenance strategy library to generate targeted maintenance work orders; combined with the power plant maintenance plan and equipment operating status, it optimizes the allocation of resources for maintenance work orders and generates maintenance implementation plans that include personnel configuration, spare parts requirements, and work schedule arrangements; based on the maintenance implementation plan and real-time operating data, it calculates the expected maintenance effect and operating parameter improvement targets, and generates an expected effect evaluation report that includes a comparison of the state before and after maintenance; and displays the maintenance work orders, maintenance implementation plans, and expected effect evaluation reports through a 3D visualized early warning interface, generating a boiler operation and maintenance management interface that integrates early warning, diagnosis, and decision-making.

[0076] It should be noted that a knowledge base has been built based on the power plant's historical maintenance records and equipment parameters over many years. A record in the knowledge base might be defined as follows: When the "risk index of tube rupture in high-temperature superheater tubes" is at the "high risk" level of 0.7 to 0.85, the corresponding maintenance plan is to "arrange a shutdown within 72 hours to conduct macroscopic inspection, thickness measurement, and eddy current testing on the target tube section. The required resources include: 2 boiler engineers, 1 metal supervision specialist, and 5 meters of spare P91 steel pipe." Similarly, specific maintenance plans are pre-set for different risk types (oxide scale, expansion) and levels.

[0077] The boiler fault warning indicator set generated in step 204 shows that the risk index for tube rupture in a certain area of ​​the high-temperature superheater is 0.82 (high risk), and the risk index for oxide scale is 0.75 (high risk). The system automatically matches the corresponding solutions from the maintenance strategy library and generates a maintenance work order. The work order content may be: "Work order number: GT20231026001; Risk location: 15th row on the southeast side of the high-temperature superheater; Risk type: High risk of tube rupture and oxide scale; Recommended measures: Shut down for macroscopic inspection, thickness measurement, and eddy current testing; Recommended completion deadline: Before 10:30 am on October 29, 2023".

[0078] Based on the power plant's plan (there will be a planned load reduction window in the coming week) and the current equipment status, the work order is optimized. It may calculate that the optimal duration for this maintenance is 8 hours, requiring engineers Wang and Zhang, and metal supervisor Li, and automatically reserving 5 meters of P91 steel pipe from the inventory system. The generated maintenance implementation plan is as follows: "Time: October 28, 2023, 02:00 to 10:00; Personnel: Wang (responsible person), Zhang, Li; Spare parts: P91 steel pipe (5 meters); Main procedures: scaffolding erection, macroscopic inspection, thickness measurement, eddy current testing, restoration." Based on real-time operational data (the current average temperature in the area is 535℃) and the maintenance plan (replacing the damaged pipe section), the system predicts that after maintenance, the temperature in the area can be reduced to approximately the design value of 525℃, and the pipe burst risk index can be reduced from 0.82 to 0.15. The report will include a comparative formula to quantify the effect: The calculation showed a total risk reduction of 15.7, which significantly improved safety.

[0079] The highlighted risk areas, pop-up maintenance work order details, Gantt charts of maintenance implementation plans, and expected effect evaluation reports in the 3D visualization early warning interface are all integrated into a single 3D visualization early warning interface. Operators can clearly see at a glance where the risks are, how to handle them, and what the effects will be, thus achieving integrated operation and maintenance management from early warning to decision-making.

[0080] In this embodiment of the invention, a distributed sensor network is used to comprehensively collect multi-physics field data of the boiler. A boiler state vector is generated using the dynamic reference scaling method, establishing a boiler operating state dataset containing spatiotemporal correlation characteristics. This dataset more accurately and comprehensively reflects the boiler operating state, providing a high-quality data foundation for subsequent analysis and helping to promptly identify potential problems. Based on computational fluid dynamics characteristic mode decomposition (EMD) technology, combustion fluctuations, convective heat transfer, and heat conduction features are accurately extracted from the boiler operating state dataset to construct a boiler flow field feature parameter set. This provides a key basis for understanding the boiler operating mechanism and predicting faults. Compared with traditional methods, this approach can more accurately capture flow field changes and improve the accuracy of fault prediction. A state evolution model based on physical conservation laws is employed, combined with the boiler flow field feature parameter set, to construct a dynamic boiler evolution model and generate multi-physics field prediction data. Considering the multiphysics coupling effect, the future state of the boiler can be predicted more scientifically and accurately, providing a reliable reference for operation and maintenance decisions and helping to take measures in advance to avoid failures. Based on the principle of non-equilibrium thermodynamic entropy generation, thermodynamic consistency correction is performed on the multiphysics prediction data, and various risk indicators are calculated to generate a set of boiler fault early warning indicators. This ensures that the prediction data conforms to thermodynamic laws, improves the accuracy and reliability of fault early warning, and can more effectively prevent boiler failures and ensure the safe operation of the boiler. Through a digital twin 3D visualization engine, the set of boiler fault early warning indicators is mapped to a 3D model of the boiler heating surface, generating a 3D visualization early warning interface with spatial positioning and risk level prompts. This enables operators to quickly and accurately locate risk areas, understand risk levels, and take timely countermeasures, improving operation and maintenance efficiency and safety.

[0081] Figure 3 This is a schematic diagram of a boiler real-time data prediction device according to an embodiment of the present invention. The boiler real-time data prediction device 300 can vary considerably due to differences in configuration or performance. The device 300 includes a transmitter 301, a receiver 302, and a processor 303. The processor 303 can also be a controller. Figure 3 The device is designated as "controller / processor 303". Optionally, the device 300 may also include a modem processor 305, which may include an encoder 306, a modulator 307, a decoder 308, and a demodulator 309.

[0082] In one example, transmitter 301 modulates (e.g., analog-to-analog conversion, filtering, amplification, and up-conversion, etc.) the output sample and generates an uplink signal, which is transmitted via an antenna to an access network device. On the downlink, the antenna receives the downlink signal transmitted by the access network device. Receiver 302 modulates (e.g., filtering, amplification, down-conversion, and digitization, etc.) the signal received from the antenna and provides an input sample. In modem processor 305, encoder 306 receives service data and signaling messages to be transmitted on the uplink and processes (e.g., formatting, encoding, and interleaving) the service data and signaling messages. Modulator 307 further processes (e.g., symbol mapping and modulation) the encoded service data and signaling messages and provides an output sample. Demodulator 309 processes (e.g., demodulates) the input sample and provides a symbol estimate. Decoder 308 processes (e.g., deinterleaving and decoding) the symbol estimate and provides decoded data and signaling messages to device 300. Encoder 306, modulator 307, demodulator 309, and decoder 308 can be implemented by a combined modem processor 305. These units perform processing according to the radio access technology adopted by the radio access network (e.g., LTE and other evolved systems access technologies). It should be noted that when device 300 does not include modem processor 305, the above-mentioned functions of modem processor 305 can also be performed by processor 303.

[0083] The processor 303 controls and manages the operation of the device 300, and is used to execute the processing procedures performed by the device 300 in the above embodiments of this disclosure. For example, the processor 303 is also used to execute various steps of the transmitting or receiving device in the above method embodiments, and / or other steps of the technical solutions described in the embodiments of this disclosure.

[0084] Furthermore, the device 300 may also include a memory 304 for storing program code and data for the device 300.

[0085] Understandable Figure 3 Only a simplified design of device 300 is shown. In practical applications, device 300 can include any number of transmitters, receivers, processors, modem processors, memory, etc., and all devices that can implement the embodiments of this disclosure are within the protection scope of the embodiments of this disclosure.

[0086] The present invention also provides a boiler real-time data prediction device, which includes a memory and a processor. The memory stores computer-readable instructions. When the computer-readable instructions are executed by the processor, the processor performs the steps of the boiler real-time data prediction system in the above embodiments.

[0087] The present invention also provides a computer-readable storage medium, which can be a non-volatile computer-readable storage medium or a volatile computer-readable storage medium, wherein the computer-readable storage medium stores instructions that, when the instructions are executed on a computer, cause the computer to perform the steps of the boiler real-time data prediction system.

[0088] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0089] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0090] The above-described embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A boiler real-time data prediction system, characterized in that, The boiler real-time data prediction system includes: The acquisition module is used to collect temperature, pressure and velocity field data of the boiler during operation, generate boiler state vectors through dynamic reference scaling method, and establish boiler operating state dataset. The decomposition module is used to extract combustion fluctuation characteristics, convective heat transfer characteristics, and heat conduction characteristics from the boiler operating status dataset to construct a set of boiler flow field characteristic parameters. The module is used to construct a dynamic evolution model of the boiler by adopting a state evolution model based on the law of physical conservation and combining it with the boiler flow field characteristic parameter set, and to generate multi-physics prediction data. The correction module is used to perform thermodynamic consistency correction based on the multiphysics prediction data, calculate the tube rupture risk index, oxide scale risk index and expansion risk index, and generate a set of boiler fault early warning indicators. The prediction module is used to map the boiler fault early warning index set to the three-dimensional model of the boiler heating surface through a digital twin three-dimensional visualization engine, and generate a three-dimensional visualization early warning interface.

2. The boiler real-time data prediction system according to claim 1, characterized in that, include: Collect boiler tube wall temperature data, steam pressure data, and flue gas velocity data to generate a raw multiphysics monitoring dataset. Establish a dynamic reference scale benchmark based on boiler design parameters, including rated temperature benchmark value, design pressure benchmark value and standard flow rate benchmark value, and generate a dynamic reference scale parameter set; The physical quantity data in the original multiphysics monitoring dataset is continuously scaled with the corresponding dynamic reference scale parameters to generate dimensionless temperature field data, dimensionless pressure field data, and dimensionless flow velocity field data. The dimensionless temperature field data, dimensionless pressure field data, and dimensionless velocity field data are integrated according to spatial location and acquisition time to construct a boiler state vector; Based on the topological relationship of the boiler heating surface structure, the boiler state vector is mapped and associated with the coordinates of the boiler three-dimensional model to establish a boiler operating state dataset.

3. The boiler real-time data prediction system according to claim 2, characterized in that, include: Based on the three-dimensional geometry and operating conditions of the boiler, a computational fluid dynamics model of the internal flow field of the boiler is established to generate a basic framework for flow field characteristic analysis. Modal decomposition is performed on the aforementioned flow field feature analysis infrastructure to extract the dominant spatial modes of the flow field inside the boiler and generate a set of flow field feature mode basis functions; The dimensionless velocity field data in the boiler operating status dataset is projected onto the flow field characteristic mode basis function set, the time coefficients corresponding to each mode are calculated, and the flow field mode spatiotemporal coefficient matrix is ​​generated. Based on the internal energy transport mechanism of the boiler, combustion fluctuation components, convective heat transfer components and heat conduction components are separated from the spatiotemporal coefficient matrix of the flow field modes to construct a multi-scale flow field characteristic parameter set; The correlation between the multi-scale flow field characteristic parameter set and the heat load distribution of the boiler heating surface is established by using correlation analysis, thus forming the boiler flow field characteristic parameter set.

4. The boiler real-time data prediction system according to claim 3, characterized in that, include: Establish a system of control equations for the boiler system and generate a basic framework for state evolution. The combustion fluctuation characteristics, convective heat transfer characteristics, and heat conduction characteristics of the boiler flow field characteristic parameter set are embedded into the state evolution basic framework to construct a boiler dynamic evolution model. Based on the current state parameters in the boiler operating status dataset, the time derivatives of the state variables are calculated using the boiler dynamic evolution model to generate a state evolution rate field. The state evolution rate field is time-progressed to generate a multiphysics state evolution sequence for future time periods; The multiphysics state evolution sequence is verified by using a set of flow field characteristic parameters, and state data that does not conform to the laws of fluid mechanics is corrected to generate multiphysics prediction data.

5. The boiler real-time data prediction system according to claim 4, characterized in that, include: Calculate the product of heat flux density and temperature gradient at each location in the multiphysics prediction data to generate entropy generation rate distribution data; Thermodynamic constraints are corrected for multiphysics prediction data based on entropy generation rate distribution data to eliminate prediction results that violate the second law of thermodynamics and generate thermodynamically consistent state data. Based on the temperature and stress field data in the thermodynamic consistency state data, the equivalent thermal stress and creep damage accumulation of the heated surface pipe are calculated to generate material damage evolution data. Calculate the oxide scale growth rate and peeling risk probability of each heated surface area to generate oxide scale risk distribution data; By combining material damage evolution data and oxide scale risk distribution data, a risk superposition model is used to calculate tube rupture risk indicators, oxide scale risk indicators, and expansion risk indicators, thereby generating a set of boiler fault early warning indicators.

6. The boiler real-time data prediction system according to claim 5, characterized in that, include: Establish a three-dimensional geometric model of the boiler heating surface, including structural models of water-cooled walls, superheaters, reheaters and economizers, and generate a three-dimensional digital twin basic model of the boiler. Based on the risk level data in the boiler fault early warning indicator set, a risk level and color mapping relationship is established to generate a risk visualization coding scheme. The spatial location data of the boiler fault early warning index set is registered with the three-dimensional digital twin basic model of the boiler to generate risk distribution mapping data. The risk distribution mapping data is combined with the risk visualization coding scheme using real-time rendering technology to generate a dynamically updated risk heat map on the three-dimensional digital twin basic model of the boiler. Integrate early warning time, risk type, and handling suggestions into a 3D visualization interface to generate a 3D visualization early warning interface.

7. The boiler real-time data prediction system according to claim 6, characterized in that, Establish a color mapping relationship when the risk value is in the medium-risk range of 0.3 to 0.7: R=255； ； B=0； Where r is the risk index value of boiler failure.

8. The boiler real-time data prediction system according to claim 6, characterized in that, It also includes an operations and maintenance module: Based on historical maintenance records and boiler equipment operating parameters, a knowledge base containing maintenance solutions corresponding to different fault types is established, and a boiler maintenance strategy library is generated. Based on the risk type and risk level in the boiler fault early warning indicator set, the corresponding maintenance plan is matched from the boiler maintenance strategy library to generate a maintenance work order; Based on the power plant maintenance plan and equipment operating status, the maintenance work orders are optimized and allocated to generate maintenance implementation plans; Based on the maintenance implementation plan and real-time operation data, calculate the expected maintenance effect and the target for improving operating parameters, and generate an expected effect evaluation report. The maintenance work orders, overhaul implementation plans, and expected effect evaluation reports are displayed through a 3D visualization early warning interface, generating a boiler operation and maintenance management interface.

9. The boiler real-time data prediction system according to claim 8, characterized in that, The expected results assessment report includes a comparison formula to quantify the results: 。

Images