Three-dimensional visual digital twinborn monitoring and early warning method and system for boiler heating surface

Abstract

The invention discloses a three-dimensional visual digital twinborn monitoring and early warning method and system for a boiler heating surface, and relates to the technical field of boiler heating surface monitoring. Constructing a three-dimensional geometric model of the boiler heating surface based on the multi-source data, and carrying out lightweight processing on the three-dimensional geometric model; associating the preprocessed multi-source data with the three-dimensional geometric model, and constructing a heating surface physical simulation model; establishing a data-driven dynamic synchronization mechanism, fusing the operation state parameters and surface state data of the boiler heating surface, which are acquired in real time, with a calculation result of the physical simulation model of the heating surface, and correcting model output by adopting a Kalman filtering algorithm; and constructing a visual interaction platform, and carrying out state monitoring and fault early warning on the heating surface through the visual interaction platform. According to the method, the troubleshooting time is shortened, the operation and maintenance efficiency is improved, the safety risk is reduced, data support is provided for optimized operation of the boiler, and the state of the heating surface is monitored in real time.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of boiler heating surface monitoring, in particular to a boiler heating surface three-dimensional visual digital twin monitoring and early warning method and system. BACKGROUND

[0002] As the core component of energy conversion system, the boiler heating surface is the key carrier for realizing the energy transfer from fuel heat to medium in the power generation and chemical industries. Its operating environment is characterized by high temperature, high pressure, strong corrosion and dynamic scouring, and the components under extreme working conditions for a long time are prone to multiple types of faults, which directly threatens the safe and stable operation of the boiler system.

[0003] The heating surface (including water wall, superheater, reheater, economizer, etc.) needs to continuously withstand the direct scouring of high-temperature flue gas, and the instantaneous temperature near some burners can exceed 1500℃. However, the tube wall is always in a high-temperature range of 350-480℃, resulting in the continuous thermal load on the material. At the same time, the steam or water medium flowing inside the heating surface can reach a pressure of 10-30MPa, and the high-pressure environment makes the tube wall always in a high-stress state, further increasing the risk of material fatigue damage. In addition, the components such as SO2, NOx and fly ash particles in the flue gas will have a dual effect of chemical corrosion and physical scouring on the tube wall. The chemical reaction between the acidic gas and the tube wall metal generates corrosion products, causing the tube wall thickness to gradually thin. The high-speed fly ash particles continuously scour the tube wall surface, especially in local areas such as tube bundle bends and reducers, significantly increasing the wear rate and forming obvious local thinning pits.

[0004] Therefore, in view of the defects of the prior art, how to provide a boiler heating surface three-dimensional visual digital twin monitoring and early warning method and system is a problem that needs to be solved by those skilled in the art. SUMMARY

[0005] Therefore, the present application provides a boiler heating surface three-dimensional visual digital twin monitoring and early warning method and system, which shortens the troubleshooting time, improves the operation and maintenance efficiency, reduces the safety risk, provides data support for the optimized operation of the boiler, and realizes the real-time monitoring of the heating surface state.

[0006] In order to achieve the above purpose, the present application adopts the following technical scheme: a boiler heating surface three-dimensional visual digital twin monitoring and early warning method, comprising: acquiring multi-source data of the boiler heating surface and pre-processing the multi-source data; constructing a three-dimensional geometric model of the boiler heating surface based on the multi-source data and performing lightweight processing on the three-dimensional geometric model; associating the pre-processed multi-source data with the three-dimensional geometric model, integrating an ANSYS physical engine, and constructing a heating surface physical simulation model; A data-driven dynamic synchronization mechanism is established to fuse the real-time collected operating state parameters and surface state data of the boiler heating surface with the calculation results of the heating surface physical simulation model, and a Kalman filtering algorithm is used to correct the model output; A visual interactive platform is constructed to monitor the state of the heating surface and provide early warning of faults through the visual interactive platform.

[0007] Preferably, the real-time collected operating state parameters and surface state data of the boiler heating surface include: temperature and pressure data of key parts of the heating surface collected by distributed temperature sensors and pressure sensors respectively; Medium flow information is collected using an ultrasonic flowmeter; Optical fiber grating sensors are used to monitor thermal stress distribution; The surface state data of the heating surface is obtained by combining image sensors with laser scanning equipment.

[0008] Preferably, the preprocessed multi-source data is associated with a three-dimensional geometric model, an ANSYS physical engine is integrated, and a heating surface physical simulation model is constructed, including: The preprocessed multi-source data is classified and mapped to establish a one-to-one correspondence between the operating state parameters of the boiler heating surface and the corresponding spatial nodes in the three-dimensional geometric model; through coordinate calibration technology, the position of the surface state data in the image or point cloud data is synchronized to the three-dimensional geometric model; When integrating the ANSYS physical engine, the corresponding physical parameter library is configured in the engine according to the material properties of the heating surface; based on the physical parameter library, a multi-physical field coupling simulation model is constructed, multi-physical field data is calculated, and the multi-physical field data is coupled to dynamically simulate the physical state changes of the heating surface under different operating conditions.

[0009] Preferably, when constructing the multi-physical field coupling simulation model, a differentiated meshing strategy is used: for key heat transfer components, a refined meshing method is used; For non-critical heat transfer components, a coarse meshing method is used.

[0010] Preferably, a data-driven dynamic synchronization mechanism is established to fuse the real-time collected operating state parameters and surface state data of the boiler heating surface with the calculation results of the heating surface physical simulation model, including: Multi-source data registration technology is used to calibrate the time and space dimensions of the real-time collected data and the model simulation results: In the time dimension, the timestamp alignment algorithm is used to ensure that the real-time data and the simulation results at the same time are matched; In the spatial dimension, a coordinate system based on a three-dimensional geometric model is used to perform secondary calibration of the spatial position of real-time data and simulation results; Construct a weighted fusion model and assign weights based on the credibility of real-time data and simulation results; The state data of the fused heated surface were obtained through weighted calculation.

[0011] Preferably, the Kalman filter algorithm is used to correct the model output, including: The state parameters output by the physical simulation model of the heated surface are used as the state vector of the Kalman filter, and the multi-source data collected and preprocessed in real time are used as the observation vector. Based on the dynamic characteristics of the physical simulation model, a state transition matrix is ​​constructed; combined with the measurement accuracy and error characteristics of the sensor, an observation matrix and an observation noise covariance matrix are constructed. During the filtering iteration process, the posterior state estimate based on time k-1 is used. and its covariance matrix Predict the prior state and covariance at time k: The Kalman gain matrix is ​​calculated, and the deviation between the prior estimate and the observed vector is fused using the Kalman gain matrix to obtain the corrected posterior state estimate. At the same time, the state error covariance matrix is ​​updated to provide error parameters for the filtering iteration at the next time step.

[0012] Preferably, a three-dimensional visualization digital twin monitoring and early warning system for boiler heating surfaces includes: The data acquisition module is used to acquire multi-source data of the boiler heating surface and preprocess the multi-source data; A three-dimensional geometric model construction module is used to construct a three-dimensional geometric model of the boiler heating surface based on the multi-source data, and to perform lightweight processing on the three-dimensional geometric model; An integration module is used to associate preprocessed multi-source data with a three-dimensional geometric model, integrate the ANSYS physics engine, and build a physical simulation model of the heated surface. The correction module is used to establish a data-driven dynamic synchronization mechanism, which integrates the real-time collected operating status parameters and surface status data of the boiler heating surface with the calculation results of the physical simulation model of the heating surface, and uses the Kalman filter algorithm to correct the model output. The visualization module is used to build a visualization interaction platform, through which the status of the heated surface is monitored and fault warnings are given.

[0013] Via the technical solution, compared with the prior art, the application provides a boiler heating surface three-dimensional visualization digital twin monitoring and early warning method and system, shortens the troubleshooting time, improves the operation and maintenance efficiency, reduces the safety risk, provides data support for boiler optimized operation, and realizes real-time monitoring of the heating surface state. BRIEF DESCRIPTION OF DRAWINGS

[0014] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the drawings needed to be used in the embodiments or prior art description will be briefly introduced as follows. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative labor on the basis of the provided drawings.

[0015] Figure 1 A boiler heating surface three-dimensional visualization digital twin monitoring and early warning method provided by the present application is shown in the flowchart.

[0016] Figure 2 A boiler heating surface three-dimensional visualization digital twin monitoring and early warning system structure provided by the present application is shown in the schematic diagram. DETAILED DESCRIPTION

[0017] The technical solutions in the embodiments of the present application will be described clearly and completely in the following with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only some embodiments of the present application, not all. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor are within the scope of protection of the present application.

[0018] The embodiments of the present application disclose a boiler heating surface three-dimensional visualization digital twin monitoring and early warning method, as shown in Figure 1 The method comprises the following steps: Obtaining multi-source data of the boiler heating surface and pre-processing the multi-source data; Obtaining material attribute data of the boiler heating surface and using a multi-sensor cooperative mode to collect real-time operation state parameters and surface state data of the boiler heating surface; wherein the operation state parameters include temperature, pressure, medium flow and thermal stress distribution data, and the surface state data includes heating surface deformation, corrosion degree, ash deposition thickness and coking range.

[0019] Based on the multi-source data, a three-dimensional geometric model of the boiler heating surface is constructed, and the three-dimensional geometric model is processed in a lightweight manner; Based on the design drawings of the boiler heating surface and the on-site point cloud scanning data, the on-site point cloud scanning data is obtained by using a laser scanner, and a 1:1 three-dimensional geometric model is constructed by using SolidWorks software, which completely restores the size parameters and assembly relationship of each component of the heating surface.

[0020] The on-site shooting surface image of the heating surface is pasted to the surface of the three-dimensional geometric model through texture mapping technology, restoring the appearance characteristics such as pipe wall color, weld texture, and equipment identification; at the same time, the model is optimized for lightweight, by simplifying the grid in non-critical areas and retaining high-precision grids in critical areas, the model accuracy is guaranteed, and the subsequent calculation resource consumption is reduced.

[0021] The preprocessed multi-source data is associated with the three-dimensional geometric model, the ANSYS physical engine is integrated, and the heating surface physical simulation model is constructed; A data-driven dynamic synchronization mechanism is established, the real-time collected operating state parameters and surface state data of the boiler heating surface are fused with the calculation results of the heating surface physical simulation model, and the Kalman filtering algorithm is used to correct the model output; ensure that the state of the virtual model is consistent with the operating state of the entity heating surface; A visual interactive platform is constructed, and the state of the heating surface is monitored and fault warning is realized through the visual interactive platform.

[0022] A visual interactive platform with interactive function is built, through which real-time monitoring and abnormal warning of the heating surface operating state are realized, the platform supports device roaming viewing, monitoring data superimposed display, abnormal hotspot query and fault development process deduction in three-dimensional scene, and has built-in multi-level warning mechanism and fault diagnosis suggestion module.

[0023] The fault diagnosis suggestion module specifically includes: built-in simulation models of typical faults (such as pipe burst, cracking, and coking), input fault parameters (such as pipe wall thinning amount and crack length), and dynamic demonstration of fault development process, intuitive display of the influence of fault on the system; The multi-level warning mechanism specifically includes: based on the heating surface design standard and historical fault data, setting safety thresholds of temperature, pressure, thickness and other parameters, when the monitoring data exceeds the threshold, the platform automatically triggers the warning (first-level warning: sound and light prompt; second-level warning: push message to operation and maintenance personnel; third-level warning: linkage control system sends shutdown suggestion), and marks the fault location on the three-dimensional geometric model, and generates a fault analysis report (including fault type, influence range, and treatment suggestion).

[0024] Specifically, the real-time collected operating state parameters and surface state data of the boiler heating surface include: temperature and pressure data of key parts of the heating surface collected by distributed temperature sensors and pressure sensors respectively; Collecting medium flow information by using an ultrasonic flowmeter; Monitoring thermal stress distribution by using a fiber grating sensor; Through the combination of an image sensor and a laser scanning device, surface state data of the heated surface is obtained to form a multi-dimensional and high-resolution original data set.

[0025] Collecting temperature data of each region of the heated surface by using a thermocouple temperature sensor, with a sampling interval of no more than 1 minute; monitoring the pressure difference inside and outside the tube panel of the heated surface by using a pressure sensor; recording the change of medium flow by using an ultrasonic flowmeter; and realizing real-time monitoring of thermal stress distribution by using a fiber grating sensor. Surface topography information is collected by a high-definition industrial camera and a laser scanner: the industrial camera is installed at a preset position inside the boiler and uses a wide-angle lens to take an overall image of the heated surface; the laser scanner obtains three-dimensional topography data of the heated surface through laser point cloud technology to accurately identify surface defects such as deformation, corrosion, dust accumulation and coking.

[0026] Specifically, the pretreated multi-source data is associated with a three-dimensional geometric model, an ANSYS physical engine is integrated, and a physical simulation model of the heated surface is constructed, including: The pretreated multi-source data is classified and mapped, and a one-to-one correspondence is established between the operating state parameters of the boiler heated surface and the corresponding spatial nodes in the three-dimensional geometric model; the real-time data of each node is accurately matched to the corresponding position of the model; through coordinate calibration technology, the position of the surface state data in the image or point cloud data is synchronized to the three-dimensional geometric model; and accurate positioning of the defect position is realized.

[0027] When integrating the ANSYS physical engine, according to the material properties of the heated surface (such as thermal conductivity, elastic modulus, Poisson's ratio, etc.), the corresponding physical parameter library is configured in the engine; providing basic data support for subsequent physical process simulation; based on the physical parameter library, a multi-physical field coupling simulation model is constructed according to the theories of heat conduction, fluid mechanics and structural mechanics, multi-physical field data is calculated, and the multi-physical field data is coupled and calculated to dynamically simulate the physical state changes of the heated surface under different operating conditions.

[0028] The multi-physical field coupling simulation model includes a heat conduction module, a fluid mechanics module and a structural mechanics module; In the heat conduction module, real-time temperature data is input, combined with the thermal conductivity of the material, the dynamic distribution of the internal temperature field of the heated surface is calculated, and the heat transfer process between the tube wall and each component is simulated; In the fluid mechanics module, medium flow and pressure data are imported, the flow state of the medium inside the heated surface is simulated, and the influence of flow rate changes on the heat transfer efficiency and stress state of the heated surface is analyzed; In the structural mechanics module, combined with the temperature field and pressure data, the stress distribution, strain condition and deformation displacement of each part of the heating surface are calculated, and the mechanical state of the stress concentration areas such as welds and pipe bends is focused on.

[0029] Through the data interaction interface between the modules, real-time sharing and coupled calculation of multi-physical field data are realized, and a complete physical simulation model of the heating surface is formed. The model can dynamically simulate the physical state changes of the heating surface under different operating conditions, and provide simulation basis for subsequent fault prediction and early warning.

[0030] Specifically, in constructing the multi-physical field coupled simulation model, a differentiated mesh division strategy is adopted: for key heat transfer components, a refined mesh division is adopted; For non-critical heat transfer components, a coarse mesh division is adopted.

[0031] In the process of constructing the multi-physical field coupled simulation module, a differentiated mesh division strategy is adopted according to the structural characteristics and operating condition differences of different components of the heating surface. For key heat transfer components such as water wall and superheater, refined mesh division (mesh size controlled at 1-2mm) is adopted to ensure the accuracy of temperature field and stress field calculation; for non-core heat transfer components such as header and support, relatively coarse mesh division (mesh size set to 5-8mm) is adopted to reduce resource consumption while ensuring calculation efficiency. At the same time, to improve the response speed of the simulation model, the calculation process of the physical engine is optimized: parallel computing technology is adopted to divide the multi-physical field coupled calculation task into multiple sub-tasks and assign them to different calculation units for synchronous processing; the simulation time step is dynamically adjusted: in the stable condition stage, the time step is appropriately increased (such as set to 10s) to reduce the calculation amount; in the stage of large fluctuation (such as start-stop furnace, load adjustment), the time step is reduced (such as set to 1s) to ensure the capture of rapidly changing physical state. In addition, a simulation result verification mechanism is established, and the temperature and stress calculation values output by the simulation model are compared with the field measured data regularly. If the deviation exceeds 5%, the physical parameters in the model (such as the correction of thermal conductivity coefficient and friction coefficient) are adjusted until the simulation accuracy meets the requirements.

[0032] Specifically, a data-driven dynamic synchronization mechanism is established to fuse the real-time collected operating state parameters and surface state data of the boiler heating surface with the calculation results of the heating surface physical simulation model, including: A real-time data transmission link is built using 5G or industrial Ethernet technology to transmit the real-time collected operating state parameters and surface state data to the data fusion processing unit at a millisecond level (transmission delay ≤100ms); at the same time, a fixed simulation result output period (consistent with the data collection period) is set to ensure that the calculation results of the physical simulation model can be synchronized to the processing unit on time.

[0033] The multi-source data registration technology is used to calibrate the time and space dimensions of the real-time collected data and the simulation results of the model: In the time dimension, the timestamp alignment algorithm is used to ensure that the real-time data and the simulation results at the same time are matched; In the spatial dimension, based on the coordinate system of the three-dimensional geometric model, the spatial positions of the real-time data and the simulation results are calibrated again; the spatial misalignment problem caused by the installation deviation of the sensor or the error of the simulation model is eliminated.

[0034] A weighted fusion model is constructed, and weights are assigned according to the reliability of the real-time data and the simulation results; for parameters such as temperature and pressure that are less affected by external interference, higher weight (0.7) is given to the real-time data and lower weight (0.3) is given to the simulation results; for parameters such as thermal stress and deformation that are difficult to measure accurately, higher weight (0.6) is given to the simulation results and lower weight (0.4) is given to the real-time data.

[0035] Through weighted calculation, the fused state data of the heated surface is obtained. This data not only retains the timeliness of the real-time data, but also incorporates the systematicness of the simulation results, and can more accurately reflect the actual operating state of the heated surface. Finally, a feedback mechanism for the fused data is established to feed back the fused state data to the three-dimensional geometric model and the physical simulation model in real time, drive the model to update the state, and realize the dynamic synchronization of the virtual model and the physical heated surface.

[0036] Specifically, in the application process of the weighted fusion model, a dynamic weight adjustment mechanism is introduced, and the weight distribution is dynamically optimized by analyzing the data deviation in real time. A deviation threshold (such as 5%) is set, when the deviation between the real-time data and the simulation results is less than the threshold, the initial weight is maintained; when the deviation exceeds the threshold, the weight is automatically adjusted: if the real-time data fluctuates greatly (such as instantaneous temperature mutation), the weight of the real-time data is reduced and the weight of the simulation results is increased; if the deviation between the simulation results and the historical measured data continuously increases, the weight of the real-time data is increased and the weight of the simulation results is decreased, to ensure that the fused data always maintains high accuracy. At the same time, an abnormal data emergency processing sub-module is constructed, when the real-time data appears to be disconnected, abnormally jumps (such as temperature suddenly rises or falls beyond the reasonable range), etc., the sub-module automatically starts the emergency plan: if the data disconnection time is short (≤30s), the simulation results under similar historical conditions are temporarily replaced to maintain the continuous operation of the model; if the data disconnection time is long or the abnormal data cannot be corrected, the data completion algorithm is triggered, based on the trend prediction of the historical data and the simulation model, to generate reasonable replacement data, and mark the data source in the fusion results to remind the operation and maintenance personnel to check the original data abnormality reason in the future.

[0037] Specifically, the Kalman filter algorithm is used to correct the model output, including: Firstly, the state equation and observation equation of Kalman filter are established. The temperature, stress, deformation and other state parameters output by the physical simulation model of the heating surface are taken as the state vector of Kalman filter, and the multi-source data collected and preprocessed in real time are taken as the observation vector; The core state parameters output by the physical simulation model of the boiler heating surface are taken as the state vector x of Kalman filter k , which covers the temperature T, thermal stress σ, deformation displacement d, pipe wall thickness h and medium pressure p of the key monitoring nodes of the heating surface, i.e. x k =[T k ,σ k ,d k ,h k ,p k ] T ; wherein k is the discrete time step (consistent with the sensor sampling period), x k represents the comprehensive state of the heating surface at time k, and x k-1 represents the state at time k-1.

[0038] According to the dynamic characteristics of the physical simulation model, the state transition matrix is constructed to describe the evolution law of the heating surface state from the current time to the next time; combined with the measurement accuracy and error characteristics of the sensor, the observation matrix and the observation noise covariance matrix are constructed to quantify the random error in the measurement process; Considering the continuous change of the heating surface state with time, combined with the dynamic characteristics of the physical simulation model, the state equation is constructed: x k =Ax k -1+w k ; Wherein A represents the state transition matrix, adopts the diagonal matrix A=DIAG(1.02,1.01,1.005,1.0,1.03), and each diagonal element is set according to the parameter change rate (such as the pressure is greatly affected by the working condition fluctuation, and the coefficient is set to 1.03; the pipe wall thickness changes slowly, and the coefficient is set to 1.0), to ensure that the state transition conforms to the actual physical law; w k represents the process noise, and the process noise obeys the zero-mean Gaussian distribution, and its covariance matrix Q k is determined according to the historical simulation error statistics, and has the following form: ; Wherein, the diagonal elements represent the process noise variance of each state parameter.

[0039] The monitoring data collected by multiple sensors in real time are taken as the observation vector z k , including the collected temperature T meas,k , the thermal stress σ meas,k collected by the fiber grating sensor, and the pipe wall thickness h meas,ket al., Construct the observation equation: z k = Hx k + v k ; H represents the observation matrix: adopt full rank matrix to ensure one-to-one correspondence between observation value and state variable, in the form of: ; v k represents observation noise: subject to zero mean Gaussian distribution, and its covariance matrix R k According to the sensor accuracy determination, in the form of: .

[0040] In the filtering iteration process, first, the prediction step: based on the posterior state estimation value at k-1 time and its covariance matrix , predict the prior state and covariance at k time: The prior state estimation is expressed as ; this formula deduces the optimal estimation value at the previous time to the current time through the state transition matrix, to obtain the initial output prediction value of the physical simulation model.

[0041] The prior covariance matrix is expressed as ; wherein, A T is the transpose of the state transition matrix, through which the influence of process noise is fused, and the uncertainty of the prior estimation is quantified.

[0042] Calculate the observation residual e k∣k-1 (the difference between the observation value and the prior prediction value) and its covariance N k∣k-1 , to evaluate the consistency of the observation data and the model prediction; The expression of the observation residual is as follows: ; The expression of the residual covariance is as follows: N k∣k-1 = Hx k∣k-1 x H T + R k ; This matrix integrates the prior covariance and the observation noise covariance, and reflects the statistical characteristics of the residual, providing a basis for subsequent filtering gain calculation.

[0043] Subsequently, an update step is performed: a Kalman gain matrix is calculated, which takes into account the sizes of the state prediction error and the observation error, and when the observation error is small, a higher weight is given to the observation data; when the prediction error is small, a higher weight is given to the prediction result. Using the Kalman gain matrix, the deviation of the prior estimate value from the observation vector is fused and calculated to obtain a corrected posterior state estimate value, which is the model output result after Kalman filtering correction. At the same time, the state error covariance matrix is updated to provide error parameters for the next time filtering iteration.

[0044] The Kalman gain calculation formula is: ; Wherein, is the inverse matrix of the residual covariance matrix. The Kalman gain determines the contribution weight of the observation data to the state correction by balancing the prior estimation error and the observation error (for example, when the observation noise is small, the gain increases, and more depends on the observation value; otherwise, more depends on the model prediction).

[0045] The posterior state estimate value (model output correction result) is represented as: ; This formula corrects the model prior prediction value by the product of the Kalman gain and the residual to obtain the optimal estimate value of the heated surface state at time k, that is, the corrected model output result.

[0046] The posterior covariance matrix is updated as: ; Wherein, I is the unit matrix. The covariance of the state estimate is updated by this formula to quantify the uncertainty of the corrected result and provide initial covariance parameters for the next time filtering iteration.

[0047] Finally, a filtering convergence judgment condition is set. When the state estimate value deviation at consecutive multiple times is less than a set threshold, it is determined that the filtering process converges and the model output correction effect is stable; if the deviation continuously exceeds the threshold, the parameter settings of the state equation and the observation equation, or the reliability of the sensor data are checked, and the model parameters are adjusted in time to ensure that the Kalman filtering algorithm can effectively correct the model output deviation and improve the consistency of the virtual model and the entity heated surface state.

[0048] Specifically, to cope with the change in noise characteristics caused by the working condition fluctuation of the heated surface (such as starting and stopping the furnace, load change), an adaptive factor λ k is introduced to dynamically adjust the process noise covariance Q k : Q k ′=λ k ×Q k . Wherein, λ k is calculated according to the root mean square (RMS) of the residual: ; RMS(e k∣k-1 ) is the root mean square of the current residual, RMS(e avg ) is the average value of the historical residual root mean square (statistical data of nearly 100 time steps). When the working condition fluctuates greatly (such as the residual RMS exceeds the average value), λ k >1, the process noise covariance is increased, and the adaptability of the model to dynamic changes is improved; otherwise, λ k =1, the filtering stability is maintained.

[0049] The Mahalanobis distance is combined to determine whether the observation data is an outlier, such as an abnormal jump caused by sensor failure. If the Mahalanobis distance exceeds the confidence interval (set as the threshold τ α corresponding to 95% confidence), the observation value is discarded, and only the model prediction value is used for state update: ; When γ k∣k-1 >τ α , the Kalman gain K k =0, and the posterior estimate , avoiding the interference of outliers on the correction result.

[0050] The correction effect of Kalman filtering is evaluated by the mean absolute error and the root mean square error: The mean absolute error (MAE) is represented as: ; Where x true,i is the offline detection value of the solid heated surface (such as periodic ultrasonic thickness measurement results); The root mean square error (RMSE) is represented as: .

[0051] Iterative optimization: if the MAE or RMSE exceeds the set threshold, adjust the filtering model parameters.

[0052] Specifically, the diagonal elements of the correction state transition matrix A are modified, and the state transition rule is optimized; The diagonal elements of the process noise covariance Q k or the observation noise covariance R k are adjusted to match the actual error characteristics; The historical data is re-counted, and the confidence threshold τ α of the Mahalanobis distance is updated. Through multiple iterations, the deviation between the output of the corrected model and the state of the solid heated surface is controlled within 5%, meeting the high-fidelity requirements of digital twinning.

[0053] Embodiments of the present application realize multi-dimensional breakthrough in the field of boiler heating surface monitoring and early warning through digital twinning and multi-disciplinary technology fusion, multi-sensor collaborative collection combined with Kalman filter correction and multi-physical field simulation, which greatly reduces the core parameter monitoring error and significantly improves the accuracy compared with the traditional single-point scheme; at the same time, relying on the linkage of high-definition imaging and laser scanning, it can identify tiny surface defects, with a coverage far exceeding manual inspection, solving the problem of defect misjudgment. The virtual-real model dynamic synchronization delay is low, and combined with a multi-level early warning mechanism, it can trigger early warning in advance before the critical state of failure, avoiding the emergency shutdown caused by the alarm lag of the traditional scheme; through dynamic data fusion and abnormal value filtering, the false alarm rate of early warning is greatly reduced, and invalid operation and maintenance work is reduced.

[0054] In one specific embodiment of the present application, a boiler heating surface three-dimensional visual digital twinning monitoring and early warning system, as shown in Figure 2 , comprises: A data acquisition module for acquiring multi-source data of the boiler heating surface and pre-processing the multi-source data; A three-dimensional geometric model construction module for constructing a three-dimensional geometric model of the boiler heating surface based on the multi-source data and performing lightweight processing on the three-dimensional geometric model; An integration module for associating the pre-processed multi-source data with the three-dimensional geometric model, integrating an ANSYS physical engine, and constructing a heating surface physical simulation model; A correction module for establishing a data-driven dynamic synchronization mechanism, fusing the real-time collected operating state parameters and surface state data of the boiler heating surface with the calculation results of the heating surface physical simulation model, and correcting the model output using a Kalman filter algorithm; A visualization module for constructing a visual interactive platform for monitoring and early warning of the heating surface through the visual interactive platform.

[0055] Each embodiment in the specification is described in a progressive manner, and each embodiment focuses on the differences from other embodiments. For the device disclosed in the embodiments, since it corresponds to the method disclosed in the embodiments, the description is relatively simple, and the relevant parts are described in the method part.

[0056] The above description of the disclosed embodiments enables a person skilled in the art to implement or use the present application. Various modifications to the embodiments will be apparent to those skilled in the art, and the general principles defined herein can be implemented in other embodiments without departing from the spirit or scope of the present application. Therefore, the present application will not be limited to the embodiments shown herein, but will conform to the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method for three-dimensional visualization digital twin monitoring and early warning of boiler heating surfaces, characterized in that, include: Acquire multi-source data of the boiler heating surface and preprocess the multi-source data; A three-dimensional geometric model of the boiler heating surface is constructed based on the multi-source data, and the three-dimensional geometric model is then subjected to lightweight processing. The preprocessed multi-source data is associated with the three-dimensional geometric model and integrated with the ANSYS physics engine to construct a physical simulation model of the heated surface. A data-driven dynamic synchronization mechanism is established to integrate the real-time collected operating status parameters and surface status data of the boiler heating surface with the calculation results of the physical simulation model of the heating surface, and the Kalman filter algorithm is used to correct the model output. A visual interactive platform is constructed to monitor the status of the heated surface and provide early warning of faults.

2. The method for three-dimensional visualization digital twin monitoring and early warning of boiler heating surfaces according to claim 1, characterized in that, The operating status parameters and surface condition data of the boiler heating surface are collected in real time, including temperature and pressure data of key parts of the heating surface collected by distributed temperature sensors and pressure sensors respectively. Using an ultrasonic flow meter to collect medium flow information; Fiber Bragg grating sensors are used to monitor thermal stress distribution; By combining an image sensor with a laser scanning device, surface condition data of the heated surface can be obtained.

3. The method for three-dimensional visualization digital twin monitoring and early warning of boiler heating surfaces according to claim 1, characterized in that, The preprocessed multi-source data is associated with the 3D geometric model and integrated with the ANSYS physics engine to construct a physical simulation model of the heated surface, including: The preprocessed multi-source data is classified and mapped to establish a one-to-one correspondence between the operating state parameters of the boiler heating surface and the corresponding spatial nodes in the three-dimensional geometric model; the position of the surface state data in the image or point cloud data is synchronized to the three-dimensional geometric model through coordinate calibration technology. When integrating the ANSYS physics engine, a corresponding physical parameter library is configured in the engine according to the material properties of the heated surface; a multiphysics coupled simulation model is constructed based on the physical parameter library, multiphysics data is calculated, and the multiphysics data is coupled for calculation to dynamically simulate the changes in the physical state of the heated surface under different operating conditions.

4. The method for three-dimensional visualization digital twin monitoring and early warning of boiler heating surfaces according to claim 3, characterized in that, When constructing a multiphysics coupled simulation model, a differentiated mesh generation strategy is adopted: for key heat transfer components, a fine mesh generation method is used; For non-critical heat transfer components, a coarsened mesh is used.

5. The method for three-dimensional visualization digital twin monitoring and early warning of boiler heating surfaces according to claim 1, characterized in that, A data-driven dynamic synchronization mechanism is established to fuse the real-time collected operating status parameters and surface state data of the boiler heating surface with the calculation results of the physical simulation model of the heating surface, including: Multi-source data registration technology is used to calibrate the real-time acquired data and model simulation results in both time and space dimensions. In the time dimension, a timestamp alignment algorithm is used to match real-time data and simulation results at the same moment. In the spatial dimension, a coordinate system based on a three-dimensional geometric model is used to perform secondary calibration of the spatial position of real-time data and simulation results; Construct a weighted fusion model and assign weights based on the credibility of real-time data and simulation results; The state data of the fused heated surface were obtained through weighted calculation.

6. The method for three-dimensional visualization digital twin monitoring and early warning of boiler heating surfaces according to claim 1, characterized in that, The Kalman filter algorithm is used to correct the model output, including: The state parameters output by the physical simulation model of the heated surface are used as the state vector of the Kalman filter, and the multi-source data collected and preprocessed in real time are used as the observation vector. Based on the dynamic characteristics of the physical simulation model, a state transition matrix is ​​constructed; combined with the measurement accuracy and error characteristics of the sensor, an observation matrix and an observation noise covariance matrix are constructed. During the filtering iteration process, the posterior state estimate based on time k-1 is used. and its covariance matrix Predict the prior state and covariance at time k; The Kalman gain matrix is ​​calculated, and the deviation between the prior estimate and the observed vector is fused using the Kalman gain matrix to obtain the corrected posterior state estimate. At the same time, the state error covariance matrix is ​​updated to provide error parameters for the filtering iteration at the next time step.

7. A three-dimensional visualization digital twin monitoring and early warning system for boiler heating surfaces, employing the three-dimensional visualization digital twin monitoring and early warning method for boiler heating surfaces as described in any one of claims 1-6, characterized in that, include: The data acquisition module is used to acquire multi-source data of the boiler heating surface and preprocess the multi-source data; A three-dimensional geometric model construction module is used to construct a three-dimensional geometric model of the boiler heating surface based on the multi-source data, and to perform lightweight processing on the three-dimensional geometric model; An integration module is used to associate preprocessed multi-source data with a three-dimensional geometric model, integrate the ANSYS physics engine, and build a physical simulation model of the heated surface. The correction module is used to establish a data-driven dynamic synchronization mechanism, which integrates the real-time collected operating status parameters and surface status data of the boiler heating surface with the calculation results of the physical simulation model of the heating surface, and uses the Kalman filter algorithm to correct the model output. The visualization module is used to build a visualization interaction platform, through which the status of the heated surface is monitored and fault warnings are given.

Images