A method for reconstructing multi-layer flame temperature field distribution of a boiler burner

Abstract

The embodiment of the application discloses a boiler burner multi-layer flame temperature field distribution reconstruction method, which comprises the following steps: setting up a temperature measurement and fire detection hardware and collecting multi-layer flame radiation and full-quantity operation parameters, fusing output spectrum and working condition fusion basic databases through time alignment, using a pre-calibrated coupling coefficient matrix for online iterative decoupling, outputting corrected temperature measurement reference data of each layer, constructing a CFD reduced-order model, correcting simulation deviation through a neural network, outputting high-precision in-furnace temperature field calibration data, three-dimensionally reconstructing the temperature field after layering and smoothing the high-precision in-furnace temperature field calibration data, establishing a prediction model based on historical three-dimensional temperature fields, and outputting final reconstruction data of the multi-layer flame temperature field; early abnormalities are identified, and the boiler is output through a standard interface to generate and execute coal quantity and air quantity control instructions. The problems that multi-layer flame coupling interference is not eliminated and temperature field reconstruction only depends on instantaneous static data are solved.

Description

Technical Field

[0001] This invention relates to the field of boiler burner technology, specifically to a method for reconstructing the temperature field distribution of a multi-layer flame in a boiler burner. Background Technology

[0002] As the core equipment in thermal power generation, the temperature field distribution of the multi-layer burner flame within the furnace of a coal-fired boiler is a key parameter for ensuring combustion safety, improving operational efficiency, and controlling pollutant emissions. Achieving online visualization and precise reconstruction of this temperature field is the core data foundation supporting intelligent optimization and adjustment of boiler combustion.

[0003] Currently, flame temperature detection technologies are mainly divided into two categories: contact and non-contact. However, existing technologies generally suffer from the following two core problems when applied to multi-layer burner scenarios: 1. Unresolved multi-layer flame coupling interference leads to distorted layered temperature reconstruction: Existing methods mostly model individual burners or single-layer flames independently, neglecting the strong thermal radiation cross-interference between upper and lower burner flames, as well as the airflow collision and coupling superposition effects formed by primary and secondary airflows. The fundamental reason is the lack of a model that can quantify the relationship between the radiation perspective factor and the airflow coupling function between each burner layer. This makes it impossible to calculate the true temperature of each flame layer online from the interfering mixed temperature measurement signals, resulting in blurred boundaries and distorted temperature gradients in the reconstructed multi-layer temperature field, making it impossible to accurately distinguish the independent combustion states of each combustion layer and the overburning layer. For example, in one power plant, due to the failure to eliminate radiation interference, the temperature measurement data of the upper flame was systematically overestimated by about 8%, while the overburning layer was underestimated by about 7% due to airflow collision, resulting in a chaotic temperature field hierarchy.

[0004] 2. Temperature field reconstruction relies solely on instantaneous static data, lacking dynamic correction based on historical time-series information, resulting in poor stability. Traditional methods rely solely on static fitting based on the current instantaneous operating conditions, failing to consider the temporal evolution of the temperature field under similar historical conditions. When faced with variable operating conditions such as coal quality fluctuations and rapid load increases or decreases, the reconstructed results exhibit significant numerical drift over time, lacking dynamic continuity and stability, making it difficult to reliably identify persistent anomalies such as flame deflection and combustion center shift. For example, when low-load blending of inferior coal, the static reconstruction method shows a maximum deviation exceeding 12% within one hour, failing to stably track changes in combustion state. Summary of the Invention

[0005] To address the technical problems of unresolved multi-layer flame coupling interference and temperature field reconstruction relying solely on instantaneous static data, this invention provides the following technical solution: A method for reconstructing the temperature field distribution of a multi-layer flame in a boiler burner includes the following steps: S1, build temperature measurement and fire detection hardware and collect multi-layer flame radiation and full operating parameters, and output the spectrum and operating condition fusion basic database after time alignment and fusion. S2, based on the spectrum and working condition fusion database output by S1, performs colorimetric temperature measurement and noise reduction, and uses a pre-calibrated coupling coefficient matrix to perform online iterative decoupling, outputting the corrected temperature measurement reference data for each layer; S3 uses the spectrum and working condition fusion basic database output by S1 and the temperature measurement reference data after correction of each layer output by S2 as inputs to construct a CFD reduced-order model. The simulation deviation is corrected by actual measurement through neural network, and high-precision furnace temperature field calibration data is output. S4 smooths the high-precision furnace temperature field calibration data output by S3 and then reconstructs the three-dimensional temperature field. Based on the historical three-dimensional temperature field, a prediction model is established, and the final reconstructed data of the multi-layer flame temperature field is output. S5, based on the final reconstructed multi-layer flame temperature field data output by S4, identifies early anomalies, outputs them to the boiler through a standard interface, and generates and executes coal and air volume control commands.

[0006] Optionally, the specific steps of S1 are as follows: S11: Configure temperature measurement and flame detection instruments, establish data transmission links, and output compliant flame radiation data transmission channels; S12: Based on the flame radiation data transmission channel output by S11, it synchronously collects narrow-band flame radiation data and full operating parameters of each burner, performs timestamp alignment and data fusion, and outputs a spectrum and operating condition fusion basic database.

[0007] Optionally, the specific steps of S11 are as follows: S111: Install narrow-band photoelectric temperature measurement and flame detection instruments at the measuring points of each layer of burners and overburning layer of the boiler, and output standardized flame detection hardware carriers. S112: A standardized fire detection hardware carrier based on the output of S111, with signal transmission lines laid out and conditioning and anti-interference modules configured to output a compliant flame radiation data transmission channel. The specific steps of S12 are as follows: S121: Based on the flame radiation data transmission channel output by S112, the flame detector probe is controlled to synchronously acquire the flame radiation intensity and scintillation characteristic parameters of each burner in two adjacent narrow bands, and output a multi-layer flame raw spectrum dataset. S122: Based on the multi-layer flame raw spectrum dataset output by S121, all operating parameters are collected synchronously, timestamp alignment and data fusion are performed, and a basic database of spectrum and operating condition fusion is output.

[0008] Optionally, the specific steps of S2 are as follows: S21: Retrieve the spectrum and operating condition fusion basic database output by S12, calculate the initial flame temperature of each single burner using the dual-band colorimetric temperature measurement formula, remove isolated outliers, and output the real-time temperature measurement results of each burner after preliminary purification. S22: Based on the real-time temperature measurement results output by S21, the temperature measurement data of each layer is iteratively solved using the pre-calibrated interlayer radiation and airflow coupling coefficient matrix, and the corrected temperature measurement reference data of each layer with interlayer interference basically eliminated is output.

[0009] Optionally, the specific steps of S21 are as follows: S211: Retrieve the spectrum and operating condition fusion basic database output by S12, use the dual-band colorimetric temperature measurement formula to calculate the initial flame temperature of each single burner from the ratio of the radiation intensity of the two narrow bands, and output the initial temperature measurement value of each burner. S212: Based on the initial temperature measurement values ​​of each burner output by S211, isolated outliers are removed by amplitude limiting filtering or the Laida criterion, and the real-time temperature measurement results of each burner after preliminary purification are output. The specific steps of S22 are as follows: S221: Through experiments or CFD simulations, pre-establish the radiation and airflow coupling coefficient matrix between each layer of burners; S222: Based on the real-time temperature measurement results output by S212 and the coefficient matrix established by S221, a nonlinear equation system is established and solved using Newton and Raphson iterations, outputting the corrected temperature measurement reference data for each layer with basically eliminated interlayer interference.

[0010] Optionally, the specific steps of S3 are as follows: S31: Using the spectrum and operating condition fusion base database output by S12 and the temperature measurement reference data after correction of each layer output by S22 as input, a CFD combustion simulation reference model is constructed and the dominant temperature mode is extracted for order reduction processing to generate an online simulation base model that can be calculated quickly, and outputs a simulation temperature field original database covering a wide range of scenarios. S32: Based on the temperature measurement reference data of each layer after correction output by S22, and combined with the original database of simulated temperature field output by S31, deviation mapping and dynamic weighted correction are performed to output high-precision furnace temperature field calibration data with continuous spatial distribution.

[0011] Optionally, the specific steps of S31 are as follows: S311: Using the spectrum output by S12 and the furnace geometry and operating conditions in the fusion database as boundary conditions, and the corrected temperature reference data of each layer output by S22 as the initial field, a full-scale CFD combustion simulation reference model is constructed. S312: A full-scale CFD combustion simulation benchmark model built on S311. It uses intrinsic orthogonal decomposition to extract the dominant temperature modes, constructs a reduced-order model, and outputs an online simulation base model that can be calculated quickly. S313: Based on the online simulation base model output by S312, it uses the spectrum output by S12 and the real-time and historical typical operating condition parameters of the fusion base database of operating conditions as boundary conditions to perform batch and rapid simulation, and outputs a raw database of simulation temperature fields covering a wide range of scenarios. The specific steps of S32 are as follows: S321: Using the corrected temperature reference data of each layer output by S22 as the target and the original database of the simulated temperature field output by S313 as the input, establish an associated correction model that can map and correct simulation deviations. S322: Based on the correlation correction model output by S321, dynamic weights are set: when the signal-to-noise ratio is high, the measured value is given higher weight; when the probe signal attenuates, the simulation weight is adaptively increased, and the output is a correction model with robust self-adjustment capability. S323: Based on the correction model output by S322, it uses a neural network to identify and filter out systematic measurement biases, and outputs a correction calculation model that has been intelligently anti-interference processed. S324: Based on the corrected calculation model output by S323, it takes the calculation results of the online simulation model with the current working condition boundary conditions as input, eliminates the deviation, and outputs high-precision furnace temperature field calibration data with continuous spatial distribution.

[0012] Optionally, the specific steps of S4 are as follows: S41: Based on the high-precision furnace temperature field calibration data output by S32, after performing layer matching and interlayer smoothing, it is mapped to the boiler three-dimensional mesh model. Through finite element fusion fitting, the current three-dimensional visualized temperature field covering the entire furnace is output. S42: Retrieve historical three-dimensional temperature fields accumulated by S41 under similar historical operating conditions as comparison samples, use their temporal evolution patterns to establish a prediction model, and output the final reconstructed data of multi-layer flame temperature fields.

[0013] Optionally, the specific steps of S41 are as follows: S411: Based on the high-precision furnace temperature field calibration data output by S32, the data is classified into layers according to burner layer, overburning layer and height section, and the basic dataset of layered temperature distribution is output. S412: Based on the basic dataset of layered temperature distribution output by S411, combined with the coefficient matrix, smoothing calibration is performed at the interlayer interface to output layered temperature data in interlayer equilibrium. S413: Based on the interlayer equilibrium layered temperature data output by S412, coordinate mapping is performed with the boiler 3D mesh model to output spatially adapted temperature layer data; S414: Based on the spatially adapted temperature layer data output by S413, finite element interpolation and fusion fitting are used to output a three-dimensional visualized temperature field covering the entire furnace at the current moment. The specific steps of S42 are as follows: S421: Based on the spectrum and operating condition fusion database output by S12, accumulate the historical three-dimensional temperature field of S414, and construct a time-series comparison sample library with operating condition tags; for the current operating condition, retrieve similar historical temperature field evolution sequences of the same type of operating condition. S422: Using the current three-dimensional visualized temperature field output by S414 as the observation value, a prediction model is established using the historical temperature field evolution sequence of the same working condition retrieved by S421. Kalman filtering is used for fusion to output the final reconstructed data of the multi-layer flame temperature field.

[0014] Optionally, the specific steps of S5 are as follows: S51: Based on the final reconstructed data of the multi-layer flame temperature field output by S4, compare it with the standard combustion temperature zone to identify early anomalies and output combustion anomaly alarm signals. S52: Based on the combustion anomaly alarm signal output by S51, combined with the equipment safety red line and emission threshold, outputs a graded combustion risk assessment result; S53: The graded combustion risk assessment results output by S52 and the final reconstructed multi-layer flame temperature field data output by S4 are output to the boiler through a standard protocol interface, supporting both open-loop guidance and closed-loop optimization control modes. S54: Based on the mode and data output by S53, generate and execute coal quantity and air volume control commands.

[0015] Compared with the prior art, the method for reconstructing the multi-layer flame temperature field distribution of a boiler burner provided by the present invention has the following beneficial effects: 1. By implementing online iterative decoupling correction based on the coupling coefficient matrix for the flame temperature measurement data of each layer, it can basically eliminate the interference of interlayer thermal radiation and airflow superposition, and make the boundary of the flame temperature and interlayer gradient reconstruction results of each layer clear and accurately reflect the independent combustion state. 2. By using multi-layer temperature field time series data of similar historical operating conditions to establish a prediction model, the current reconstruction results are optimized and corrected by online filtering. This has the effect of suppressing the time series drift of reconstruction results under varying operating conditions and ensuring the dynamic continuity and long-term comparability of temperature field data during continuous operation. 3. By integrating decoupling correction and online timing calibration, the final output is high-precision and high-stability multi-layer flame three-dimensional temperature field data, which can provide a stable and reliable data benchmark for downstream combustion anomaly diagnosis and closed-loop optimization control, and support the complete closed-loop effect of combustion management from accurate perception to closed-loop optimization. Attached Figure Description

[0016] Figure 1 This is a flowchart illustrating a method for reconstructing the temperature field distribution of a multi-layer flame in a boiler burner, according to an embodiment of this disclosure. Detailed Implementation

[0017] To make the objectives, technical solutions, and advantages of the present invention clearer, the embodiments of the present invention will be described in further detail below.

[0018] like Figure 1 As shown, the present invention provides a method for reconstructing the temperature field distribution of a multi-layer flame in a boiler burner, comprising the following steps: S1: Build temperature measurement and fire detection hardware and collect multi-layer flame radiation and full operating parameters. After time alignment and fusion, output the spectrum and operating condition fusion basic database.

[0019] Specifically, this step includes configuring temperature-measuring flame detectors, establishing data transmission links, and outputting compliant flame radiation data transmission channels; installing photoelectric temperature-measuring flame detectors capable of acquiring flame radiation signals in specific narrow bands at the required measurement points on each burner layer and overburning layer of the boiler, completing the precise layout of all measurement points, and outputting standardized flame detection hardware carriers; based on the standardized flame detection hardware carriers, laying signal transmission lines from each flame detection probe to the back-end intelligent analysis unit, configuring signal conditioning and anti-interference modules, and outputting compliant flame radiation data transmission channels.

[0020] Based on the flame radiation data transmission channel, narrow-band flame radiation data and full operating parameters of each burner are synchronously collected, and timestamp alignment and data fusion are performed to output a spectral and operating condition fusion basic database. Based on the flame radiation data transmission channel, the temperature and flame detection probes are controlled to synchronously collect raw data of flame radiation intensity of each burner in two adjacent narrow bands, and obtain basic characteristic parameters such as flame flicker frequency and amplitude, outputting a multi-layer flame raw spectral dataset. Based on the multi-layer flame raw spectral dataset, full operating parameters such as unit load, industrial analysis data of coal quality entering the furnace, primary air volume, secondary air volume, and coal feed distribution at corresponding times are synchronously collected through the unit DCS interface, and timestamp alignment and binding and data fusion are performed to output a spectral and operating condition fusion basic database.

[0021] S2, based on the spectrum and operating condition fusion database output by S1, performs colorimetric temperature measurement and noise reduction, and uses a pre-calibrated coupling coefficient matrix for online iterative decoupling to output the corrected temperature measurement reference data for each layer.

[0022] Specifically, this step includes: retrieving the spectral and operating condition fusion database; calculating the initial flame temperature of each individual burner using the dual-band colorimetric thermometry formula; removing isolated outliers; and outputting the real-time temperature measurement results of each burner after preliminary purification. Next, retrieving the spectral and operating condition fusion database again; using the dual-band colorimetric thermometry formula derived from Wien's displacement law; calculating the initial flame temperature of each individual burner based on the ratio of the radiation intensities of the two narrow bands; and outputting the initial temperature measurement values ​​of each burner. Based on the initial temperature measurement values ​​of each burner, using fast algorithms such as amplitude limiting filtering or the Laida criterion, isolated outliers caused by instantaneous ash accumulation on the probe or signal jumps are removed; and the real-time temperature measurement results of each burner after preliminary purification are output. Based on the real-time temperature measurement results, using a pre-calibrated interlayer radiation and airflow coupling coefficient matrix, the temperature measurement data of each layer are iteratively solved to output the corrected temperature reference data for each layer, which has largely eliminated interlayer interference. Through specialized experiments or high-precision CFD simulations, a coefficient matrix characterizing the radiation interference and airflow coupling relationships between burners in different layers under a specific furnace structure is pre-established. The specific method involves allowing each burner layer to operate independently in the simulation environment, calculating the radiation perspective factor and convection coupling function of that layer to monitoring points in other layers, and summing all inter-layer coefficients to form a coupling coefficient matrix. This matrix can be adaptively interpolated and retrieved based on macroscopic parameters such as load and airflow. Based on real-time temperature measurement results and the established coefficient matrix, a set of nonlinear equations between the measured temperature vector and the true temperature vector of each layer is established. The Newton and Raphson iterative algorithms are used to solve the equations online, and the interference of interlayer thermal radiation and airflow superposition is removed layer by layer. The corrected temperature measurement reference data of each layer with the interference of interlayer interference is output.

[0023] S3 uses the spectrum and operating condition fusion database output by S1 and the temperature measurement reference data after correction of each layer output by S2 as inputs to construct a CFD reduced-order model. The simulation deviation is corrected by actual measurement through neural network, and high-precision furnace temperature field calibration data is output.

[0024] Specifically, this step includes using a spectral and operating condition fusion database and corrected temperature reference data for each layer as input to construct a CFD combustion simulation benchmark model and extract the dominant temperature modes for order reduction processing to generate an online simulation base model that can be calculated quickly, and outputting a simulation temperature field original database covering a wide range of scenarios; using the furnace geometry and operating conditions in the spectral and operating condition fusion database as boundary conditions and the corrected temperature reference data for each layer as the initial field, embedding gas-solid two-phase flow, pulverized coal combustion and radiative heat transfer models to construct a full-scale CFD combustion simulation benchmark model.

[0025] Based on the constructed full-scale CFD combustion simulation benchmark model, the intrinsic orthogonal decomposition (POD) algorithm is applied to extract several dominant temperature modes that are most sensitive to changes in operating parameters such as load, coal quality, and air distribution. A low-dimensional reduced-order model is constructed, and an online simulation base model that can be calculated quickly is output. Based on the online simulation base model, real-time and historical typical operating parameters from the spectral and operating condition fusion database are dynamically assigned to the model as boundary conditions to drive the reduced-order model to perform batch and rapid simulation calculations, and output a raw database of simulation temperature fields covering a wide range of scenarios.

[0026] Using the calibrated temperature measurement reference data of each layer as a benchmark, and combining it with the original database of the simulated temperature field, deviation mapping and dynamic weighted correction are performed to output high-precision furnace temperature field calibration data with continuous spatial distribution. Using the calibrated temperature measurement reference data of each layer as the training target, and the simulation results of the corresponding working conditions in the original database of the simulated temperature field as input, a neural network machine learning algorithm is used for supervised training to establish an associated correction model that can map and correct simulation deviations. Based on the associated correction model, a dynamic weight adjustment mechanism is set: when the signal-to-noise ratio of the fire detector signal is high, a higher correction weight is assigned to the measured value to calibrate the system deviation of the simulation model; when some probes experience signal attenuation or failure due to contamination or aging, the weight of the simulation model is adaptively increased to compensate, outputting a correction model with robust self-adjustment capabilities. Based on the correction model, the neural network's ability to learn the laws of combustion physics is utilized to identify and mark systematic measurement deviations that contradict the overall temperature field distribution caused by long-term probe contamination and other reasons. The influence of such deviations is automatically reduced or filtered out during model calculation, and the corrected calculation model with intelligent anti-interference processing is output. Based on the corrected calculation model, the constructed online simulation base model is loaded with the current real-time operating condition boundary conditions for online calculation. Using the simulated temperature field as input, the correction model is used to instantly eliminate the deviation between the simulation and the actual combustion conditions, and output high-precision furnace temperature field calibration data with continuous spatial distribution.

[0027] S4 performs layered smoothing of the high-precision furnace temperature field calibration data output by S3, and then reconstructs the three-dimensional temperature field. Based on the historical three-dimensional temperature field, a prediction model is established, and the final reconstructed data of the multi-layer flame temperature field is output.

[0028] Specifically, this step includes: based on high-precision furnace temperature field calibration data, performing layered matching and interlayer smoothing, mapping it to the boiler's 3D mesh model, and outputting a 3D visualized temperature field covering the entire furnace at the current moment through finite element fusion fitting; based on the high-precision furnace temperature field calibration data, classifying it into layers according to burner layer, overburning layer, and different height sections of the furnace, outputting a basic dataset of layered temperature distribution; based on the basic dataset of layered temperature distribution, combining the coefficient matrix, performing data smoothing and deviation calibration at the interface of adjacent layers to ensure continuous temperature gradient without abrupt changes, outputting balanced layered temperature data; based on the balanced layered temperature data, performing coordinate association with the 3D geometric mesh model of the boiler furnace to complete the accurate mapping from discrete measuring points to spatially continuous layers, outputting spatially adapted temperature layer data; based on the spatially adapted temperature layer data, using a finite element interpolation algorithm to perform full-field fusion fitting of multiple layers and multiple cross-sections of the temperature layers, outputting a 3D visualized temperature field covering the entire furnace at the current moment.

[0029] Historical 3D temperature fields accumulated under similar operating conditions are retrieved as comparison samples. A predictive model is established using their temporal evolution patterns, outputting the final reconstructed multi-layer flame temperature field data. Based on operating condition tags in the spectral and operating condition fusion database, 3D visualized temperature field data output at different historical periods are continuously accumulated, constructing a temporal comparison sample library with operating condition tags. For the current real-time operating condition, the most similar historical temperature field evolution sequence under the same operating condition is retrieved from the library. Based on the current 3D visualized temperature field, it serves as the observed value of the system state. Using the retrieved historical temperature field evolution sequence under similar operating conditions, a state equation for the temporal evolution of the temperature field is constructed as a predictive model. A Kalman filter framework is used to optimally fuse the observed and predicted values, effectively suppressing instantaneous jumps and noise under varying operating conditions, outputting the final reconstructed multi-layer flame temperature field data that is continuous, stable, and has long-term comparability in the time dimension.

[0030] S5, based on the final reconstructed multi-layer flame temperature field data output by S4, identifies early anomalies, outputs them to the boiler through a standard interface, and generates and executes coal and air volume control commands.

[0031] Specifically, in this step, based on the final reconstructed data of the multi-layer flame temperature field, it is compared in real time with the preset standard combustion temperature zones of each layer. Utilizing its high precision and stability, it sensitively identifies early anomalies such as flame deviation, combustion center shift, and local overheating, and outputs combustion anomaly alarm signals. Based on the combustion anomaly alarm signals, combined with equipment safety red lines and environmental emission thresholds, the anomaly risk is classified and graded, and a graded combustion risk assessment result is output. Based on the graded combustion risk assessment result, together with the final reconstructed data of the multi-layer flame temperature field, it is output to the boiler DCS through standard protocol interfaces such as OPC, supporting both open-loop guidance for operators and closed-loop optimization control of the system. Based on the mode adaptation status and temperature field data, the DCS or upper-level optimization system uses the final reconstructed temperature as precise feedback, combined with the operating status such as pulverizing and air volume, to generate and execute refined control instructions for coal and air volume distribution, realizing the precise perception of the furnace flame status and the closed-loop optimization of the combustion process.

[0032] The method for reconstructing the temperature field distribution of a multi-layer flame in a boiler burner provided by this invention has the following beneficial effects: 1. By implementing online iterative decoupling correction based on the coupling coefficient matrix for the flame temperature measurement data of each layer, it can basically eliminate the interference of interlayer thermal radiation and airflow superposition, and make the boundary of the flame temperature and interlayer gradient reconstruction results of each layer clear and accurately reflect the independent combustion state. 2. By using multi-layer temperature field time series data of similar historical operating conditions to establish a prediction model, the current reconstruction results are optimized and corrected by online filtering. This has the effect of suppressing the time series drift of reconstruction results under varying operating conditions and ensuring the dynamic continuity and long-term comparability of temperature field data during continuous operation. 3. By integrating decoupling correction and online timing calibration, the final output is high-precision and high-stability multi-layer flame three-dimensional temperature field data, which can provide a stable and reliable data benchmark for downstream combustion anomaly diagnosis and closed-loop optimization control, and support the complete closed-loop effect of combustion management from accurate perception to closed-loop optimization.

[0033] It is understood that the above embodiments are merely exemplary implementations used to illustrate the principles of the present invention, and the present invention is not limited thereto. For those skilled in the art, various modifications and improvements can be made without departing from the spirit and essence of the present invention, and these modifications and improvements are also considered to be within the scope of protection of the present invention.

Claims

1. A method for reconstructing a multi-layer flame temperature field distribution of a boiler burner, characterized in that, Includes the following steps: S1, build temperature measurement and fire detection hardware and collect multi-layer flame radiation and full operating parameters, and output the spectrum and operating condition fusion basic database after time alignment and fusion. S2, based on the spectrum and working condition fusion database output by S1, performs colorimetric temperature measurement and noise reduction, and uses a pre-calibrated coupling coefficient matrix to perform online iterative decoupling, outputting the corrected temperature measurement reference data for each layer; S3 uses the spectrum and working condition fusion basic database output by S1 and the temperature measurement reference data after correction of each layer output by S2 as inputs to construct a CFD reduced-order model. The simulation deviation is corrected by actual measurement through neural network, and high-precision furnace temperature field calibration data is output. S4 smooths the high-precision furnace temperature field calibration data output by S3 and then reconstructs the three-dimensional temperature field. Based on the historical three-dimensional temperature field, a prediction model is established, and the final reconstructed data of the multi-layer flame temperature field is output. S5, based on the final reconstructed multi-layer flame temperature field data output by S4, identifies early anomalies, outputs them to the boiler through a standard interface, and generates and executes coal and air volume control commands.

2. A method of reconstructing a multi-layer flame temperature field distribution of a boiler burner according to claim 1, characterized in that, The specific steps of S1 are as follows: S11: Configure temperature measurement and flame detection instruments, establish data transmission links, and output compliant flame radiation data transmission channels; S12: Based on the flame radiation data transmission channel output by S11, it synchronously collects narrow-band flame radiation data and full operating parameters of each burner, performs timestamp alignment and data fusion, and outputs a spectrum and operating condition fusion basic database.

3. A method of reconstructing a multi-layer flame temperature field distribution of a boiler burner according to claim 2, characterized in that, The specific steps of S11 are as follows: S111: Install narrow-band photoelectric temperature measurement and flame detection instruments at the measuring points of each layer of burners and overburning layer of the boiler, and output standardized flame detection hardware carriers. S112: A standardized fire detection hardware carrier based on the output of S111, with signal transmission lines laid out and conditioning and anti-interference modules configured to output a compliant flame radiation data transmission channel. The specific steps of S12 are as follows: S121: Based on the flame radiation data transmission channel output by S112, the flame detector probe is controlled to synchronously acquire the flame radiation intensity and scintillation characteristic parameters of each burner in two adjacent narrow bands, and output a multi-layer flame raw spectrum dataset. S122: Based on the multi-layer flame raw spectrum dataset output by S121, all operating parameters are collected synchronously, timestamp alignment and data fusion are performed, and a basic database of spectrum and operating condition fusion is output.

4. The method of claim 2, wherein the method further comprises: The specific steps of S2 are as follows: S21: Retrieve the spectrum and operating condition fusion basic database output by S12, calculate the initial flame temperature of each single burner using the dual-band colorimetric temperature measurement formula, remove isolated outliers, and output the real-time temperature measurement results of each burner after preliminary purification. S22: Based on the real-time temperature measurement results output by S21, the temperature measurement data of each layer is iteratively solved using the pre-calibrated interlayer radiation and airflow coupling coefficient matrix, and the corrected temperature measurement reference data of each layer with interlayer interference basically eliminated is output.

5. A method of reconstructing a multi-layer flame temperature field distribution of a boiler burner according to claim 4, characterized in that, The specific steps of S21 are as follows: S211: Retrieve the spectrum and operating condition fusion basic database output by S12, use the dual-band colorimetric temperature measurement formula to calculate the initial flame temperature of each single burner from the ratio of the radiation intensity of the two narrow bands, and output the initial temperature measurement value of each burner. S212: Based on the initial temperature measurement values ​​of each burner output by S211, isolated outliers are removed by amplitude limiting filtering or the Laida criterion, and the real-time temperature measurement results of each burner after preliminary purification are output. The specific steps of S22 are as follows: S221: Through experiments or CFD simulations, pre-establish the radiation and airflow coupling coefficient matrix between each layer of burners; S222: Based on the real-time temperature measurement results output by S212 and the coefficient matrix established by S221, a nonlinear equation system is established and solved using Newton and Raphson iterations, outputting the corrected temperature measurement reference data for each layer with basically eliminated interlayer interference.

6. The method of claim 4, wherein the method further comprises: The specific steps of S3 are as follows: S31: Using the spectrum and operating condition fusion base database output by S12 and the temperature measurement reference data after correction of each layer output by S22 as input, a CFD combustion simulation reference model is constructed and the dominant temperature mode is extracted for order reduction processing to generate an online simulation base model that can be calculated quickly, and outputs a simulation temperature field original database covering a wide range of scenarios. S32: Based on the temperature measurement reference data of each layer after correction output by S22, and combined with the original database of simulated temperature field output by S31, deviation mapping and dynamic weighted correction are performed to output high-precision furnace temperature field calibration data with continuous spatial distribution.

7. The method for reconstructing the temperature field distribution of a multi-layer flame in a boiler burner according to claim 6, characterized in that, The specific steps of S31 are as follows: S311: Using the spectrum output by S12 and the furnace geometry and operating conditions in the fusion database as boundary conditions, and the corrected temperature reference data of each layer output by S22 as the initial field, a full-scale CFD combustion simulation reference model is constructed. S312: A full-scale CFD combustion simulation benchmark model built on S311. It uses intrinsic orthogonal decomposition to extract the dominant temperature modes, constructs a reduced-order model, and outputs an online simulation base model that can be calculated quickly. S313: Based on the online simulation base model output by S312, it uses the spectrum output by S12 and the real-time and historical typical operating condition parameters of the fusion base database of operating conditions as boundary conditions to perform batch and rapid simulation, and outputs a raw database of simulation temperature fields covering a wide range of scenarios. The specific steps of S32 are as follows: S321: Using the corrected temperature reference data of each layer output by S22 as the target and the original database of the simulated temperature field output by S313 as the input, establish an associated correction model that can map and correct simulation deviations. S322: Based on the correlation correction model output by S321, dynamic weights are set: when the signal-to-noise ratio is high, the measured value is given higher weight; when the probe signal attenuates, the simulation weight is adaptively increased, and the output is a correction model with robust self-adjustment capability. S323: Based on the correction model output by S322, it uses a neural network to identify and filter out systematic measurement biases, and outputs a correction calculation model that has been intelligently anti-interference processed. S324: Based on the corrected calculation model output by S323, it takes the calculation results of the online simulation model with the current working condition boundary conditions as input, eliminates the deviation, and outputs high-precision furnace temperature field calibration data with continuous spatial distribution.

8. The method for reconstructing the temperature field distribution of a multi-layer flame in a boiler burner according to claim 6, characterized in that, The specific steps of S4 are as follows: S41: Based on the high-precision furnace temperature field calibration data output by S32, after performing layer matching and interlayer smoothing, it is mapped to the boiler three-dimensional mesh model. Through finite element fusion fitting, the current three-dimensional visualized temperature field covering the entire furnace is output. S42: Retrieve historical three-dimensional temperature fields accumulated by S41 under similar historical operating conditions as comparison samples, use their temporal evolution patterns to establish a prediction model, and output the final reconstructed data of multi-layer flame temperature fields.

9. The method for reconstructing the temperature field distribution of a multi-layer flame in a boiler burner according to claim 8, characterized in that, The specific steps of S41 are as follows: S411: Based on the high-precision furnace temperature field calibration data output by S32, the data is classified into layers according to burner layer, overburning layer and height section, and the basic dataset of layered temperature distribution is output. S412: Based on the basic dataset of layered temperature distribution output by S411, combined with the coefficient matrix, smoothing calibration is performed at the interlayer interface to output layered temperature data in interlayer equilibrium. S413: Based on the interlayer equilibrium layered temperature data output by S412, coordinate mapping is performed with the boiler 3D mesh model to output spatially adapted temperature layer data; S414: Based on the spatially adapted temperature layer data output by S413, finite element interpolation and fusion fitting are used to output a three-dimensional visualized temperature field covering the entire furnace at the current moment. The specific steps of S42 are as follows: S421: Based on the spectrum and operating condition fusion database output by S12, accumulate the historical three-dimensional temperature field of S414, and construct a time-series comparison sample library with operating condition tags; for the current operating condition, retrieve similar historical temperature field evolution sequences of the same type of operating condition. S422: Using the current three-dimensional visualized temperature field output by S414 as the observation value, a prediction model is established using the historical temperature field evolution sequence of the same working condition retrieved by S421. Kalman filtering is used for fusion to output the final reconstructed data of the multi-layer flame temperature field.

10. The method for reconstructing the temperature field distribution of a multi-layer flame in a boiler burner according to claim 1, characterized in that, The specific steps of S5 are as follows: S51: Based on the final reconstructed data of the multi-layer flame temperature field output by S4, compare it with the standard combustion temperature zone to identify early anomalies and output combustion anomaly alarm signals. S52: Based on the combustion anomaly alarm signal output by S51, combined with the equipment safety red line and emission threshold, outputs a graded combustion risk assessment result; S53: The graded combustion risk assessment results output by S52 and the final reconstructed multi-layer flame temperature field data output by S4 are output to the boiler through a standard protocol interface, supporting both open-loop guidance and closed-loop optimization control modes. S54: Based on the mode and data output by S53, generate and execute coal quantity and air volume control commands.

Images