Method and system for on-line monitoring of temperature field of thermal power boiler based on acoustic wave temperature measurement

Abstract

The application provides a kind of temperature field on-line monitoring method and system of thermal power boiler based on acoustic temperature measurement, it is related to thermal power boiler temperature monitoring technical field, comprising: obtaining acoustic signal extracts time of flight and attenuation amplitude and inverts linear integral temperature value;According to the smoke flow vector field, the sound wave propagation path deviation is calculated and corrected;Obtain heat release distribution and boundary heat flux density, combine combustion comprehensive constraint to calculate temperature distribution and reconstruct temperature field;Time series correlation analysis is carried out to determine abnormal area and generate optimization suggestion.The application improves the temperature field monitoring precision, enhances the measurement stability, reduces the system interference sensitivity.

Description

Technical Field

[0001] This invention relates to the field of temperature monitoring technology for thermal power boilers, and in particular to an online monitoring method and system for the temperature field of thermal power boilers based on acoustic wave temperature measurement. Background Technology

[0002] As a core component of power generation, the temperature distribution within the combustion chamber of a thermal power boiler directly impacts its operating efficiency, safety performance, and pollutant emissions. Accurate monitoring of the combustion chamber temperature distribution is crucial for optimizing combustion control, extending equipment lifespan, and reducing environmental pollution.

[0003] Traditional temperature field monitoring in thermal power boilers mainly relies on contact-type temperature measuring devices such as thermocouples and resistance temperature detectors (RTDs), which can only obtain temperature information of the combustion chamber wall or local areas, and cannot achieve comprehensive monitoring of the entire combustion chamber temperature field. With the development of acoustic temperature measurement technology, non-contact temperature field monitoring using the propagation characteristics of sound waves has become possible, providing a new technical approach for online temperature field monitoring of thermal power boilers.

[0004] However, existing technologies still have problems such as not fully considering the influence of particulate matter and flue gas flow in the combustion chamber on the sound wave propagation path, resulting in large deviations in temperature measurement results, lack of constraints on the physical processes of combustion, relying solely on acoustic information to reconstruct the temperature field, failing to fully utilize the physical laws of the combustion process, and lacking the ability to intelligently identify and analyze abnormal fluctuations in the temperature field, making it impossible to achieve correlation analysis between the temperature field and combustion control parameters, and making it difficult to provide effective decision support for combustion condition optimization. Summary of the Invention

[0005] This invention provides a method and system for online monitoring of the temperature field of a thermal power boiler based on acoustic temperature measurement, which can at least solve some of the problems existing in the prior art.

[0006] A first aspect of this invention provides a method for online monitoring of the temperature field of a thermal power boiler based on acoustic wave thermometry, comprising: Acquire acoustic signals from the combustion chamber of a thermal power boiler and extract the flight time and attenuation amplitude. Based on the flight time, retrieve the linear integral temperature value. Based on the attenuation amplitude, calculate the attenuation coefficient and construct the particulate matter concentration distribution field to solve the flue gas flow vector field. The spatial offset of the sound wave propagation path is calculated and corrected based on the flue gas flow vector field to obtain an optimized propagation path. The linear integral temperature value is corrected based on the particulate matter concentration distribution field to obtain an optimized linear integral temperature value. Discretized temperature units are obtained by using an adaptive mesh partitioning algorithm based on the optimized propagation path. The spatial distribution law of heat release and the heat flux density of the radiation heat transfer boundary are obtained, and the energy conservation constraint value and the heat transfer boundary constraint value are calculated. The combustion comprehensive constraint is determined by combining the spatial smoothing constraint value corresponding to the flue gas convection heat transfer principle. The temperature value distribution of the discretized temperature unit is calculated by combining the optimized linear integral temperature value and the continuous temperature field is reconstructed. The continuous temperature field is correlated with the pre-acquired historical temperature field data to identify abnormal fluctuation areas. Historical combustion control parameters are obtained to determine the causal relationship between the temperature field response and the combustion control parameters, and the dominant control parameter combination is inferred. The temperature field monitoring results and combustion condition optimization suggestions are generated by combining the abnormal fluctuation areas.

[0007] In one alternative implementation, Acquiring acoustic signals from the combustion chamber of a thermal power boiler and extracting the time of flight and attenuation amplitude, retrieving linear integral temperature values ​​based on the time of flight, calculating attenuation coefficients based on the attenuation amplitude, and constructing a particulate matter concentration distribution field to solve the flue gas flow vector field includes: Sound wave pulses are emitted by a pre-arranged sound wave device, and the propagated sound wave signals are received. The emission time and arrival time are recorded to determine the flight time, and the emission amplitude and reception amplitude are recorded to determine the waveform attenuation amplitude. The average sound velocity is determined based on the flight time and sound wave propagation path length, and then converted into a linear integral temperature value according to the correlation between the molar mass of the flue gas components and temperature. The total attenuation is determined by performing a logarithmic transformation on the waveform attenuation amplitude and combining it with the sound wave propagation path length. Based on the relationship between flue gas density and temperature and the relationship between particulate matter scattering cross section and wavelength, the medium density attenuation coefficient and the particulate matter scattering attenuation coefficient are separated from the total attenuation. The combustion chamber space is divided into grid cells. The particulate concentration of each grid cell is calculated based on the particulate scattering attenuation coefficient to form a particulate concentration distribution field. The flue gas density of each grid cell is calculated based on the medium density attenuation coefficient to construct a flue gas density distribution field. The three-dimensional spatial gradient of the flue gas density distribution field is calculated to obtain the density gradient field. The flue gas flow vector field is calculated by combining the principle of mass conservation.

[0008] In one alternative implementation, The spatial offset of the sound wave propagation path is calculated based on the flue gas flow vector field and corrected to obtain an optimized propagation path. The linear integral temperature value is then corrected based on the particulate matter concentration distribution field to obtain an optimized linear integral temperature value. Based on the optimized propagation path, discretized temperature units are obtained through an adaptive mesh generation algorithm, including: The flue gas velocity vector of each grid cell in the flue gas flow vector field is extracted and line integraled along each sound wave propagation path to obtain the cumulative offset vector. The cumulative offset vector is decomposed to obtain the transverse component perpendicular to the sound wave propagation path, which is used as the spatial offset. The sound wave propagation path is corrected according to the spatial offset to obtain the optimized propagation path. The particle concentration distribution field is linearly integrated along the sound wave propagation path to obtain the particle concentration integral value. The medium non-uniformity correction coefficient is calculated based on the particle concentration integral value, and the optimized linear integral temperature value is obtained by combining the linear integral temperature value. The spatial distance between different optimized propagation paths is calculated and the coverage density of the optimized propagation paths in each region of the combustion chamber is determined. Regions with an optimized propagation path coverage density higher than a preset density threshold are marked as dense regions, and regions with a density density not higher than the density threshold are marked as sparse regions. The dense regions are meshed to obtain refined mesh cells, and the sparse regions are meshed to obtain coarse mesh cells. Based on the optimized propagation paths, the refined mesh cells and coarse mesh cells are corrected to obtain discretized temperature cells.

[0009] In one alternative implementation, The spatial distribution of heat release and the boundary heat flux density of radiative heat transfer are obtained, and the energy conservation constraint value and heat transfer boundary constraint value are calculated. Combined with the spatial smoothing constraint value corresponding to the flue gas convection heat transfer principle, the comprehensive combustion constraints are determined, including: The fuel supply and air supply are obtained, flame radiation images of the combustion chamber wall are acquired and flame temperature is calculated through spectral analysis, fuel heating power is calculated based on the fuel supply, oxygen consumption rate is calculated based on the air supply and flame temperature, and heat release rate of each discretized temperature unit is determined. The spatial distribution law of heat release is obtained based on the heat release rate and fuel heating power. The wall surface heat flux density and wall surface flue gas temperature are obtained. The heat transfer coefficient is calculated based on the wall surface heat flux density. The radiative heat transfer boundary heat flux density is obtained based on the heat transfer coefficient and the wall surface flue gas temperature. The heat transfer boundary constraint value is obtained by combining the temperature gradient heat flux density of adjacent discretized temperature units. The heat release rate of each discretized temperature unit is extracted as a heat source term. The flue gas flow velocity vector is extracted and combined with the temperature gradient corresponding to the optimized linear integral temperature value to obtain the convection term. Based on the heat source term and the convection term, and combined with the thermal conductivity in the furnace, the energy conservation constraint value is obtained. The spatial gradient is calculated based on the temperature difference and spacing between adjacent discretized temperature units, and the curvature value is obtained by performing second-order differentiation. The curvature value is then smoothed to obtain the spatial smoothing constraint value. Combustion comprehensive constraint is obtained by combining the energy conservation constraint value and the heat transfer boundary constraint value.

[0010] In one alternative implementation, The temperature distribution of the discretized temperature unit is calculated by combining the optimized linear integral temperature values, and the continuous temperature field is reconstructed, including: Extract the optimized linear integral temperature value corresponding to each sound wave propagation path, discretize the sound wave propagation path into multiple path segments and calculate the overlap length between each path segment and the corresponding discretized temperature unit. Based on the overlap length, decompose the optimized linear integral temperature into the temperature contribution of each discretized temperature unit and calculate the cumulative contribution. Initialize the temperature value of each discretized temperature unit, calculate the physical deviation based on the combustion comprehensive constraint and the current temperature value to obtain the physical adjustment amount, calculate the measurement deviation based on the cumulative contribution and the current temperature value to obtain the measurement adjustment amount, solve for the comprehensive adjustment amount based on the physical adjustment amount and the measurement adjustment amount and apply it to the corresponding discretized temperature unit, repeat the adjustment until the physical deviation and the measurement deviation converge to obtain the temperature value distribution corresponding to the discretized temperature unit; The spatial coordinates and temperature values ​​of each discretized temperature unit in the temperature value distribution are extracted. A three-dimensional mesh is constructed based on the spatial coordinates, and mesh nodes are determined. The temperature values ​​are assigned to the corresponding mesh nodes. The spatial positions between adjacent mesh nodes are interpolated based on the temperature values ​​of the mesh nodes to obtain interpolated temperatures. The temperature values ​​of the mesh nodes and the interpolated temperatures are combined to obtain a continuous temperature field.

[0011] In one alternative implementation, The determination of abnormal fluctuation regions involves performing time-series correlation analysis between the continuous temperature field and pre-acquired historical temperature field data, including: Extract the temperature values ​​at each spatial location in the continuous temperature field and align them temporally with the historical temperature sequences corresponding to the spatial locations in the pre-acquired historical temperature field data to obtain temporally aligned data. The temperature deviation is calculated between the temperature value at each spatial location in the time-series aligned data and the historical temperature sequence. Based on the temperature deviation, the deviation distribution of each spatial location is constructed. Spatial clustering is performed on the deviation distribution to identify regions where the deviation exceeds a preset range and is spatially continuous, thus obtaining candidate regions. Historical temperature sequences of each spatial location in the candidate region are extracted and the corresponding temporal variation trend is calculated. The temperature values ​​of each spatial location in the candidate region in the continuous temperature field are matched with the temporal variation trend and the trend conformity is calculated. The candidate region is then filtered based on the trend conformity and a pre-set conformity threshold to obtain abnormal fluctuation regions.

[0012] In one alternative implementation, Historical combustion control parameters are obtained to determine the causal relationship between temperature field response and combustion control parameters, and the dominant control parameter combination is deduced. Combined with the abnormal fluctuation region, temperature field monitoring results and combustion condition optimization suggestions are generated, including: Historical combustion control parameters and corresponding historical temperature field data are obtained. The temperature change at each spatial location in the historical temperature field data is extracted and aligned with the historical combustion control parameters to obtain associated data. The cross-correlation strength between historical combustion control parameters and temperature changes in the associated data under different time delays is calculated, and the optimal time delay is identified. Based on the optimal time delay, the global response of each historical combustion control parameter is calculated. The global response is nonlinearly decoupled and decomposed to extract the interaction components, and the coupling strength between different historical combustion control parameters is calculated. The coupling strength is eigenvalue decomposed to extract the dominant feature vector and mapped back to the historical combustion control parameter space to obtain the dominant control parameter combination. The historical combustion control parameters are calculated to respond to temperature disturbances in the abnormal fluctuation region under optimal time delay conditions, and the disturbance contribution is determined. The historical combustion control parameter with the largest disturbance contribution is used as the driving parameter. The historical regulation trajectory of the dominant control parameter combination and the temperature evolution trajectory of the abnormal fluctuation region are extracted and dynamically matched in time to obtain the decoupling deviation period. The historical combustion control parameter combination with the largest coupling strength in the decoupling deviation period is used as the coordination parameter. Combined with the driving parameter, the temperature field monitoring results and combustion condition optimization suggestions are obtained.

[0013] A second aspect of the present invention provides an online temperature field monitoring system for thermal power boilers based on acoustic wave temperature measurement, comprising: The acoustic temperature measurement module is used to acquire acoustic signals in the combustion chamber of a thermal power boiler and extract the flight time and attenuation amplitude. Based on the flight time, the linear integral temperature value is retrieved. Based on the attenuation amplitude, the attenuation coefficient is calculated and a particulate matter concentration distribution field is constructed to solve the flue gas flow vector field. The path temperature optimization module is used to calculate the spatial offset of the sound wave propagation path based on the flue gas flow vector field and correct it to obtain an optimized propagation path. Based on the particulate matter concentration distribution field, the linear integral temperature value is corrected to obtain an optimized linear integral temperature value. Based on the optimized propagation path, discretized temperature units are obtained by using an adaptive mesh partitioning algorithm. The temperature field reconstruction module is used to obtain the spatial distribution law of heat release and the heat flux density of radiation heat transfer boundary, and calculate the energy conservation constraint value and the heat transfer boundary constraint value. Combined with the spatial smoothing constraint value corresponding to the flue gas convection heat transfer principle, the combustion comprehensive constraint is determined. Combined with the optimized linear integral temperature value, the temperature value distribution of the discretized temperature unit is calculated and reconstructed to obtain the continuous temperature field. The anomaly identification module is used to perform time-series correlation analysis between the continuous temperature field and the pre-acquired historical temperature field data to determine the abnormal fluctuation area, obtain historical combustion control parameters to determine the causal relationship between the temperature field response and the combustion control parameters, and infer the dominant control parameter combination. Combined with the abnormal fluctuation area, it generates temperature field monitoring results and combustion condition optimization suggestions.

[0014] A third aspect of the present invention provides an electronic device, comprising: A processor and a memory for storing processor-executable instructions, wherein the processor is configured to invoke instructions stored in the memory to perform the aforementioned method.

[0015] A fourth aspect of the present invention provides a computer-readable storage medium having stored thereon computer program instructions that, when executed by a processor, implement the aforementioned method.

[0016] In this invention, by combining the sound wave flight time and waveform attenuation amplitude, not only is the linear integral temperature value obtained, but the flue gas flow vector field can also be calculated, solving the measurement deviation problem caused by ignoring the influence of the flow field. By introducing spatial offset correction and particulate matter concentration distribution field correction, the sound wave propagation path and linear integral temperature value are optimized, significantly improving the temperature measurement accuracy. It is particularly suitable for environments with high ash content and high concentration of particulate matter. By adopting an adaptive mesh generation algorithm and combining comprehensive combustion constraints such as energy conservation constraints, heat transfer boundary constraints, and spatial smoothing constraints, the discretized temperature unit is accurately reconstructed into a continuous temperature field, overcoming the problem of inaccurate boundaries. Through temporal correlation analysis and causal correlation reasoning, a mapping relationship between abnormal temperature field fluctuations and combustion control parameters is established, which can intelligently identify abnormal areas and provide targeted combustion condition optimization suggestions. Attached Figure Description

[0017] Figure 1 This is a schematic flowchart of an embodiment of the present invention for an online monitoring method of temperature field in a thermal power boiler based on acoustic wave thermometry; Figure 2 This is a flowchart illustrating the processing of the flue gas flow vector field and particulate matter concentration distribution field in the online monitoring method for temperature field of thermal power boilers based on acoustic temperature measurement, as described in an embodiment of the present invention. Detailed Implementation

[0018] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0019] The technical solution of the present invention will be described in detail below with reference to specific embodiments. These specific embodiments can be combined with each other, and the same or similar concepts or processes may not be described again in some embodiments.

[0020] Figure 1 This is a flowchart illustrating the online temperature field monitoring method for thermal power boilers based on acoustic wave temperature measurement, as described in an embodiment of the present invention. Figure 1 As shown, the method includes: Acquire acoustic signals from the combustion chamber of a thermal power boiler and extract the flight time and attenuation amplitude. Based on the flight time, retrieve the linear integral temperature value. Based on the attenuation amplitude, calculate the attenuation coefficient and construct the particulate matter concentration distribution field to solve the flue gas flow vector field. The spatial offset of the sound wave propagation path is calculated and corrected based on the flue gas flow vector field to obtain an optimized propagation path. The linear integral temperature value is corrected based on the particulate matter concentration distribution field to obtain an optimized linear integral temperature value. Discretized temperature units are obtained by using an adaptive mesh partitioning algorithm based on the optimized propagation path. The spatial distribution law of heat release and the heat flux density of the radiation heat transfer boundary are obtained, and the energy conservation constraint value and the heat transfer boundary constraint value are calculated. The combustion comprehensive constraint is determined by combining the spatial smoothing constraint value corresponding to the flue gas convection heat transfer principle. The temperature value distribution of the discretized temperature unit is calculated by combining the optimized linear integral temperature value and the continuous temperature field is reconstructed. The continuous temperature field is correlated with the pre-acquired historical temperature field data to identify abnormal fluctuation areas. Historical combustion control parameters are obtained to determine the causal relationship between the temperature field response and the combustion control parameters, and the dominant control parameter combination is inferred. The temperature field monitoring results and combustion condition optimization suggestions are generated by combining the abnormal fluctuation areas.

[0021] In one alternative implementation, Acquiring acoustic signals from the combustion chamber of a thermal power boiler and extracting the time of flight and attenuation amplitude, retrieving linear integral temperature values ​​based on the time of flight, calculating attenuation coefficients based on the attenuation amplitude, and constructing a particulate matter concentration distribution field to solve the flue gas flow vector field includes: Sound wave pulses are emitted by a pre-arranged sound wave device, and the propagated sound wave signals are received. The emission time and arrival time are recorded to determine the flight time, and the emission amplitude and reception amplitude are recorded to determine the waveform attenuation amplitude. The average sound velocity is determined based on the flight time and sound wave propagation path length, and then converted into a linear integral temperature value according to the correlation between the molar mass of the flue gas components and temperature. The total attenuation is determined by performing a logarithmic transformation on the waveform attenuation amplitude and combining it with the sound wave propagation path length. Based on the relationship between flue gas density and temperature and the relationship between particulate matter scattering cross section and wavelength, the medium density attenuation coefficient and the particulate matter scattering attenuation coefficient are separated from the total attenuation. The combustion chamber space is divided into grid cells. The particulate concentration of each grid cell is calculated based on the particulate scattering attenuation coefficient to form a particulate concentration distribution field. The flue gas density of each grid cell is calculated based on the medium density attenuation coefficient to construct a flue gas density distribution field. The three-dimensional spatial gradient of the flue gas density distribution field is calculated to obtain the density gradient field. The flue gas flow vector field is calculated by combining the principle of mass conservation.

[0022] Multiple acoustic wave devices are pre-arranged on the combustion chamber wall of a thermal power boiler. Each device includes a transmitting unit and a receiving unit. The transmitting unit emits acoustic pulses at a frequency of 1000 Hz, and the receiving unit receives the acoustic signals after propagation through the combustion chamber. The acoustic wave devices are evenly distributed along the boiler wall to ensure the formation of a network of acoustic wave propagation paths covering the entire combustion chamber space. For example, in the combustion chamber of a 300 MW thermal power boiler, 12 acoustic wave devices can be installed on the four walls, forming at least 30 effective acoustic wave propagation paths.

[0023] The moment the sound wave is emitted is recorded as time zero, and the moment the receiving unit receives the sound wave signal is recorded as the arrival time. The difference between the two is the sound wave's flight time. Simultaneously, the amplitude of the received sound wave signal and the emitted amplitude are recorded, and the waveform attenuation is calculated. For example, when a sound wave travels 60 meters through the combustion chamber, its flight time is 120 milliseconds, and the received amplitude is 0.3 times the emitted amplitude.

[0024] The average speed of sound is calculated based on the measured flight time and the known path length of the sound wave. The average speed of sound equals the path length divided by the flight time. In the previous example, the average speed of sound was 500 m / s. There is a specific correlation between the speed of sound in flue gas and the molar mass and temperature of the flue gas components. For a typical flue gas composition of a thermal power boiler, a conversion relationship between speed of sound and temperature can be established: temperature is approximately equal to the square of the average speed of sound divided by a constant factor. The constant factor is set according to the actual situation, resulting in an average temperature of 1250 degrees Celsius along the propagation path. The average temperature is considered as a linear integral temperature value along the path and used for subsequent temperature field reconstruction.

[0025] Sound waves attenuate during propagation due to factors including changes in medium density and scattering by particles. By performing a logarithmic transformation on the waveform attenuation amplitude and combining this with the sound wave propagation path length, the total attenuation can be determined. In the previous example, a logarithmic transformation on the attenuation amplitude of 0.3 yields a logarithmic attenuation value of approximately -1.2. Dividing this by the propagation path length of 60 meters gives an attenuation of -0.02 dB / m.

[0026] The total attenuation is separated into media density attenuation and particulate scattering attenuation. Media density attenuation is closely related to flue gas temperature; the higher the temperature, the lower the density, and the smaller the attenuation. Particulate scattering attenuation, on the other hand, is related to particulate concentration and sound wave wavelength. For a known sound wave frequency, an attenuation model can be established to separate these two effects. Assuming the total attenuation is -0.02 dB / m along a certain path, the model calculates that the media density attenuation coefficient is -0.015 dB / m, and the particulate scattering attenuation coefficient is -0.005 dB / m.

[0027] To reconstruct the temperature and particulate matter concentration fields, the combustion chamber space was divided into grid cells. For a 300 MW boiler, this can be achieved using a 10×10×20 cubic grid, with each grid cell having a side length of approximately 1 meter. Based on the particulate matter scattering attenuation coefficients obtained along multiple acoustic paths, algebraic reconstruction techniques were used to calculate the particulate matter concentration in each grid cell. The particulate matter concentration value of each grid cell was adjusted iteratively to minimize the error between the calculated path attenuation and the actual measured value. The reconstructed particulate matter concentration field showed that the pulverized coal concentration in the lower part of the combustion chamber reached 20 g / m³, while the upper part decreased to below 5 g / m³.

[0028] Similarly, the flue gas density distribution field is reconstructed based on the medium density attenuation coefficient. Using the medium density attenuation coefficients obtained from acoustic measurements along multiple paths, an algebraic reconstruction algorithm is used to calculate the flue gas density of each grid cell. In the example boiler, the flue gas density at the bottom of the combustion chamber is approximately 0.5 kg / m³, and in the top region, it is approximately 0.2 kg / m³.

[0029] The reconstructed flue gas density distribution field is subjected to three-dimensional spatial gradient calculation to obtain the density gradient field. The density gradient field shows the rate of change of flue gas density in each grid cell along the three coordinate axes. For example, the upward density gradient in the central region is approximately -0.02 kg / m³ / m, indicating that the density decreases by 0.02 kg / m³ for every 1 meter of upward movement.

[0030] By applying the principle of mass conservation, the flue gas velocity in each grid cell is calculated, yielding the flue gas flow vector field. The relationship between density gradient and velocity is determined based on the continuity equation. While maintaining mass conservation, regions with decreasing density correspond to regions with increasing velocity. By solving the relevant equations, the velocity vector for each grid cell is obtained. For example, the upward airflow velocity in the central region of the combustion chamber is approximately 10 m / s, while the downward airflow velocity at the edge regions may be 3-5 m / s.

[0031] In this embodiment, by simultaneously utilizing the sound wave flight time and waveform attenuation information, the decoupled acquisition of temperature information and medium state information is achieved in the same sound wave measurement process. This avoids the dependence of temperature measurement on optical conditions, probe high-temperature resistance performance, or assumptions about a single physical quantity, and improves the reliability and stability of measurement in high-temperature, high-dust, and strongly disturbed combustion environments. By separating the medium density attenuation and particulate matter scattering attenuation in the sound wave attenuation, synchronous inversion of flue gas density change and particulate matter distribution is achieved, effectively reducing uncertainty and improving the accuracy of pollutant distribution assessment. By performing three-dimensional spatial gradient analysis on the flue gas density distribution field and combining it with the mass conservation relationship to calculate the flue gas flow vector field, the indirect acquisition of the flue gas flow state in the combustion chamber is achieved, enhancing the online identification capability of combustion organization state, local stagnation zones, and abnormal flow structures.

[0032] In one alternative implementation, The spatial offset of the sound wave propagation path is calculated based on the flue gas flow vector field and corrected to obtain an optimized propagation path. The linear integral temperature value is then corrected based on the particulate matter concentration distribution field to obtain an optimized linear integral temperature value. Based on the optimized propagation path, discretized temperature units are obtained through an adaptive mesh generation algorithm, including: The flue gas velocity vector of each grid cell in the flue gas flow vector field is extracted and line integraled along each sound wave propagation path to obtain the cumulative offset vector. The cumulative offset vector is decomposed to obtain the transverse component perpendicular to the sound wave propagation path, which is used as the spatial offset. The sound wave propagation path is corrected according to the spatial offset to obtain the optimized propagation path. The particle concentration distribution field is linearly integrated along the sound wave propagation path to obtain the particle concentration integral value. The medium non-uniformity correction coefficient is calculated based on the particle concentration integral value, and the optimized linear integral temperature value is obtained by combining the linear integral temperature value. The spatial distance between different optimized propagation paths is calculated and the coverage density of the optimized propagation paths in each region of the combustion chamber is determined. Regions with an optimized propagation path coverage density higher than a preset density threshold are marked as dense regions, and regions with a density density not higher than the density threshold are marked as sparse regions. The dense regions are meshed to obtain refined mesh cells, and the sparse regions are meshed to obtain coarse mesh cells. Based on the optimized propagation paths, the refined mesh cells and coarse mesh cells are corrected to obtain discretized temperature cells.

[0033] The flue gas velocity vector of each grid cell is extracted from the flue gas flow vector field to accurately describe the flue gas flow state within the combustion chamber. Taking a 300 MW boiler combustion chamber as an example, the velocity vector in the central rising airflow region is 10 m / s vertically upward, while the velocity vector in the recirculation region near the wall is 3 m / s diagonally downward. The cumulative offset vector is calculated by line integral along each sound wave propagation path; this represents the cumulative offset effect caused by the flue gas flow during sound wave propagation. For a 60-meter-long sound wave propagation path, if the average flue gas velocity along the path is 5 m / s and forms a 45-degree angle with the sound wave propagation direction, the cumulative offset vector is approximately 0.6 meters. The cumulative offset vector is decomposed into a longitudinal component parallel to the sound wave propagation path and a transverse component perpendicular to the propagation path. The longitudinal component affects the sound wave propagation time, while the transverse component causes a spatial offset in the sound wave propagation path. In the aforementioned example, the transverse component is approximately 0.4 meters, representing the transverse offset of the actual sound wave propagation path relative to the assumed straight path.

[0034] The sound wave propagation path is corrected based on the calculated spatial offset to obtain an optimized propagation path. The optimized propagation path is closer to the actual propagation trajectory of the sound wave in the flowing flue gas, exhibiting a slightly curved shape. For example, the propagation path originally assumed to be a straight line may be offset by 0.2 to 0.4 meters in the flow direction in the central region after correction, more accurately reflecting the propagation characteristics of the sound wave in a non-uniform flow field.

[0035] The particulate matter concentration values ​​of each grid cell are extracted from the particulate matter concentration distribution field to describe the spatial distribution of particulate matter within the combustion chamber. The particulate matter concentration is high at the bottom of the boiler combustion chamber near the burner, reaching up to 20 g / m³, gradually decreasing to below 5 g / m³ in the upper region. Line integrals of the particulate matter concentration values ​​are calculated along the optimized acoustic wave propagation path to obtain the integral particulate matter concentration value. For acoustic wave propagation paths passing through high-concentration areas, the integral particulate matter concentration value may reach 600 g / m², while the integral value for paths mainly passing through low-concentration areas may only be 150 g / m².

[0036] The medium inhomogeneity correction coefficient is calculated based on the integral value of particulate matter concentration. The inhomogeneity of particulate matter concentration distribution can cause scattering effects on sound wave propagation, thus affecting temperature measurement accuracy. The medium inhomogeneity correction coefficient has an approximately non-linear relationship with the integral value of particulate matter concentration. When the integral value of particulate matter concentration is 200 g / m², the correction coefficient is approximately 0.95; when the integral value increases to 500 g / m², the correction coefficient decreases to 0.85, indicating that sound wave propagation is subject to stronger scattering. The medium inhomogeneity correction coefficient is combined with the previously obtained linear integral temperature value, and the optimized linear integral temperature value is obtained through iterative calculation. For example, with an original linear integral temperature value of 1250 degrees Celsius and a correction coefficient of 0.92, the calculated optimized linear integral temperature value is approximately 1160 degrees Celsius, more accurately reflecting the actual temperature.

[0037] Calculate the spatial distance between each optimized propagation path and evaluate the path coverage density in each region of the combustion chamber. For a 300 MW boiler combustion chamber, divide the space into several regions of 10m × 10m × 20m and calculate the number of optimized propagation paths passing through each region. Typically, the path coverage density is higher in the central region of the combustion chamber, with 0.5 to 1 path per cubic meter of space, while the path coverage density is lower in the corner regions, with only 0.1 to 0.2 paths per cubic meter of space.

[0038] The optimized propagation path coverage density is compared with a preset density threshold, typically set at 0.3 paths / cubic meter. Areas with a coverage density higher than the threshold are marked as dense areas, while areas with a density lower than the threshold are classified as sparse areas. In the combustion chamber, the central combustion zone and flame development zone are typically dense areas, while the corner areas and areas near the exit are often sparse areas.

[0039] Mesh refinement is performed on densely populated areas, refining the original 1m×1m×1m mesh into 0.5m×0.5m×0.5m mesh cells to improve the accuracy of temperature field reconstruction. Mesh merging is performed on sparsely populated areas, merging the original mesh into coarser 2m×2m×2m mesh cells to reduce computational load and smooth out noise effects. Adaptive meshing technology enables the rational allocation of computational resources, focusing on improving the accuracy of temperature field reconstruction in key areas.

[0040] Discretized temperature elements are obtained by modifying the refined and coarse mesh elements based on optimized propagation paths. Algebraic reconstruction techniques are then used to assign the optimized linear integral temperature values ​​along each propagation path to the mesh elements it traverses. For refined mesh elements in dense regions, the reconstruction accuracy can reach ±20 degrees Celsius due to sufficient path coverage; for coarse mesh elements in sparse regions, the reconstruction accuracy is approximately ±50 degrees Celsius, which still meets the requirements of engineering applications.

[0041] In this embodiment, by using the flue gas flow vector field to perform line integral analysis on the sound wave propagation process and correcting the spatial offset of the sound wave propagation path, the refraction and offset errors of the sound wave caused by the flue gas flow can be effectively compensated, making the sound wave propagation path closer to the real propagation trajectory. This significantly improves the spatial accuracy of the inversion results based on the sound wave flight time and attenuation information. By integrating the particulate matter concentration along the optimized sound wave propagation path and constructing a medium non-uniformity correction coefficient, the linear integral temperature value is corrected, so that the temperature inversion results can explicitly reflect the influence of particulate matter distribution and medium non-uniformity on the sound wave propagation characteristics. This improves the reliability and stability of temperature measurement in high dust and highly non-uniform combustion environments. By analyzing and optimizing the coverage density of the propagation path in the combustion chamber space, and accordingly using mesh densification or mesh merging for adaptive discretization of different regions, the problem of local over-coarseness or overall over-density in fixed mesh division is effectively alleviated, achieving a balance between accuracy and computational efficiency.

[0042] Figure 2 This is a flowchart illustrating the processing of the flue gas flow vector field and particulate matter concentration distribution field in the online monitoring method for temperature field of thermal power boilers based on acoustic temperature measurement, as described in an embodiment of the present invention.

[0043] In one alternative implementation, The spatial distribution of heat release and the boundary heat flux density of radiative heat transfer are obtained, and the energy conservation constraint value and heat transfer boundary constraint value are calculated. Combined with the spatial smoothing constraint value corresponding to the flue gas convection heat transfer principle, the comprehensive combustion constraints are determined, including: The fuel supply and air supply are obtained, flame radiation images of the combustion chamber wall are acquired and flame temperature is calculated through spectral analysis, fuel heating power is calculated based on the fuel supply, oxygen consumption rate is calculated based on the air supply and flame temperature, and heat release rate of each discretized temperature unit is determined. The spatial distribution law of heat release is obtained based on the heat release rate and fuel heating power. The wall surface heat flux density and wall surface flue gas temperature are obtained. The heat transfer coefficient is calculated based on the wall surface heat flux density. The radiative heat transfer boundary heat flux density is obtained based on the heat transfer coefficient and the wall surface flue gas temperature. The heat transfer boundary constraint value is obtained by combining the temperature gradient heat flux density of adjacent discretized temperature units. The heat release rate of each discretized temperature unit is extracted as a heat source term. The flue gas flow velocity vector is extracted and combined with the temperature gradient corresponding to the optimized linear integral temperature value to obtain the convection term. Based on the heat source term and the convection term, and combined with the thermal conductivity in the furnace, the energy conservation constraint value is obtained. The spatial gradient is calculated based on the temperature difference and spacing between adjacent discretized temperature units, and the curvature value is obtained by performing second-order differentiation. The curvature value is then smoothed to obtain the spatial smoothing constraint value. Combustion comprehensive constraint is obtained by combining the energy conservation constraint value and the heat transfer boundary constraint value.

[0044] Real-time fuel and air supply data are acquired from the boiler control system. For a 300 MW thermal power boiler, the typical pulverized coal supply is approximately 30 tons / hour, and the air supply is approximately 300,000 standard cubic meters / hour. Flame radiation images of the combustion chamber are acquired using a flame monitoring camera mounted on the boiler wall. The image resolution is 1280×960 pixels, and the frame rate is 25 frames / second. Spectral analysis is performed on the acquired flame images to extract radiation spectrum curves in the wavelength range of 400 to 900 nanometers. The flame temperature is calculated by analyzing the characteristic peak values ​​and intensity ratios of the radiation spectrum. For example, in the main combustion zone, the flame temperature obtained from the intensity ratio analysis of hydrocarbon combustion characteristic spectral lines is approximately 1600 degrees Celsius, while the temperature in the secondary combustion zone is approximately 1300 degrees Celsius.

[0045] The fuel heating power is calculated based on the fuel supply and corresponding lower heating value. Taking bituminous coal as an example, with a lower heating value of 23,000 kJ / kg and a fuel supply of 30 tons / hour, the calculated fuel heating power is 192 MW. This value represents the total heat entering the boiler, providing a benchmark for heat balance analysis. The oxygen consumption rate is calculated based on the air supply and flame temperature. Under an air supply of 300,000 standard cubic meters / hour, the oxygen input is approximately 63,000 standard cubic meters / hour. Combining flame temperature and combustion kinetics, the calculated oxygen consumption rate in the main combustion zone is approximately 15,000 standard cubic meters / hour·m³, while that in the secondary combustion zone is approximately 5,000 standard cubic meters / hour·m³.

[0046] The heat release rate of each discretized temperature unit was determined based on the oxygen consumption rate. During pulverized coal combustion, approximately 12,000 kJ of heat is released for every 1 standard cubic meter of oxygen consumed. Therefore, the heat release rate in the main combustion zone is approximately 180,000 kJ / h·m³, and in the secondary combustion zone, it is approximately 60,000 kJ / h·m³. Based on a comparative analysis of the heat release rate and the fuel's heating power, the spatial distribution of heat release was determined. The results show that approximately 70% of the heat is released in the main combustion zone, 20% in the secondary combustion zone, and the remaining 10% in the burnout zone.

[0047] Heat flux density data of the boiler wall was obtained using a wall heat flux meter. At the water-cooled wall corresponding to the main combustion zone, the heat flux density was approximately 150 kW / m², while at the water-cooled wall corresponding to the secondary combustion zone, the heat flux density was approximately 100 kW / m². The flue gas temperature near the wall was obtained based on an optimized propagation path. Using the sound wave propagation path near the wall, the flue gas temperature near the main combustion zone was measured to be approximately 1400°C, and near the secondary combustion zone, it was approximately 1200°C. Combining the wall heat flux density and flue gas temperature, a logarithmic temperature distribution model was used to calculate the heat transfer coefficient. The heat transfer coefficient corresponding to the main combustion zone was approximately 150 W / m²·°C, and the corresponding coefficient for the secondary combustion zone was approximately 110 W / m²·°C.

[0048] The radiative heat transfer boundary heat flux density is calculated based on the heat transfer coefficient and the flue gas temperature at the wall. Considering that the boiler primarily experiences radiative heat transfer, the calculated heat transfer coefficient, combined with the Stepan-Boltzmann law, yields a radiative heat transfer boundary heat flux density of approximately 140 kW / m² for the main combustion zone and approximately 90 kW / m² for the secondary combustion zone. Simultaneously, the heat flux density is calculated by incorporating the temperature gradient between adjacent discretized temperature cells. The temperature gradient is typically radial, reaching up to 100°C / cm² near the wall. Combining the radiative heat transfer heat flux density with the temperature gradient heat flux density yields the heat transfer boundary constraint value, which is used to constrain the temperature field reconstruction process.

[0049] The heat release rate of each discretized temperature unit is extracted as a heat source term, representing the heat released per unit volume. The heat source term is spatially non-uniformly distributed, reaching 180,000 kJ / h·m³ in the main combustion zone, while areas far from the burner may only have 20,000 kJ / h·m³. The flue gas velocity vector is extracted and combined with the temperature gradient corresponding to the optimized linear integral temperature value to solve for the convection term. The flue gas velocity can reach 10 m / s in the central rising region, with a temperature gradient of approximately 20°C / m, resulting in a calculated convective heat transfer of approximately 12,000 kJ / h·m³.

[0050] Based on the heat source and convection terms, and combined with the pre-obtained thermal conductivity within the furnace, the energy conservation constraint value is solved. The thermal conductivity of the flue gas within the furnace varies with temperature, reaching approximately 0.15 W / m·°C at 1500°C. By solving the energy conservation equation, the energy input and output balance within each discretized temperature unit is ensured, yielding the energy conservation constraint value, which is used to correct the temperature field distribution.

[0051] The spatial gradient is calculated based on the temperature difference and spacing between adjacent discretized temperature units. For example, the temperature difference between the main combustion region and the surrounding region can reach 300 degrees Celsius, with a spacing of 1 meter, resulting in a spatial gradient of 300 degrees Celsius / meter. The second derivative of the spatial gradient is used to calculate the curvature value, representing the degree of bending of the temperature field. In the flame boundary region, the curvature value is relatively large, reaching 50 degrees Celsius / square meter, while in the temperature-uniform region, the curvature value is close to 0. The curvature value is smoothed using a Gaussian filtering algorithm with a filtering radius of 1.5 meters to obtain a smoothed curvature field, from which the spatial smoothing constraint value is calculated.

[0052] The energy conservation constraint, heat transfer boundary constraint, and spatial smoothing constraint are weighted and fused to obtain the comprehensive combustion constraint. The weight of the energy conservation constraint is set to 0.6, the weight of the heat transfer boundary constraint is 0.3, and the weight of the spatial smoothing constraint is 0.1.

[0053] In this embodiment, by combining the fuel supply, lower heating value, and flame temperature obtained from flame radiation image inversion, a spatial correlation between fuel heating power and local heat release rate is established. This ensures that the heat release distribution is consistent with the actual combustion intensity and fuel-air matching state, improving the accuracy and reliability of combustion intensity distribution identification. By combining wall flue gas temperature, wall heat flux density, and heat transfer coefficient in the construction of heat transfer boundary constraints, the combustion chamber boundary conditions can dynamically reflect the actual heat transfer state, avoiding systematic deviations caused by fixed or simplified boundary conditions. This enhances the continuity and physical rationality of the furnace temperature field near the heated surface, improving the ability to identify local overheating and abnormal heat transfer areas. By using the heat release rate of the discretized temperature unit as a heat source term and combining it with the flue gas flow velocity vector to construct a convection term, energy conservation constraints are introduced to perform overall consistency correction of the temperature field. This ensures that the inverted temperature distribution not only meets the measurement information constraints but also conforms to the actual energy transfer process within the furnace, effectively reducing non-physical fluctuations and energy imbalance problems.

[0054] In one alternative implementation, The temperature distribution of the discretized temperature unit is calculated by combining the optimized linear integral temperature values, and the continuous temperature field is reconstructed, including: Extract the optimized linear integral temperature value corresponding to each sound wave propagation path, discretize the sound wave propagation path into multiple path segments and calculate the overlap length between each path segment and the corresponding discretized temperature unit. Based on the overlap length, decompose the optimized linear integral temperature into the temperature contribution of each discretized temperature unit and calculate the cumulative contribution. Initialize the temperature value of each discretized temperature unit, calculate the physical deviation based on the combustion comprehensive constraint and the current temperature value to obtain the physical adjustment amount, calculate the measurement deviation based on the cumulative contribution and the current temperature value to obtain the measurement adjustment amount, solve for the comprehensive adjustment amount based on the physical adjustment amount and the measurement adjustment amount and apply it to the corresponding discretized temperature unit, repeat the adjustment until the physical deviation and the measurement deviation converge to obtain the temperature value distribution corresponding to the discretized temperature unit; The spatial coordinates and temperature values ​​of each discretized temperature unit in the temperature value distribution are extracted. A three-dimensional mesh is constructed based on the spatial coordinates, and mesh nodes are determined. The temperature values ​​are assigned to the corresponding mesh nodes. The spatial positions between adjacent mesh nodes are interpolated based on the temperature values ​​of the mesh nodes to obtain interpolated temperatures. The temperature values ​​of the mesh nodes and the interpolated temperatures are combined to obtain a continuous temperature field.

[0055] The optimized linear integral temperature values ​​corresponding to each acoustic wave propagation path are extracted from the measurement system. For a 300 MW thermal power boiler, 60 to 80 acoustic wave propagation paths are arranged to cover the main area of ​​the combustion chamber. Each acoustic wave propagation path has a unique identifier and a corresponding optimized linear integral temperature value. For example, the optimized linear integral temperature value of an acoustic wave propagation path passing through the main combustion zone might be 1450 degrees Celsius, while that of a path passing through the secondary combustion zone might be 1200 degrees Celsius, and that of a path passing through the burnout zone might be 900 degrees Celsius. The acoustic wave propagation paths are discretized into multiple path segments, with the discretization accuracy determined according to the grid size. For an acoustic wave propagation path with a length of 40 meters, if a discretization accuracy of 1 meter is used, it is discretized into 40 path segments. For each path segment, the overlap length with the discretized temperature cells it passes through is calculated. Since the acoustic wave propagation path does not strictly travel along the grid lines, a single path segment may overlap with multiple discretized temperature cells simultaneously. For example, a path segment with a length of 1 meter may overlap with the main discretized temperature unit it passes through by 0.8 meters and with the adjacent unit by 0.2 meters.

[0056] The optimized linear integral temperature is decomposed into the temperature contribution of each discretized temperature element based on the overlap length. The larger the overlap length, the greater the contribution of the discretized temperature element to the linear integral temperature. For a sound wave propagation path with an optimized linear integral temperature of 1450 degrees Celsius, if the overlap length between a discretized temperature element and the path accounts for 5% of the total path length, then the contribution of that element to the linear integral temperature is 72.5 degrees Celsius. The temperature contributions of the same discretized temperature element on all sound wave propagation paths passing through it are summarized to obtain the corresponding cumulative contribution. For discretized temperature elements located in the intersection region of multiple paths, the cumulative contribution value is larger and the constraint is stronger; while for elements located in the edge region, the cumulative contribution is smaller and the constraint is weaker.

[0057] Initialize the temperature value of each discretized temperature unit. Initialization can be done using a uniform distribution or an empirically based hierarchical distribution. A uniform distribution initializes all units to the same temperature value, such as 1000 degrees Celsius. A hierarchical distribution allocates different initial temperatures based on spatial height, such as initializing the bottom region to 1300 degrees Celsius, the middle region to 1100 degrees Celsius, and the upper region to 900 degrees Celsius. Calculate the physical deviations based on the combustion constraints and the current temperature values. Physical deviations include energy conservation deviations, heat transfer boundary deviations, and spatial smoothing deviations. For energy conservation deviations, if a discretized temperature unit has a heat source term of 150,000 kJ / h·m³, a convection term of 10,000 kJ / h·m³, and a heat conduction term of 20,000 kJ / h·m³, energy conservation requires that the algebraic sum of these three terms be zero. If the calculated result is 120,000 kJ / h·m³, then an energy conservation deviation exists.

[0058] The physical adjustment is calculated based on the physical deviation. The physical adjustment is proportional to the physical deviation, with a proportionality coefficient typically set between 0.5 and 0.8. For example, taking energy conservation deviation as an example, if the deviation is 120,000 kJ / h·m³ and the proportionality coefficient is 0.6, the corresponding temperature adjustment is approximately 30 degrees Celsius. The measurement deviation is calculated based on the cumulative contribution and the current temperature value. The measurement deviation reflects the difference between the calculated linear integral temperature and the measured optimized linear integral temperature under the current temperature distribution. If the calculated linear integral temperature for a certain sound wave propagation path under the current temperature distribution is 1380 degrees Celsius, while the measured value is 1450 degrees Celsius, then the measurement deviation is 70 degrees Celsius. The measurement adjustment is calculated based on the measurement deviation. The measurement adjustment is allocated to each unit according to the overlap length ratio between the sound wave propagation path and each discretized temperature unit. If the overlap length ratio of a unit on the path is 5%, then the measurement adjustment obtained by that unit is 3.5 degrees Celsius.

[0059] The comprehensive adjustment is calculated based on the physical and measurement adjustment values. The comprehensive adjustment is a weighted sum of the physical and measurement adjustment values, with the weights dynamically adjusted according to reliability. Typically, in the initial iteration phase, the measurement adjustment has a higher weight (0.7), while the physical adjustment has a weight of 0.3. As iterations progress, the weight of the physical adjustment gradually increases to 0.5. The comprehensive adjustment is applied to the corresponding discretized temperature unit, updating the temperature value. This adjustment process is repeated until both the physical and measurement deviations converge to below a preset threshold. Typically, the physical deviation convergence threshold is set to 3% of the initial value, and the measurement deviation convergence threshold is set to 5 degrees Celsius. After approximately 20 to 30 iterations, the temperature value distribution of the discretized temperature unit tends to stabilize, yielding a temperature field solution that satisfies both physical and measurement constraints.

[0060] Extract the spatial coordinates and temperature values ​​of each discretized temperature cell in the temperature value distribution. The spatial coordinates of a discretized temperature cell are usually represented by its center point. For example, if the center point of a cell is 10 meters horizontally, 8 meters vertically, and 15 meters high, the corresponding temperature value is 1350 degrees Celsius. Construct a three-dimensional mesh based on the spatial coordinates and determine the mesh nodes. For boiler combustion chambers, a structured mesh is typically used, with horizontal and vertical spacing of 0.5 to 2 meters and vertical spacing of 1 to 3 meters. Assign the temperature values ​​of the discretized temperature cells to the corresponding mesh nodes. If the center of a discretized temperature cell does not coincide with the location of a mesh node, an inverse distance weighted interpolation method is used to calculate the node temperature value. For example, if a mesh node is surrounded by four discretized temperature cells with temperatures of 1300, 1350, 1280, and 1320 degrees Celsius, and distances of 0.6, 0.8, 0.5, and 0.7 meters respectively, the temperature value of that node is approximately 1310 degrees Celsius.

[0061] Interpolated temperatures are obtained by interpolating the spatial positions of adjacent grid nodes based on their temperature values. Trilinear interpolation or spline interpolation methods are used to ensure a smooth spatial transition of the temperature field. For example, for two adjacent grid nodes with temperatures of 1310°C and 1350°C respectively, and a distance of 1 meter, the interpolated temperature at their midpoint is 1330°C. Combining the grid node temperatures with the interpolated temperatures yields a continuous temperature field.

[0062] In this embodiment, by decomposing the optimized linear integral temperature along the sound wave propagation path into an accumulated contribution to each discretized temperature unit, the measurement information can be accurately mapped to the spatial unit in a weighted manner. This improves the utilization efficiency and spatial directivity of the sound wave measurement information in temperature field inversion. By simultaneously introducing the physical adjustment amount formed by combustion comprehensive constraints and the measurement adjustment amount formed by sound wave measurement consistency, the temperature values ​​of the discretized temperature units are jointly iteratively corrected. This ensures that the final temperature distribution satisfies both the multi-path sound wave measurement results and the physical laws such as energy conservation, heat transfer boundary, and spatial smoothness. This effectively suppresses non-physical temperature fluctuations caused by measurement noise and ill-posed inversion, significantly improving the stability and reliability of the temperature inversion results. Through the dual-bias convergence criterion in the iteration process, the synchronous convergence of temperature values ​​in both measurement consistency and physical consistency dimensions is guaranteed, reducing the risk of local overfitting of measurement data or excessive reliance on physical models. This gives the temperature field results better robustness and adaptability under different operating conditions.

[0063] In one alternative implementation, The determination of abnormal fluctuation regions involves performing time-series correlation analysis between the continuous temperature field and pre-acquired historical temperature field data, including: Extract the temperature values ​​at each spatial location in the continuous temperature field and align them temporally with the historical temperature sequences corresponding to the spatial locations in the pre-acquired historical temperature field data to obtain temporally aligned data. The temperature deviation is calculated between the temperature value at each spatial location in the time-series aligned data and the historical temperature sequence. Based on the temperature deviation, the deviation distribution of each spatial location is constructed. Spatial clustering is performed on the deviation distribution to identify regions where the deviation exceeds a preset range and is spatially continuous, thus obtaining candidate regions. Historical temperature sequences of each spatial location in the candidate region are extracted and the corresponding temporal variation trend is calculated. The temperature values ​​of each spatial location in the candidate region in the continuous temperature field are matched with the temporal variation trend and the trend conformity is calculated. The candidate region is then filtered based on the trend conformity and a pre-set conformity threshold to obtain abnormal fluctuation regions.

[0064] Temperature data for each spatial location is extracted from the continuous temperature field. For a 300 MW thermal power boiler combustion chamber, the continuous temperature field typically contains tens of thousands of spatial sampling points, with a sampling density of 4 to 8 points per cubic meter in key monitoring areas. Each spatial location is represented by three-dimensional coordinates, such as 10.5 meters horizontally, 8.2 meters vertically, and 15.3 meters high, corresponding to a real-time temperature of 1320 degrees Celsius. Historical temperature data corresponding to the current continuous temperature field is retrieved from the historical temperature field database. The historical database records temperature changes at each spatial location over a past period, typically retaining data from the most recent 168 hours, with a sampling interval of 10 minutes. For the aforementioned spatial location, the historical temperature sequence might be 1310 degrees Celsius one hour ago, 1315 degrees Celsius two hours ago, 1305 degrees Celsius three hours ago, etc.

[0065] The current continuous temperature field is time-series aligned with historical temperature fields. This alignment considers factors such as boiler load variations and fuel characteristic changes to ensure that the compared time points have similar or identical operating conditions. The alignment method employs a dynamic time warping algorithm, determining the alignment window based on key parameters such as the boiler load curve and main steam temperature curve. For example, if the current boiler load is 270 MW, time points in historical data with loads between 260 and 280 MW are selected for comparison. The resulting time-series aligned data includes spatial location, current temperature value, and corresponding historical temperature sequences, forming a complete spatiotemporal temperature dataset.

[0066] Temperature deviation is calculated for each spatial location in the time-series aligned data and compared to the historical temperature sequence. The temperature deviation is calculated by dividing the difference between the current temperature value and the average of the historical temperature sequence by the standard deviation of the historical temperature fluctuations. For a spatial location with a historical average temperature of 1310 degrees Celsius and a standard deviation of 15 degrees Celsius, if the current temperature is 1350 degrees Celsius, the temperature deviation is 2.67, meaning the current temperature deviates from the historical average by 2.67 standard deviations. A larger temperature deviation indicates a greater deviation from the historical normal range and a higher probability of an anomaly.

[0067] A deviation distribution for each spatial location is constructed based on temperature deviation. This deviation distribution is presented as a scalar field in three-dimensional space, intuitively displaying the spatial distribution of temperature anomalies. Spatial clustering is performed on the deviation distribution to identify regions where the deviation exceeds a preset range and is spatially continuous. Density clustering is used for spatial clustering, with key parameters including a deviation threshold and a spatial continuity threshold. The deviation threshold is typically set to 2.0 to 3.0, indicating that a temperature deviation exceeding 2 to 3 standard deviations is considered a potential anomaly; the spatial continuity threshold is set to a minimum cluster volume of 0.5 cubic meters to ensure that the identified anomaly regions have a certain scale and are not noise points.

[0068] Clustering results yielded candidate anomaly regions. For example, during the operation of a 300 MW boiler, 3 to 5 candidate anomaly regions might be identified: the region near the burner has a mean deviation of 2.8 and a volume of approximately 2 cubic meters; the region above the secondary air inlet has a mean deviation of 3.2 and a volume of approximately 1.5 cubic meters; and the region in the upper middle part of the furnace has a mean deviation of 2.5 and a volume of approximately 3 cubic meters. These candidate regions represent potential temperature anomaly areas and require further analysis and confirmation.

[0069] Historical temperature sequences for each spatial location within the candidate region are extracted, and the corresponding temporal trends are calculated. The temporal trends are calculated using a sliding window method, with the window width typically set to 6 to 12 time points to reflect the direction and rate of recent temperature changes. Temporal trends can be categorized into three types: upward trend, downward trend, and stable trend. An upward trend indicates a continuous increase in temperature with a slope greater than 0.5 degrees Celsius per hour; a downward trend indicates a continuous decrease in temperature with a slope less than -0.5 degrees Celsius per hour; and a stable trend indicates small temperature fluctuations with a slope between -0.5 and 0.5 degrees Celsius per hour. For the candidate region above the secondary wind inlet, the historical temperature sequence may exhibit a significant upward trend, with an average warming rate of 2.5 degrees Celsius per hour over the past 6 hours.

[0070] The temperature values ​​at each spatial location in a candidate region within a continuous temperature field are matched with the temporal trend to calculate the trend conformity. The trend conformity assesses whether the current temperature value matches the historical trend prediction. The conformity is calculated as the relative deviation between the current temperature value and the predicted value based on the historical trend. For an upward trend region, if the historical trend predicts the current temperature to be 1340 degrees Celsius, and the actual measured temperature is 1350 degrees Celsius, the relative deviation is 0.75%, indicating a high trend conformity. However, if the actual temperature is 1400 degrees Celsius, the relative deviation is 4.48%, indicating a low trend conformity and suggesting abnormal temperature rise.

[0071] Candidate areas are filtered based on trend conformity and a pre-set conformity threshold to identify areas of abnormal fluctuation. The conformity threshold is typically set at 3%; candidate areas below this threshold are considered normal temperature fluctuations, while areas above it are identified as areas of abnormal fluctuation. In the example above, the trend conformity of the area above the secondary air inlet is 4.48%, exceeding the threshold and thus identified as an area of ​​abnormal fluctuation; while the trend conformity of the area near the burner is 2.8%, below the threshold and thus identified as an area of ​​normal fluctuation.

[0072] In this embodiment, by aligning the current temperature field with historical temperature sequences over time, temperature data under different operating conditions and at different times are made comparable. This avoids misjudging normal temperature fluctuations caused by load changes or operating condition switching as abnormalities, thereby improving the rationality and consistency of temperature deviation assessment. By constructing a temperature deviation distribution in the spatial dimension and performing spatial clustering, only areas with deviations exceeding a preset range and exhibiting continuous distribution are identified as candidate areas. This effectively suppresses discrete anomalies caused by local noise, single-point measurement errors, or interpolation errors, significantly reducing the false alarm rate and making the anomaly identification results more stable and reliable. By introducing the temporal change trend of historical temperature sequences and matching the temperature values ​​in the current temperature field with this trend, the candidate areas are further screened using trend conformity. This ensures that the finally identified abnormal fluctuation areas not only deviate significantly in amplitude but also have obvious differences in evolutionary characteristics from historical normal or abnormal patterns, enhancing the ability to distinguish real abnormal operating conditions.

[0073] In one alternative implementation, Historical combustion control parameters are obtained to determine the causal relationship between temperature field response and combustion control parameters, and the dominant control parameter combination is deduced. Combined with the abnormal fluctuation region, temperature field monitoring results and combustion condition optimization suggestions are generated, including: Historical combustion control parameters and corresponding historical temperature field data are obtained. The temperature change at each spatial location in the historical temperature field data is extracted and aligned with the historical combustion control parameters to obtain associated data. The cross-correlation strength between historical combustion control parameters and temperature changes in the associated data under different time delays is calculated, and the optimal time delay is identified. Based on the optimal time delay, the global response of each historical combustion control parameter is calculated. The global response is nonlinearly decoupled and decomposed to extract the interaction components, and the coupling strength between different historical combustion control parameters is calculated. The coupling strength is eigenvalue decomposed to extract the dominant feature vector and mapped back to the historical combustion control parameter space to obtain the dominant control parameter combination. The historical combustion control parameters are calculated to respond to temperature disturbances in the abnormal fluctuation region under optimal time delay conditions, and the disturbance contribution is determined. The historical combustion control parameter with the largest disturbance contribution is used as the driving parameter. The historical regulation trajectory of the dominant control parameter combination and the temperature evolution trajectory of the abnormal fluctuation region are extracted and dynamically matched in time to obtain the decoupling deviation period. The historical combustion control parameter combination with the largest coupling strength in the decoupling deviation period is used as the coordination parameter. Combined with the driving parameter, the temperature field monitoring results and combustion condition optimization suggestions are obtained.

[0074] Historical combustion control parameter data is obtained from the boiler distributed control system database. For a 300 MW thermal power boiler, historical combustion control parameters mainly include key parameters such as primary air volume, secondary air volume, coal feed rate, swirl intensity, and air-to-coal ratio. Typically, historical data from the last 720 hours is collected, with a sampling interval of 1 minute. Each set of historical combustion control parameters includes detailed timestamp information; for example, at 10:30 AM on January 15th, the primary air volume is 70 kg / s, the secondary air volume is 120 kg / s, the coal feed rate is 35 tons / hour, the swirl intensity is 0.85, and the air-to-coal ratio is 1.6. Historical temperature field data corresponding to the historical combustion control parameter times is extracted from the historical temperature field database. The historical temperature field data has a resolution of 4 sampling points per cubic meter, covering the entire spatial location of the boiler combustion chamber, and includes records of temperature changes at each point over time.

[0075] The temperature change at each spatial location in the historical temperature field data is calculated. The temperature change is obtained by differencing adjacent time points, reflecting the dynamic characteristics of temperature change. For example, if the temperature at a spatial location is 1350 degrees Celsius at 10:30 and 1355 degrees Celsius at 10:31, the corresponding temperature change is 5 degrees Celsius per minute. The temperature change data is aligned with historical combustion control parameter data by timestamp to form a correlated dataset. Each record in the correlated dataset contains time information, combustion control parameter values, and the temperature change at each spatial location, establishing a mapping relationship between control parameters and temperature response.

[0076] The cross-correlation strength between historical combustion control parameters and temperature changes in the associated data was calculated under different time delays. The cross-correlation strength was calculated using a sliding window method with a window width of 60 minutes, a step size of 1 minute, and a time delay range of 0 to 30 minutes. For each spatial location and each combustion control parameter, the cross-correlation coefficient was calculated under different time delays. For example, analyzing the relationship between primary air volume changes and temperature changes in the middle of the furnace revealed that the cross-correlation coefficient reached its maximum value of 0.82 at a time delay of 7 minutes, indicating that adjusting the primary air volume approximately 7 minutes later had the most significant impact on the temperature at that location. Based on the cross-correlation analysis results, the optimal time delay for each spatial location with respect to different combustion control parameters was identified. For the bottom region of the furnace, the optimal time delay was typically 3 to 8 minutes; for the middle region, it was 7 to 15 minutes; and for the upper region, it was 15 to 25 minutes, reflecting the transmission characteristics of the combustion process.

[0077] The direct and indirect responses to each historical combustion control parameter are calculated based on optimal time delay. The direct response represents the temperature change caused by a single control parameter change, calculated using a univariate conditional cross-correlation method. For example, under relatively stable conditions, a 10% increase in primary air volume results in a direct temperature change, potentially raising the temperature by 20 degrees Celsius in a certain area. The indirect response represents the temperature change caused by the interaction between control parameters, calculated using a multivariate partial correlation analysis method. For example, a 10% increase in primary air volume causing a change in the air-fuel ratio, leading to an indirect temperature change, might result in a 5-degree Celsius temperature decrease. The vector sum of the direct and indirect responses yields the global response, reflecting the temperature field characteristics under the combined influence of the control parameters.

[0078] The global response is nonlinearly decoupled and decomposed to extract interaction components. Nonlinear decoupling is performed using kernel principal component analysis (KPCA), with the radial basis function chosen as the kernel function and a bandwidth parameter set to 0.5. The interaction components represent the influence of nonlinear interactions between different control parameters on the temperature field. The coupling strength between different historical combustion control parameters is calculated. The row and column elements of the coupling strength matrix represent the interaction strength between the corresponding two control parameters. For example, the coupling strength between primary air volume and secondary air volume is 0.78, indicating a strong interaction; the coupling strength between primary air volume and coal feed rate is 0.92, indicating a very strong interaction; and the coupling strength between secondary air volume and swirl intensity is 0.65, indicating a moderate interaction.

[0079] Eigenvalue decomposition is performed on the coupling strength matrix to extract the dominant eigenvectors. The eigenvalues ​​are sorted by magnitude; the top three eigenvalues ​​and their corresponding eigenvectors typically explain over 80% of the system variability. The dominant eigenvectors are mapped back to the historical combustion control parameter space to obtain the dominant control parameter combinations. These combinations represent the set of parameters with the greatest impact on the system and their weights. For a typical 300 MW boiler, the dominant control parameter combinations might be: the first combination includes a coal feed weight of 0.65 and a primary air volume weight of 0.32; the second combination includes a secondary air volume weight of 0.58 and a swirl intensity weight of 0.39; and the third combination includes an air-to-coal ratio weight of 0.75 and a burner tilt angle weight of 0.22.

[0080] The historical combustion control parameters were calculated to respond to temperature disturbances at various spatial locations within the abnormal fluctuation region under optimal time delay conditions, determining the disturbance contribution. The disturbance response was calculated using Granger causality analysis with a time window of 120 minutes and an order of 10. The disturbance contribution represents the explanatory power of each control parameter for the abnormal temperature fluctuations. For example, for the abnormal region above the secondary air inlet, the primary air volume disturbance contributed 15%, the secondary air volume disturbance contributed 65%, the coal feed rate disturbance contributed 12%, and other parameters contributed 8%. The historical combustion control parameter with the largest disturbance contribution was identified as the driving parameter; in the aforementioned example, the secondary air volume was identified as the driving parameter.

[0081] Historical control trajectory of the dominant control parameter combination and temperature evolution trajectory of the abnormal fluctuation area are extracted. The control trajectory represents the change process of control parameters over the past 24 hours, and the temperature evolution trajectory represents the change process of temperature within the corresponding time period. Dynamic time-series matching is performed on the two types of trajectories to identify decoupling deviation periods. Dynamic time-series matching adopts a dynamic time warping algorithm with a window width of 30 minutes and a step size of 5 minutes. Decoupling deviation periods are those periods in which the change of control parameters and the temperature response do not conform to the normal pattern. For example, between 10:45 and 11:15, although the secondary air volume increased by 12%, the temperature in the abnormal area did not decrease as expected, but instead increased by 35 degrees Celsius. This period is marked as a decoupling deviation period.

[0082] The historical combustion control parameter combinations with the strongest coupling strength during the decoupling deviation period were analyzed and identified as cooperative parameters. Cooperative parameters represent a set of parameters that have a strong coupling effect with the driving parameters. In the aforementioned example, the parameters strongly coupled with the secondary air volume are swirl intensity and burner tilt angle, with coupling strengths of 0.65 and 0.55, respectively, and are thus identified as cooperative parameters. Combining the driving parameters and cooperative parameters, an adjustment scheme is solved according to the control objective. The adjustment scheme is generated using a fuzzy decision tree algorithm with a decision tree depth of 3 and 8 leaf nodes. For example, for the problem of abnormally high temperature in the area above the secondary air inlet, the adjustment scheme might be: increase the secondary air volume by 15%, while decreasing the swirl intensity by 0.1 and adjusting the burner tilt angle by 5 degrees, which is expected to reduce the temperature in the abnormal area by 40 degrees Celsius to the normal range within 10 minutes.

[0083] Generate a report on temperature field monitoring results and combustion condition optimization suggestions. The report includes a description of the abnormal temperature area, identification of key driving parameters, analysis of synergistic parameters, suggested adjustment schemes, and an assessment of expected effects. For example: The abnormal area is located above the secondary air inlet, extending 8.5 to 10 meters horizontally, 7 to 8.5 meters vertically, and 14 to 15.5 meters high. The current temperature is 1400 degrees Celsius, 60 degrees Celsius higher than normal, and has lasted for 2.5 hours. The driving parameter is the secondary air volume, currently at 120 kg / s, deviating from normal conditions by 15%. The synergistic parameters are swirl intensity and burner angle. The optimization suggestion is to increase the secondary air volume to 138 kg / s, decrease the swirl intensity to 0.75, and adjust the burner angle to 25 degrees. The expected temperature drop to 1360 degrees Celsius within 10 minutes, maintaining stable combustion.

[0084] In this embodiment, by introducing cross-correlation analysis under different time delay conditions in the time dimension and identifying the optimal time delay, the response relationship between combustion control parameters and temperature changes can accurately reflect the transmission delay characteristics in the combustion process. This avoids the problem of distorted causal judgment caused by ignoring dynamic lag, and improves the accuracy of control parameter impact assessment. By uniformly modeling the direct and indirect responses and forming a global response, and then performing nonlinear decoupling decomposition on the global response, the coupling effects caused by the interaction of multiple control parameters are effectively separated, significantly enhancing the analytical capability for complex combustion regulation relationships. By performing eigenvalue decomposition on the coupling strength and extracting the dominant feature vector, the combination of dominant control parameters that has the most significant impact on the temperature field can be automatically identified from high-dimensional historical control parameters, greatly reducing parameter redundancy and misjudgment risk, and improving the efficiency and reliability of control parameter combination identification.

[0085] A second aspect of the present invention provides an online temperature field monitoring system for thermal power boilers based on acoustic wave temperature measurement, comprising: The acoustic temperature measurement module is used to acquire acoustic signals in the combustion chamber of a thermal power boiler and extract the flight time and attenuation amplitude. Based on the flight time, the linear integral temperature value is retrieved. Based on the attenuation amplitude, the attenuation coefficient is calculated and a particulate matter concentration distribution field is constructed to solve the flue gas flow vector field. The path temperature optimization module is used to calculate the spatial offset of the sound wave propagation path based on the flue gas flow vector field and correct it to obtain an optimized propagation path. Based on the particulate matter concentration distribution field, the linear integral temperature value is corrected to obtain an optimized linear integral temperature value. Based on the optimized propagation path, discretized temperature units are obtained by using an adaptive mesh partitioning algorithm. The temperature field reconstruction module is used to obtain the spatial distribution law of heat release and the heat flux density of radiation heat transfer boundary, and calculate the energy conservation constraint value and the heat transfer boundary constraint value. Combined with the spatial smoothing constraint value corresponding to the flue gas convection heat transfer principle, the combustion comprehensive constraint is determined. Combined with the optimized linear integral temperature value, the temperature value distribution of the discretized temperature unit is calculated and reconstructed to obtain the continuous temperature field. The anomaly identification module is used to perform time-series correlation analysis between the continuous temperature field and the pre-acquired historical temperature field data to determine the abnormal fluctuation area, obtain historical combustion control parameters to determine the causal relationship between the temperature field response and the combustion control parameters, and infer the dominant control parameter combination. Combined with the abnormal fluctuation area, it generates temperature field monitoring results and combustion condition optimization suggestions.

[0086] A third aspect of the present invention provides an electronic device, comprising: A processor and a memory for storing processor-executable instructions, wherein the processor is configured to invoke instructions stored in the memory to perform the aforementioned method.

[0087] A fourth aspect of the present invention provides a computer-readable storage medium having stored thereon computer program instructions that, when executed by a processor, implement the aforementioned method.

[0088] This invention can be a method, apparatus, system, and / or computer program product. The computer program product may include a computer-readable storage medium having computer-readable program instructions loaded thereon for performing various aspects of the invention.

[0089] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for online monitoring of the temperature field of a thermal power boiler based on acoustic wave thermometry, characterized in that, include: Acquire acoustic signals from the combustion chamber of a thermal power boiler and extract the flight time and attenuation amplitude. Based on the flight time, retrieve the linear integral temperature value. Based on the attenuation amplitude, calculate the attenuation coefficient and construct the particulate matter concentration distribution field to solve the flue gas flow vector field. The spatial offset of the sound wave propagation path is calculated and corrected based on the flue gas flow vector field to obtain an optimized propagation path. The linear integral temperature value is corrected based on the particulate matter concentration distribution field to obtain an optimized linear integral temperature value. Discretized temperature units are obtained by using an adaptive mesh partitioning algorithm based on the optimized propagation path. The spatial distribution law of heat release and the heat flux density of the radiation heat transfer boundary are obtained, and the energy conservation constraint value and the heat transfer boundary constraint value are calculated. The combustion comprehensive constraint is determined by combining the spatial smoothing constraint value corresponding to the flue gas convection heat transfer principle. The temperature value distribution of the discretized temperature unit is calculated by combining the optimized linear integral temperature value and the continuous temperature field is reconstructed. The continuous temperature field is correlated with the pre-acquired historical temperature field data to identify abnormal fluctuation areas. Historical combustion control parameters are obtained to determine the causal relationship between the temperature field response and the combustion control parameters, and the dominant control parameter combination is inferred. The temperature field monitoring results and combustion condition optimization suggestions are generated by combining the abnormal fluctuation areas.

2. The method according to claim 1, characterized in that, Acquiring acoustic signals from the combustion chamber of a thermal power boiler and extracting the time of flight and attenuation amplitude, retrieving linear integral temperature values ​​based on the time of flight, calculating attenuation coefficients based on the attenuation amplitude, and constructing a particulate matter concentration distribution field to solve the flue gas flow vector field includes: Sound wave pulses are emitted by a pre-arranged sound wave device, and the propagated sound wave signals are received. The emission time and arrival time are recorded to determine the flight time, and the emission amplitude and reception amplitude are recorded to determine the waveform attenuation amplitude. The average sound velocity is determined based on the flight time and sound wave propagation path length, and then converted into a linear integral temperature value according to the correlation between the molar mass of the flue gas components and temperature. The total attenuation is determined by performing a logarithmic transformation on the waveform attenuation amplitude and combining it with the sound wave propagation path length. Based on the relationship between flue gas density and temperature and the relationship between particulate matter scattering cross section and wavelength, the medium density attenuation coefficient and the particulate matter scattering attenuation coefficient are separated from the total attenuation. The combustion chamber space is divided into grid cells. The particulate concentration of each grid cell is calculated based on the particulate scattering attenuation coefficient to form a particulate concentration distribution field. The flue gas density of each grid cell is calculated based on the medium density attenuation coefficient to construct a flue gas density distribution field. The three-dimensional spatial gradient of the flue gas density distribution field is calculated to obtain the density gradient field. The flue gas flow vector field is calculated by combining the principle of mass conservation.

3. The method according to claim 1, characterized in that, The spatial offset of the sound wave propagation path is calculated based on the flue gas flow vector field and corrected to obtain an optimized propagation path. The linear integral temperature value is then corrected based on the particulate matter concentration distribution field to obtain an optimized linear integral temperature value. Based on the optimized propagation path, discretized temperature units are obtained through an adaptive mesh generation algorithm, including: The flue gas velocity vector of each grid cell in the flue gas flow vector field is extracted and line integraled along each sound wave propagation path to obtain the cumulative offset vector. The cumulative offset vector is decomposed to obtain the transverse component perpendicular to the sound wave propagation path, which is used as the spatial offset. The sound wave propagation path is corrected according to the spatial offset to obtain the optimized propagation path. The particle concentration distribution field is linearly integrated along the sound wave propagation path to obtain the particle concentration integral value. The medium non-uniformity correction coefficient is calculated based on the particle concentration integral value, and the optimized linear integral temperature value is obtained by combining the linear integral temperature value. The spatial distance between different optimized propagation paths is calculated and the coverage density of the optimized propagation paths in each region of the combustion chamber is determined. Regions with an optimized propagation path coverage density higher than a preset density threshold are marked as dense regions, and regions with a density density not higher than the density threshold are marked as sparse regions. The dense regions are meshed to obtain refined mesh cells, and the sparse regions are meshed to obtain coarse mesh cells. Based on the optimized propagation paths, the refined mesh cells and coarse mesh cells are corrected to obtain discretized temperature cells.

4. The method according to claim 1, characterized in that, The spatial distribution of heat release and the boundary heat flux density of radiative heat transfer are obtained, and the energy conservation constraint value and heat transfer boundary constraint value are calculated. Combined with the spatial smoothing constraint value corresponding to the flue gas convection heat transfer principle, the comprehensive combustion constraints are determined, including: The fuel supply and air supply are obtained, flame radiation images of the combustion chamber wall are acquired and flame temperature is calculated through spectral analysis, fuel heating power is calculated based on the fuel supply, oxygen consumption rate is calculated based on the air supply and flame temperature, and heat release rate of each discretized temperature unit is determined. The spatial distribution law of heat release is obtained based on the heat release rate and fuel heating power. The wall surface heat flux density and wall surface flue gas temperature are obtained. The heat transfer coefficient is calculated based on the wall surface heat flux density. The radiative heat transfer boundary heat flux density is obtained based on the heat transfer coefficient and the wall surface flue gas temperature. The heat transfer boundary constraint value is obtained by combining the temperature gradient heat flux density of adjacent discretized temperature units. The heat release rate of each discretized temperature unit is extracted as a heat source term. The flue gas flow velocity vector is extracted and combined with the temperature gradient corresponding to the optimized linear integral temperature value to obtain the convection term. Based on the heat source term and the convection term, and combined with the thermal conductivity in the furnace, the energy conservation constraint value is obtained. The spatial gradient is calculated based on the temperature difference and spacing between adjacent discretized temperature units, and the curvature value is obtained by performing second-order differentiation. The curvature value is then smoothed to obtain the spatial smoothing constraint value. Combustion comprehensive constraint is obtained by combining the energy conservation constraint value and the heat transfer boundary constraint value.

5. The method according to claim 1, characterized in that, The temperature distribution of the discretized temperature unit is calculated by combining the optimized linear integral temperature values, and the continuous temperature field is reconstructed, including: Extract the optimized linear integral temperature value corresponding to each sound wave propagation path, discretize the sound wave propagation path into multiple path segments and calculate the overlap length between each path segment and the corresponding discretized temperature unit. Based on the overlap length, decompose the optimized linear integral temperature into the temperature contribution of each discretized temperature unit and calculate the cumulative contribution. Initialize the temperature value of each discretized temperature unit, calculate the physical deviation based on the combustion comprehensive constraint and the current temperature value to obtain the physical adjustment amount, calculate the measurement deviation based on the cumulative contribution and the current temperature value to obtain the measurement adjustment amount, solve for the comprehensive adjustment amount based on the physical adjustment amount and the measurement adjustment amount and apply it to the corresponding discretized temperature unit, repeat the adjustment until the physical deviation and the measurement deviation converge to obtain the temperature value distribution corresponding to the discretized temperature unit; The spatial coordinates and temperature values ​​of each discretized temperature unit in the temperature value distribution are extracted. A three-dimensional mesh is constructed based on the spatial coordinates, and mesh nodes are determined. The temperature values ​​are assigned to the corresponding mesh nodes. The spatial positions between adjacent mesh nodes are interpolated based on the temperature values ​​of the mesh nodes to obtain interpolated temperatures. The temperature values ​​of the mesh nodes and the interpolated temperatures are combined to obtain a continuous temperature field.

6. The method according to claim 1, characterized in that, The determination of abnormal fluctuation regions involves performing time-series correlation analysis between the continuous temperature field and pre-acquired historical temperature field data, including: Extract the temperature values ​​at each spatial location in the continuous temperature field and align them temporally with the historical temperature sequences corresponding to the spatial locations in the pre-acquired historical temperature field data to obtain temporally aligned data. The temperature deviation is calculated between the temperature value at each spatial location in the time-series aligned data and the historical temperature sequence. Based on the temperature deviation, the deviation distribution of each spatial location is constructed. Spatial clustering is performed on the deviation distribution to identify regions where the deviation exceeds a preset range and is spatially continuous, thus obtaining candidate regions. Historical temperature sequences of each spatial location in the candidate region are extracted and the corresponding temporal variation trend is calculated. The temperature values ​​of each spatial location in the candidate region in the continuous temperature field are matched with the temporal variation trend and the trend conformity is calculated. The candidate region is then filtered based on the trend conformity and a pre-set conformity threshold to obtain abnormal fluctuation regions.

7. The method according to claim 1, characterized in that, Historical combustion control parameters are obtained to determine the causal relationship between temperature field response and combustion control parameters, and the dominant control parameter combination is deduced. Combined with the abnormal fluctuation region, temperature field monitoring results and combustion condition optimization suggestions are generated, including: Historical combustion control parameters and corresponding historical temperature field data are obtained. The temperature change at each spatial location in the historical temperature field data is extracted and aligned with the historical combustion control parameters to obtain associated data. The cross-correlation strength between historical combustion control parameters and temperature changes in the associated data under different time delays is calculated, and the optimal time delay is identified. Based on the optimal time delay, the global response of each historical combustion control parameter is calculated. The global response is nonlinearly decoupled and decomposed to extract the interaction components, and the coupling strength between different historical combustion control parameters is calculated. The coupling strength is eigenvalue decomposed to extract the dominant feature vector and mapped back to the historical combustion control parameter space to obtain the dominant control parameter combination. The historical combustion control parameters are calculated to respond to temperature disturbances in the abnormal fluctuation region under optimal time delay conditions, and the disturbance contribution is determined. The historical combustion control parameter with the largest disturbance contribution is used as the driving parameter. The historical regulation trajectory of the dominant control parameter combination and the temperature evolution trajectory of the abnormal fluctuation region are extracted and dynamically matched in time to obtain the decoupling deviation period. The historical combustion control parameter combination with the largest coupling strength in the decoupling deviation period is used as the coordination parameter. Combined with the driving parameter, the temperature field monitoring results and combustion condition optimization suggestions are obtained.

8. An online temperature field monitoring system for thermal power boilers based on acoustic wave thermometry, used to implement the method described in any one of claims 1-7, characterized in that, include: The acoustic temperature measurement module is used to acquire acoustic signals in the combustion chamber of a thermal power boiler and extract the flight time and attenuation amplitude. Based on the flight time, the linear integral temperature value is retrieved. Based on the attenuation amplitude, the attenuation coefficient is calculated and a particulate matter concentration distribution field is constructed to solve the flue gas flow vector field. The path temperature optimization module is used to calculate the spatial offset of the sound wave propagation path based on the flue gas flow vector field and correct it to obtain an optimized propagation path. Based on the particulate matter concentration distribution field, the linear integral temperature value is corrected to obtain an optimized linear integral temperature value. Based on the optimized propagation path, discretized temperature units are obtained by using an adaptive mesh partitioning algorithm. The temperature field reconstruction module is used to obtain the spatial distribution law of heat release and the heat flux density of radiation heat transfer boundary, and calculate the energy conservation constraint value and the heat transfer boundary constraint value. Combined with the spatial smoothing constraint value corresponding to the flue gas convection heat transfer principle, the combustion comprehensive constraint is determined. Combined with the optimized linear integral temperature value, the temperature value distribution of the discretized temperature unit is calculated and reconstructed to obtain the continuous temperature field. The anomaly identification module is used to perform time-series correlation analysis between the continuous temperature field and the pre-acquired historical temperature field data to determine the abnormal fluctuation area, obtain historical combustion control parameters to determine the causal relationship between the temperature field response and the combustion control parameters, and infer the dominant control parameter combination. Combined with the abnormal fluctuation area, it generates temperature field monitoring results and combustion condition optimization suggestions.

9. An electronic device, characterized in that, include: processor; Memory used to store processor-executable instructions; The processor is configured to invoke instructions stored in the memory to execute the method according to any one of claims 1 to 7.

10. A computer-readable storage medium having computer program instructions stored thereon, characterized in that, When the computer program instructions are executed by the processor, they implement the method described in any one of claims 1 to 7.

Images