Correcting system and method for flame deflection of corner tangential boiler

Abstract

A corner tangential boiler flame deflection correction system disclosed by the present invention comprises a flame image acquisition module, the flame image acquisition module is connected with a flame center identification module, the flame center identification module is connected with a correction control module, and the correction control module is internally provided with a flame deflection-air door adjustment mapping model. And the correction control module is connected with an execution module. The invention further discloses a flame deflection correcting method for the corner tangential boiler. According to the flame deflection correction system and method for the corner tangential boiler, the problems that in the prior art, traditional flame deflection correction depends on manual operation, and flame center space coordinates cannot be accurately quantified are solved.

Description

Technical Field

[0001] This invention belongs to the field of power plant boiler combustion control technology, specifically relating to a four-corner tangential boiler flame deflection correction system, and also to the correction method of the correction system. Background Technology

[0002] Four-corner tangential coal-fired boilers are widely used in thermal power units due to their high combustion efficiency and compact structure. Their working principle involves injecting pulverized coal gas in an imaginary tangential direction through burner nozzles arranged on the four walls, forming a rotating fireball inside the furnace. This intense swirling motion enhances the mixing of pulverized coal and air, prolonging the residence time of the pulverized coal and ensuring a high burnout rate. However, in actual operation, the flame center often deviates from the geometric center of the furnace due to various factors, a phenomenon known as "flame deflection."

[0003] The main causes of flame deflection include: uneven coal powder distribution due to differences in resistance of the coal mill outlet pulverizer pipes; airflow deflection caused by coking or wear of the burner nozzles; momentum imbalance caused by deviations in the opening of the primary and secondary dampers; flow field disturbance caused by fluctuations in furnace negative pressure; and temperature gradients formed by uneven heat absorption on the sidewalls. Flame deflection can lead to a series of safety and economic problems: the water-cooled wall on the side closer to the flame is affected by high-temperature radiation, resulting in local temperature increases, reduced ash viscosity, and a significant increase in coking rate, even leading to high-temperature corrosion of the water-cooled wall; the flue gas temperature on the side farther from the flame decreases, causing deviations in the main steam temperature and reheat steam temperature, which in severe cases triggers steam temperature protection and forces the unit to reduce load; long-term deflection can also exacerbate thermal fatigue of the water-cooled wall, increase the risk of tube rupture, and significantly reduce the safety and economy of boiler operation.

[0004] Traditional flame deflection correction relies on manual operation. Operators judge the deflection status by observing the flue gas temperature at the furnace outlet, the water-cooled wall temperature, or visually inspecting the flame through observation holes, and then manually adjust the opening of the dampers at each level and the burner angle. This method has significant drawbacks: First, the judgment is limited to single-point temperature or subjective visual observation, making it impossible to quantify the spatial coordinates of the flame center and accurately assess the degree of deflection; second, the adjustment process relies on personal experience, lacks scientific quantitative guidance, requires repeated trials, and has low adjustment accuracy; third, the manual response cycle is long, usually intervening only after significant steam temperature deviation or coking occurs, which is a passive and delayed adjustment and cannot prevent accidents. Summary of the Invention

[0005] The primary objective of this invention is to provide a four-corner tangential boiler flame deflection correction system, which solves the problem that traditional flame deflection correction relies on manual operation and cannot accurately quantify the spatial coordinates of the flame center.

[0006] The second objective of this invention is to provide a method for correcting flame deflection in a four-corner tangential boiler.

[0007] The first technical solution adopted in this invention is a four-corner tangential boiler flame deflection correction system, including a flame image acquisition module, a flame center identification module connected to the flame image acquisition module, a correction control module connected to the flame center identification module, a built-in flame deflection-damper adjustment mapping model in the correction control module, and an execution module connected to the correction control module.

[0008] The first technical solution of this invention is also characterized in that, The flame image acquisition module consists of high-temperature cameras installed at the four corners of the furnace, with the high-temperature cameras facing the center of the furnace to acquire real-time dynamic images of the flame. The flame center recognition module is used to preprocess flame images, extract features, and reconstruct 3D, and calculate the spatial offset parameters of the flame center, including the offset direction and offset distance.

[0009] The flame deflection-damper adjustment mapping model generates opening adjustment commands for each burner damper based on the offset parameters; The flame deflection-damper adjustment mapping model is trained using a BP neural network. The input is the offset parameter, and the output is the opening correction value of the damper for each burner layer. The training samples include the offset parameters of historical deflection cases and the corresponding optimal damper adjustment amount.

[0010] The execution module includes a servo motor and a transmission mechanism, which are used to receive adjustment commands and drive the damper to perform opening adjustment.

[0011] The execution module also includes a position feedback unit, which is used to collect the actual opening degree of the damper in real time and feed it back to the correction control module to form a closed-loop control.

[0012] The high-temperature camera uses a water-cooled protective cover, and the lens is equipped with an infrared filter. It can withstand the external ambient temperature of the furnace from -20℃ to 80℃, with an image resolution of no less than 1920×1080 and a frame rate of no less than 25 fps.

[0013] The second technical solution adopted in this invention is a method for correcting flame deflection in a four-corner tangential boiler, the specific steps of which are as follows: S1: Simultaneously collect dynamic images of the flame through high-temperature cameras at the four corners of the furnace and transmit them to the flame center recognition module; S2: The flame center recognition module preprocesses, segments, and reconstructs the image in three dimensions, and calculates the spatial offset parameters of the flame center; S3: The correction control module inputs the spatial offset parameters into the flame deflection-damper adjustment mapping model and outputs the opening adjustment command of each burner damper; S4: The execution module drives the damper to adjust according to the adjustment command and realizes closed-loop control through position feedback; S5: Repeat steps S1-S4 until the flame center offset distance is less than the preset threshold.

[0014] The second technical solution of the present invention is further characterized in that, In S2, the 3D reconstruction adopts a binocular vision matching algorithm. The 3D coordinates of the flame feature points are calculated by the calibration parameters of the four corner cameras. The calibration parameters include the intrinsic parameter matrix, distortion coefficients and extrinsic parameter matrix.

[0015] The generation rule for the opening adjustment command in S3 is as follows: when the flame shifts to a certain angle, the secondary air opening of the burner corresponding to that angle is increased, the secondary air opening of the burner at the opposite angle is decreased, and the primary air opening is adjusted proportionally in coordination. The specific process of constructing the flame deflection-damper adjustment mapping model in S3 is as follows: A backpropagation (BP) neural network is used, consisting of an input layer, hidden layers, and an output layer. The input layer is specifically as follows: Input parameters are normalized to eliminate the impact of dimensional differences on training. , (1) In the formula, the offset distance D and the offset direction θ; The hidden layers are specifically designed as follows: two hidden layers are used, with the first hidden layer containing 12 neurons and the second hidden layer containing 8 neurons. The output layer has n neurons, where n equals the total number of damper layers at the four corners of the four-corner burner. The output parameters are the opening correction values ​​for each damper. The unit is %, the range is [-5%, +5%], and the single adjustment range is ≤5%; The output values ​​are denormalized to the actual physical range, and values ​​exceeding [-5, +5] are forcibly truncated. (2) In the formula, y i The output layer normalized value [0,1] is multiplied by 10 and subtracted by 5 and then mapped to [-5, +5].

[0016] The preset threshold in S5 is set according to the boiler model, and the value range is 0~300 mm. When the offset distance is less than the threshold for more than 10 seconds, the system enters standby mode.

[0017] The beneficial effects of this invention are: (1) The present invention uses an industrial camera to replace the acoustic temperature measurement array, which reduces the cost of a single system and eliminates the need for frequent replacement of vulnerable parts, resulting in low maintenance costs.

[0018] (2) The present invention adopts a response time of ≤2 s for the entire process from image acquisition to damper adjustment, which achieves real-time correction compared to the 5 min response cycle of manual adjustment.

[0019] (3) The present invention uses three-dimensional reconstruction technology to obtain the spatial coordinates of the flame center, with a positioning error of ≤50 mm, which overcomes the limitations of traditional single-point temperature measurement.

[0020] (4) This invention adopts full-process automation from monitoring, judgment to adjustment, without manual intervention, reducing the labor intensity of operators, effectively solving the problem of monitoring and correction of flame deviation in four-corner tangential boilers, significantly reducing the risk of coking and tube rupture of water-cooled walls, reducing steam temperature deviation, improving the safety and economy of boiler operation, and has broad engineering application value. Attached Figure Description

[0021] Figure 1 This is a schematic diagram of the structure of the four-corner tangential boiler flame deflection correction system of the present invention.

[0022] Figure 2 This is a schematic diagram of the BP neural network algorithm of the present invention. Detailed Implementation

[0023] The present invention will now be described in detail with reference to the accompanying drawings and specific embodiments.

[0024] Example 1 This invention provides a four-corner tangential boiler flame deflection correction system, such as... Figure 1 As shown, it includes a flame image acquisition module, a flame center recognition module connected to the flame image acquisition module, a correction control module connected to the flame center recognition module, a built-in flame deflection-damper adjustment mapping model in the correction control module, and an execution module connected to the correction control module.

[0025] Example 2 Based on Example 1, the flame image acquisition module consists of high-temperature cameras installed at the four corners of the furnace, with the high-temperature cameras facing the center area of ​​the furnace, for real-time acquisition of dynamic flame images; the flame center recognition module is used to preprocess the flame images, extract features and reconstruct three dimensions, and calculate the spatial offset parameters of the flame center, including the offset direction and offset distance.

[0026] Example 3 Based on the above embodiments, the flame deflection-damper adjustment mapping model generates opening adjustment commands for each burner damper according to the offset parameters. The model is trained using a BP neural network, with the offset parameters as input and the opening correction values ​​for each layer of burner dampers as output. Training samples include the offset parameters of historical deflection cases and the corresponding optimal damper adjustment amounts. The execution module includes a servo motor and transmission mechanism for receiving adjustment commands and driving the dampers to perform opening adjustments. The execution module also includes a position feedback unit for real-time acquisition of the actual damper opening and feedback to the correction control module, forming a closed-loop control. The high-temperature camera uses a water-cooled protective cover, and the lens is equipped with an infrared filter, capable of withstanding furnace external ambient temperatures of -20℃ to 80℃, with an image resolution of no less than 1920×1080 and a frame rate of no less than 25 fps.

[0027] In this embodiment, after the system is started, the high-temperature camera continuously acquires flame images (25 frames / second), the flame center recognition module outputs the flame center coordinates (X, Y, Z) every 100 ms, and the correction control module calculates the offset parameters (D, θ) in real time. When D>D0 and the duration is ≥3 s, the system determines that the flame is deflected and enters the correction mode; if D≤D0 or the duration is less than 3 s, the system maintains the standby state.

[0028] The correction control module inputs (D, θ) into the mapping model to calculate the correction value of each damper opening; generates adjustment commands and sends them to the execution module in the order of "diagonal first, then adjacent angle"; the execution module drives the damper adjustment, the position sensor provides real-time feedback on the opening, and the correction control module performs secondary fine-tuning based on the deviation.

[0029] When D≤D0 and the duration≥10s, the system determines that the correction is complete and enters the stable monitoring state, outputting a status report every 5s; if D is still greater than D0 after adjustment, the correction process is repeated until stability is achieved or an alarm is triggered.

[0030] Example 4 This embodiment provides a method for correcting flame deflection in a four-corner tangential boiler. The specific steps are as follows: S1: Simultaneously collect dynamic images of the flame through high-temperature cameras at the four corners of the furnace and transmit them to the flame center recognition module; S2: The flame center recognition module preprocesses, segments, and reconstructs the image in three dimensions, and calculates the spatial offset parameters of the flame center; S3: The correction control module inputs the spatial offset parameters into the flame deflection-damper adjustment mapping model and outputs the opening adjustment command of each burner damper; S4: The execution module drives the damper to adjust according to the adjustment command and realizes closed-loop control through position feedback; S5: Repeat steps S1-S4 until the flame center offset distance is less than the preset threshold.

[0031] The preset threshold in S5 is set according to the boiler model, and the value range is 0~300 mm. When the offset distance is less than the threshold for more than 10 seconds, the system enters standby mode.

[0032] Example 5 Based on Example 4, in S2, the three-dimensional reconstruction adopts a binocular vision matching algorithm. The three-dimensional coordinates of the flame feature points are calculated through the calibration parameters of the four corner cameras. The calibration parameters include the intrinsic parameter matrix, distortion coefficients and extrinsic parameter matrix.

[0033] The generation rule for the opening adjustment command in S3 is as follows: when the flame shifts to a certain angle, the secondary air opening of the burner corresponding to that angle is increased, the secondary air opening of the burner at the opposite angle is decreased, and the primary air opening is adjusted proportionally in coordination. The specific process of constructing the flame deflection-damper adjustment mapping model in S3 is as follows: like Figure 2 As shown, a BP neural network is used, including an input layer, a hidden layer, and an output layer; The input layer is specifically as follows: Input parameters are normalized to eliminate the impact of dimensional differences on training. , (1) In the formula, the offset distance D and the offset direction θ; The hidden layers are specifically designed as follows: two hidden layers are used, with the first hidden layer containing 12 neurons and the second hidden layer containing 8 neurons. The output layer has n neurons, where n equals the total number of damper layers at the four corners of the four-corner burner. The output parameters are the opening correction values ​​for each damper. The unit is %, the range is [-5%, +5%], and the single adjustment range is ≤5%; The output values ​​are denormalized to the actual physical range, and values ​​exceeding [-5, +5] are forcibly truncated. (2) In the formula, y i The output layer normalized value [0,1] is multiplied by 10 and subtracted by 5 and then mapped to [-5, +5].

[0034] The preset threshold in S5 is set according to the boiler model, and the value range is 0~300 mm. When the offset distance is less than the threshold for more than 10 seconds, the system enters standby mode.

[0035] Example 6 This embodiment provides a four-corner tangential boiler flame deflection correction system and its correction method, such as Figure 1 As shown, the system includes a flame image acquisition module, a flame center recognition module, a correction control module, and an execution module; The flame image acquisition module consists of four high-temperature industrial cameras, installed at the observation holes at the four corners of the furnace. The camera optical axes point towards the geometric center of the furnace, forming multi-angle coverage of the flame area. The cameras utilize water-cooled protective covers connected to the boiler cooling water pipeline, ensuring an inlet water pressure of 0.3~0.5 MPa and a flow rate ≥10 L / min. They can withstand external furnace temperatures of -20 ℃ to 80 ℃, and the lenses are equipped with 800~1000 nm infrared filters, enabling them to penetrate high-temperature flue gas and capture images of the flame core area. Each camera has a 1920×1080 resolution and a 25 fps frame rate, transmitting image data to the flame center recognition module via fiber optic cable with a transmission latency ≤100 ms.

[0036] The flame center identification module is an industrial control computer with built-in image processing and analysis software, which performs the following functions: Image preprocessing: Gaussian filtering is used to remove noise, histogram equalization is used to enhance the contrast between the flame and the background, and camera calibration parameters are used to correct lens distortion; Flame segmentation: Semantic segmentation of images is performed based on the U-Net deep learning model to extract the contour of the flame region, with a segmentation accuracy of ≥95%; Feature extraction: Calculate the geometric center (mean pixel coordinates) of the flame region, the centroid and edge feature points (inflection points, vertices) of the high-temperature core region (the first 30% of the grayscale value); 3D Reconstruction: The intrinsic parameters (focal length, principal point coordinates) and extrinsic parameters (rotation matrix, translation vector) of the four cameras are obtained through the Zhang Zhengyou calibration method. The 3D coordinates of the feature points are calculated based on the binocular vision matching algorithm, and the spatial position (X, Y, Z) of the flame center is finally determined with a positioning error ≤50 mm.

[0037] The correction control module is a PLC control system with a built-in flame deflection-damper adjustment mapping model and control logic. Using the furnace geometric center (0, 0, 0) as a reference, the offset parameters of the flame center are calculated: offset distance. The offset direction θ = arctan(Y / X) (angle range 0°~360°). The model is constructed using a BP neural network, with the input layer being (D, θ) and the output layer being the opening correction value of each damper at each angle burner (angles A / B / C / D). The neural network is trained using historical operating data, including skew cases (offset parameters and corresponding optimal damper adjustment amounts), with a training error ≤3%. The single damper adjustment amplitude is ≤5% to avoid drastic disturbances; when the offset distance D>500 mm, an emergency stop protection is triggered, emitting an audible and visual alarm.

[0038] The flame deflection-damper adjustment mapping model uses a BP neural network, and its specific construction is as follows: (1) Input Layer Offset distance D, offset direction θ.

[0039] The input parameters are normalized (mapped to the [0,1] interval) to eliminate the impact of dimensional differences on training. , (1) In the formula, the offset distance D and the offset direction θ; (2) Hidden Layer Two hidden layers are used: the first hidden layer has 12 neurons (activation function: ReLU, to solve the gradient vanishing problem), and the second hidden layer has 8 neurons (activation function: ReLU).

[0040] The input dimension is low (2-dimensional) but the output dimension is high (depending on the number of dampers). The nonlinear mapping capability is enhanced by two hidden layers. The number of neurons is designed based on the empirical formula of "input layer × 6~8 times".

[0041] (3) Output Layer Number of neurons: n (n equals the total number of damper layers of the four-corner burner (corners A / B / C / D). For example, if there is one damper at each of the four corners in each layer, and there are a total of 6 layers, then n=24.

[0042] Output parameters: Opening correction values ​​for each damper (Unit: %, range [-5%, +5%], subject to safety constraint "single adjustment range ≤ 5%") Post-processing: Output values ​​are denormalized and mapped to the actual physical range, and values ​​exceeding [-5, +5] are forcibly truncated. (2) (Note: y i The output layer normalized value [0,1] is multiplied by 10 and subtracted by 5 to map to [-5, +5] The network training process is as follows: (1) Dataset construction Sample source: "skew cases" in historical operation data, including flame offset parameters (D, θ), and the "optimal damper opening correction value" for the corresponding cases that has been manually optimized or experimentally verified.

[0043] Sample size: It is recommended to have no less than 5,000 groups (including scenarios with different loads, coal types, and degrees of skewness to ensure generalization).

[0044] Dataset split: training set (70%), validation set (20%), test set (10%).

[0045] (2) Initialize parameters Weights and biases: Initialized using Xavier Learning rate: The initial value is set to 0.001 (optimized using the Adam adaptive learning rate algorithm and dynamically adjusted).

[0046] Training rounds: Initially set to 1000 rounds.

[0047] (3) Forward propagation calculation Hidden layer output: Hidden layer 1: (W1 is the weight matrix from the input layer to the first hidden layer, and b1 is the bias).

[0048] Second hidden layer: (W2 is the weight matrix of the first to second hidden layers, b2) is the bias.

[0049] Output layer output: (W3 is the weight matrix from the second hidden layer to the output layer, b3 is the bias, there is no activation function, and the output is a continuous value directly).

[0050] (4) Definition of loss function Mean squared error (MSE) is used to measure the deviation between the predicted correction value and the actual optimal correction value, with a target training error ≤ 3%. (3) In the formula, m is the number of samples, n is the number of output neurons, and the error is expressed as a percentage: Loss% = Loss × 100) (5) Backpropagation and parameters The partial derivatives of the loss function with respect to the weights (W1, W2, W3) and biases (b1, b2, b3) of each layer are solved using the chain rule. The Adam algorithm (which combines momentum and adaptive learning rate to accelerate convergence and avoid local optima) is used, and an L2 regularization term is added to suppress overfitting.

[0051] The execution module consists of a servo motor, a reduction gearbox, a position sensor, and a transmission mechanism, and interfaces with the existing mechanical structure of the boiler's damper. The servo motor has a response frequency ≥10 Hz and a position control accuracy of ±0.5%FS, receiving opening commands from the correction control module via a 4~20 mA signal. The position sensor collects the actual damper opening in real time and feeds it back to the correction control module as a digital signal, forming a closed-loop control to ensure adjustment accuracy.

[0052] The correction principle is as follows: the rotational momentum of the flame in a tangential boiler is determined by the combined momentum of the four airflows. When the airflow at one corner is too large, the flame center will shift towards the opposite corner. Based on this, the correction logic of this invention is as follows: When the flame deviates towards angle A (θ=0°~90°), increase the secondary air opening at angle A (increase its momentum), decrease the secondary air opening at angle C (diagonal) (decrease its momentum), and at the same time adjust the primary air opening proportionally to make the combined momentum return to the center of the furnace. The opening adjustment amount is positively correlated with the offset distance D. The larger D is, the larger the adjustment range, and the maximum single adjustment amount does not exceed 5%. For vertical (Z-axis) offset, correction is achieved by adjusting the swing angle of the upper and lower burners, with an adjustment range of ±30°.

Claims

1. A four-corner tangential boiler flame deflection correction system, characterized in that, It includes a flame image acquisition module, which is connected to a flame center recognition module. The flame center recognition module is connected to a correction control module. The correction control module has a built-in flame deflection-damper adjustment mapping model and is connected to an execution module.

2. The four-corner tangential boiler flame deflection correction system according to claim 1, characterized in that, The flame image acquisition module consists of high-temperature cameras installed at the four corners of the furnace, with the high-temperature cameras facing the center of the furnace, for real-time acquisition of dynamic flame images; The flame center identification module is used to preprocess flame images, extract features, and reconstruct three dimensions, and calculate the spatial offset parameters of the flame center, including the offset direction and offset distance.

3. The four-corner tangential boiler flame deflection correction system according to claim 2, characterized in that, The flame deflection-damper adjustment mapping model generates opening adjustment commands for each burner damper based on the offset parameters. The flame deflection-damper adjustment mapping model is trained using a BP neural network. The input is the offset parameter, and the output is the opening correction value of the damper for each burner layer. The training samples include the offset parameters of historical deflection cases and the corresponding optimal damper adjustment amount.

4. The four-corner tangential boiler flame deflection correction system according to claim 3, characterized in that, The execution module includes a servo motor and a transmission mechanism, used to receive the adjustment command and drive the damper to perform opening adjustment.

5. The four-corner tangential boiler flame deflection correction system according to claim 4, characterized in that, The execution module also includes a position feedback unit, which is used to collect the actual opening degree of the damper in real time and feed it back to the correction control module to form a closed-loop control.

6. The four-corner tangential boiler flame deflection correction system according to claim 2, characterized in that, The high-temperature camera uses a water-cooled protective cover, and the lens is equipped with an infrared filter. It can withstand the external ambient temperature of the furnace from -20 ℃ to 80 ℃, with an image resolution of no less than 1920×1080 and a frame rate of no less than 25 fps.

7. A method for correcting flame deflection in a four-corner tangential boiler, characterized in that, Using the four-corner tangential boiler flame deflection correction system as described in claim 5 includes the following steps: S1: Simultaneously collect dynamic images of the flame through high-temperature cameras at the four corners of the furnace and transmit them to the flame center recognition module; S2: The flame center recognition module preprocesses, segments, and reconstructs the image in three dimensions, and calculates the spatial offset parameters of the flame center; S3: The correction control module inputs the spatial offset parameters into the flame deflection-damper adjustment mapping model and outputs the opening adjustment command of each burner damper; S4: The execution module drives the damper to adjust according to the adjustment command and realizes closed-loop control through position feedback; S5: Repeat steps S1-S4 until the flame center offset distance is less than the preset threshold.

8. The method for correcting flame deviation in a four-corner tangential boiler according to claim 7, characterized in that, In S2, the three-dimensional reconstruction adopts a binocular vision matching algorithm, which calculates the three-dimensional coordinates of flame feature points through the calibration parameters of the four corner cameras. The calibration parameters include intrinsic parameter matrix, distortion coefficient and extrinsic parameter matrix.

9. The method for correcting flame deviation in a tangential boiler according to claim 7, characterized in that, The generation rule for the opening adjustment command mentioned in S3 is as follows: when the flame shifts to a certain angle, the secondary air opening of the burner corresponding to that angle is increased, the secondary air opening of the burner at the opposite angle is decreased, and the primary air opening is adjusted proportionally in coordination. The specific process of constructing the flame deflection-damper adjustment mapping model described in S3 is as follows: A backpropagation (BP) neural network is used, consisting of an input layer, hidden layers, and an output layer. The input layer is specifically: Input parameters are normalized to eliminate the impact of dimensional differences on training. ， （1） In the formula, the offset distance D and the offset direction θ; The hidden layers specifically consist of two layers: the first hidden layer contains 12 neurons, and the second hidden layer contains 8 neurons. The output layer specifically consists of n neurons, where n is equal to the total number of damper layers at the four corners of the four-corner burner. The output parameters are the opening correction values ​​for each damper. The unit is %, the range is [-5%, +5%], and the single adjustment range is ≤5%; The output values ​​are denormalized to the actual physical range, and values ​​exceeding [-5, +5] are forcibly truncated. （2） In the formula, y i The output layer normalized value [0,1] is multiplied by 10 and subtracted by 5 and then mapped to [-5, +5].

10. The method for correcting flame deflection in a tangential boiler according to claim 7, characterized in that, The preset threshold mentioned in S5 is set according to the boiler model and has a value range of 0~300 mm. When the offset distance is less than the threshold for more than 10 seconds, the system enters standby mode.

Images