Boiler wall-adhering gas conveying method and system for reducing ash deposition and slag bonding

Abstract

The invention discloses a boiler wall-adhering gas conveying method and system for reducing ash deposition and slag bonding, and relates to the field of boiler cleaning.The method comprises the steps that a pipeline ash deposition detection sample is collected, and pipeline ash deposition thickness distribution and pipeline ash deposition adhesive force distribution are counted; a boiler wall-adhering gas conveying scheme is configured; the rotational flow wind speed, the rotational flow wind direction, the pressure in the pipe, the cleaning duration, the pipeline ash deposition thickness distribution and the pipeline ash deposition adhesion distribution are processed through the residual ash thickness prediction model, and the residual ash thickness distribution is output; processing the rotational flow wind speed, the rotational flow wind direction, the in-pipe pressure and the cleaning duration through the energy consumption prediction model, and outputting a cleaning energy consumption prediction value; and when the residual ash thickness at any position of the residual ash thickness distribution is smaller than or equal to the residual ash thickness threshold value and the cleaning energy consumption predicted value is smaller than or equal to the energy consumption threshold value, boiler wall-adhering gas conveying control is conducted according to the boiler wall-adhering gas conveying scheme, and pipeline cleaning in the boiler is executed. The problem that residual coal ash in peripheral pipelines cannot be cleaned in the prior art is solved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of boiler cleaning, in particular to a boiler wall-attached gas delivery method and system for reducing ash and slag accumulation. BACKGROUND

[0002] With the development of industry, boilers are increasingly widely used in various production activities. However, the traditional boiler cleaning method uses a straight-through blowing device, which can only clean some obvious ash, but cannot effectively clean the residual ash in the surrounding pipes. The residual ash attached to the pipes for a long time causes pipe corrosion, pipe perforation and water leakage. If the pipe leaks after water leakage, the boiler needs to be shut down for maintenance, causing unnecessary losses, shortening the service life of the boiler, and even causing safety accidents. SUMMARY

[0003] The embodiments of the present application provide a boiler wall-attached gas delivery method and system for reducing ash and slag accumulation, which solves the technical problem that the prior art cannot clean a large amount of residual ash in the surrounding pipes.

[0004] The technical solution of the present application to solve the above technical problems is as follows: In a first aspect, the present application provides a boiler wall-attached gas delivery method for reducing ash and slag accumulation, comprising: Collecting a pipe ash detection sample set and counting pipe ash thickness distribution and pipe ash adhesion force distribution, with the boiler operating parameters, boiler model and boiler airflow monitoring parameters of a target boiler as constraints; Configuring a boiler wall-attached gas delivery scheme, wherein the boiler wall-attached gas delivery scheme includes cyclone wind speed, cyclone wind direction, pipe internal pressure and cleaning duration; Processing the cyclone wind speed, the cyclone wind direction, the pipe internal pressure, the cleaning duration, the pipe ash thickness distribution and the pipe ash adhesion force distribution through a residual ash thickness prediction model bound to the target boiler, and outputting a residual ash thickness distribution; Processing the cyclone wind speed, the cyclone wind direction, the pipe internal pressure and the cleaning duration through an energy consumption prediction model bound to the target boiler, and outputting a cleaning energy consumption prediction value; In a second aspect, the present application provides a boiler wall-attached gas delivery system for reducing ash and slag accumulation, comprising: A sample collection module for collecting a pipe ash detection sample set and counting pipe ash thickness distribution and pipe ash adhesion force distribution, with the boiler operating parameters, boiler model and boiler airflow monitoring parameters of a target boiler as constraints; A scheme configuration module for configuring a boiler wall-attached gas delivery scheme, wherein the boiler wall-attached gas delivery scheme includes cyclone wind speed, cyclone wind direction, pipe internal pressure and cleaning duration; a residual ash thickness prediction module configured to process the rotational flow speed, the rotational flow direction, the pressure inside the tube, the cleaning duration, the tube ash thickness distribution, and the tube ash adhesion force distribution by a residual ash thickness prediction model bound to the target boiler, and output a residual ash thickness distribution; an energy consumption output module configured to process the rotational flow speed, the rotational flow direction, the pressure inside the tube, and the cleaning duration by an energy consumption prediction model bound to the target boiler, and output a cleaning energy consumption prediction value; a scheme execution module configured to perform a boiler wall-attached gas delivery control according to the boiler wall-attached gas delivery scheme when residual ash thickness at any position of the residual ash thickness distribution is less than or equal to a residual ash thickness threshold value and the cleaning energy consumption prediction value is less than or equal to an energy consumption threshold value, and perform in-tube cleaning.

[0005] The present application provides one or more technical solutions, at least having the following technical effects or advantages: The boiler wall-attached gas delivery method and system provided by the present application can effectively control energy consumption while ensuring cleaning effect by comprehensively considering parameters of boiler operation, statistically analyzing tube ash deposition, and predicting residual ash thickness and cleaning energy consumption by using prediction models. When residual ash thickness and energy consumption both meet the requirements, the boiler wall-attached gas delivery control is performed to realize efficient and energy-saving in-tube cleaning. When the threshold conditions are not met, the fitness of the scheme is calculated by configuring weights, and the particle swarm optimization method is combined to continuously optimize the gas delivery scheme, so that a more suitable target gas delivery scheme is finally found, further improving the cleaning effect and energy utilization efficiency.

[0006] Through the above technical solutions, the tube ash thickness and tube ash adhesion force distribution of the target boiler are collected, and the boiler wall-attached gas delivery scheme is configured. Then, the residual ash thickness distribution and the cleaning energy consumption prediction value are output by using the residual ash thickness prediction model and the energy consumption prediction model, respectively. The problem that the residual coal ash in the surrounding tube cannot be cleaned in the prior art is solved, the occurrence of tube corrosion, perforation, and other situations is reduced, the loss caused by shutdown for maintenance due to tube leakage is avoided, the service life of the boiler is prolonged, the risk of safety accidents is reduced, and the stable operation of the boiler in industrial production is ensured. BRIEF DESCRIPTION OF DRAWINGS

[0007] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings needed in the embodiment description will be briefly introduced. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative labor.

[0008] Figure 1is a flowchart of a boiler wall-attached gas conveying method for reducing ash deposition and slagging provided by an embodiment of the present application; Figure 2 is a structural diagram of a boiler wall-attached gas conveying system for reducing ash deposition and slagging provided by an embodiment of the present application.

[0009] In the drawings, the components represented by various reference numerals are as follows: The sample collection module 11, the scheme configuration module 12, the ash thickness prediction module 13, the energy consumption output module 14, and the scheme execution module 15. DETAILED DESCRIPTION

[0010] The embodiment of the present application provides the boiler wall-attached gas conveying method for reducing ash deposition and slagging, which is used for solving the technical problem that a large amount of residual coal ash in the peripheral pipeline cannot be cleaned in the prior art.

[0011] The technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments of the present application. Based on the embodiments in the present application, all other embodiments obtained by a person skilled in the art without creative work fall within the scope of protection of the present application.

[0012] In the description of the present application, the terms "first", "second" are used only for the purpose of description, and cannot be understood as indicating or implying relative importance or implicitly indicating the number of the indicated technical features. Therefore, the features defined as "first", "second" can explicitly or implicitly include one or more of the features. In the description of the present application, the meaning of "multiple" is two or more, unless otherwise specifically limited.

[0013] In the description of the present application, the term "for example" is used to indicate "as an example, illustration or explanation". Any embodiment described as "for example" in the present application is not necessarily interpreted as more preferred or more advantageous than other embodiments. In order to enable any person skilled in the art to implement and use the present application, the following description is given. In the following description, details are listed for the purpose of explanation. It should be understood that a person skilled in the art can realize the present application without using these specific details. In other examples, well-known structures and processes will not be described in detail to avoid unnecessary details making the description of the present application obscure. Therefore, the present application is not intended to be limited to the shown embodiments, but is consistent with the broadest scope in accordance with the principles and characteristics disclosed in the present application.

[0014] Embodiment one, as Figure 1 shown, the embodiment of the present application provides a boiler wall-attached gas conveying method for reducing ash deposition and slagging, comprising: S10: Collecting a pipe ash detection sample set with the boiler operation parameters, the boiler model and the boiler airflow monitoring parameters of the target boiler as constraints, and counting the pipe ash thickness distribution and the pipe ash adhesion force distribution; In the embodiment of the application, first, the pipe ash detection sample set is collected with the boiler operation parameters, the boiler model and the boiler airflow monitoring parameters of the target boiler as constraints, and data quality inspection is performed on the collected pipe ash detection sample set to check whether there is data missing, abnormal value and the like. If there is data missing, the mean value, median value and the like can be used to fill in according to the distribution of other data in the sample set; if there is abnormal value, the abnormal data can be identified and removed through statistical analysis method, for example, based on standard deviation.

[0015] The boiler operation parameters are the operation parameters after the last cleaning, and the boiler airflow monitoring parameters are the airflow monitoring parameters after the last cleaning.

[0016] Secondly, the pipe ash thickness distribution and the pipe ash adhesion force distribution obtained by the statistical pipe ash are visually displayed. For example, the column chart is used to display the pipe ash thickness at different positions, and the line chart is used to display the change trend of the pipe ash adhesion force.

[0017] Specifically, step S10 in the method comprises: Collecting a pipe ash detection sample to be analyzed with the boiler model as a constraint, wherein the pipe ash detection sample to be analyzed has a label identifying the boiler operation parameters and a label identifying the boiler airflow monitoring parameters; When the boiler operation parameters and any attribute boiler operation parameter deviation of the label identifying the boiler operation parameters are all less than or equal to the corresponding attribute boiler operation parameter deviation threshold, the boiler operation parameters are considered to be consistent; When the boiler airflow monitoring parameters and any attribute boiler airflow monitoring parameter deviation of the label identifying the boiler airflow monitoring parameters are all less than or equal to the corresponding attribute boiler airflow monitoring parameter deviation threshold, the boiler airflow monitoring parameters are considered to be consistent; When the boiler operation parameters are consistent and the boiler airflow monitoring parameters are consistent, the pipe ash detection sample to be analyzed is added to the pipe ash detection sample set.

[0018] In the embodiment of the application, the pipe ash detection sample to be analyzed is collected with the boiler model as a constraint, and the pipe ash detection sample to be analyzed has a label identifying the boiler operation parameters and a label identifying the boiler airflow monitoring parameters.

[0019] After the sample of the collected pipeline ash is analyzed, the label data is further verified. By comparing with the historical operation data and the airflow monitoring data of the boiler, it is checked whether the label data is consistent with the normal variation range and regularity. If it is found that the label data has obvious unreasonable situation, for example, a certain operating parameter exceeds the normal range of the boiler of this type, or the airflow monitoring parameter does not match the actual pipeline layout and gas conveying situation, the sample needs to be further verified or rejected.

[0020] If the deviation of the boiler operating parameter from any attribute boiler operating parameter of the label of the boiler operating parameter is less than or equal to the corresponding attribute boiler operating parameter deviation threshold, the boiler operating parameter is considered to be consistent; if the deviation of the boiler airflow monitoring parameter from any attribute boiler airflow monitoring parameter of the label of the boiler airflow monitoring parameter is less than or equal to the corresponding attribute boiler airflow monitoring parameter deviation threshold, the boiler airflow monitoring parameter is considered to be consistent.

[0021] Exemplarily, it is assumed that the boiler operating parameters include temperature, pressure, water level and the like, and the corresponding attribute boiler operating parameter deviation thresholds are set to temperature deviation ±5℃, pressure deviation ±0.1MPa and water level deviation ±2cm. When the label of the collected sample of the pipeline ash to be analyzed shows that the temperature is 150℃, the pressure is 2.0MPa and the water level is 50cm, while the actual operating parameters of the target boiler are temperature 152℃, pressure 2.05MPa and water level 51cm. At this time, the temperature deviation is 2℃, which is less than the temperature deviation threshold 5℃; the pressure deviation is 0.05MPa, which is less than the pressure deviation threshold 0.1MPa; the water level deviation is 1cm, which is less than the water level deviation threshold 2cm, and the boiler operating parameters are considered to be consistent.

[0022] Similarly, it is assumed that the boiler airflow monitoring parameters include wind speed, wind direction, wind pressure and the like, and the corresponding attribute boiler airflow monitoring parameter deviation thresholds are set to wind speed deviation ±1m / s, wind direction deviation ±5° and wind pressure deviation ±0.05MPa. If the label of the collected sample of the pipeline ash to be analyzed shows that the wind speed is 5m / s, the wind direction is due east and the wind pressure is 1.5MPa, while the actual airflow monitoring parameters of the target boiler are wind speed 5.2m / s, wind direction east-northeast 3° and wind pressure 1.52MPa. The wind speed deviation is 0.2m / s, which is less than the wind speed deviation threshold 1m / s; the wind direction deviation is 3°, which is less than the wind direction deviation threshold 5°; the wind pressure deviation is 0.02MPa, which is less than the wind pressure deviation threshold 0.05MPa, and the boiler airflow monitoring parameters are considered to be consistent.

[0023] When the boiler operating parameters are consistent and the boiler airflow monitoring parameters are consistent, the to-be-analyzed pipeline ash detection sample is added to the pipeline ash detection sample set. Through the data screening process, the accuracy and reliability of the collected pipeline ash detection sample set are ensured, laying a foundation for subsequent accurate statistics of the pipeline ash thickness distribution and the pipeline ash adhesion force distribution, so that the boiler wall-attached gas delivery scheme generated based on the data is more reasonable, effectively improving the cleaning effect of the boiler pipeline and the energy utilization efficiency.

[0024] The method comprises the following steps: The inner wall of the target boiler pipeline is divided into a grid distribution by a preset edge length grid. Based on the grid distribution of the inner wall of the boiler pipeline, the pipeline ash detection sample set is divided to obtain a first pipeline grid distribution ash state to a Qth pipeline grid distribution ash state, where Q represents the total number of samples, and Q is greater than or equal to 30. The first pipeline grid distribution ash state to the Qth pipeline grid distribution ash state are integrated to obtain a first grid ash state set to an Mth grid ash state set, where M represents the total number of grids. The first grid ash state set to the Mth grid ash state set are traversed to perform centroid ash state analysis to obtain the pipeline ash thickness distribution and the pipeline ash adhesion force distribution.

[0025] In the embodiments of the present application, the inner wall of the target boiler pipeline is divided into a grid distribution by a preset edge length grid, which discretizes the complex pipeline inner wall space. For example, the inner wall of the target boiler pipeline is divided into small grids of 3 cm x 3 cm to obtain the grid distribution of the inner wall of the boiler pipeline.

[0026] First, after obtaining the grid distribution of the inner wall of the boiler pipeline, the pipeline ash detection sample set is divided based on the grid distribution, each sample corresponds to a pipeline grid distribution ash state, and the sample data is associated with a specific grid position.

[0027] Secondly, the first pipeline grid distribution ash state to the Qth pipeline grid distribution ash state are obtained, where Q represents the total number of samples, and Q is greater than or equal to 30. The larger the value of Q, the more accurate the calculation. This step can eliminate the random error between samples and make the ash state of each grid more accurate and stable.

[0028] Thirdly, the first pipeline grid distribution ash state to the Qth pipeline grid distribution ash state are integrated to obtain a first grid ash state set to an Mth grid ash state set. Through integration, the ash states of multiple samples in the same grid are combined and counted. Wherein, M represents the total number of grids.

[0029] Finally, the centroid soot state analysis is performed on the set of states, which can find the typical characteristics of each grid soot state, such as the average soot thickness, the average soot adhesion, etc. By analyzing and summarizing the centroid soot states of all grids, the pipe soot thickness distribution and the pipe soot adhesion distribution can be finally obtained.

[0030] In performing the centroid soot state analysis, statistical methods are adopted. For example, for the soot thickness data in each grid soot state set, the mean, median, standard deviation, etc. are calculated to obtain the concentration trend and dispersion degree of the soot thickness. For the soot adhesion data, the same method is used for analysis.

[0031] Illustratively, the average soot thickness of 50 soot thickness data in a certain grid soot state set is 2mm. The median is calculated by first sorting the data from small to large. If the number of data is odd, the middle number is the median; if the number of data is even, the average of the two middle numbers is the median. Assuming that the two numbers in the middle are 2mm and 2.1mm after sorting, the median is 2.05mm.

[0032] The standard deviation reflects the dispersion degree of the data, which is obtained by calculating the square root of the average of the square sum of the difference between each data and the mean. If the standard deviation is small, the data is relatively concentrated, and the soot thickness is relatively stable; if the standard deviation is large, the data is relatively dispersed, and the soot thickness fluctuates greatly.

[0033] For the soot adhesion data, the same statistical analysis is performed. For example, there are 50 soot adhesion data in a certain grid soot state set, which are 0.5N, 0.6N, 0.4N, etc. The mean, median and standard deviation are calculated to understand the concentration trend and dispersion degree of the grid soot adhesion.

[0034] The pipe soot thickness distribution and the pipe soot adhesion distribution obtained by the above steps provide detailed and accurate data support for subsequent configuration of the boiler wall gas conveying scheme.

[0035] Further, the centroid soot state analysis is performed on the set of states, which can find the typical characteristics of each grid soot state, such as the average soot thickness, the average soot adhesion, etc. By analyzing and summarizing the centroid soot states of all grids, the pipe soot thickness distribution and the pipe soot adhesion distribution can be finally obtained. From the first grid soot state set, a number of soot states are extracted, wherein any one of the number of soot states includes soot thickness and soot adhesion; The soot thickness normalized value is taken as the first-dimensional coordinate, and the soot adhesion normalized value is taken as the second-dimensional coordinate. The distribution of the number of soot states is performed to obtain a number of soot state distribution coordinates; Based on the several deposition state distribution coordinates, LOF outlier factor minimum value extraction is performed to obtain a centroid deposition state distribution coordinate; Based on the centroid deposition state distribution coordinate, a first grid pipeline deposition thickness and a first grid pipeline deposition adhesion are extracted; When the Mth grid pipeline deposition thickness and the Mth grid pipeline adhesion are obtained, the first grid pipeline deposition thickness to the Mth grid pipeline deposition thickness is integrated to obtain the pipeline deposition thickness distribution, and the first grid pipeline deposition adhesion to the Mth grid pipeline adhesion is integrated to obtain the pipeline deposition adhesion distribution.

[0036] In the embodiments of the present application, when the several deposition states are extracted from the first grid deposition state set, it is ensured that the extracted deposition states are representative and can reflect the overall characteristics of the grid deposition state. For deposition thickness and deposition adhesion, since the dimensions and numerical ranges may be different, normalization processing is required for subsequent analysis and processing. There are many normalization methods, for example, "minimum-maximum" normalization, which maps the deposition thickness and deposition adhesion data to the [0, 1] interval.

[0037] Further, the deposition thickness normalized value is taken as the first-dimensional coordinate, the deposition adhesion normalized value is taken as the second-dimensional coordinate, and the distribution of the several deposition states is performed to obtain several deposition state distribution coordinates, and the deposition state is visualized and displayed on a two-dimensional plane.

[0038] Exemplarily, 10 deposition states are extracted from a certain grid deposition state set, the deposition thickness normalized values are 0.2, 0.3, 0.4, etc., and the deposition adhesion normalized values are 0.1, 0.2, 0.3, etc. Taking the deposition thickness normalized value as the abscissa and the deposition adhesion normalized value as the ordinate, the 10 deposition state distribution coordinates can be obtained, such as (0.7, 0.1), (0.3, 0.2), etc.

[0039] Based on the several deposition state distribution coordinates, LOF outlier factor minimum value extraction is performed. LOF local outlier factor is an algorithm for identifying outliers in a data set. By calculating the local outlier factor of each point, the point that is relatively isolated from the surrounding points is found. The point corresponding to the LOF outlier factor minimum value is extracted, which is the centroid deposition state distribution coordinate.

[0040] Exemplarily, the No. 1 sample deposition state distribution coordinate is (0.7, 0.1), the No. 2 sample deposition state distribution coordinate is (0.3, 0.2), and the No. 3 sample deposition state distribution coordinate is (0.99, 0.8).

[0041] LOF outlier factor minimum extraction is performed, LOF = LRD average value of sample k neighbors / LRD of sample itself, the smaller the LOF value, the weaker the outlier, the closer to the local center. Calculation ≈1.0936, ≈0.9572, ≈0.9572, After comparison, the LOF values of the second sample and the third sample are the smallest and equal, which are local non-outlier samples. The centroid soot state distribution coordinates = (0.3, 0.2) are obtained, that is, the coordinates of the second sample.

[0042] Further, based on the centroid soot state distribution coordinates, the first grid pipeline soot thickness and the first grid pipeline soot adhesion are extracted, for example, according to the coordinates (0.3, 0.2) of the second sample, the first grid pipeline soot thickness and the first grid pipeline soot adhesion are 0.3 and 0.2 respectively.

[0043] Finally, when the Mth grid pipeline soot thickness and the Mth grid pipeline adhesion are obtained, the first grid pipeline soot thickness to the Mth grid pipeline soot thickness are integrated to obtain the pipeline soot thickness distribution.

[0044] The first grid pipeline soot adhesion to the Mth grid pipeline adhesion are integrated to obtain the pipeline soot adhesion distribution.

[0045] By integrating the pipeline soot thickness distribution and the pipeline soot adhesion distribution, when configuring the boiler wall gas conveying scheme subsequently, according to the soot conditions at different positions, the direction, speed and pressure and other parameters of gas conveying are accurately adjusted. For example, for the grid position with larger soot thickness and stronger adhesion, the pressure and speed of gas conveying are increased to improve the cleaning effect; for the position with thinner soot and weaker adhesion, the intensity of gas conveying is appropriately reduced to avoid waste of energy.

[0046] S20: configuring a boiler wall gas conveying scheme, wherein the boiler wall gas conveying scheme includes cyclone wind speed, cyclone wind direction, pipe pressure and cleaning duration; In the embodiment of the application, the boiler wall gas conveying scheme is configured in combination with the obtained pipeline soot thickness distribution and pipeline soot adhesion distribution. The boiler wall gas conveying scheme includes cyclone wind speed, cyclone wind direction, pipe pressure and cleaning duration; For the cyclone wind speed, in the area with large soot thickness and strong adhesion, higher cyclone wind speed is needed to generate sufficient impact force to remove soot. For example, when the pipeline soot thickness of a certain area is greater than 3 mm and the soot adhesion is greater than 0.8 N, the cyclone wind speed can be set to 20 m / s; while in the area with thinner soot and weaker adhesion, such as the pipeline soot thickness being less than 1 mm and the soot adhesion being less than 0.2 N, the cyclone wind speed can be set to 10 m / s.

[0047] The swirl air direction is determined according to the layout of the pipeline and the distribution of the accumulated ash. For the area where there is an accumulated ash dead angle, the swirl air direction is adjusted to directly act on the accumulated ash part. For example, the accumulated ash is serious at the bend of the pipeline, and the swirl air direction needs to be adjusted to be consistent with the tangent direction of the bend to enhance the scouring effect on the accumulated ash at the bend.

[0048] For the pressure in the pipe, in the area where the accumulated ash is serious, appropriately increasing the pressure in the pipe can increase the impact force of the gas. For example, when the accumulated ash thickness of the pipeline exceeds 2 mm, the pressure in the pipe is set to 0.5 MPa; in the area where the accumulated ash is thin, the pressure in the pipe is reduced to 0.2 MPa to avoid unnecessary pressure damage to the pipeline.

[0049] For the cleaning time, the overall situation of the accumulated ash and the operation arrangement of the boiler need to be considered. For example, for the pipeline where the accumulated ash is evenly distributed and the accumulated ash thickness is moderate, the cleaning time can be set to 3 hours; if the accumulated ash is unevenly distributed and the accumulated ash is serious in some areas, the cleaning time is extended to 5 hours. At the same time, it is necessary to avoid the influence of the cleaning time on the normal operation of the boiler, and it is necessary to reasonably arrange the cleaning time, for example, the accumulated ash cleaning can be carried out during the low load operation period of the boiler.

[0050] By reasonably configuring the swirl air speed, the swirl air direction, the pressure in the pipe and the cleaning time, a more scientific and effective boiler wall-attached gas conveying scheme can be developed, thereby improving the cleaning effect of the boiler pipeline, reducing the influence of the accumulated ash and slag on the operation of the boiler, and further improving the energy utilization efficiency and the operation stability of the boiler.

[0051] S30: processing the swirl air speed, the swirl air direction, the pressure in the pipe, the cleaning time, the pipeline accumulated ash thickness distribution and the pipeline accumulated ash adhesion force distribution through the residual ash thickness prediction model bound with the target boiler, and outputting a residual ash thickness distribution; In the embodiment of the present application, a residual ash thickness prediction model is constructed, and the swirl air speed, the swirl air direction, the pressure in the pipe, the cleaning time, the pipeline accumulated ash thickness distribution and the pipeline accumulated ash adhesion force distribution are input to accurately predict the residual ash thickness distribution in the boiler pipeline after cleaning.

[0052] For example, in the input data, the pipeline accumulated ash thickness distribution of a certain area shows that the accumulated ash is thick and the accumulated ash adhesion force is strong, and at the same time, the swirl air speed is set to a high value, the pressure in the pipe is large, and the cleaning time is long, the model predicts that the residual ash thickness of this area after cleaning may be relatively thick; while for the area where the accumulated ash is thin and the adhesion force is weak, if the swirl air speed is low, the pressure in the pipe is small and the cleaning time is short, the model will predict that the residual ash thickness of this area is relatively thin.

[0053] Specifically, step S30 in the method comprises: The graph neural network architecture is constructed based on the grid distribution of the inner wall of the boiler pipeline, wherein the input nodes of the graph neural network architecture include grid input nodes, gas delivery input nodes and grid output nodes, wherein the grid input nodes and the grid output nodes are one-to-one corresponding to the grid distribution of the inner wall of the boiler pipeline, the gas delivery input nodes correspond to the air inlet position of the pipeline, the grid input nodes are used to receive the pipeline ash deposition thickness distribution and the pipeline ash deposition adhesion force distribution, the gas delivery input nodes are used to receive the cyclone wind speed, the cyclone wind direction, the pipeline pressure, the cleaning time length, and the grid output nodes are used to output the residual ash thickness distribution. Based on the boiler model, the cyclone wind speed record data, the cyclone wind direction record data, the pipeline pressure record data, the cleaning time length record data, the pipeline ash deposition thickness distribution record information, the pipeline ash deposition adhesion force distribution record information and the label identifying the residual ash thickness distribution are collected, and the graph neural network architecture is trained to generate the residual ash thickness prediction model.

[0054] In the embodiments of the present application, based on the grid distribution of the inner wall of the boiler pipeline, the neural network has strong non-linear mapping capability, and the graph neural network architecture is constructed based on the neural network. Based on the effective grid input nodes, the gas delivery input nodes and the grid output nodes, the residual ash thickness prediction model is obtained through iterative training.

[0055] Exemplarily, the iterative training of the residual ash thickness prediction model can be realized through the following technical path: First, data preparation, the input nodes of the graph neural network architecture include grid input nodes, gas delivery input nodes and grid output nodes, wherein the grid input nodes and the grid output nodes are one-to-one corresponding to the grid distribution of the inner wall of the boiler pipeline, the gas delivery input nodes correspond to the air inlet position of the pipeline, the grid input nodes are used to receive the pipeline ash deposition thickness distribution and the pipeline ash deposition adhesion force distribution, the gas delivery input nodes are used to receive the cyclone wind speed, the cyclone wind direction, the pipeline pressure, the cleaning time length, and the grid output nodes are used to output the residual ash thickness distribution.

[0056] Second, model building, the number of nodes of the input layer is equal to the dimension of the input features, such as 10 features including the cyclone wind speed record data, the cyclone wind direction record data, the pipeline pressure record data, the cleaning time length record data, the pipeline ash deposition thickness distribution record information, the pipeline ash deposition adhesion force distribution record information and the label identifying the residual ash thickness distribution, so the input layer contains 10 nodes; 1-3 hidden layers are set, the number of nodes of each layer is adjusted through experiments, such as 64, 32, etc., and the activation function is selected as ReLU; the number of nodes of the output layer is equal to the number of prediction targets, such as 1 node for predicting only the time consumption, and 2 nodes for simultaneously predicting the time consumption and the energy consumption, and the output layer generally does not use the activation function, and directly outputs continuous values.

[0057] Again, model training, residual ash thickness prediction as output, corresponding prediction sample in training set as supervision label, using Adam optimizer and mean square error (MSE) loss function to build training framework, setting batch size to 32, total training rounds to 50, and introducing early stopping mechanism (patience = 5), when the validation set loss does not appear for 5 consecutive rounds, the training process is automatically terminated, and the trained residual ash thickness prediction model is obtained, which effectively avoids model overfitting while ensuring that the model reaches a convergent state. Bind the trained residual ash thickness prediction model to the target boiler.

[0058] S40: Through the energy consumption prediction model bound to the target boiler, the cyclone wind speed, the cyclone wind direction, the pressure in the pipe, and the cleaning time are processed, and a cleaning energy consumption prediction value is output. In the embodiments of the present application, an energy consumption prediction model is constructed, and the cyclone wind speed, cyclone wind direction, pressure in the pipe, and cleaning time are input to accurately predict the energy consumption in the boiler wall-attached gas conveying and cleaning process.

[0059] The neural network model can automatically learn the complex mapping relationship between the input variables and the output variables through the combination of multiple layers of neurons. Similarly, an energy consumption prediction model is constructed based on a neural network, and the cyclone wind speed, cyclone wind direction, pressure in the pipe, and cleaning time are processed to predict the cleaning energy consumption.

[0060] Specifically, step S40 in the method comprises: With the boiler model as a constraint, load the cyclone wind speed, cyclone wind direction, pressure in the pipe, cleaning time, and label of identified energy consumption value, and train a plurality of energy consumption prediction models. The integrated output is the mean value of the outputs of the plurality of energy consumption prediction base models, the plurality of energy consumption prediction base models are combined to obtain the energy consumption prediction model.

[0061] In the embodiments of the present application, a plurality of energy consumption prediction base models are trained based on a neural network with the boiler model as a constraint.

[0062] Exemplarily, the training of the energy consumption prediction base model can be realized through the following technical path: First, data preparation, the input nodes of the energy consumption prediction model include the cyclone wind speed, cyclone wind direction, pressure in the pipe, cleaning time, and label of identified energy consumption value.

[0063] Secondly, model building, the input layer is the node number equal to the dimension of input features, such as 10 features including cyclone wind speed, cyclone wind direction, pipe pressure, cleaning time and label of energy consumption value, so the input layer contains 10 nodes; 1-3 layers of hidden layers are set, the number of nodes of each layer is adjusted through experiments, such as 64, 32, etc., and the activation function is selected as ReLU; the number of nodes of the output layer is equal to the number of prediction targets, such as 1 node for predicting only the time consumption, and 2 nodes for predicting the time consumption and energy consumption at the same time, and the output layer generally does not use the activation function, and directly outputs continuous values.

[0064] Thirdly, model training, the predicted energy consumption is taken as the output, the corresponding prediction sample in the training set is taken as the supervision label, the Adam optimizer and the mean square error (MSE) loss function are used to build the training framework, the batch size is set to 32, the total training rounds are set to 50, and the early stopping mechanism (patience = 5) is introduced, when the loss of the validation set does not appear for 5 consecutive rounds, the training process is automatically terminated, and the trained energy consumption prediction model is obtained, so that the model overfitting is effectively avoided, and the model reaches the convergence state.

[0065] When the energy consumption prediction base model is trained, the mean value of the outputs of a plurality of energy consumption prediction base models is taken as the integrated output, a plurality of energy consumption prediction base models are combined to obtain an energy consumption prediction model. The energy consumption prediction model is bound with the target boiler, and the model will output the cleaning energy consumption prediction value. According to the prediction value, energy allocation and cost budget are prepared in advance to avoid waste of energy and unnecessary cost expenditure.

[0066] S50: When the residual ash thickness of any place of the residual ash thickness distribution is less than or equal to the residual ash thickness threshold value, and the cleaning energy consumption prediction value is less than or equal to the energy consumption threshold value, the boiler wall-attached gas conveying control is performed according to the boiler wall-attached gas conveying scheme, and the in-furnace pipeline cleaning is performed.

[0067] In the embodiment of the application, the residual ash thickness threshold value and the energy consumption threshold value are set, the residual ash thickness threshold value is set according to the normal operation requirement and safety standard of the boiler, if the residual ash thickness is too high, the thermal efficiency and the operation stability of the boiler will be affected; the energy consumption threshold value is determined based on the energy consumption index and the cost control target of the boiler, and the high energy consumption will increase the operation cost.

[0068] When the residual ash thickness of any place of the residual ash thickness distribution is less than or equal to the residual ash thickness threshold value, and the cleaning energy consumption prediction value is less than or equal to the energy consumption threshold value, it is indicated that the current boiler wall-attached gas conveying scheme can effectively clean the in-furnace pipeline ash and control the energy consumption within a reasonable range. At this time, the boiler wall-attached gas conveying control is performed according to the scheme, the corresponding gas conveying equipment is started, and the parameters such as cyclone wind speed, cyclone wind direction, pipe pressure and cleaning time are accurately adjusted to comprehensively clean the in-furnace pipeline.

[0069] Specifically, step S50 in the method comprises: When the residual ash thickness at any position of the residual ash thickness distribution is greater than the residual ash thickness threshold value, or / and the cleaning energy consumption prediction value is greater than the energy consumption threshold value, a first weight is configured for the normalized parameter of the mean value of the residual ash thickness distribution, and a second weight is configured for the normalized parameter of the cleaning energy consumption prediction value, wherein the first weight ∈ [0.7, 1], the second weight ∈ [0, 0.3], and the sum of the first weight and the second weight is equal to 1; Based on the first weight and the second weight, the residual ash thickness distribution mean value normalized parameter and the cleaning energy consumption prediction value normalized parameter are added to obtain a scheme fitness; The scheme fitness is stored in association with the boiler wall-attached gas delivery scheme, and is added to the analyzed gas delivery scheme group; When the number of schemes in the analyzed gas delivery scheme group is greater than or equal to a scheme number threshold value, and a convergent scheme has not been obtained, particle swarm optimization is performed based on the analyzed gas delivery scheme group to obtain a target gas delivery scheme for boiler wall-attached gas delivery control and in-furnace pipeline cleaning.

[0070] In the embodiments of the present application, when one or more residual ash thicknesses in the residual ash thickness distribution are greater than the residual ash thickness threshold value, or / and the cleaning energy consumption prediction value is greater than the energy consumption threshold value, it indicates that the current boiler wall-attached gas delivery scheme may have deficiencies and needs to be further adjusted.

[0071] The mean value normalized parameter of the residual ash thickness distribution reflects the average thickness of the residual ash in the pipeline after cleaning, and the smaller the value, the thinner the residual ash thickness and the better the cleaning effect; the cleaning energy consumption prediction value normalized parameter reflects the energy consumption expected in the cleaning process, and the smaller the value, the lower the energy consumption and the better the economy.

[0072] Further, a first weight is configured for the normalized parameter of the mean value of the residual ash thickness distribution, and a second weight is configured for the normalized parameter of the cleaning energy consumption prediction value. The first weight has a value range of [0.7, 1], the second weight has a value range of [0, 0.3], and the sum of the two is equal to 1. For example, the first weight is configured for the normalized parameter of the mean value of the residual ash thickness distribution, and has a value of 0.7; the second weight is configured for the normalized parameter of the cleaning energy consumption prediction value, and has a value of 0.3.

[0073] Based on the first weight and the second weight, the residual ash thickness distribution mean value normalized parameter and the cleaning energy consumption prediction value normalized parameter are added to obtain a scheme fitness. The higher the value of the scheme fitness, the better the comprehensive performance of the scheme in terms of cleaning of the ash and control of the energy consumption.

[0074] Exemplarily, assuming that for the soot cleaning of a certain boiler pipeline, schemes A and B are proposed, the residual ash thickness distribution mean value normalization parameter of scheme A is 0.2, and the cleaning energy consumption prediction value normalization parameter is 0.6; the residual ash thickness distribution mean value normalization parameter of scheme B is 0.4, and the cleaning energy consumption prediction value normalization parameter is 0.2. After weighted calculation, the scheme fitness of scheme A is 0.32, and the scheme fitness of scheme B is 0.34. Comparison shows that A is a better scheme.

[0075] The scheme fitness is stored in association with the corresponding boiler wall-attached gas delivery scheme, and is added to the analyzed gas delivery scheme group. The analyzed gas delivery scheme group is used to record and manage all evaluated schemes, facilitating subsequent analysis and comparison.

[0076] When the number of schemes in the analyzed gas delivery scheme group is greater than or equal to the scheme number threshold, if a convergent scheme has not been obtained, that is, no scheme can make the residual ash thickness at any point of the residual ash thickness distribution less than or equal to the residual ash thickness threshold and the cleaning energy consumption prediction value less than or equal to the energy consumption threshold, particle swarm optimization needs to be performed based on the analyzed gas delivery scheme group.

[0077] Particle swarm optimization is an intelligent optimization algorithm that searches for the optimal solution in the solution space by simulating the group behavior of bird or fish swarms. In this application, the particle swarm optimization algorithm continuously adjusts the position and velocity of particles based on the scheme information in the analyzed gas delivery scheme group to find the target gas delivery scheme. The target gas delivery scheme obtained by optimization is used for boiler wall-attached gas delivery control, the corresponding gas delivery equipment is started, and parameters such as cyclone wind speed, cyclone wind direction, pipe pressure, and cleaning duration are accurately adjusted to comprehensively clean the pipes in the furnace, thereby improving the cleaning effect of the boiler pipeline, reducing the impact of soot and slag accumulation on the operation of the boiler, and improving the energy utilization efficiency and operational stability of the boiler.

[0078] Exemplarily, the core of particle swarm optimization is to find the optimal solution that minimizes the "scheme fitness" in the parameter space by simulating the cooperative search behavior of the "particle swarm". The residual ash thickness normalization parameter and the cleaning energy consumption normalization parameter are used as the optimization variables of the two dimensions of particle swarm optimization, and real number coding is adopted. Each particle corresponds to a set of parameter combinations, representing a potential soot cleaning scheme. Taking "population size N=20, maximum iteration number T=50" as an example, 20 initial positions of particles are randomly generated, the initial fitness values of each particle are calculated, the A and B schemes described above are used as two particles in the initial population, 18 randomly generated particles are added, and after 50 iterations, the optimal position is converged, or 3-5 independent particle swarm optimizations are performed. If the optimal fitness values of the results of multiple optimizations differ by less than 5%, the results are considered stable and reliable.

[0079] In summary, compared with the prior art, the boiler wall gas conveying method of the present application accurately predicts the residual ash thickness distribution in the boiler pipeline after cleaning and the energy consumption during the cleaning process by constructing a residual ash thickness prediction model and an energy consumption prediction model. Using a graph neural network architecture, factors such as cyclone wind speed, cyclone wind direction, pipeline pressure, cleaning time, pipeline ash thickness distribution, and pipeline ash adhesion force distribution are fully considered, making the prediction results more accurate and reliable. Through iterative training, a converged residual ash thickness prediction model is obtained, providing solid data support for subsequent cleaning scheme evaluation. With the boiler model as a constraint, several energy consumption prediction base models are trained, and the output mean is taken as the integrated output to obtain an energy consumption prediction model. This model can accurately predict the energy consumption during the boiler wall gas conveying cleaning process, helping to make energy allocation and cost budget in advance, and avoiding waste of energy and unnecessary cost expenditure.

[0080] In addition, the present application also sets residual ash thickness threshold and energy consumption threshold. When the residual ash thickness distribution and the cleaning energy consumption prediction value meet the threshold requirements, the cleaning can be directly carried out according to the boiler wall gas conveying scheme; if the requirements are not met, the scheme fitness is calculated and associated with the storage, and if a converged scheme is still not obtained, particle swarm optimization is performed to further optimize the gas conveying scheme.

[0081] Embodiment two, as shown in Figure 2 Based on the same inventive concept as the boiler wall gas conveying method for reducing ash deposition and slagging provided in embodiment one, the present application embodiment also provides a boiler wall gas conveying system for reducing ash deposition and slagging, comprising: A sample collection module 11 is used to collect a pipeline ash detection sample set with the boiler operating parameters, boiler model and boiler airflow monitoring parameters of a target boiler as constraints, and to count the pipeline ash thickness distribution and pipeline ash adhesion force distribution. A scheme configuration module 12 is used to configure a boiler wall gas conveying scheme, wherein the boiler wall gas conveying scheme includes cyclone wind speed, cyclone wind direction, pipeline pressure and cleaning time. An ash thickness prediction module 13 is used to process the cyclone wind speed, the cyclone wind direction, the pipeline pressure, the cleaning time, the pipeline ash thickness distribution and the pipeline ash adhesion force distribution through a residual ash thickness prediction model bound to the target boiler, and to output a residual ash thickness distribution. An energy consumption output module 14 is used to process the cyclone wind speed, the cyclone wind direction, the pipeline pressure, and the cleaning time through an energy consumption prediction model bound to the target boiler, and to output a cleaning energy consumption prediction value. The scheme execution module 15 is configured to execute in-situ tube cleaning according to the boiler wall-attached gas conveying scheme when any residual ash thickness of the residual ash thickness distribution is less than or equal to a residual ash thickness threshold value and the cleaning energy consumption prediction value is less than or equal to an energy consumption threshold value.

[0082] The information collection module 11 is configured to: Collect a to-be-analyzed tube ash deposition detection sample with the boiler model as a constraint, wherein the to-be-analyzed tube ash deposition detection sample has a label of boiler operation parameters and a label of boiler airflow monitoring parameters; When any attribute boiler operation parameter deviation between the boiler operation parameters and the label of boiler operation parameters is less than or equal to a corresponding attribute boiler operation parameter deviation threshold value, the boiler operation parameters are considered to be consistent. When any attribute boiler airflow monitoring parameter deviation between the boiler airflow monitoring parameters and the label of boiler airflow monitoring parameters is less than or equal to a corresponding attribute boiler airflow monitoring parameter deviation threshold value, the boiler airflow monitoring parameters are considered to be consistent. When the boiler operation parameters are consistent and the boiler airflow monitoring parameters are consistent, the to-be-analyzed tube ash deposition detection sample is added to the tube ash deposition detection sample set.

[0083] In one application embodiment, the tube ash deposition detection sample set is collected, and the tube ash thickness distribution and the tube ash adhesion force distribution are counted, including: The inner wall of the target boiler tube is divided into a grid by a preset side length grid to obtain a boiler tube inner wall grid distribution; Based on the boiler tube inner wall grid distribution, the tube ash deposition detection sample set is traversed to obtain a first tube grid distribution ash deposition state to a Qth tube grid distribution ash deposition state, wherein Q represents a total number of samples, and Q is greater than or equal to 30. The first tube grid distribution ash deposition state to the Qth tube grid distribution ash deposition state are integrated into a first grid ash deposition state set to an Mth grid ash deposition state set, wherein M represents a total number of grids. The first grid ash deposition state set to the Mth grid ash deposition state set are traversed to perform centroid ash deposition state analysis to obtain the tube ash thickness distribution and the tube ash adhesion force distribution.

[0084] Specifically, the ash thickness prediction model 13 is configured to: From the first grid ash deposition state set, a plurality of ash deposition states are extracted, wherein any one of the plurality of ash deposition states includes an ash deposition thickness and an ash deposition adhesion force. ​Taking the normalized value of the ash deposition thickness as the first-dimensional coordinate and taking the normalized value of the ash deposition adhesion as the second-dimensional coordinate, a distribution of the plurality of ash deposition states is performed to obtain a plurality of ash deposition state distribution coordinates; Based on the plurality of ash deposition state distribution coordinates, a LOF outlier factor minimum value is extracted to obtain a centroid ash deposition state distribution coordinate; Based on the centroid ash deposition state distribution coordinate, a first grid pipeline ash deposition thickness and a first grid pipeline ash deposition adhesion are extracted; When the Mth grid pipeline ash deposition thickness and the Mth grid pipeline adhesion are obtained, the first grid pipeline ash deposition thickness is integrated until the Mth grid pipeline ash deposition thickness to obtain the pipeline ash deposition thickness distribution, and the first grid pipeline ash deposition adhesion is integrated until the Mth grid pipeline adhesion to obtain the pipeline ash deposition adhesion distribution.

[0085] Further, the "processing the rotational flow speed, the rotational flow direction, the pipeline pressure, the cleaning time, the pipeline ash deposition thickness distribution and the pipeline ash deposition adhesion distribution through the residual ash thickness prediction model bound with the target boiler, and outputting a residual ash thickness distribution" includes: Based on the boiler pipeline inner wall grid distribution, a graph neural network architecture is constructed, wherein the input nodes of the graph neural network architecture include grid input nodes, gas delivery input nodes and grid output nodes, wherein the grid input nodes and the grid output nodes are one-to-one corresponding to the boiler pipeline inner wall grid distribution, the gas delivery input nodes correspond to the pipeline air inlet position, the grid input nodes are used to receive the pipeline ash deposition thickness distribution and the pipeline ash deposition adhesion distribution, the gas delivery input nodes are used to receive the rotational flow speed, the rotational flow direction, the pipeline pressure and the cleaning time, and the grid output nodes are used to output the residual ash thickness distribution; Based on the boiler model, rotational flow speed record data, rotational flow direction record data, pipeline pressure record data, cleaning time record data, pipeline ash deposition thickness distribution record information, pipeline ash deposition adhesion distribution record information and labels identifying residual ash thickness distribution are collected to train the graph neural network architecture and generate the residual ash thickness prediction model bound with the target boiler.

[0086] The energy consumption output module 14 is specifically configured to: With the boiler model as a constraint, the rotational flow speed, the rotational flow direction, the pipeline pressure, the cleaning time and the labels identifying the energy consumption value are loaded to train a plurality of energy consumption prediction base models; The plurality of energy consumption prediction base models are combined to obtain the energy consumption prediction model, taking the output mean value of the plurality of energy consumption prediction base models as the integrated output.

[0087] The scheme execution module 15 is specifically configured to: When the residual ash thickness at any position of the residual ash thickness distribution is greater than a residual ash thickness threshold value, or / and the cleaning energy consumption prediction value is greater than an energy consumption threshold value, a first weight is configured for a normalized parameter of the mean value of the residual ash thickness distribution, and a second weight is configured for a normalized parameter of the cleaning energy consumption prediction value, wherein the first weight ∈ [0.7, 1], the second weight ∈ [0, 0.3], and the sum of the first weight and the second weight is equal to 1; Based on the first weight and the second weight, the residual ash thickness distribution mean value normalized parameter and the cleaning energy consumption prediction value normalized parameter are added to obtain a scheme fitness; The scheme fitness is stored in association with the boiler wall-attached gas delivery scheme, and is added to the analyzed gas delivery scheme group; When the number of schemes in the analyzed gas delivery scheme group is greater than or equal to a scheme number threshold value, and a convergent scheme has not been obtained, particle swarm optimization is performed based on the analyzed gas delivery scheme group to obtain a target gas delivery scheme for boiler wall-attached gas delivery control and in-furnace pipeline cleaning.

[0088] To sum up, the embodiments of the present application have at least the following technical effects: The embodiments of the present application provide a boiler wall-attached gas delivery method and system for reducing ash deposition and slagging, comprehensively consider parameters of boiler operation, statistically analyze pipeline ash deposition, and predict residual ash thickness and cleaning energy consumption by using a prediction model, so that the cleaning effect can be ensured while the energy consumption is effectively controlled. By setting a threshold value, when the residual ash thickness and the energy consumption both meet the requirements, the boiler wall-attached gas delivery control is performed to realize efficient and energy-saving in-furnace pipeline cleaning. When the threshold condition is not met, the scheme fitness is calculated by configuring weights, and the particle swarm optimization is combined to continuously optimize the gas delivery scheme, so that a more suitable target gas delivery scheme is finally found to further improve the cleaning effect and energy utilization efficiency. Through the above technical solutions, the target boiler pipeline ash thickness and pipeline ash adhesion force distribution are collected, and a boiler wall-attached gas delivery scheme is configured. Then, the residual ash thickness distribution and the cleaning energy consumption prediction value are output by using a residual ash thickness prediction model and an energy consumption prediction model, respectively. The problem that the residual ash in the surrounding pipeline cannot be cleaned in the prior art is solved, the occurrence of pipeline corrosion, perforation, and other situations is reduced, the loss caused by furnace shutdown for maintenance due to pipeline leakage is avoided, the service life of the boiler is prolonged, the risk of safety accidents is reduced, and the stable operation of the boiler in industrial production is ensured.

[0089] It should be noted that the above-mentioned sequence of the embodiments of the present application is only for description, and does not represent the advantages and disadvantages of the embodiments. Moreover, the above describes specific embodiments of the present application. In addition, the processes depicted in the drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multi-task processing and parallel processing are also possible or can be advantageous.

[0090] The above description is merely exemplary of the application, one skilled in the art will readily devise many variations and modifications of the application without departing from the scope of the application as defined by the following claims. Accordingly, all such variations and modifications are intended to be included within the scope of the application.

[0091] The description and drawings are merely illustrative of the application and do not limit the scope of the application as do the appended claims, which by themselves are illustrative of the application. Consequently, any modifications, changes, combinations, or equivalents of the application falling within the scope of the application are intended to be included in the scope of the application.

Claims

1. A method for reducing wall deposition in a furnace by means of a gas delivery against the wall, characterized in that, The application is applied to a wall-attached wind control device, comprising: With the boiler operation parameters, the boiler model and the boiler airflow monitoring parameters of the target boiler as constraints, a pipe ash deposition detection sample set is collected, and a pipe ash deposition thickness distribution and a pipe ash deposition adhesion force distribution are counted; A boiler wall-attached gas delivery scheme is configured, wherein the boiler wall-attached gas delivery scheme comprises a rotational flow wind speed, a rotational flow wind direction, a pipe internal pressure and a cleaning duration; Through a residual ash thickness prediction model bound with the target boiler, the rotational flow wind speed, the rotational flow wind direction, the pipe internal pressure, the cleaning duration, the pipe ash deposition thickness distribution and the pipe ash deposition adhesion force distribution are processed, and a residual ash thickness distribution is output; Through an energy consumption prediction model bound with the target boiler, the rotational flow wind speed, the rotational flow wind direction, the pipe internal pressure and the cleaning duration are processed, and a cleaning energy consumption prediction value is output; When the residual ash thickness of any place of the residual ash thickness distribution is less than or equal to a residual ash thickness threshold value, and the cleaning energy consumption prediction value is less than or equal to an energy consumption threshold value, a boiler wall-attached gas delivery control is performed according to the boiler wall-attached gas delivery scheme, and an in-furnace pipe cleaning is executed.

2. The method of claim 1, wherein, With the boiler operation parameters, the boiler model and the boiler airflow monitoring parameters of the target boiler as constraints, a pipe ash deposition detection sample set is collected, and a pipe ash deposition thickness distribution and a pipe ash deposition adhesion force distribution are counted, comprising: With the boiler model as a constraint, a to-be-analyzed pipe ash deposition detection sample is collected, wherein the to-be-analyzed pipe ash deposition detection sample has a label identifying a boiler operation parameter and a label identifying a boiler airflow monitoring parameter; When the boiler operation parameters and any attribute boiler operation parameter deviation of the label identifying the boiler operation parameter are less than or equal to a corresponding attribute boiler operation parameter deviation threshold value, the boiler operation parameters are considered to be consistent; When the boiler airflow monitoring parameters and any attribute boiler airflow monitoring parameter deviation of the label identifying the boiler airflow monitoring parameter are less than or equal to a corresponding attribute boiler airflow monitoring parameter deviation threshold value, the boiler airflow monitoring parameters are considered to be consistent; When the boiler operation parameters are consistent and the boiler airflow monitoring parameters are consistent, the to-be-analyzed pipe ash deposition detection sample is added to the pipe ash deposition detection sample set.

3. The method of claim 1, wherein, A pipe ash deposition detection sample set is collected, and a pipe ash deposition thickness distribution and a pipe ash deposition adhesion force distribution are counted, comprising: A target boiler pipe inner wall is grid-divided through a preset side length grid, and a boiler pipe inner wall grid distribution is obtained; Based on the boiler pipe inner wall grid distribution, the pipe ash deposition detection sample set is divided through iteration, and a first pipe grid distribution ash deposition state to a Qth pipe grid distribution ash deposition state is obtained, wherein Q represents a total number of samples, and Q is greater than or equal to 30; The first pipe grid distribution ash deposition state to the Qth pipe grid distribution ash deposition state is integrated in the same grid state, and a first grid ash deposition state set to an Mth grid ash deposition state set is obtained, wherein M represents a total number of grids; The first grid ash deposition state set to the Mth grid ash deposition state set is iterated, and a centroid ash deposition state analysis is performed, and the pipe ash deposition thickness distribution and the pipe ash deposition adhesion force distribution are obtained.

4. The method of claim 3, wherein, The centroid soot state analysis is performed by traversing the first grid soot state set to the Mth grid soot state set to obtain the pipe soot thickness distribution and the pipe soot adhesion force distribution, including: Extracting a plurality of soot states from the first grid soot state set, wherein any one of the plurality of soot states includes a soot thickness and a soot adhesion force; Performing distribution on the plurality of soot states with the soot thickness normalized value as the first dimension coordinate and the soot adhesion force normalized value as the second dimension coordinate to obtain a plurality of soot state distribution coordinates; Based on the plurality of soot state distribution coordinates, performing LOF outlier factor minimum value extraction to obtain a centroid soot state distribution coordinate; Based on the centroid soot state distribution coordinate, extracting a first grid pipe soot thickness and a first grid pipe soot adhesion force; When the Mth grid pipe soot thickness and the Mth grid pipe adhesion force are obtained, integrating the first grid pipe soot thickness to the Mth grid pipe soot thickness to obtain the pipe soot thickness distribution, and integrating the first grid pipe soot adhesion force to the Mth grid pipe adhesion force to obtain the pipe soot adhesion force distribution.

5. The method of claim 3, wherein, Processing the rotational flow speed, the rotational flow direction, the pipe internal pressure, the cleaning time length, the pipe soot thickness distribution and the pipe soot adhesion force distribution through the residual ash thickness prediction model bound with the target boiler to output a residual ash thickness distribution, including: Based on the boiler pipe inner wall grid distribution, constructing a graph neural network architecture, wherein the input nodes of the graph neural network architecture include grid input nodes, gas delivery input nodes and grid output nodes, wherein the grid input nodes and the grid output nodes are one-to-one corresponding to the boiler pipe inner wall grid distribution, the gas delivery input nodes correspond to the pipe inlet position, the grid input nodes are used to receive the pipe soot thickness distribution and the pipe soot adhesion force distribution, the gas delivery input nodes are used to receive the rotational flow speed, the rotational flow direction, the pipe internal pressure and the cleaning time length, and the grid output nodes are used to output the residual ash thickness distribution; Based on the boiler model, collecting rotational flow speed record data, rotational flow direction record data, pipe internal pressure record data, cleaning time length record data, pipe soot thickness distribution record information, pipe soot adhesion force distribution record information and a label identifying the residual ash thickness distribution, training the graph neural network architecture to generate the residual ash thickness prediction model bound with the target boiler.

6. The method of claim 1, wherein, Processing the rotational flow speed, the rotational flow direction, the pipe internal pressure and the cleaning time length through the energy consumption prediction model bound with the target boiler to output a cleaning energy consumption prediction value, including: Loading the rotational flow speed, the rotational flow direction, the pipe internal pressure, the cleaning time length and the label identifying the energy consumption value as constraints to train a plurality of energy consumption prediction base models; Combining the plurality of energy consumption prediction base models to obtain the energy consumption prediction model with the output mean value of the plurality of energy consumption prediction base models as the integrated output.

7. The method of claim 1, wherein, Further comprising: When the residual ash thickness at any position of the residual ash thickness distribution is greater than a residual ash thickness threshold, or / and the cleaning energy consumption prediction value is greater than an energy consumption threshold, a first weight is configured for a normalized parameter of the mean value of the residual ash thickness distribution, and a second weight is configured for a normalized parameter of the cleaning energy consumption prediction value, wherein the first weight ∈ [0.7, 1], the second weight ∈ [0, 0.3], and the sum of the first weight and the second weight is equal to 1; Based on the first weight and the second weight, the residual ash thickness distribution mean value normalized parameter and the cleaning energy consumption prediction value normalized parameter are added to obtain a scheme fitness; The scheme fitness is stored in association with the boiler wall-attached gas delivery scheme, and is added to the analyzed gas delivery scheme group; When the number of schemes in the analyzed gas delivery scheme group is greater than or equal to a scheme number threshold, and a convergent scheme has not been obtained, particle swarm optimization is performed based on the analyzed gas delivery scheme group to obtain a target gas delivery scheme for boiler wall-attached gas delivery control and in-furnace pipeline cleaning.

8. A furnace wall-attached gas delivery system for reducing deposition and slagging, characterized by, The method for performing any one of claims 1-7 comprises: A sample collection module for collecting a pipeline ash deposition detection sample set with the target boiler operation parameters, boiler model, and boiler airflow monitoring parameters as constraints, and for counting the pipeline ash deposition thickness distribution and the pipeline ash deposition adhesion force distribution; A scheme configuration module for configuring a boiler wall-attached gas delivery scheme, wherein the boiler wall-attached gas delivery scheme includes a rotational flow wind speed, a rotational flow wind direction, a pipeline internal pressure, and a cleaning duration; An ash thickness prediction module for processing the rotational flow wind speed, the rotational flow wind direction, the pipeline internal pressure, the cleaning duration, the pipeline ash deposition thickness distribution, and the pipeline ash deposition adhesion force distribution through a residual ash thickness prediction model bound to the target boiler, and for outputting a residual ash thickness distribution; An energy consumption output module for processing the rotational flow wind speed, the rotational flow wind direction, the pipeline internal pressure, and the cleaning duration through an energy consumption prediction model bound to the target boiler, and for outputting a cleaning energy consumption prediction value; A scheme execution module for performing boiler wall-attached gas delivery control according to the boiler wall-attached gas delivery scheme when the residual ash thickness at any position of the residual ash thickness distribution is less than or equal to a residual ash thickness threshold, and the cleaning energy consumption prediction value is less than or equal to an energy consumption threshold, and for performing in-furnace pipeline cleaning.

Images