Soot blowing control method, device and equipment for boiler and storage medium

Abstract

The invention provides a soot blowing control method, device and equipment for a boiler and a storage medium, and the method comprises the steps: collecting a soot deposition state image of a heating surface of the boiler in real time through visual equipment, carrying out the segmentation and feature extraction of the image, and generating an initial detail feature vector; reconstructing a dust deposition feature image based on the feature vectors, further segmenting and extracting basic local feature vectors, and generating a final dust deposition state image; and classifying the ash deposition state images corresponding to all the visual devices, calculating the ash deposition thermal resistance value corresponding to the target image, and comparing the ash deposition thermal resistance value with a preset threshold value so as to intelligently decide whether to start the ash blower to remove the deposited ash. By the adoption of the technical scheme, redundant soot blowing operation is reduced, the steam consumption of the soot blower and the blowing damage risk of the heating surface are effectively reduced, meanwhile, safety accidents caused by soot deposition are avoided, and the safety and economical efficiency of boiler operation are improved.

Description

Technical Field

[0001] The present application relates to the technical field of soot blowing control for boiler heating surfaces, and in particular, to a soot blowing control method, device, equipment and storage medium for boilers. Background Art

[0002] In the current power production field, energy conservation and pollution reduction are key tasks that need to be solved urgently. Due to its large reserves and high economy, Xinjiang coal has become an important raw material for thermal power generation. However, when burning high-alkali coal in Xinjiang, it is easy to cause ash deposition, fouling and corrosion on the boiler heating surface, seriously affecting the boiler thermal efficiency and operation stability. The problem of ash deposition on the boiler heating surface cannot be completely solved through structural transformation and operation adjustment. Soot blowing is an effective method to maintain cleanliness. The traditional fixed-time and fixed-quantity soot blowing method does not match the actual demand, resulting in under-blowing or over-blowing. Therefore, it is necessary to effectively judge the degree of ash deposition and formulate an intelligent on-demand soot blowing strategy. Currently, based on the heat transfer principle, a large number of image recognition studies on the ash deposition condition of the heating surface have been carried out using big data algorithms.

[0003] However, most of the existing image recognition methods for ash deposition on the heating surface are limited to simple backgrounds, and some subtle image features are easily ignored during the actual recognition process. This limitation in feature extraction leads to the loss of a large amount of detailed information, and then causes significant deviations in the recognition results of similar images, seriously restricting the improvement of classification performance. In addition, the complex spatial layout inside the boiler and the staggered arrangement of pipelines and equipment make it difficult to comprehensively and accurately reflect the true ash deposition condition in the furnace only relying on image recognition technology. Summary of the Invention

[0004] In view of this, the embodiments of the present application provide a soot blowing control method, device, equipment and storage medium for boilers. The present application monitors the actual pollution condition of the heating surface in real time, dynamically optimizes the soot blowing strategy, realizes on-demand soot blowing, reduces redundant soot blowing operations, effectively reduces the steam consumption of the soot blower and the risk of heating surface blow damage, and at the same time avoids safety accidents caused by ash deposition, improving the safety and economy of boiler operation.

[0005] The present application mainly includes the following aspects:

[0006] In the first aspect, the embodiments of the present application provide a soot blowing control method for a boiler. The boiler is provided with at least one visual device for collecting images of the ash deposition state of the boiler heating surface. The soot blowing control method includes:

[0007] For each visual device, obtain the current image of the ash fouling state of the boiler heating surface collected by the visual device; segment the current image of the ash fouling state of the boiler heating surface to obtain a plurality of initial ash fouling state images, and for each initial ash fouling state image, perform feature extraction on the initial ash fouling state image to obtain a plurality of initial detailed feature vectors of the initial ash fouling state image; use the plurality of initial detailed feature vectors of the initial ash fouling state image to generate an ash fouling feature reconstruction image corresponding to the visual device; segment the ash fouling feature reconstruction image corresponding to the visual device to obtain a plurality of initial ash fouling feature reconstruction images, and for each initial ash fouling feature reconstruction image, perform feature extraction on the initial ash fouling feature reconstruction image to obtain a plurality of basic local feature vectors of the initial ash fouling feature reconstruction image; use the plurality of basic local feature vectors of the initial ash fouling feature reconstruction image to generate an ash fouling state image corresponding to the visual device; perform state classification on the ash fouling state images corresponding to all visual devices, and based on the result of the state classification, determine the ash fouling thermal resistance value corresponding to at least one target ash fouling state image; compare the ash fouling thermal resistance values corresponding to all target ash fouling state images with the ash fouling thermal resistance threshold respectively, and based on the comparison result, determine whether to start the soot blower to remove the ash fouling on the corresponding boiler heating surface.

[0008] Further, the step of using the plurality of initial detailed feature vectors of the initial ash fouling state image to generate an ash fouling feature reconstruction image corresponding to the visual device includes:

[0009] Perform convolution processing on the plurality of initial detailed feature vectors of the initial ash fouling state image to obtain a plurality of key feature vectors of the initial ash fouling state image; perform deconvolution processing on the plurality of key feature vectors of the initial ash fouling state image to obtain a plurality of detailed feature vectors of the initial ash fouling state image; compare the plurality of initial detailed feature vectors of the initial ash fouling state image with the plurality of detailed feature vectors of the initial ash fouling state image to obtain a plurality of differential detailed feature vectors; determine each differential detailed feature vector as a key feature vector of the initial ash fouling state image; for each initial detailed feature vector in the plurality of initial detailed feature vectors of the initial ash fouling state image, calculate the correlation between the initial detailed feature vector and each initial detailed feature vector and each key feature vector respectively to obtain the weights of the initial detailed feature vector with respect to each initial detailed feature vector and each key feature vector; based on the weights of the initial detailed feature vector with respect to each initial detailed feature vector and each key feature vector, perform weighted fusion on the initial detailed feature vector, each initial detailed feature vector and each key feature vector to obtain a fused key feature vector of the initial ash fouling state image; combine all the fused key feature vectors to generate an ash fouling feature reconstruction image corresponding to the visual device.

[0010] Further, the generating of the ash fouling state image corresponding to the visual device by reconstructing a plurality of basic local feature vectors of the image using the initial ash fouling feature includes:

[0011] Performing convolution processing on a plurality of basic local feature vectors of the reconstructed image of the initial ash fouling feature to obtain a plurality of core feature vectors of the reconstructed image of the initial ash fouling feature; performing deconvolution processing on the plurality of core feature vectors of the reconstructed image of the initial ash fouling feature to obtain a plurality of local feature vectors of the reconstructed image of the initial ash fouling feature; comparing the plurality of initial local feature vectors of the reconstructed image of the initial ash fouling feature with the plurality of local feature vectors of the reconstructed image of the initial ash fouling feature to obtain a plurality of differential local feature vectors; for each initial local feature vector among the plurality of initial local feature vectors of the reconstructed image of the initial ash fouling feature, calculating the correlation between the initial local feature vector and each initial local feature vector and each core feature vector respectively to obtain the weights of the initial local feature vector with respect to each initial local feature vector and each core feature vector; based on the weights of the initial local feature vector with respect to each initial local feature vector and each core feature vector, performing weighted fusion on the initial local feature vector, each initial local feature vector, and each core feature vector to obtain the fused core feature vectors of the reconstructed image of the initial ash fouling feature; combining all the fused core feature vectors to generate the ash fouling state image corresponding to the visual device.

[0012] Further, the classifying the ash fouling state images corresponding to all visual devices and determining the ash fouling thermal resistance value corresponding to at least one target ash fouling state image based on the result of the state classification includes:

[0013] Inputting the fused core feature vectors of the ash fouling state images corresponding to all visual devices into a preset classification model, so that the preset classification model classifies the ash fouling state images corresponding to all visual devices, and obtaining at least one target ash fouling state image output by the preset classification model; determining the ash fouling thermal resistance value corresponding to at least one target ash fouling state image based on the mapping relationship between the ash fouling state image and the ash fouling thermal resistance value.

[0014] Further, the determining whether to start the soot blower to remove the ash fouling on the corresponding boiler heating surface based on the comparison result includes:

[0015] For each target ash fouling state image, if the ash fouling thermal resistance value corresponding to the target ash fouling state image is greater than or equal to the ash fouling thermal resistance threshold, start the soot blower to remove the ash fouling on the boiler heating surface corresponding to the target ash fouling state image; if the ash fouling thermal resistance value corresponding to the target ash fouling state image is less than the ash fouling thermal resistance threshold, do not start the soot blower.

[0016] Further, the soot blowing control method further includes:

[0017] Obtain the current load of the boiler; input the current load of the boiler into the ash fouling trend prediction model to obtain the ash fouling distribution data of multiple boiler blind area heating surfaces located in the blind area of the visual device output by the ash fouling trend prediction model; for each boiler blind area heating surface of the multiple boiler blind area heating surfaces, input the ash fouling distribution data of this boiler blind area heating surface into the ash fouling and slagging degree evaluation and warning model to obtain the ash fouling and slagging degree level of this boiler blind area heating surface output by the ash fouling and slagging degree evaluation and warning model; if the ash fouling and slagging degree level of this boiler blind area heating surface is greater than or equal to the preset level threshold, start the soot blower to remove the ash fouling on this boiler blind area heating surface; if the ash fouling and slagging degree level of this boiler blind area heating surface is less than the preset level threshold, do not start the soot blower.

[0018] Further, the ash fouling prediction model is constructed through the following steps:

[0019] Based on the physical properties of ash fouling, construct a coal type ash fouling model for the boiler; based on the geometric structure of the boiler, construct the boiler heating surface structure; based on the coal type ash fouling model and the boiler heating surface structure of the boiler, construct an ash fouling numerical simulation model of the boiler heating surface; input multiple load and coal type parameters into the ash fouling numerical simulation model to obtain the running state of ash fouling particles in the boiler output by the ash fouling numerical simulation model; input the running state of ash fouling particles in the boiler into the coal type ash fouling model to obtain the ash fouling distribution data of multiple boiler heating surfaces corresponding to each load among multiple loads output by the coal type ash fouling model; based on the ash fouling distribution data of multiple boiler heating surfaces corresponding to each load among multiple loads, construct an ash fouling prediction model.

[0020] In a second aspect, an embodiment of the present application further provides a soot blowing device for a boiler, and the soot blowing control device includes:

[0021] An image acquisition module, for each visual device, obtain the current image of the ash fouling state of the boiler heating surface collected by this visual device;

[0022] A detailed feature extraction module, segment the current image of the ash fouling state of the boiler heating surface to obtain multiple initial ash fouling state images, and for each initial ash fouling state image, perform feature extraction on this initial ash fouling state image to obtain multiple initial detailed feature vectors of this initial ash fouling state image;

[0023] A detailed feature reconstruction module, use the multiple initial detailed feature vectors of this initial ash fouling state image to generate an ash fouling feature reconstruction image corresponding to this visual device;

[0024] The local feature extraction module segments the reconstructed image of the ash accumulation feature corresponding to the visual device to obtain multiple initial reconstructed images of the ash accumulation feature. For each initial reconstructed image of the ash accumulation feature, feature extraction is performed on the initial reconstructed image of the ash accumulation feature to obtain multiple basic local feature vectors of the initial reconstructed image of the ash accumulation feature;

[0025] The local feature reconstruction module uses the multiple basic local feature vectors of the initial reconstructed image of the ash accumulation feature to generate an ash accumulation state image corresponding to the visual device;

[0026] The thermal resistance calculation module classifies the states of the ash accumulation state images corresponding to all visual devices. Based on the result of the state classification, the ash accumulation thermal resistance value corresponding to at least one target ash accumulation state image is determined;

[0027] The first sootblowing decision module compares the ash accumulation thermal resistance values corresponding to all target ash accumulation state images with the ash accumulation thermal resistance threshold respectively. Based on the comparison result, it is determined whether to start the sootblower to remove the ash accumulation on the corresponding boiler heating surface.

[0028] In a third aspect, an embodiment of the present application further provides an electronic device, including: a processor, a memory, and a bus. The memory stores machine-readable instructions executable by the processor. When the electronic device runs, communication is carried out between the processor and the memory through the bus. When the machine-readable instructions are run by the processor, the steps of the sootblowing control method for the boiler described in the first aspect or any possible implementation manner in the first aspect are executed.

[0029] In a fourth aspect, an embodiment of the present application further provides a computer-readable storage medium. A computer program is stored on the computer-readable storage medium. When the computer program is run by a processor, the steps of the sootblowing control method for the boiler described in the first aspect or any possible implementation manner in the first aspect are executed.

[0030] A sootblowing control method, device, equipment, and storage medium for a boiler provided by an embodiment of the present application. First, the ash accumulation state image of the boiler heating surface is collected in real time through a visual device, and the image is segmented and feature-extracted to generate initial detailed feature vectors; then, based on these feature vectors, an ash accumulation feature image is reconstructed, further segmented, and basic local feature vectors are extracted to generate a final ash accumulation state image; then, the ash accumulation state images corresponding to all visual devices are classified, the ash accumulation thermal resistance value corresponding to the target image is calculated, and compared with a preset threshold, so as to intelligently decide whether to start the sootblower to remove the ash accumulation, realizing accurate identification and efficient cleaning of the ash accumulation state on the boiler heating surface.

[0031] In this way, the actual pollution condition of the heating surface is monitored in real time, the soot blowing strategy is dynamically optimized, soot blowing on demand is realized, redundant soot blowing operations are reduced, the steam consumption of the soot blower and the risk of damage to the heating surface are effectively reduced, safety accidents caused by ash accumulation are avoided, and the safety and economy of boiler operation are improved.

[0032] To make the above objects, features, and advantages of the present application more obvious and understandable, the following specifically gives preferred embodiments and, in conjunction with the accompanying drawings, makes a detailed description as follows. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] To more clearly illustrate the technical solutions of the embodiments of the present application, the following briefly introduces the drawings required to be used in the embodiments. It should be understood that the following drawings only show some embodiments of the present application and should not be regarded as limiting the scope. For those of ordinary skill in the art, other related drawings can be obtained based on these drawings without creative efforts.

[0034] Figure 1 Shows one of the flowcharts of a soot blowing control method for a boiler provided by an embodiment of the present application;

[0035] Figure 2 Shows an example diagram of the boiler structure and the installation position of the visual device;

[0036] Figure 3 Shows another flowchart of a soot blowing control method for a boiler provided by an embodiment of the present application;

[0037] Figure 4 Shows a third flowchart of a soot blowing control method for a boiler provided by an embodiment of the present application;

[0038] Figure 5 Shows a fourth flowchart of a soot blowing control method for a boiler provided by an embodiment of the present application;

[0039] Figure 6 Shows an example diagram of the relationship between the heat transfer coefficient of the boiler heating surface and the ash fouling thermal resistance of the boiler heating surface;

[0040] Figure 7 Shows an example diagram of the relationship between the ash fouling thermal resistance of the boiler heating surface and the ash fouling amount of the boiler heating surface;

[0041] Figure 8 Shows a fifth flowchart of a soot blowing control method for a boiler provided by an embodiment of the present application;

[0042] Figure 9 Shows the flowchart of another soot blowing control method for a boiler provided by an embodiment of the present application;

[0043] Figure 10 FIG. 1 shows one of the schematic structural diagrams of a soot blowing control device for a boiler provided by an embodiment of the present application;

[0044] Figure 11 FIG. 2 shows another schematic structural diagram of a soot blowing device for a boiler provided by an embodiment of the present application;

[0045] Figure 12 FIG. 3 shows the schematic structural diagram of an electronic device provided by an embodiment of the present application. Detailed implementation manners

[0046] To make the objectives, technical solutions, and advantages of the embodiments of the present application clearer, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present application. It should be understood that the accompanying drawings in the present application are only for the purposes of illustration and description, and are not used to limit the protection scope of the present application. In addition, it should be understood that the schematic drawings are not drawn to actual scale. The flowcharts used in the present application illustrate the operations implemented according to some embodiments of the present application. It should be understood that the operations in the flowchart may not be implemented in sequence, and steps without logical context relationships may be reversed or implemented simultaneously. In addition, those skilled in the art may add one or more other operations to the flowchart or remove one or more operations from the flowchart under the guidance of the content of the present application.

[0047] In addition, the described embodiments are only some of the embodiments of the present application, rather than all of the embodiments. The components of the embodiments of the present application generally described and illustrated in the accompanying drawings here may be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of the present application provided in the accompanying drawings is not intended to limit the scope of the present application to be protected, but merely represents the selected embodiments of the present application. All other embodiments obtained by those skilled in the art based on the embodiments of the present application without creative efforts fall within the protection scope of the present application.

[0048] The following methods, devices, equipment, or computer-readable storage media in the embodiments of the present application can be applied to any scenario where boiler soot blowing is required. The embodiments of the present application do not limit the specific application scenarios, and any solution using the soot blowing control method and device for a boiler provided by the embodiments of the present application falls within the protection scope of the present application.

[0049] It should be noted that in the current field of power production, energy conservation and pollution reduction are key tasks that need to be addressed urgently. Xinjiang coal has become an important raw material for thermal power generation due to its large reserves and high economy. However, when burning high-alkali coal in Xinjiang, it is easy to cause ash fouling, contamination and corrosion of the boiler heating surface, seriously affecting the boiler thermal efficiency and operation stability. The ash fouling problem of the boiler heating surface cannot be completely solved through structural transformation and operation adjustment. Soot blowing is an efficient method to maintain cleanliness. The traditional fixed-time and fixed-quantity soot blowing method does not match the actual demand, resulting in under-blowing or over-blowing. Therefore, it is necessary to effectively judge the degree of ash fouling and formulate an intelligent on-demand soot blowing strategy. Currently, based on the principles of heat transfer, a large number of image recognition studies on the ash fouling condition of the heating surface have been carried out using big data algorithms. However, most of the existing image recognition methods for ash fouling on the heating surface are limited to simple backgrounds and are prone to ignoring some subtle image features during the actual recognition process. This limitation in feature extraction leads to the loss of a large amount of detailed information, and then causes significant deviations in the recognition results of similar images, seriously restricting the improvement of classification performance. In addition, the complex spatial layout inside the boiler and the staggered arrangement of pipelines and equipment make it difficult to comprehensively and accurately reflect the true ash fouling condition in the furnace only relying on image recognition technology.

[0050] In view of the above problems, the embodiments of the present application propose a soot blowing control method, device, equipment and storage medium for boilers. The present application monitors the actual pollution condition of the heating surface in real time, dynamically optimizes the soot blowing strategy, realizes on-demand soot blowing, reduces redundant soot blowing operations, effectively reduces the steam consumption of the soot blower and the risk of heating surface blow damage, and at the same time avoids safety accidents caused by ash fouling, improving the safety and economy of boiler operation.

[0051] To facilitate the understanding of the present application, the technical solutions provided by the present application will be described in detail below with reference to specific embodiments.

[0052] Please refer to Figure 1 , Figure 1 which is one of the flowcharts of a soot blowing control method for boilers provided by the embodiments of the present application.

[0053] As shown in Figure 1 , for the soot blowing control method for boilers provided by the embodiments of the present application, the boiler is provided with at least one visual device, and the visual device is used to collect images of the ash fouling state of the boiler heating surface.

[0054] Here, according to the historical data of the boiler operation condition, the type of visual device suitable for monitoring inside the boiler is selected. As an example, the type of visual device can be selected according to the historical data of the boiler combustion temperature. An example diagram of the boiler structure and the installation position of the visual device is as shown in Figure 2 .

[0055] The number and installation position of the visual devices are adjusted through the following steps:

[0056] First, the ash accumulation condition inside the boiler sight hole can be detected through a visual device, and the quantity and installation position of the visual device can be preliminarily determined.

[0057] Then, based on the physical properties of the ash accumulation, an ash accumulation model of the coal type for the boiler is constructed. Here, the physical properties of the ash accumulation used for constructing the ash accumulation model of the coal type for the boiler may include but are not limited to: the component concentration and melting point of the ash accumulation.

[0058] Next, based on the geometric structure of the boiler, a boiler heating surface structure is constructed. Here, according to the actual structure of the boiler, a three-dimensional boiler heating surface structure is constructed. Among them, the heating surface is usually composed of a series of pipes, so the thickness of the boiler heating surface is represented by the pipe diameter; the heating surface area is constructed according to the actual projected area.

[0059] Subsequently, based on the ash accumulation model of the coal type for the boiler and the boiler heating surface structure, an ash accumulation numerical simulation model of the boiler heating surface is constructed. Here, the computational fluid dynamics (CFD) technology is used to construct the ash accumulation numerical simulation model of the boiler heating surface. The core of the CFD numerical simulation technology is to simulate and analyze the actual operation and combustion conditions inside the boiler in detail by constructing an accurate mathematical model, so as to achieve accurate prediction of the ash accumulation trend of the boiler.

[0060] After that, multiple load and coal type parameters are input into the ash accumulation numerical simulation model, and the running state of the ash accumulation particles inside the boiler output by the ash accumulation numerical simulation model is obtained. Here, the ash accumulation numerical simulation model can simulate the flow field (airflow distribution), temperature field (temperature distribution), component concentration field (gas composition distribution), and particle trace (movement trajectory of particles) inside the furnace and the vertical shaft flue. Through the ash accumulation numerical simulation model, the temperature distribution, component concentration distribution, and particle running state of different boiler heating surfaces can be obtained under a wide range of loads and different coal type parameters. Among them, the wide range of loads is different load ranges of the boiler.

[0061] Secondly, the running state of the ash accumulation particles inside the boiler is input into the ash accumulation model of the coal type, and the ash accumulation distribution data of multiple boiler heating surfaces corresponding to each load among multiple loads are obtained. Here, the movement trajectory of the ash particles in the flue gas is calculated through the ash accumulation model of the coal type, and the deposition position and deposition amount of the particles on each heating surface are determined.

[0062] Finally, based on the ash accumulation distribution data of multiple boiler heating surfaces corresponding to each load among multiple loads, the quantity and installation position of the visual device are optimized. An in-furnace visual system for the ash accumulation state image of each heating surface of the boiler is built based on the coupling of the in-furnace image pre-monitoring technology and the CFD numerical simulation technology.

[0063] After the installation and debugging of the visual device are completed, ensure that the real-time stream image can be successfully accessed into the industrial TV large screen system to achieve efficient and stable monitoring and display functions.

[0064] In this application, the soot blowing control method for boilers includes the following steps:

[0065] Step S101, for each visual device, obtain the current image of the ash accumulation state of the boiler heating surface collected by this visual device.

[0066] Here, the visual device captures the current image of the ash accumulation state of the boiler heating surface frame by frame. Decode and re-encode the current image of the ash accumulation state of the boiler heating surface into an image in the required format, and then use image enhancement techniques for de-blurring and denoising to preprocess the image to be recognized formed after re-encoding to highlight the key ash accumulation features in the image to be recognized, and determine the preprocessed image to be recognized as the current image of the ash accumulation state of the boiler heating surface.

[0067] Step S102, segment the current image of the ash accumulation state of the boiler heating surface to obtain multiple initial ash accumulation state images, and for each initial ash accumulation state image, extract features from this initial ash accumulation state image to obtain multiple initial detailed feature vectors of this initial ash accumulation state image.

[0068] Here, for the first time, patch embedding is performed on the current image of the ash accumulation state of the boiler heating surface according to a preset size, that is, the current image of the ash accumulation state of the boiler heating surface is divided into multiple small pieces. Extract detailed features through all-round detailed feature scanning. As an example, this application uses a convolutional neural network (CNN) model to extract image features.

[0069] Step S103, use the multiple initial detailed feature vectors of this initial ash accumulation state image to generate an ash accumulation feature reconstruction image corresponding to this visual device.

[0070] The following combines Figure 3 to illustrate how to use the multiple initial detailed feature vectors of this initial ash accumulation state image to generate an ash accumulation feature reconstruction image corresponding to this visual device.

[0071] Please refer to Figure 3 , Figure 3 which is the second flowchart of a soot blowing control method for boilers provided by an embodiment of this application.

[0072] As Figure 3 shown, regarding step S103, in specific implementation, as an example, it may include the following steps:

[0073] Step S1031: Perform convolution processing on multiple initial detailed feature vectors of the initial ash accumulation state image to obtain multiple key feature vectors of the initial ash accumulation state image.

[0074] Here, standardize the multiple key feature vectors to ensure consistent data distribution.

[0075] Step S1032: Perform deconvolution processing on multiple key feature vectors of the initial ash accumulation state image to obtain multiple detailed feature vectors of the initial ash accumulation state image.

[0076] Step S1033: Compare multiple initial detailed feature vectors of the initial ash accumulation state image with multiple detailed feature vectors of the initial ash accumulation state image to obtain multiple differential detailed feature vectors.

[0077] Here, the purpose of obtaining multiple differential detailed feature vectors is to make up for unrecognized details and enhance the comprehensiveness of image recognition.

[0078] Step S1034: Determine each differential detailed feature vector as a key feature vector of the initial ash accumulation state image.

[0079] Step S1035: For each initial detailed feature vector among multiple initial detailed feature vectors of the initial ash accumulation state image, calculate the correlation between the initial detailed feature vector and each initial detailed feature vector and each key feature vector respectively to obtain the weights of the initial detailed feature vector with respect to each initial detailed feature vector and each key feature vector.

[0080] Here, as an example, the self-attention fusion technology can be used to assign higher weights to the ash accumulation areas and lower weights to the background areas. The mechanism of the self-attention fusion technology can effectively highlight the importance of key features by dynamically allocating weights, while filtering out redundant information with low weights and adaptively adjusting the weight distribution according to different working conditions, so as to realize the preliminary feature recognition of the image of the ash accumulation state on the heating surface.

[0081] Step S1036: Based on the weights of the initial detailed feature vector with respect to each initial detailed feature vector and each key feature vector, perform weighted fusion on the initial detailed feature vector, each initial detailed feature vector and each key feature vector to obtain the fused key feature vector of the initial ash accumulation state image.

[0082] Step S1037: Combine all fused key feature vectors to generate an ash accumulation feature reconstruction image corresponding to the visual device.

[0083] Refer again to Figure 1, in step S104, segment the reconstructed image of the ash accumulation feature corresponding to the visual device to obtain multiple initial reconstructed images of the ash accumulation feature. For each initial reconstructed image of the ash accumulation feature, perform feature extraction on the initial reconstructed image of the ash accumulation feature to obtain multiple basic local feature vectors of the initial reconstructed image of the ash accumulation feature.

[0084] Here, perform secondary embedding patching on the current image of the ash accumulation state of the boiler heating surface according to a preset size.

[0085] In step S105, use the multiple basic local feature vectors of the initial reconstructed image of the ash accumulation feature to generate an image of the ash accumulation state corresponding to the visual device.

[0086] Next, in combination with Figure 4 illustrate how to generate a reconstructed image of the ash accumulation feature corresponding to the visual device by using the multiple initial detailed feature vectors of the initial image of the ash accumulation state.

[0087] Please refer to Figure 4 , Figure 4 which is the third flowchart of a sootblowing control method for a boiler provided by an embodiment of the present application.

[0088] As shown in Figure 4 , regarding step S105, in specific implementation, as an example, it may include the following steps:

[0089] In step S1051, perform convolution processing on the multiple basic local feature vectors of the initial reconstructed image of the ash accumulation feature to obtain multiple core feature vectors of the initial reconstructed image of the ash accumulation feature.

[0090] In step S1052, perform deconvolution processing on the multiple core feature vectors of the initial reconstructed image of the ash accumulation feature to obtain multiple local feature vectors of the initial reconstructed image of the ash accumulation feature.

[0091] In step S1053, compare the multiple initial local feature vectors of the initial reconstructed image of the ash accumulation feature with the multiple local feature vectors of the initial reconstructed image of the ash accumulation feature to obtain multiple differential local feature vectors.

[0092] In step S1054, determine each differential local feature vector as the core feature vector of the initial reconstructed image of the ash accumulation feature.

[0093] In step S1055, for each initial local feature vector among the multiple initial local feature vectors of the initial reconstructed image of the ash accumulation feature, perform correlation calculation on the initial local feature vector with each initial local feature vector and each core feature vector respectively to obtain the weights of the initial local feature vector with each initial local feature vector and each core feature vector.

[0094] Step S1056: Based on the weights of the initial local feature vector with each initial local feature vector and each core feature vector respectively, perform weighted fusion on the initial local feature vector, each initial local feature vector, and each core feature vector to obtain the fusion core feature vector of the initial fouling feature reconstruction image.

[0095] Here, as an example, the self-attention fusion technology can be used to assign higher weights to the fouling areas and lower weights to the background areas.

[0096] In the embodiment of the present application, after obtaining all the fusion core feature vectors, perform numerical expansion on all the fusion core feature vectors, that is, use the pooling layer to reduce the dimensions of all the fusion core feature vectors. Translate, adjust, and scale the fusion core feature vectors to ensure that the feature vectors are unified into a specific size in the convolutional layer and pooling layer of the CNN model for convenient standardized comparison. At the same time, adjust the image features from different angles to comprehensively analyze the diversity of the image features from different perspectives.

[0097] Step S1056: Combine all the fusion core feature vectors to generate the fouling state image corresponding to the visual device.

[0098] Regarding steps S102 - S105, two times of segmentation, convolution, deconvolution, and fusion are performed to obtain more accurate image feature information of the fouling state.

[0099] Refer to again Figure 1 , step S106: Classify the fouling state images corresponding to all the visual devices, and based on the result of the state classification, determine the fouling thermal resistance value corresponding to at least one target fouling state image.

[0100] The following combines Figure 5 to illustrate how to classify the fouling state images corresponding to all the visual devices and determine the fouling thermal resistance value corresponding to at least one target fouling state image based on the result of the state classification.

[0101] Please refer to Figure 5 , Figure 5 This is the fifth flowchart of a soot blowing control method for a boiler provided by the embodiment of the present application.

[0102] As Figure 5 shown, regarding step S106, in specific implementation, as an example, it may include the following steps:

[0103] Step S1061: Input the fusion core feature vectors of the dust accumulation state images corresponding to all visible devices into a preset classification model, so that the preset classification model classifies the dust accumulation state images corresponding to all visible devices, and obtain at least one target dust accumulation state image output by the preset classification model.

[0104] Here, as an example, the preset classification model can be a CNN model. Repeatedly train and optimize the fusion core feature vectors of the dust accumulation state images corresponding to all visible devices, classify the dust accumulation state images, accurately identify the dust accumulation state of the boiler heating surface, and select the dust accumulation state images with poor dust accumulation state, that is, select the target dust accumulation state images.

[0105] Step S1062: Determine the dust accumulation thermal resistance values corresponding to at least one target dust accumulation state image based on the mapping relationship between the dust accumulation state image and the dust accumulation thermal resistance value.

[0106] Here, to improve the accuracy of the CNN model in identifying the dust accumulation state of each heating surface of the boiler, further carry out theoretical calculations in combination with CFD numerical simulation technology and related calculation features, and continuously optimize the image recognition results based on the theoretical calculation results to improve the image recognition analysis model. That is, through the mutual verification and complementation of the theoretical calculation results and the numerical simulation data, the accurate determination of the dust accumulation state parameters of each heating surface of the boiler is realized.

[0107] Such as Figure 6 and Figure 7 shown in, the change of the heat transfer coefficient of the boiler heating surface has a significant impact on the dust accumulation thermal resistance, and the change of the dust accumulation thermal resistance is directly related to the dust accumulation amount. Therefore, the dust accumulation thermal resistance can be used as a key index to judge whether it is necessary to start the soot blower to remove the dust.

[0108] The mapping relationship between the dust accumulation state image and the dust accumulation thermal resistance value is determined through the following steps:

[0109] First, determine the logarithmic mean temperature difference ΔT under the clean state of the boiler heating surface m0 . As an example, most of the heat exchangers on the boiler heating surface are arranged in countercurrent, so the logarithmic mean temperature difference under the clean state of the boiler heating surface can be determined by formula (1).

[0110]

[0111] Among them, T h10 , T h20 are the inlet and outlet temperatures of the hot fluid (flue gas) under the clean state of the boiler heating surface respectively; T c10 , T c20 are the inlet and outlet temperatures of the cold fluid (water or steam) under the clean state of the boiler heating surface respectively.

[0112] Then, determine the heat transfer coefficient under the fouling condition of the boiler heating surface. Specifically, according to the heat balance principle, establish the heat transfer amount under the clean condition of the boiler heating surface. As an example, the heat transfer amount under the clean condition of the boiler heating surface can be determined by formula (2).

[0113] Q 0 =K 0 AΔT m0 (2)

[0114] Wherein, Q 0 is the heat transfer amount under the clean condition of the boiler heating surface; K 0 is the heat transfer coefficient under the clean condition of the boiler heating surface; A is the heat transfer area of the boiler heating surface. It can be seen from formula (2) that the heat transfer coefficient under the clean condition of the boiler heating surface can be determined by formula (3).

[0115]

[0116] Next, determine the heat transfer thermal resistance under the clean condition of the boiler heating surface. As an example, the heat transfer thermal resistance under the clean condition of the boiler heating surface can be determined by formula (4).

[0117]

[0118] Wherein, R 0 is the heat transfer thermal resistance under the clean condition of the boiler heating surface.

[0119] Subsequently, the logarithmic mean temperature difference under the fouling condition of the boiler heating surface. As an example, the logarithmic mean temperature difference under the fouling condition of the boiler heating surface can be determined by formula (5).

[0120]

[0121] Wherein, ΔT m1 is the logarithmic mean temperature difference under the fouling condition of the boiler heating surface, T h11 and T h21 are the inlet and outlet temperatures of the hot fluid (flue gas) under the fouling condition of the boiler heating surface respectively; T c11 and T c21 are the inlet and outlet temperatures of the cold fluid (water or steam) under the fouling condition of the boiler heating surface respectively.

[0122] After that, determine the heat transfer coefficient under the fouling condition of the boiler heating surface. As an example, the logarithmic mean temperature difference under the fouling condition of the boiler heating surface can be determined by formula (6).

[0123] Q 1 =K 1 AΔT m1 (6)

[0124] Wherein, Q1 is the heat transfer amount under the fouling state of the boiler heating surface; K 1 is the heat transfer coefficient under the fouling state of the boiler heating surface. It can be known from formula (6) that the heat transfer coefficient under the clean state of the boiler heating surface can be determined by formula (7).

[0125]

[0126] Secondly, determine the heat transfer resistance under the fouling state of the boiler heating surface. As an example, the heat transfer resistance under the fouling state of the boiler heating surface can be determined by formula (8).

[0127]

[0128] Finally, determine the fouling heat resistance value. Here, the difference between the heat transfer resistance under the fouling state of the boiler heating surface and the heat transfer resistance under the clean state is determined as the fouling heat resistance generated by fouling. As an example, the heat transfer resistance under the fouling state of the boiler heating surface can be determined by formula (9).

[0129]

[0130] wherein, R ash is the fouling heat resistance value.

[0131] Step S107: Compare the fouling heat resistance values corresponding to all the target fouling state images with the fouling heat resistance threshold respectively, and based on the comparison results, determine whether to start the soot blower to remove the fouling on the corresponding boiler heating surface.

[0132] Here, as an example, the fouling heat resistance threshold is 0.0025 m 2 ·K / W.

[0133] The following combines Figure 8 to illustrate how to determine whether to start the soot blower to remove the fouling on the corresponding boiler heating surface based on the comparison results.

[0134] Please refer to Figure 8 , Figure 8 which is the fifth flowchart of a soot blowing control method for a boiler provided by an embodiment of the present application.

[0135] As shown in Figure 8 , regarding determining whether to start the soot blower to remove the fouling on the corresponding boiler heating surface based on the comparison results in step S107, in specific implementation, as an example, it may include the following steps:

[0136] Step S1071: For each target fouling state image, if the fouling heat resistance value corresponding to the target fouling state image is greater than or equal to the fouling heat resistance threshold, start the soot blower to remove the fouling on the boiler heating surface corresponding to the target fouling state image.

[0137] Here, after starting the soot blower to remove the soot on the boiler heating surface corresponding to the target soot state image, return to execute step S101.

[0138] Step S1072, if the soot heat resistance value corresponding to the target soot state image is less than the soot heat resistance threshold, do not start the soot blower.

[0139] Here, if the soot blower is not started, return to execute step S101.

[0140] Please refer to Figure 9 , Figure 9 which is a flowchart of another soot blowing control method for a boiler provided by an embodiment of the present application.

[0141] As Figure 9 shown in , the internal structure of the boiler is complex and there are many visual blind areas. It is difficult to fully present the overall soot condition of the boiler only through visual devices. Therefore, for the visual blind areas, the soot blowing control method further includes:

[0142] Step S201, obtain the current load of the boiler.

[0143] Step S202, input the current load of the boiler into the soot trend prediction model, and obtain the soot distribution data of multiple boiler blind area heating surfaces located in the blind area of the visual device output by the soot trend prediction model.

[0144] Here, taking the soot distribution data of multiple heating surfaces corresponding to each load in the above multiple loads as data samples, a soot trend prediction model for the heating surfaces of the boiler visual blind area based on the CNN model is constructed to predict the soot trend of the heating surfaces in the boiler visual blind area.

[0145] Step S203, for each boiler blind area heating surface of multiple boiler blind area heating surfaces, input the soot distribution data of the boiler blind area heating surface into the soot coking degree evaluation and warning model, and obtain the soot coking degree level of the boiler blind area heating surface output by the soot coking degree evaluation and warning model.

[0146] Here, the evaluation and early warning model for the degree of ash accumulation and coking is obtained through the following steps: First, determine the typical evaluation index system for the degree of ash accumulation and coking on the boiler heating surface. Among them, the evaluation indexes can include but are not limited to: ash accumulation thickness and coking area. Then, judge the data type of each index, and normalize different types of data to simplify the complexity of data processing of different types in the comprehensive evaluation process. Next, standardize each type of data after normalization to eliminate the influence of different dimensions of different data types. Subsequently, define the optimal value and the worst value of the index to calculate the optimal distance and the worst distance. Among them, the optimal distance is the Euclidean distance between the evaluation index values of the degree of ash accumulation and coking on the boiler heating surface and the optimal value, and the worst distance is the Euclidean distance between the evaluation index values of the degree of ash accumulation and coking on the boiler heating surface and the worst value. As an example, assume that the evaluation index is the ash accumulation thickness, then the worst value may be 10 mm, indicating that the heating surface is severely ash-accumulated and affects the boiler operation efficiency. After that, integrate the standardized data into a probability matrix. Secondly, based on the probability matrix, determine the information entropy of each index. Further, for the information entropy of each index, determine the index weight corresponding to the information entropy of the index. Finally, based on each weight, the optimal distance and the worst distance, determine multiple comprehensive proximity degrees, and based on the multiple comprehensive proximity degrees, classify the degree of ash accumulation and coking on the boiler heating surface. Based on the classification structure, construct an evaluation and early warning model for the degree of ash accumulation and coking on the heating surface of the boiler vision blind area based on the entropy weight - technique for order preference by similarity to an ideal solution (TOPSIS). As an example, when 0.8 ≤ comprehensive proximity degree ≤ 1, the ash accumulation and coking level of the heating surface in the boiler vision blind area is grade one; when 0.5 ≤ comprehensive proximity degree ≤ 0.7, the ash accumulation and coking level of the heating surface in the boiler vision blind area is grade two; when 0 ≤ comprehensive proximity degree ≤ 0.4, the ash accumulation and coking level of the heating surface in the boiler vision blind area is grade three.

[0147] Step S204, if the degree of ash accumulation and coking on the heating surface of the boiler blind area is greater than or equal to the preset level threshold, start the soot blower to remove the ash on the heating surface of the boiler blind area.

[0148] Here, following the above example, the preset level threshold is grade two.

[0149] Step S205, if the degree of ash accumulation and coking on the heating surface of the boiler blind area is less than the preset level threshold, do not start the soot blower.

[0150] A soot blowing control method for a boiler provided by an embodiment of the present application, through the method, the actual pollution condition of the heating surface is monitored in real time, the soot blowing strategy is dynamically optimized, on-demand soot blowing is realized, redundant soot blowing operations are reduced, the steam consumption of the soot blower and the risk of heating surface blowing damage are effectively reduced, and at the same time, safety accidents caused by ash accumulation are avoided, and the safety and economy of boiler operation are improved.

[0151] Based on the same application concept, an ash blowing device for a boiler corresponding to the ash blowing control method for a boiler provided in the above embodiment is further provided in the embodiment of the present application. Since the principle of solving problems by the device in the embodiment of the present application is similar to that of the ash blowing control method for a boiler in the above embodiment of the present application, the implementation of the device can refer to the implementation of the method, and the repeated parts will not be elaborated.

[0152] Please refer to Figure 10 and Figure 11 , Figure 10 which is one of the schematic structural diagrams of an ash blowing device for a boiler provided in the embodiment of the present application. Figure 11 which is the second schematic structural diagram of an ash blowing device for a boiler provided in the embodiment of the present application.

[0153] As Figure 10 shown, the ash blowing control device 310 for a boiler provided in the embodiment of the present application includes:

[0154] An image acquisition module 311, for each visual device, acquires a current image of the ash accumulation state of the boiler heating surface collected by the visual device;

[0155] A detailed feature extraction module 312 divides the current image of the ash accumulation state of the boiler heating surface to obtain a plurality of initial ash accumulation state images, and for each initial ash accumulation state image, performs feature extraction on the initial ash accumulation state image to obtain a plurality of initial detailed feature vectors of the initial ash accumulation state image;

[0156] A detailed feature reconstruction module 313 generates an ash accumulation feature reconstruction image corresponding to the visual device by using the plurality of initial detailed feature vectors of the initial ash accumulation state image;

[0157] A local feature extraction module 314 divides the ash accumulation feature reconstruction image corresponding to the visual device to obtain a plurality of initial ash accumulation feature reconstruction images, and for each initial ash accumulation feature reconstruction image, performs feature extraction on the initial ash accumulation feature reconstruction image to obtain a plurality of basic local feature vectors of the initial ash accumulation feature reconstruction image;

[0158] A local feature reconstruction module 315 generates an ash accumulation state image corresponding to the visual device by using the plurality of basic local feature vectors of the initial ash accumulation feature reconstruction image;

[0159] A thermal resistance calculation module 316 classifies the states of the ash accumulation state images corresponding to all visual devices, and based on the result of the state classification, determines the ash accumulation thermal resistance value corresponding to at least one target ash accumulation state image;

[0160] The first soot blowing decision-making module 317 compares the soot thermal resistance values corresponding to all target soot accumulation state images with the soot thermal resistance threshold respectively. Based on the comparison results, it determines whether to start the soot blower to remove the soot on the corresponding boiler heating surface.

[0161] Further, the detailed feature reconstruction module 313 is specifically configured to:

[0162] Perform convolution processing on multiple initial detailed feature vectors of the initial soot accumulation state image to obtain multiple key feature vectors of the initial soot accumulation state image; perform deconvolution processing on the multiple key feature vectors of the initial soot accumulation state image to obtain multiple detailed feature vectors of the initial soot accumulation state image; compare the multiple initial detailed feature vectors of the initial soot accumulation state image with the multiple detailed feature vectors of the initial soot accumulation state image to obtain multiple differential detailed feature vectors; determine each differential detailed feature vector as a key feature vector of the initial soot accumulation state image; for each initial detailed feature vector in the multiple initial detailed feature vectors of the initial soot accumulation state image, calculate the correlation between the initial detailed feature vector and each initial detailed feature vector and each key feature vector respectively to obtain the weights of the initial detailed feature vector with respect to each initial detailed feature vector and each key feature vector; based on the weights of the initial detailed feature vector with respect to each initial detailed feature vector and each key feature vector, perform weighted fusion on the initial detailed feature vector, each initial detailed feature vector and each key feature vector to obtain a fused key feature vector of the initial soot accumulation state image; combine all the fused key feature vectors to generate a soot accumulation feature reconstruction image corresponding to the visual device.

[0163] Further, the local feature reconstruction module 315 is specifically configured to:

[0164] Perform convolution processing on multiple basic local feature vectors of the reconstructed image of the initial fouling characteristics to obtain multiple core feature vectors of the reconstructed image of the initial fouling characteristics; perform deconvolution processing on the multiple core feature vectors of the reconstructed image of the initial fouling characteristics to obtain multiple local feature vectors of the reconstructed image of the initial fouling characteristics; compare the multiple initial local feature vectors of the reconstructed image of the initial fouling characteristics with the multiple local feature vectors of the reconstructed image of the initial fouling characteristics to obtain multiple differential local feature vectors; for each initial local feature vector among the multiple initial local feature vectors of the reconstructed image of the initial fouling characteristics, calculate the correlation between the initial local feature vector and each initial local feature vector and each core feature vector respectively to obtain the weights of the initial local feature vector with respect to each initial local feature vector and each core feature vector; based on the weights of the initial local feature vector with respect to each initial local feature vector and each core feature vector, perform weighted fusion on the initial local feature vector, each initial local feature vector, and each core feature vector to obtain the fused core feature vector of the reconstructed image of the initial fouling characteristics; combine all the fused core feature vectors to generate the fouling state image corresponding to the visual device.

[0165] Further, the thermal resistance calculation module 316 is specifically configured to:

[0166] Input the fused core feature vectors of the fouling state images corresponding to all visual devices into a preset classification model, so that the preset classification model classifies the states of the fouling state images corresponding to all visual devices, and obtain at least one target fouling state image output by the preset classification model; based on the mapping relationship between the fouling state image and the fouling thermal resistance value, determine the fouling thermal resistance value corresponding to at least one target fouling state image.

[0167] Further, when the first soot blowing decision module 317 is used to determine whether to start the soot blower to remove the fouling on the corresponding boiler heating surface based on the comparison result, it is further specifically configured to:

[0168] For each target fouling state image, if the fouling thermal resistance value corresponding to the target fouling state image is greater than or equal to the fouling thermal resistance threshold, start the soot blower to remove the fouling on the boiler heating surface corresponding to the target fouling state image; if the fouling thermal resistance value corresponding to the target fouling state image is less than the fouling thermal resistance threshold, do not start the soot blower.

[0169] As Figure 11 shown, further, the soot blowing control device further includes:

[0170] A load acquisition module 318, which acquires the current load of the boiler;

[0171] The ash accumulation trend prediction module 319 inputs the current load of the boiler into the ash accumulation trend prediction model, and obtains the ash accumulation distribution data of multiple heat transfer surfaces in the blind area of the visual device output by the ash accumulation trend prediction model;

[0172] The level determination module 320 evaluates each heat transfer surface in the blind area of the boiler. It inputs the ash accumulation distribution data of this heat transfer surface in the blind area of the boiler into the ash fouling and slagging degree evaluation and warning model, and obtains the ash fouling and slagging degree level of this heat transfer surface in the blind area of the boiler output by the ash fouling and slagging degree evaluation and warning model;

[0173] The second soot blowing decision module 321, if the ash fouling and slagging degree level of this heat transfer surface in the blind area of the boiler is greater than or equal to the preset level threshold, starts the soot blower to remove the ash accumulation on this heat transfer surface in the blind area of the boiler;

[0174] The third soot blowing decision module 322, if the ash fouling and slagging degree level of this heat transfer surface in the blind area of the boiler is less than the preset level threshold, does not start the soot blower.

[0175] Furthermore, the ash accumulation trend prediction module 319 constructs the ash accumulation prediction model through the following steps:

[0176] Based on the physical properties of ash accumulation, construct a coal type ash accumulation model for the boiler;

[0177] Based on the geometric structure of the boiler, construct the boiler heat transfer surface structure; based on the coal type ash accumulation model of the boiler and the boiler heat transfer surface structure, construct the ash accumulation numerical simulation model of the boiler heat transfer surface; input multiple load and coal type parameters into the ash accumulation numerical simulation model, and obtain the running state of ash accumulation particles in the boiler output by the ash accumulation numerical simulation model; input the running state of ash accumulation particles in the boiler into the coal type ash accumulation model, and obtain the ash accumulation distribution data of multiple heat transfer surfaces in the boiler corresponding to each load among multiple loads; based on the ash accumulation distribution data of multiple heat transfer surfaces in the boiler corresponding to each load among multiple loads, construct the ash accumulation prediction model.

[0178] A soot blowing control device for a boiler provided by an embodiment of the present application, through this device, can monitor the actual pollution condition of the heat transfer surface in real time, dynamically optimize the soot blowing strategy, achieve on-demand soot blowing, reduce redundant soot blowing operations, effectively reduce the steam consumption of the soot blower and the risk of heat transfer surface blowing damage, and at the same time avoid safety accidents caused by ash accumulation, improving the safety and economy of boiler operation.

[0179] Please refer to Figure 12 , Figure 12 which is a schematic structural diagram of an electronic device provided by an embodiment of the present application.

[0180] As Figure 12As shown in the figure, the electronic device 400 includes a processor 410, a memory 420, and a bus 430.

[0181] The memory 420 stores machine-readable instructions executable by the processor 410. When the electronic device 400 runs, the processor 410 communicates with the memory 420 through the bus 430. When the machine-readable instructions are executed by the processor 410, they can execute the steps of the soot blowing control method for a boiler as described above Figure 1 , Figure 3 , Figure 4 , Figure 5 , Figure 8 and Figure 9 shown. For the specific implementation manner, reference may be made to the method embodiments, which will not be elaborated herein.

[0182] The embodiments of the present application further provide a computer-readable storage medium. A computer program is stored on the computer-readable storage medium. When the computer program is run by a processor, it can execute the steps of the soot blowing control method for a boiler as described above Figure 1 , Figure 3 , Figure 4 , Figure 5 , Figure 8 and Figure 9 shown. For the specific implementation manner, reference may be made to the method embodiments, which will not be elaborated herein.

[0183] Those skilled in the art can clearly understand that for the convenience and brevity of description, the specific working processes of the above-described systems and devices can refer to the corresponding processes in the foregoing method embodiments, which will not be elaborated herein.

[0184] In several embodiments provided by the present application, it should be understood that the disclosed systems, devices, and methods can be implemented in other ways. The device embodiments described above are merely illustrative. For example, the division of the units is only a logical function division, and there can be other division methods in actual implementation. For another example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the couplings or direct couplings or communication connections shown or discussed with each other can be through some communication interfaces. The indirect couplings or communication connections of the devices or units can be in electrical, mechanical, or other forms.

[0185] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they can be located in one place, or they can be distributed to multiple network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0186] In addition, in each embodiment of the present application, each functional unit may be integrated into one processing unit, may exist physically alone for each unit, or two or more units may be integrated into one unit.

[0187] If the above-mentioned function is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a non-volatile computer-readable storage medium executable by a processor. Based on such an understanding, the technical solution of the present application, in essence, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes several instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the method described in each embodiment of the present application. The aforementioned storage medium includes: various media that can store program codes, such as USB flash drives, mobile hard disks, read-only memories (ROM), random access memories (RAM), magnetic disks, or optical discs.

[0188] The above is only the specific implementation manner of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art within the technical scope disclosed by the present application can easily think of changes or substitutions, which should all be covered by the protection scope of the present application. Therefore, the protection scope of the present application should be subject to the protection scope of the claims.

Claims

1. A sootblowing control method for a boiler, characterized in that: The boiler is provided with at least one visual device, and the visual device is used to collect images of the sooting state of the boiler heating surface; the sootblowing control method comprises: For each visual device, a current image of the dusty state of the boiler heating surface collected by the visual device is obtained; Segmenting the current image of the dust accumulation state of the boiler heating surface to obtain a plurality of initial dust accumulation state images, and extracting features of each initial dust accumulation state image to obtain a plurality of initial detail feature vectors of the initial dust accumulation state image; Using a plurality of initial detail feature vectors of the initial dust accumulation state image, a dust accumulation feature reconstructed image corresponding to the visual device is generated; Segmenting the dust accumulation feature reconstructed image corresponding to the visual device to obtain a plurality of initial dust accumulation feature reconstructed images, and extracting features from each initial dust accumulation feature reconstructed image to obtain a plurality of basic local feature vectors of the initial dust accumulation feature reconstructed image; Reconstructing a plurality of basic local feature vectors of the image using the initial dust accumulation feature to generate a dust accumulation state image corresponding to the visual device; Classify the dust accumulation status images corresponding to all visual devices, and determine the dust accumulation thermal resistance value corresponding to at least one target dust accumulation status image based on the status classification result; The ash accumulation thermal resistance values ​​corresponding to all target ash accumulation state images are compared with the ash accumulation thermal resistance thresholds respectively, and based on the comparison results, it is determined whether to start the soot blower to remove the ash accumulation on the corresponding boiler heating surface.

2. The sootblowing control method according to claim 1, characterized in that: The step of using a plurality of initial detail feature vectors of the initial dust accumulation state image to generate a dust accumulation feature reconstructed image corresponding to the visual device includes: Performing convolution processing on a plurality of initial detail feature vectors of the initial dust accumulation state image to obtain a plurality of key feature vectors of the initial dust accumulation state image; Performing deconvolution processing on multiple key feature vectors of the initial dust accumulation state image to obtain multiple detail feature vectors of the initial dust accumulation state image; Comparing a plurality of initial detail feature vectors of the initial dust accumulation state image with a plurality of detail feature vectors of the initial dust accumulation state image to obtain a plurality of difference detail feature vectors; Each difference detail feature vector is determined as a key feature vector of the initial dust accumulation state image; For each initial detail feature vector among a plurality of initial detail feature vectors of the initial dust accumulation state image, respectively, the initial detail feature vector is correlated with each initial detail feature vector and each key feature vector to obtain weights of the initial detail feature vector, each initial detail feature vector and each key feature vector; Based on the weights of the initial detail feature vector, each initial detail feature vector and each key feature vector, the initial detail feature vector, each initial detail feature vector and each key feature vector are weightedly fused to obtain a fused key feature vector of the initial dust accumulation state image; All fused key feature vectors are combined to generate a dust accumulation feature reconstructed image corresponding to the visual device.

3. The sootblowing control method according to claim 1, characterized in that: The step of reconstructing a plurality of basic local feature vectors of the image using the initial dust accumulation feature to generate a dust accumulation state image corresponding to the visual device includes: Performing convolution processing on a plurality of basic local feature vectors of the initial dust accumulation feature reconstructed image to obtain a plurality of core feature vectors of the initial dust accumulation feature reconstructed image; Performing deconvolution processing on multiple core feature vectors of the initial dust accumulation feature reconstructed image to obtain multiple local feature vectors of the initial dust accumulation feature reconstructed image; Comparing a plurality of initial local feature vectors of the initial dust accumulation feature reconstructed image with a plurality of local feature vectors of the initial dust accumulation feature reconstructed image to obtain a plurality of difference local feature vectors; Each difference local feature vector is determined as the core feature vector of the initial dust accumulation feature reconstructed image; For each initial local feature vector of a plurality of initial local feature vectors of the initial dust accumulation feature reconstructed image, correlation calculation is performed on the initial local feature vector with each initial local feature vector and each core feature vector to obtain weights of the initial local feature vector with each initial local feature vector and each core feature vector; Based on the weights of the initial local feature vector, each initial local feature vector and each core feature vector, the initial local feature vector, each initial local feature vector and each core feature vector are weightedly fused to obtain a fused core feature vector of the initial dust accumulation feature reconstructed image; All fused core feature vectors are combined to generate a dust accumulation state image corresponding to the visual device.

4. The sootblowing control method according to claim 1, characterized in that: The step of classifying the dust accumulation status images corresponding to all visual devices and determining the dust accumulation thermal resistance value corresponding to at least one target dust accumulation status image based on the result of the status classification includes: Inputting the fused core feature vectors of the dust accumulation state images corresponding to all visual devices into a preset classification model, so that the preset classification model performs state classification on the dust accumulation state images corresponding to all visual devices, and obtaining at least one target dust accumulation state image output by the preset classification model; Based on the mapping relationship between the dust accumulation state image and the dust accumulation thermal resistance value, a dust accumulation thermal resistance value corresponding to at least one target dust accumulation state image is determined.

5. The sootblowing control method according to claim 1, characterized in that: The step of determining whether to start a sootblower to remove the accumulated soot on the corresponding boiler heating surface based on the comparison result includes: For each target ash accumulation state image, if the ash accumulation thermal resistance value corresponding to the target ash accumulation state image is greater than or equal to the ash accumulation thermal resistance threshold, then start the soot blower to remove the ash accumulation on the boiler heating surface corresponding to the target ash accumulation state image; If the ash accumulation thermal resistance value corresponding to the target ash accumulation state image is less than the ash accumulation thermal resistance threshold, the soot blower is not started.

6. The sootblowing control method according to claim 1, characterized in that: The sootblowing control method further comprises: Get the current load of the boiler; Inputting the current load of the boiler into the ash accumulation trend prediction model, and obtaining the ash accumulation distribution data of the blind heating surfaces of the boilers located in the blind area of ​​the visual device output by the ash accumulation trend prediction model; For each of the multiple boiler blind heating surfaces, input the ash accumulation distribution data of the boiler blind heating surface into the ash accumulation and coking degree evaluation and early warning model, and obtain the ash accumulation and coking degree level of the boiler blind heating surface output by the ash accumulation and coking degree evaluation and early warning model; If the ash and coking degree level of the heating surface of the blind area of ​​the boiler is greater than or equal to the preset level threshold, the soot blower is started to remove the ash accumulation on the heating surface of the blind area of ​​the boiler; If the ash and coking degree level of the heating surface in the blind area of ​​the boiler is lower than the preset level threshold, the soot blower will not be started.

7. The sootblowing control method according to claim 6, characterized in that: The dust accumulation prediction model is constructed by following the steps below: Based on the physical properties of ash accumulation, the coal ash accumulation model of the boiler is constructed; Based on the geometric structure of the boiler, construct the boiler heating surface structure; Based on the coal type ash accumulation model of the boiler and the boiler heating surface structure, a numerical simulation model of ash accumulation on the boiler heating surface is constructed; Inputting a plurality of load and coal type parameters into the ash accumulation numerical simulation model, and obtaining the operating state of ash accumulation particles in the boiler output by the ash accumulation numerical simulation model; Inputting the operating state of the ash particles in the boiler into the coal type ash accumulation model, obtaining the ash accumulation distribution data of multiple boiler heating surfaces corresponding to each of the multiple loads output by the coal type ash accumulation model; An ash accumulation prediction model is constructed based on ash accumulation distribution data of multiple boiler heating surfaces corresponding to each of the multiple loads.

8. A sootblowing control device for a boiler, characterized in that: The sootblowing control device comprises: An image acquisition module, for each visual device, acquires a current image of the dust state of the boiler heating surface acquired by the visual device; A detail feature extraction module is used to segment the current image of the dust accumulation state of the boiler heating surface to obtain a plurality of initial dust accumulation state images, and for each initial dust accumulation state image, perform feature extraction on the initial dust accumulation state image to obtain a plurality of initial detail feature vectors of the initial dust accumulation state image; A detail feature reconstruction module generates a dust accumulation feature reconstruction image corresponding to the visual device using a plurality of initial detail feature vectors of the initial dust accumulation state image; A local feature extraction module is used to segment the dust accumulation feature reconstructed image corresponding to the visual device to obtain a plurality of initial dust accumulation feature reconstructed images, and for each initial dust accumulation feature reconstructed image, perform feature extraction on the initial dust accumulation feature reconstructed image to obtain a plurality of basic local feature vectors of the initial dust accumulation feature reconstructed image; A local feature reconstruction module, which reconstructs a plurality of basic local feature vectors of the image using the initial dust accumulation feature to generate a dust accumulation state image corresponding to the visual device; The thermal resistance calculation module classifies the dust accumulation state images corresponding to all visual devices, and determines the dust accumulation thermal resistance value corresponding to at least one target dust accumulation state image based on the state classification result; The first sootblowing decision module compares the soot thermal resistance values ​​corresponding to all target soot state images with the soot thermal resistance threshold values, and determines whether to start the sootblower to remove the soot on the corresponding boiler heating surface based on the comparison result.

9. An electronic device, characterized in that: include: A processor, a memory and a bus, wherein the memory stores machine-readable instructions executable by the processor, and when the electronic device is running, the processor and the memory communicate through the bus, and the machine-readable instructions are executed by the processor to execute the steps of the sootblowing control method for a boiler as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps of the sootblowing control method for a boiler according to any one of claims 1 to 7 are executed.

Images