Image-based method, device, equipment, and medium for bag breakage identification and localization in baghouse dust collectors.

Abstract

This application discloses an image-based method, apparatus, device, and medium for identifying and locating bag ruptures in a baghouse dust collector, relating to the field of dust collection equipment technology. The method includes: acquiring image information of the bag opening of the baghouse dust collector; preprocessing the bag opening image information to generate target image information; inputting the target image information into a preset deep learning model for bag rupture location identification to obtain the row and column numbers of the rupture openings; storing the row and column numbers of the rupture openings; and sending an alarm. By acquiring images of the bag openings of the baghouse dust collector, preprocessing and identifying ruptures using a deep learning model, and directly outputting the row and column numbers of the rupture openings, rapid location and alarm functions are achieved, improving detection efficiency and accuracy, and facilitating real-time monitoring and maintenance.

Description

Technical Field

[0001] This invention relates to the field of dust removal equipment technology, and in particular to an image-based method, device, equipment, and medium for identifying and locating bag breaks in a bag filter. Background Technology

[0002] Baghouse dust collectors are widely used in various industrial sectors, including steel, cement, aluminum, non-ferrous metallurgy, and waste incineration, due to their high dust removal efficiency, large air volume handling capacity, and operational reliability. As the core component of a baghouse dust collector, the service life and condition of the filter bag directly affect the equipment's operating efficiency and environmental compliance. The cost of the filter bag can account for 10%-30% of the total equipment cost, with a warranty period typically of 1.5-3 years. Damaged filter bags can lead to excessive dust emission concentrations in flue gas, potentially causing a forced shutdown of the entire plant and resulting in significant economic losses.

[0003] Currently, the main methods for detecting filter bag damage include the following categories: methods based on changes in dust concentration, which use the increase in dust concentration in the clean air chamber to determine whether the filter bag is damaged; detection using pressure difference changes, which identifies filter bag damage by monitoring the change in pressure difference between the clean air chamber and the ash hopper before and after the bag is damaged; and monitoring the filter bag status using advanced technologies such as distributed fiber optic sensors or lidar. Other solutions utilize gas flow rate changes, charge methods, and visual monitoring to detect filter bag damage.

[0004] However, many sensor-based technologies cannot accurately detect the location of filter bag damage in real time, leading to delayed problem detection. While some advanced detection technologies (such as distributed fiber optic sensors) offer high accuracy, they are difficult to implement, requiring professional technicians for installation and maintenance, increasing costs. Furthermore, most detection technologies can only indicate the location of the filter bag malfunction, without providing specific row and column numbers, resulting in low repair efficiency. Therefore, a more efficient and direct method for locating filter bag damage is needed. Summary of the Invention

[0005] The main objective of this application is to provide an image-based method, apparatus, equipment, and medium for identifying and locating bag breaks in a baghouse dust collector, aiming to solve the technical problem of how to directly obtain the location of the bag break opening.

[0006] To achieve the above objectives, this application proposes an image-based method for identifying and locating bag breaks in a baghouse dust collector, the method comprising:

[0007] Obtain image information of the bag opening of the bag filter;

[0008] The bag opening image information is preprocessed to generate target image information;

[0009] The target image information is input into a preset bag-breaking location deep learning model for identification to obtain the row and column numbers of the bag-breaking holes.

[0010] The data is stored and an alarm is sent based on the row and column numbers of the bag openings.

[0011] In one embodiment, the preset bag-breaking localization deep learning model includes a first preset module, a second preset module, a third preset module, a fourth preset module, a fifth preset module, and a sixth preset module. The first preset module and the second preset module each include two convolutional layers, one max pooling layer, and a coordinate attention mechanism submodule. The third preset module includes three convolutional layers, one max pooling layer, and a coordinate attention mechanism submodule. The fourth preset module includes three convolutional layers and a coordinate attention mechanism submodule. The fifth preset module includes five convolutional layers and a coordinate attention mechanism submodule. The sixth preset module includes a fully connected layer, an activation function layer, and a reconstruction layer.

[0012] In one embodiment, before inputting the target image information into a preset bag-breaking location deep learning model for identification to obtain the row and column numbers of the bag-breaking opening, the following steps are included:

[0013] Acquire sample image information of the bag opening and establish an initial deep learning model for bag breakage localization;

[0014] The initial bag-breaking location deep learning model is trained based on the bag opening sample image information to obtain the preset bag-breaking location deep learning model.

[0015] In one embodiment, the step of training an initial deep learning model for bag breakage localization based on the bag opening sample image information to obtain a preset deep learning model for bag breakage localization includes:

[0016] Initialize the model parameters of the initial bag-breaking localization deep learning model;

[0017] Obtain the sample image information and actual coordinate information of the bag opening;

[0018] The bag opening sample image information is normalized to obtain the processed image information;

[0019] The processed image information is input into the initial neural network model to obtain predicted coordinate information;

[0020] The error value between the predicted coordinate information and the true coordinate information is calculated based on the loss function.

[0021] Obtain the learning rate;

[0022] The model parameters are iteratively updated using an optimization algorithm based on the learning rate until the maximum number of iterations is reached or the error value calculated by the loss function converges to a preset threshold, thus obtaining a preset bag-breaking localization deep learning model.

[0023] In one embodiment, the step of inputting the target image information into a preset bag-breaking location deep learning model for identification to obtain the row and column numbers of the bag-breaking opening includes:

[0024] The target image information is used to extract features and generate feature vectors by using the first, second, third, fourth and fifth preset modules of the preset bag-breaking positioning deep learning model.

[0025] The feature vector is identified by the sixth preset module of the preset bag-breaking localization deep learning model to obtain the row and column numbers of the bag-breaking opening.

[0026] In one embodiment, the step of identifying the feature vector by the sixth preset module of the preset bag-breaking localization deep learning model to obtain the row and column numbers of the bag-breaking opening includes:

[0027] The feature vector is mapped through the fully connected layer of the sixth preset module to obtain a one-dimensional vector;

[0028] The one-dimensional vector is passed through the activation function layer of the sixth preset module to obtain the damage values ​​of multiple holes;

[0029] When the damage value is a preset value, it is determined to be a bag breakage, and a one-dimensional vector of the corresponding bag breakage hole is output;

[0030] The one-dimensional vector is transformed through the reconstruction layer of the sixth preset module to obtain a two-dimensional vector, wherein the two-dimensional vector is a vector of length I×J, where I represents the number of rows of the clean air chamber opening of the bag filter, J represents the number of columns of the clean air chamber opening of the bag filter, and the size of the I×J array corresponds to the row and column layout size of the clean air chamber opening.

[0031] The corresponding row and column numbers of the bag-breaking holes are obtained from the two-dimensional vector.

[0032] In one embodiment, the step of preprocessing the bag opening image information to generate target image information includes:

[0033] Based on the bag opening image information, the resolution of the bag opening image is adjusted to a preset size to obtain an image with adjusted size;

[0034] The pixels of the resized image are normalized to a preset range to obtain the processed image information;

[0035] The processed image information is used as the target image information.

[0036] Furthermore, to achieve the above objectives, this application also proposes an image-based bag filter bag breakage identification and positioning device, which includes:

[0037] The acquisition module is used to acquire image information of the bag opening of the bag filter;

[0038] The processing module is used to preprocess the bag opening image information to generate target image information;

[0039] The identification module is used to input the target image information into a preset bag-breaking location deep learning model for identification, and obtain the row and column numbers of the bag-breaking holes;

[0040] The results module is used to store data based on the row and column numbers of the bag openings and to send alarms.

[0041] In addition, to achieve the above objectives, this application also proposes a medium, which is a computer-readable medium, on which a computer program is stored, and when the computer program is executed by a processor, it implements the steps of the image-based bag filter bag breakage identification and positioning method described above.

[0042] In addition, to achieve the above objectives, this application also provides a computer program product, which includes a computer program that, when executed by a processor, implements the steps of the image-based bag filter bag breakage identification and positioning method described above.

[0043] This application acquires image information of the bag opening of a bag filter, preprocesses the image information to generate target image information, and inputs the target image information into a preset deep learning model for bag rupture location to obtain the row and column numbers of the rupture openings. Based on these row and column numbers, the data is stored and an alarm is sent. By acquiring bag opening images of the bag filter, preprocessing and identifying ruptures using a deep learning model, and directly outputting the row and column numbers of the rupture openings, rapid location and alarm functions are achieved, improving detection efficiency and accuracy, and facilitating real-time monitoring and maintenance. Attached Figure Description

[0044] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, those skilled in the art can obtain other drawings based on these drawings without creative effort.

[0045] Figure 1 This is a flowchart illustrating the first embodiment of the image-based bag filter bag breakage identification and positioning method of this application;

[0046] Figure 2 This is a distribution diagram of filter bag openings in the first embodiment of the image-based bag filter bag breakage identification and positioning method of this application.

[0047] Figure 3 This image shows the detection result of a single hole in a bag breakage in the first embodiment of the image-based bag filter bag breakage identification and localization method of this application.

[0048] Figure 4 This is a diagram showing the array of single-hole bag rupture results from the first embodiment of the image-based bag filter rupture identification and localization method of this application;

[0049] Figure 5 This image shows the detection results of multiple holes in the bag breakage detection method of the first embodiment of the image-based bag filter bag breakage identification and positioning method of this application.

[0050] Figure 6 This is an image showing the result of an array of multiple holes in a bag-breaking area in the first embodiment of the image-based bag filter bag breakage identification and localization method of this application.

[0051] Figure 7 This is a flowchart illustrating the second embodiment of the image-based bag filter bag breakage identification and positioning method of this application;

[0052] Figure 8 This is a flowchart illustrating the third embodiment of the image-based bag filter bag breakage identification and positioning method of this application;

[0053] Figure 9 This is a schematic diagram of the module structure of the image-based bag filter bag breakage identification and positioning device according to an embodiment of this application;

[0054] Figure 10 This is a schematic diagram of the equipment structure of the hardware operating environment involved in the image-based bag filter bag breakage identification and positioning method in the embodiments of this application.

[0055] The purpose, features, and advantages of this application will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0056] It should be understood that the specific embodiments described herein are merely illustrative of the technical solutions of this application and are not intended to limit this application.

[0057] To better understand the technical solution of this application, a detailed description will be provided below in conjunction with the accompanying drawings and specific implementation methods.

[0058] The main solution of this application embodiment is: to acquire bag opening image information of a bag filter, preprocess the bag opening image information to generate target image information, input the target image information into a preset bag breakage location deep learning model for identification, obtain the row and column number of the bag breakage opening, store the bag breakage opening based on the row and column number of the bag breakage opening, and send an alarm.

[0059] Based on this, embodiments of this application provide an image-based method for identifying and locating bag breaks in a baghouse dust collector, referring to... Figure 1 , Figure 1 This is a flowchart illustrating the first embodiment of the image-based bag filter bag breakage identification and positioning method of this application.

[0060] In this embodiment, the image-based bag filter bag breakage identification and positioning method includes steps S10 to S40:

[0061] Step S10: Obtain the bag opening image information of the bag filter.

[0062] It should be noted that by using a high-resolution camera mounted on the top of the air purification chamber to capture images of the bag openings, the working status of each filter bag can be observed, obtaining image information of each bag's opening. Furthermore, the camera should possess wide dynamic range (WDR) functionality to maintain image clarity even in extreme lighting conditions, avoiding overexposure or underexposure that could affect recognition performance. To ensure image quality, a supplementary lighting device is also included, enabling clear images even in poor lighting conditions. This supplementary lighting device typically uses adjustable-brightness LEDs, whose color temperature and brightness automatically adjust according to the actual environment, ensuring optimal shooting results under any lighting conditions. Simultaneously, to avoid reflections or shadows that may be caused by direct light sources, the supplementary lighting device is designed with special optical structures, such as diffusers or reflectors, to distribute light evenly throughout the entire shooting area.

[0063] Step S20: Preprocess the bag opening image information to generate target image information.

[0064] It's important to note that the resolution differences between images captured by different cameras can impact the performance and efficiency of deep learning models. Therefore, all images need to be resized to a uniform size. However, even after resizing, variations in brightness and contrast between images can still interfere with model training and prediction. Therefore, the pixel values ​​of the resized images need to be normalized, mapping them to a preset range.

[0065] Further, based on the bag opening image information, the resolution of the bag opening image is adjusted to a preset size to obtain an adjusted image. Then, the pixels of the adjusted image are normalized to a preset range to obtain processed image information, which is used as the target image information. Specifically, firstly, based on the bag opening image information, the resolution of the acquired image is adjusted to a preset size. This process is achieved through an image scaling algorithm to ensure that the aspect ratio of the image remains consistent, avoiding feature distortion caused by inconsistent sizes. The selection of the preset size is usually optimized according to the input requirements of the deep learning model, such as common input sizes of 1024×1024 pixels or 512×512 pixels. Although the adjusted image meets the model input requirements in the spatial dimension, the pixel value range of the image may vary greatly due to the complex and variable lighting conditions in the actual shooting environment. Therefore, it is necessary to further normalize the pixels of the adjusted image. The normalization operation maps the pixel values ​​of the image from the original range (such as 0-255) to a preset range, such as [0,1] or [-1,1]. This process is achieved through a simple linear transformation, which can effectively eliminate the interference of light intensity differences on image analysis and enhance the model's robustness to different shooting conditions.

[0066] The image information after normalization is the target image information. This target image information, while maintaining the original image features, possesses better consistency and stability, providing high-quality input data for subsequent deep learning model recognition and bag breakage localization.

[0067] Step S30: Input the target image information into the preset bag-breaking localization deep learning model for recognition to obtain the row and column numbers of the bag-breaking holes.

[0068] It should be noted that the preprocessed target image information is input into the deep learning model. The model extracts and analyzes image features through multiple modules to obtain the row and column numbers of the hole in the bag. In this embodiment, the aforementioned preset deep learning model for locating the hole in the bag is constructed based on the Coordinate Attention (CA) submodule, which can extract image features and directly output the row and column numbers of the hole.

[0069] Specifically, the CA submodule is a computational unit designed to enhance feature representation capabilities, particularly suitable for lightweight networks. It addresses the problem of traditional channel attention ignoring location information by embedding positional information into channel attention. In traditional channel attention methods, the input is typically transformed into a single feature vector through 2D global pooling. While this method effectively captures the relationships between channels, it loses crucial spatial location information. The coordinate attention mechanism solves this problem by embedding location information into channel attention, allowing any intermediate tensor X = [X1, X2, ..., X...] to be processed.C ]∈R C×H×W As input, it outputs an output of the same size with enhanced representation capabilities, Y = [Y1, Y2, ..., Y]. C The CA submodule encodes channel relationships and long-range dependencies using precise location information, and consists of two steps: coordinate information embedding and coordinate attention generation. For the coordinate information embedding part, a pair of 1D feature encoding operations are performed on the input tensor X of size C×H×W, using pooling windows of sizes (H, 1) and (1, W) along the horizontal and vertical directions respectively, resulting in two feature maps with location information. The output of the c-th channel with height h is expressed as:

[0070]

[0071] The output of the c-th channel with width w is expressed as:

[0072]

[0073] Next, to better utilize the representation with a global receptive field and precise location information generated by the coordinate information embedding module, a coordinate attention generation operation is designed to generate an attention map. First, the two feature maps generated by the previous module are concatenated, and then a shared 1x1 convolution is used to perform an F1 transformation. The generated f represents the intermediate feature maps of spatial information in the horizontal and vertical directions, expressed by the formula:

[0074] f=δ(F1(z h ,z w ))

[0075] Next, f is divided into two separate tensors along the spatial dimension. h ∈R C / r×H and f h ∈R C / r×W Then use two 1x1 convolutions F h and F w feature map f h f w Transforming the input X to have the same number of channels yields the following result:

[0076] g h =σ(F h (f h ))

[0077] g w =σ(F w (f w ))

[0078] Then for g h and g wExtending this further, as attention weights, the final output of the CA submodule is represented as follows:

[0079]

[0080] Therefore, the coordinate attention mechanism submodule can not only capture cross-channel information, but also capture direction awareness and position awareness information, which can help the model more accurately locate and identify targets of interest. Secondly, the coordinate attention mechanism module is flexible and lightweight, and can be easily inserted into classic modules to enhance features by strengthening information representation.

[0081] Furthermore, the aforementioned preset bag-breaking localization deep learning model includes a first preset module, a second preset module, a third preset module, a fourth preset module, a fifth preset module, and a sixth preset module. The first and second preset modules each include two convolutional layers, one max pooling layer, and a coordinate attention mechanism submodule. The third preset module includes three convolutional layers, one max pooling layer, and a coordinate attention mechanism submodule. The fourth preset module includes three convolutional layers and a coordinate attention mechanism submodule. The fifth preset module includes five convolutional layers and a coordinate attention mechanism submodule. The sixth preset module includes one fully connected layer, one activation function layer, and one reconstruction layer. The specific framework composition can be seen from Table 1, the composition table of the preset bag-breaking localization deep learning model.

[0082] Table 1. Composition of the Pre-set Bag Breaking and Localization Deep Learning Model

[0083]

[0084]

[0085] First, the target image information is fed into the first layer of a pre-defined deep learning model for bag breakage localization for initial feature extraction. During this process, the model uses convolutional layers (conv) to perform multiple convolution operations on the image to capture the basic morphology of the filter bag surface and potential signs of breakage. Through max pooling layers, the model further reduces the spatial size of the feature maps while retaining the most important feature information. Based on this, a coordinate attention mechanism (CA) is introduced, which aggregates features along the horizontal and vertical directions, generating a pair of orientation-aware and position-sensitive feature maps. These feature maps not only enhance the representation of the region of interest (i.e., the possible location of the breakage) but also retain precise positional information, which is crucial for subsequent localization. As data is passed layer by layer in the network, the model gradually refines its understanding of the image, from the overall layout to local details, until the specific location of the breakage is finally determined. Each module contains a series of convolutional layers, max pooling layers, and coordinate attention mechanisms, enabling the model to capture long-range dependencies and positional information at different scales. For example, in the third module, three convolutional operations are followed by a max-pooling operation and a coordinate attention mechanism, aiming to directly minimize the feature vector size while extracting more detailed features, corresponding to the specific details of the filter bag openings. When the data reaches the last layer of the network, the fully connected layer (FC-I*J) predicts the row and column positions of the overall bag rupture based on the previously extracted features. Here, I represents the number of rows of the air chamber openings, and J represents the number of columns. The output of the fully connected layer is converted to a 0 or 1 value by the Sigmoid function, where "1" represents a ruptured opening and "0" represents normal. Finally, the Reshape function is used to reshape the output into a two-dimensional array consistent with the actual opening arrangement, directly indicating the specific location of the ruptured filter bag.

[0086] Furthermore, during the bag design and manufacturing stages, a unique row and column number is assigned to each potential point of damage. The surface of the filter bag is divided into multiple small openings, each given a unique row and column number, such as... Figure 2 The diagram shown depicts the distribution of filter bag openings. The grid is 14 rows and 14 columns (14x14 grid), with each circle representing a grid cell. Rows are numbered 1 to 14 from top to bottom, and columns are numbered 1 to 14 from left to right. Figure 3 The image shows the detection results of a single hole in a bag. If the hole is located in row 7, column 5, then the coordinates of that location can be represented as (i7, j5) and can be determined by... Figure 4 The output shows a single-hole broken filter bag array. The provided array is a 14x14 two-dimensional array used to represent the state of the filter bag layout in a hypothetical clean air chamber. Each element represents the state of the filter bag at a specific location: 0 indicates normal, while 1 indicates that the filter bag at that location is broken. This illustrates a scenario where only one filter bag is broken.

[0087] Furthermore, such as Figure 5 The image shows the results of multiple bag-breaking detections. If the location of the bag-breaking hole is in row 7, column 5 and row 11, column 11, then the coordinates of this location can be represented as (i7, j5) and (i11, j11), and can be obtained by... Figure 6 The output shows an array of multiple hole-damaged filter bags. These numbers can be directly attached to the surface of the filter bags through printing, labels, or other means. When a filter bag is damaged during transportation or storage, the specific location of the damage can be quickly determined based on the row and column numbers near the hole.

[0088] Step S40: Store the data based on the row and column numbers of the bag opening and send an alarm.

[0089] It's important to note that IoT technology, combined with sensors and smart devices, is used to monitor the filter bags' status in real time. Once an anomaly is detected, such as a damaged filter bag opening, the row and column number of the opening is automatically recorded and the information is uploaded to the cloud server. Simultaneously, an alarm mechanism is triggered, sending notifications to relevant personnel, including detailed information such as the damaged bag's ID, specific location, and time. This not only improves response speed but also facilitates subsequent tracking and handling.

[0090] This embodiment provides an image-based method for identifying and locating bag ruptures in a baghouse dust collector. It acquires image information of the bag opening, preprocesses this image to generate target image information, and inputs this target image information into a pre-defined deep learning model for rupture location. This model identifies the row and column numbers of the rupture openings, stores these numbers, and sends an alarm. By acquiring images of the bag openings, preprocessing and identifying ruptures using a deep learning model, and directly outputting the row and column numbers of the rupture openings, this method achieves rapid location and alarm activation, improving detection efficiency and accuracy, and facilitating real-time monitoring and maintenance.

[0091] Based on the first embodiment of this application, in the second embodiment of this application, the content that is the same as or similar to that in Embodiment 1 above can be referred to the above description, and will not be repeated hereafter. Based on this, please refer to... Figure 7 Before step S30 of the image-based bag filter bag breakage identification and positioning method, steps S201 to S202 are also included:

[0092] Step S201: Obtain sample image information of the bag opening and establish an initial deep learning model for bag breaking localization.

[0093] It should be noted that the process of acquiring sample images includes photographing filter bags under different working conditions, covering normal conditions and varying degrees of damage. To ensure data diversity and comprehensiveness, images should be acquired from multiple angles and under different lighting conditions, simulating various situations in the actual operating environment as much as possible. Furthermore, variations in filter bag material, size, and installation location should be considered to increase the representativeness of the samples. The acquired images should contain a clear bag opening area for subsequent processing and analysis. An initial deep learning model for bag breakage localization is constructed based on the Coordinate Attention (CA) mechanism. This model extracts image features through a multi-layer convolutional neural network and utilizes the CA module to enhance the feature representation sensitive to direction and position. Specifically, the model structure can include an input layer, several convolutional layers containing CA modules, pooling layers, fully connected layers, and an output layer. The input layer receives a preprocessed standard-sized image; the intermediate layers progressively extract and enhance image features through convolutional operations; finally, the fully connected layer combines the sigmoid function to predict the specific row and column numbers of the hole.

[0094] Step S202: Train the initial bag-breaking location deep learning model based on the bag opening sample image information to obtain the preset bag-breaking location deep learning model.

[0095] It should be noted that training the initial bag-breaking location deep learning model involves first initializing the model parameters, acquiring sample images of the bag opening and their actual coordinates, normalizing the sample images to obtain processed image information, inputting this processed image information into the deep learning model to obtain predicted coordinate information, calculating the error between the predicted and actual coordinates based on the loss function, obtaining the learning rate, and finally iteratively updating the model parameters based on the learning rate using an optimization algorithm until the maximum number of iterations is reached or the error value calculated by the loss function converges to a preset threshold, thus obtaining the target bag-breaking location deep learning model. Specifically, firstly, a large number of representative bag opening sample images need to be collected, covering normal conditions and different degrees of damage. Each image needs to be strictly labeled, clearly marking the location of all damage (row and column numbers). This process is crucial because these labels will serve as supervisory signals to guide the model's learning process. In addition, image preprocessing is required, including but not limited to cropping, scaling to a uniform size, and pixel value normalization, to improve the model's training efficiency and stability. Next, the initial model is trained using the labeled sample images. During training, a batch input method is adopted, with a certain number of images and their corresponding labels input each time. The network weights are updated through the backpropagation algorithm, gradually reducing the error between the predicted results and the true labels. To prevent overfitting, a validation set is usually used to evaluate the model performance, and the model structure or hyperparameter settings are adjusted accordingly.

[0096] After obtaining the target bag breaking localization deep learning model, the process also includes: evaluating the target bag breaking localization deep learning model and obtaining the evaluation results; if the evaluation results do not meet the detection requirements, the target bag breaking localization deep learning model is retrained until the detection requirements are met.

[0097] This embodiment collects and labels sample images of bag openings, constructs an initial deep learning model using a coordinate attention mechanism, and obtains a preset model through training and optimization. This enables precise positioning of damaged filter bags, improves detection efficiency and accuracy, reduces manual intervention, and facilitates timely repair.

[0098] Based on the first embodiment of this application, in the third embodiment of this application, the content that is the same as or similar to that in the first embodiment described above can be referred to the above description, and will not be repeated hereafter. Based on this, please refer to... Figure 8 The image-based bag filter bag breakage identification and positioning method, step S30, further includes steps S301 to S302:

[0099] The aforementioned pre-defined bag-breaking localization deep learning model includes a first pre-defined module, a second pre-defined module, a third pre-defined module, a fourth pre-defined module, a fifth pre-defined module, and a sixth pre-defined module. The first and second pre-defined modules each include two convolutional layers, one max pooling layer, and a coordinate attention mechanism sub-module. The third pre-defined module includes three convolutional layers, one max pooling layer, and a coordinate attention mechanism sub-module. The fourth pre-defined module includes three convolutional layers and a coordinate attention mechanism sub-module. The fifth pre-defined module includes five convolutional layers and a coordinate attention mechanism sub-module. The sixth pre-defined module includes one fully connected layer, one activation function layer, and one reconstruction layer.

[0100] Step S301: The target image information is used to extract features and generate feature vectors by using the first preset module, the second preset module, the third preset module, the fourth preset module and the fifth preset module of the preset bag breaking localization deep learning model.

[0101] It's important to note that the target image is input into the first preset module. This module is primarily responsible for initial feature extraction and size reduction. Through two convolution operations, one max pooling operation, and a coordinate attention (CA) mechanism, the original image is transformed into a feature map with feature information but a reduced size. This processing not only quickly reduces computational complexity but also preserves crucial spatial location information, providing a solid foundation for subsequent modules. Next, the feature map enters the second preset module, which further refines feature extraction. During this process, two more convolution operations, one max pooling operation, and a CA operation are performed again. Compared to the first module, the feature map here has a lower resolution but more prominent details. This helps capture more specific morphological features of the filter bag opening, improving the accuracy of subsequent localization. The third preset module is responsible for in-depth feature mining. It includes three convolution operations, one max pooling operation, and a CA operation, further reducing the feature map size while enhancing detailed features. The goal of this stage is to ensure that the model can identify even tiny damaged areas, which is crucial for accurately determining the filter bag's condition. The fourth preset module focuses on reducing the number of channels in the feature map to lower computational costs and prevent overfitting. By employing three convolution operations and one CA operation, the model structure is simplified while maintaining high-precision feature representation. Finally, the fifth pre-defined module transforms the feature map into a low-dimensional feature vector through multiple convolution operations and one CA operation.

[0102] Through the coordinated operation of the five preset modules, highly representative and discriminative feature vectors can be effectively extracted from the target image. This not only improves the accuracy of bag breakage location but also greatly enhances the system's robustness, enabling stable operation even under complex and variable actual working conditions. It allows for timely detection and location of damaged filter bags, preventing excessive dust emissions caused by filter bag breakage.

[0103] Step S302: The feature vector is identified by the sixth preset module of the preset bag-breaking localization deep learning model to obtain the row and column numbers of the bag-breaking hole.

[0104] It should be noted that the sixth preset module performs the final identification and classification of the feature vectors extracted by the first five modules to determine the specific row and column numbers of the hole in the bag. This process is implemented through fully connected layers, Sigmoid function processing, and Reshape operation.

[0105] Further, step S302 also includes: mapping the feature vector through the fully connected layer of the sixth preset module to obtain a one-dimensional vector, and then passing the one-dimensional vector through the activation function layer of the sixth preset module to obtain the damage values ​​of multiple holes. When the damage value is a preset value, it is determined to be a bag breakage, and the corresponding one-dimensional vector of the bag breakage hole is output. Specifically, the fully connected layer maps the feature vector to a one-dimensional vector with a length of I×J through a series of weights and bias parameters, where I and J represent the number of rows and columns of the air vent holes, respectively. This process essentially decodes the state information of each potential hole from the extracted feature vector to form a predicted value for the state of each hole. Next, this one-dimensional vector will be processed by the activation function layer. Commonly used activation functions such as the Sigmoid function are used here to convert each element into a value between 0 and 1. These values ​​can be understood as the probability estimate of the corresponding hole breakage: values ​​close to 1 indicate that the hole is very likely to be broken, while values ​​close to 0 indicate that the hole is in a normal state. This probabilistic output method not only provides information about the location of the damage but also gives the confidence level of each judgment, which helps with subsequent analysis and decision-making. Specifically, assuming we have 14x14 holes, the output after the fully connected layer is a one-dimensional vector of 196 values. After processing with the Sigmoid function, if the value of the 7th row and 5th column (i.e., the 82nd element, assuming row-major order) is 0.95, it means that there is a high probability of damage at that location.

[0106] Furthermore, the one-dimensional vector is transformed into a two-dimensional vector through the reconstruction layer of the sixth preset module. This two-dimensional vector is of length I×J, where I represents the number of rows of the clean air chamber openings in the bag filter, and J represents the number of columns. The size of the I×J array corresponds to the row and column layout of the clean air chamber openings. The corresponding row and column numbers of the bag breakage openings are obtained from the two-dimensional vector. Specifically, the data processed by the Sigmoid function is entered into the reconstruction layer for dimensional transformation, rearranging it into a two-dimensional array of size I×J. This transformation not only makes the data structure match the actual physical layout but also facilitates the intuitive identification of specific damage locations. For example, in a 14x14 clean air chamber layout, if the result of the one-dimensional vector processed by the Sigmoid function contains two significant high-value points, located at the 82nd and 169th positions respectively (assuming row-major order), then the Reshape operation can transform this one-dimensional vector into a 14x14 two-dimensional array. In this new array, the 82nd element corresponds to the 7th row, 5th column (I7, J5), while the 169th element corresponds to the 14th row, 1st column (I14, J1). This makes it clear that the filter bags at these two positions are at risk of breakage.

[0107] This approach not only allows for quick and accurate location of damaged filter bags but also prioritizes maintenance tasks based on confidence levels, thereby improving maintenance efficiency and reducing unnecessary downtime. Furthermore, this method supports simultaneous multi-point detection, effectively addressing various damage scenarios under complex operating conditions, further enhancing the system's practicality and reliability.

[0108] This embodiment utilizes the first five modules of a pre-defined bag-breaking location deep learning model to extract target image features and generate feature vectors. The sixth module identifies these vectors to determine the row and column numbers of the bag-breaking openings, thereby improving the accuracy and efficiency of damage detection, effectively preventing excessive dust emissions, and ensuring stable equipment operation.

[0109] This application also provides an image-based bag filter bag breakage identification and positioning device, please refer to... Figure 9 The device includes:

[0110] The acquisition module 10 is used to acquire image information of the bag opening of the bag filter.

[0111] The processing module 20 is used to preprocess the bag opening image information to generate target image information.

[0112] The recognition module 30 is used to input the target image information into the preset bag-breaking positioning deep learning model for recognition, and obtain the row and column numbers of the bag-breaking hole.

[0113] Result module 40 is used to store and send alarms based on the row and column numbers of the bag openings.

[0114] The image-based bag filter bag breakage identification and positioning device provided in this application adopts the image-based bag filter bag breakage identification and positioning method in the above embodiments, which can solve the technical problem of how to directly obtain the location of the bag breakage opening. Compared with the prior art, the beneficial effects of the image-based bag filter bag breakage identification and positioning device provided in this application are the same as the beneficial effects of the image-based bag filter bag breakage identification and positioning method provided in the above embodiments, and other technical features in the image-based bag filter bag breakage identification and positioning device are the same as the features disclosed in the above embodiment methods, and will not be repeated here.

[0115] In one embodiment, the processing module 20 is further configured to acquire bag opening sample image information and establish an initial bag breaking location deep learning model; and train the initial bag breaking location deep learning model based on the bag opening sample image information to obtain a preset bag breaking location deep learning model.

[0116] In one embodiment, the processing module 20 is further configured to initialize the model parameters of the initial bag-breaking location deep learning model; acquire bag opening sample image information and actual coordinate information; normalize the bag opening sample image information to obtain processed image information; input the processed image information into the initial neural network model to obtain predicted coordinate information; calculate the error value between the predicted coordinate information and the actual coordinate information according to the loss function; acquire the learning rate; and iteratively update the model parameters according to the learning rate through an optimization algorithm until the maximum number of iterations is reached or the error value calculated by the loss function converges to a preset threshold, thereby obtaining the preset bag-breaking location deep learning model.

[0117] In one embodiment, the result module 40 is further configured to extract features from the target image information through the first preset module, the second preset module, the third preset module, the fourth preset module, and the fifth preset module of the preset bag-breaking positioning deep learning model, and generate feature vectors; and to identify the feature vectors through the sixth preset module of the preset bag-breaking positioning deep learning model to obtain the row and column numbers of the bag-breaking openings.

[0118] In one embodiment, the result module 40 is further configured to map the feature vector through the fully connected layer of the sixth preset module to obtain a one-dimensional vector; pass the one-dimensional vector through the activation function layer of the sixth preset module to obtain the damage values ​​of multiple holes; when the damage value is a preset value, it is determined to be a bag breakage, and the corresponding one-dimensional vector of the bag breakage hole is output; and the one-dimensional vector is converted through the reconstruction layer of the sixth preset module to obtain a two-dimensional vector.

[0119] In one embodiment, the result module 40 is further configured to adjust the resolution of the bag opening image to a preset size based on the bag opening image information to obtain an image with adjusted size; normalize the pixels of the image with adjusted size to a preset range to obtain processed image information; and use the processed image information as target image information.

[0120] This application provides an image-based bag filter bag breakage identification and positioning device, which includes: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to execute the image-based bag filter bag breakage identification and positioning method in the above embodiment 1.

[0121] The following is for reference. Figure 10 This document illustrates a structural schematic diagram of an image-based bag filter bag breakage identification and positioning device suitable for implementing embodiments of this application. The image-based bag filter bag breakage identification and positioning device in this application embodiment may include, but is not limited to, mobile terminals such as mobile phones, laptops, digital radio receivers, PDAs (Personal Digital Assistants), PADs (Portable Application Description), PMPs (Portable Media Players), and vehicle terminals (e.g., vehicle navigation terminals), as well as fixed terminals such as digital TVs and desktop computers. Figure 10 The image-based bag filter bag breakage identification and positioning device shown is merely an example and should not impose any limitations on the functionality and scope of use of the embodiments of this application.

[0122] like Figure 10The image-based bag filter bag breakage identification and positioning device may include a processing unit 1001 (e.g., a central processing unit, a graphics processing unit, etc.), which can perform various appropriate actions and processes according to a program stored in ROM (Read Only Memory) 1002 or a program loaded from storage device 1003 into RAM (Random Access Memory) 1004. RAM 1004 also stores various programs and data required for the operation of the image-based bag filter bag breakage identification and positioning device. The processing unit 1001, ROM 1002, and RAM 1004 are interconnected via bus 1005. Input / output (I / O) interface 1006 is also connected to the bus. Typically, the following systems can be connected to I / O interface 1006: input devices 1007 including, for example, touchscreens, touchpads, keyboards, mice, image sensors, microphones, accelerometers, gyroscopes, etc.; output devices 1008 including, for example, liquid crystal displays (LCDs), speakers, vibrators, etc.; storage devices 1003 including, for example, magnetic tapes, hard disks, etc.; and communication devices 1009. Communication device 1009 allows the image-based bag filter bag breakage identification and positioning device to exchange data with other devices wirelessly or via wired communication. Although the figure shows an image-based bag filter bag breakage identification and positioning device with various systems, it should be understood that it is not required to implement or possess all the systems shown. More or fewer systems can be implemented alternatively.

[0123] Specifically, according to the embodiments disclosed in this application, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, embodiments disclosed in this application include a computer program product comprising a computer program carried on a computer-readable medium, the computer program containing program code for performing the methods described in the flowcharts. In such embodiments, the computer program can be downloaded and installed from a network via a communication device, or installed from storage device 1003, or installed from ROM 1002. When the computer program is executed by processing device 1001, it performs the functions defined in the methods of the embodiments disclosed in this application.

[0124] The image-based bag filter bag breakage identification and positioning device provided in this application, employing the image-based bag filter bag breakage identification and positioning method described in the above embodiments, solves the technical problem of how to directly obtain the location of the bag breakage opening. Compared with the prior art, the beneficial effects of the image-based bag filter bag breakage identification and positioning device provided in this application are the same as those of the image-based bag filter bag breakage identification and positioning method provided in the above embodiments, and other technical features of this image-based bag filter bag breakage identification and positioning device are the same as those disclosed in the previous embodiment method, and will not be repeated here.

[0125] It should be understood that the various parts disclosed in this application can be implemented using hardware, software, firmware, or a combination thereof. In the description of the above embodiments, specific features, structures, materials, or characteristics can be combined in any suitable manner in one or more embodiments or examples.

[0126] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

[0127] This application provides a computer-readable medium having computer-readable program instructions (i.e., a computer program) stored thereon that perform calculations to obtain computer-readable program instructions for executing the image-based bag filter bag breakage identification and positioning method in the above embodiments.

[0128] The computer-readable medium provided in this application may be, for example, a USB flash drive, but is not limited to, electrical, magnetic, optical, electromagnetic, infrared, or semiconductor systems, devices, or any combination thereof. More specific examples of computer-readable media may include, but are not limited to: electrical connections having one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fibers, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof. Calculations performed in this embodiment show that the computer-readable medium can be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, system, or device. The program code contained on the computer-readable medium can be transmitted using any suitable medium, including but not limited to: wires, optical cables, RF (Radio Frequency), etc., or any suitable combination thereof.

[0129] The aforementioned computer-readable medium may be included in an image-based bag filter bag breakage identification and positioning device; or it may exist independently and not assembled into an image-based bag filter bag breakage identification and positioning device.

[0130] The aforementioned computer-readable medium carries one or more programs that, when executed by the image-based bag filter bag breakage identification and positioning device, enable the image-based bag filter bag breakage identification and positioning device to write computer program code for performing the operations of this application in one or more programming languages ​​or a combination thereof. These programming languages ​​include object-oriented programming languages—such as Java, Smalltalk, and C++—and conventional procedural programming languages—such as the "C" language or similar programming languages. The program code can be executed entirely on the user's computer, partially on the user's computer, as a standalone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving remote computers, the remote computer can be connected to the user's computer via any type of network—including a local area network (LAN) or a wide area network (WAN)—or can be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0131] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various embodiments of this application. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions indicated in the blocks may occur in a different order than those indicated in the drawings. For example, two consecutively indicated blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, can be implemented using a dedicated hardware-based system that performs the specified function or operation, or using a combination of dedicated hardware and computer instructions.

[0132] The modules described in the embodiments of this application can be implemented in software or hardware. The names of the modules do not necessarily limit the functionality of the unit itself.

[0133] The readable medium provided in this application is a computer-readable medium, which stores computer-readable program instructions (i.e., a computer program) for executing the above-described image-based bag filter bag breakage identification and positioning method, and can solve the technical problem of how to directly obtain the location of the bag breakage opening. Compared with the prior art, the beneficial effects of the computer-readable medium provided in this application are the same as the beneficial effects of the image-based bag filter bag breakage identification and positioning method provided in the above embodiments, and will not be repeated here.

[0134] This application also provides a computer program product, including a computer program that, when executed by a processor, implements the steps of the image-based bag filter bag breakage identification and positioning method described above.

[0135] The computer program product provided in this application can solve the technical problem of how to directly obtain the location of the bag rupture opening. Compared with the prior art, the beneficial effects of the computer program product provided in this application are the same as those of the image-based bag filter rupture identification and positioning method provided in the above embodiments, and will not be repeated here.

[0136] The above description is only a part of the embodiments of this application and does not limit the patent scope of this application. All equivalent structural transformations made under the technical concept of this application and using the contents of the specification and drawings of this application, or direct / indirect applications in other related technical fields, are included in the patent protection scope of this application.

Claims

1. An image-based method for identifying and locating bag breaks in a baghouse dust collector, characterized in that, The method includes: Obtain image information of the bag opening of the bag filter; The bag opening image information is preprocessed to generate target image information; The target image information is input into a preset bag-breaking localization deep learning model for recognition to obtain the row and column numbers of the bag-breaking opening. The preset bag-breaking localization deep learning model includes a first preset module, a second preset module, a third preset module, a fourth preset module, a fifth preset module, and a sixth preset module. The first preset module and the second preset module each include two convolutional layers, one max pooling layer, and a coordinate attention mechanism submodule. The third preset module includes three convolutional layers, one max pooling layer, and a coordinate attention mechanism submodule. The fourth preset module includes three convolutional layers and a coordinate attention mechanism submodule. The fifth preset module includes five convolutional layers and a coordinate attention mechanism submodule. The sixth preset module includes a fully connected layer, an activation function layer, and a reconstruction layer. The data is stored and an alarm is sent based on the row and column numbers of the bag openings; The step of inputting the target image information into a preset bag-breaking location deep learning model for identification to obtain the row and column numbers of the bag-breaking opening includes: The target image information is used to extract features and generate feature vectors by using the first, second, third, fourth and fifth preset modules of the preset bag-breaking localization deep learning model. The feature vector is identified by the sixth preset module of the preset bag-breaking localization deep learning model to obtain the row and column numbers of the bag-breaking hole; The step of identifying the feature vector and obtaining the row and column numbers of the bag opening through the sixth preset module of the preset bag-breaking localization deep learning model includes: The feature vector is mapped through the fully connected layer of the sixth preset module to obtain a one-dimensional vector; The one-dimensional vector is passed through the activation function layer of the sixth preset module to obtain the damage values ​​of multiple holes; When the damage value is a preset value, it is determined to be a bag breakage, and a one-dimensional vector of the corresponding bag breakage hole is output; The one-dimensional vector is transformed through the reconstruction layer of the sixth preset module to obtain a two-dimensional vector, wherein the two-dimensional vector is of length. The vector, This indicates the number of rows of openings in the clean air chamber of a bag filter. This indicates the column number of the clean air chamber openings in a bag filter. The size of the array corresponds to the row and column layout size of the air purification chamber opening; The corresponding row and column numbers of the bag-breaking holes are obtained from the two-dimensional vector.

2. The method as described in claim 1, characterized in that, Before inputting the target image information into a preset bag-breaking location deep learning model for identification and obtaining the row and column numbers of the bag-breaking opening, the process includes: Acquire sample image information of the bag opening and establish an initial deep learning model for bag breakage localization; The initial bag-breaking location deep learning model is trained based on the bag opening sample image information to obtain the preset bag-breaking location deep learning model.

3. The method as described in claim 2, characterized in that, The step of training the initial bag-breaking location deep learning model based on the bag opening sample image information to obtain the preset bag-breaking location deep learning model includes: Initialize the model parameters of the initial bag-breaking localization deep learning model; Obtain the sample image information and actual coordinate information of the bag opening; The bag opening sample image information is normalized to obtain the processed image information; The processed image information is input into the initial bag-breaking localization deep learning model to obtain predicted coordinate information; The error value between the predicted coordinate information and the actual coordinate information is calculated based on the loss function. Obtain the learning rate; The model parameters are iteratively updated using an optimization algorithm based on the learning rate until the maximum number of iterations is reached or the error value calculated by the loss function converges to a preset threshold, thus obtaining a preset bag-breaking localization deep learning model.

4. The method as described in claim 1, characterized in that, The step of preprocessing the bag opening image information to generate target image information includes: Based on the bag opening image information, the resolution of the bag opening image is adjusted to a preset size to obtain an image with adjusted size; The pixels of the resized image are normalized to a preset range to obtain the processed image information; The processed image information is used as the target image information.

5. An image-based bag filter dust collector bag breakage identification and positioning device, characterized in that, The device is applied to the image-based bag filter bag breakage identification and positioning method as described in any one of claims 1-4, and the device comprises: The acquisition module is used to acquire image information of the bag opening of the bag filter; The processing module is used to preprocess the bag opening image information to generate target image information; The identification module is used to input the target image information into a preset bag-breaking location deep learning model for identification, and obtain the row and column numbers of the bag-breaking holes; The results module is used to store data based on the row and column numbers of the bag openings and to send alarms.

6. An image-based bag filter dust collector bag breakage identification and positioning device, characterized in that, The device includes: a memory, a processor, and an image-based bag filter bag breakage identification and positioning program stored in the memory and running on the processor, the image-based bag filter bag breakage identification and positioning program being configured to implement the steps of the image-based bag filter bag breakage identification and positioning method as described in any one of claims 1-4.

7. A medium, characterized in that, The medium stores an image-based bag filter bag breakage identification and positioning program, which, when executed by a processor, implements the steps of the image-based bag filter bag breakage identification and positioning method as described in any one of claims 1-4.

Images