Blast furnace molten iron flow state detection method, device and equipment and storage medium

Abstract

The invention provides a blast furnace molten iron flow state detection method and device, equipment and a storage medium, and the method comprises the steps: obtaining a molten iron flow image of blast furnace tapping, carrying out foreground region recognition and image feature extraction, carrying out boundary extraction on the foreground region image, and obtaining an initial upper boundary and an initial lower boundary of a molten iron flow stream; performing curve fitting to obtain an upper boundary fitting curve and a lower boundary fitting curve, extracting boundary morphological features and time sequence dynamic features of the molten iron flow stream according to the initial upper boundary, the initial lower boundary, the upper boundary fitting curve and the lower boundary fitting curve, and performing feature fusion on the boundary morphological features and the time sequence dynamic features and the molten iron flow image features; state recognition is carried out based on the molten iron flow fusion features to obtain current state information; according to the method, the precision limitation of the recognition area is improved through foreground area recognition, interference of high-temperature radiation, smoke shielding and illumination fluctuation can be overcome by combining multi-dimensional analysis of boundary morphological characteristics and time sequence dynamic characteristics, and accurate sensing of the molten iron flow state is achieved.

Description

Technical Field

[0001] This application relates to the field of steel production testing, and in particular to a method, apparatus, equipment and storage medium for detecting the flow status of molten iron in a blast furnace. Background Technology

[0002] Blast furnace ironmaking is the core process of modern steel industry. The stable operation of its tapping process is directly related to the quality of the final product, production safety and energy efficiency. During the tapping process, high-temperature molten iron flows out of the taphole to form a continuous stream of molten iron. In actual operation, various abnormal states may occur, such as flow interruption, splashing, and deviation. These are usually important indicators of abnormal furnace conditions, taphole blockage or improper operating parameters. Therefore, in order to ensure the safe and stable operation of the tapping process, real-time and accurate detection of the molten iron flow status is crucial.

[0003] Currently, monitoring of the blast furnace tapping process mainly relies on manual observation based on experience or basic video monitoring methods. This is easily affected by the subjective influence of operational experience and is difficult to achieve continuous monitoring around the clock, resulting in a significant lag in the identification of abnormal states such as flow interruption and splashing. Although some technologies use image processing methods for detection, existing image detection methods are mostly based on fixed threshold segmentation strategies, which are poorly adaptable to complex environmental factors such as high-temperature radiation, smoke interference, and light fluctuations at the tapping site. This leads to discontinuous and inaccurate extraction of the molten iron flow boundary, making it impossible to accurately capture the actual morphological changes of the flow stream. Furthermore, traditional methods fail to effectively capture the dynamic evolution of the flow stream over time, limiting the accuracy and reliability of state identification and making it difficult to meet the production requirements for precise perception of the molten iron flow state. Summary of the Invention

[0004] The purpose of this application is to provide a method, apparatus, equipment, and storage medium for detecting the flow state of molten iron in a blast furnace, so as to solve the above-mentioned technical problems.

[0005] This application provides a method for detecting the state of molten iron flow in a blast furnace. The method includes: acquiring an image of the molten iron flow during blast furnace tapping, identifying the foreground region of the molten iron flow image to obtain a foreground region image, and extracting image features from the molten iron flow image to obtain molten iron flow image features; extracting the boundaries of the molten iron flow stream based on the foreground region image to obtain an initial upper boundary and an initial lower boundary of the molten iron flow stream, and performing curve fitting on the initial upper boundary and the initial lower boundary to obtain corresponding upper boundary fitting curves and lower boundary fitting curves; extracting boundary morphological features and temporal dynamic features of the molten iron flow stream based on the initial upper boundary, the initial lower boundary, the upper boundary fitting curve, and the lower boundary fitting curve, and fusing the molten iron flow image features with the boundary morphological features and temporal dynamic features to obtain molten iron flow fusion features; and performing state recognition based on the molten iron flow fusion features to obtain the current state information of the molten iron flow stream.

[0006] In one embodiment of this application, foreground region identification of the molten iron flow image to obtain a foreground region image includes: calculating the distance between each pixel to be processed in the molten iron flow image and each sample pixel in a corresponding preset background sample set, wherein the preset background sample set is constructed by collecting neighboring pixels of each pixel position in multiple consecutive frames of images, and the preset background sample set includes multiple sample pixels; determining the pixel category of the pixel to be processed based on the distance between the pixel to be processed and each sample pixel, wherein the pixel category includes foreground region pixels and background pixels; extracting the pixels to be processed that are classified as foreground region pixels to obtain an initial foreground image, and performing binarization processing on the initial foreground image to obtain a binarized foreground mask; detecting all connected regions in the binarized foreground mask, and selecting the connected region with the largest area to generate the foreground region image.

[0007] In one embodiment of this application, the boundary extraction of the molten iron flow stream based on the foreground region image to obtain the initial upper and lower boundaries of the molten iron flow stream includes: determining the boundary of the foreground region image as the initial contour curve of the active contour model, and constructing an energy function, the energy function including an external energy term and an internal energy term, the external energy term being used to drive the contour curve to move towards the molten iron flow boundary, and the internal energy term being used to constrain the continuity and smoothness of the contour curve; iteratively optimizing the initial contour curve with the goal of minimizing the energy function, and obtaining the initial upper and lower boundaries of the molten iron flow stream when the energy function is minimized.

[0008] In one embodiment of this application, curve fitting is performed on the initial upper boundary and the initial lower boundary to obtain the corresponding upper boundary fitting curve and lower boundary fitting curve, respectively. This includes: collecting multiple upper boundary discrete feature points from the initial upper boundary, collecting multiple lower boundary discrete feature points from the initial lower boundary, and assigning boundary weight coefficients to each upper boundary discrete feature point and each lower boundary discrete feature point. The boundary weight coefficients are inversely proportional to the Euclidean distance from the upper boundary discrete feature point or the lower boundary discrete feature point to the tap hole position; and performing parabolic trajectory fitting on the upper boundary discrete feature points and lower boundary discrete feature points with boundary weight coefficients based on a weighted least squares fitting algorithm to obtain the corresponding upper boundary fitting curve and lower boundary fitting curve.

[0009] In one embodiment of this application, extracting the boundary morphology features and temporal dynamic features of the molten iron stream based on the initial upper boundary, initial lower boundary, upper boundary fitting curve, and lower boundary fitting curve includes: determining the ideal jet area by calculating the area of ​​the region formed by the upper boundary fitting curve and the lower boundary fitting curve; calculating the stream divergence coefficient by integrating and averaging the ratio of the difference in the ordinates of the upper and lower boundary fitting curves to the jet width at the taphole; determining the stream boundary fluctuation coefficient by calculating the average deviation between the ordinates of the initial upper and lower boundaries and the corresponding fitting curves at the same abscissa; determining the boundary displacement change rate based on the change in the ordinate value per unit time of the initial upper and lower boundaries at the same abscissa; obtaining the boundary fluctuation amplitude by calculating the standard deviation of the ordinates at the same abscissa of the initial upper and lower boundaries; and determining the ideal jet area, stream divergence coefficient, and stream boundary fluctuation coefficient as boundary morphology features, and the boundary displacement change rate and boundary fluctuation amplitude as temporal dynamic features.

[0010] In one embodiment of this application, the process of extracting image features from the molten iron flow image to obtain molten iron flow image features includes: inputting the molten iron flow image into a pre-trained structural feature extraction network, extracting structural features from the molten iron flow image to obtain image structural features, and determining the image structural features as molten iron flow image features, wherein the molten iron flow image features include texture features, shape features, and contour space features.

[0011] In one embodiment of this application, the current state information of the molten iron flow stream is obtained by performing state identification based on the molten iron flow fusion characteristics. This includes: inputting the molten iron flow fusion characteristics into a preset state identification model, and performing time-series correlation analysis on the molten iron flow fusion characteristics based on the long short-term memory network in the preset state identification model to obtain the time-series correlation analysis results; determining the probability distribution of the molten iron flow stream in each state category based on the time-series correlation analysis results, and determining the state category with the highest probability distribution as the current state information of the molten iron flow stream, wherein the current state information includes normal, interrupted flow, splashing, and jittering.

[0012] This application embodiment also provides a blast furnace molten iron flow state detection device, which includes: an image acquisition and processing module, used to acquire an image of molten iron flow during blast furnace tapping, and to perform foreground region identification on the molten iron flow image to obtain a foreground region image, and to extract image features from the molten iron flow image to obtain molten iron flow image features; a boundary feature extraction module, used to extract the boundary of the molten iron flow stream based on the foreground region image to obtain the initial upper boundary and initial lower boundary of the molten iron flow stream, and to perform curve fitting on the initial upper boundary and initial lower boundary respectively to obtain the corresponding upper boundary fitting curve and lower boundary fitting curve; to extract the boundary morphological features and temporal dynamic features of the molten iron flow stream based on the initial upper boundary, initial lower boundary, upper boundary fitting curve and lower boundary fitting curve, and to fuse the molten iron flow image features with the boundary morphological features and temporal dynamic features to obtain molten iron flow fusion features; and a flow stream state recognition module, used to perform state recognition based on the molten iron flow fusion features to obtain the current state information of the molten iron flow stream.

[0013] This application also provides an electronic device, including: one or more processors; and a storage device for storing one or more programs, which, when executed by the one or more processors, enable the electronic device to implement the blast furnace molten iron flow state detection method as described in any of the above embodiments.

[0014] This application also provides a computer-readable storage medium storing computer-readable instructions, which, when executed by a computer's processor, cause the computer to perform the blast furnace molten iron flow state detection method as described in any of the above embodiments.

[0015] The beneficial effects of this application are as follows: This application provides a method, apparatus, equipment, and storage medium for detecting the state of molten iron flow in a blast furnace. It acquires images of the molten iron flow during blast furnace tapping, identifies the foreground region, and extracts image features. Based on the foreground region image, it extracts the boundary of the molten iron flow stream, obtaining the initial upper and lower boundaries. Curve fitting is then performed to obtain corresponding upper and lower boundary fitting curves. Based on the initial upper and lower boundaries, the upper and lower boundary fitting curves, the boundary morphology and temporal dynamic features of the molten iron flow stream are extracted and fused with the molten iron flow image features. The current state information is obtained by identifying the state based on the fused molten iron flow features. This application overcomes the interference of high-temperature radiation, smoke and dust obstruction, and light fluctuations on boundary extraction through multi-dimensional analysis of foreground region identification, boundary morphology features, and temporal dynamic features. This achieves accurate perception of the molten iron flow state, improves the accuracy and robustness of molten iron flow boundary identification, and enhances the reliability of molten iron flow stream state identification.

[0016] It should be understood that the above general description and the following detailed description are exemplary and explanatory only, and do not limit this application. Attached Figure Description

[0017] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this application and, together with the description, serve to explain the principles of this application. It is obvious that the drawings described below are merely some embodiments of this application, and those skilled in the art can obtain other drawings based on these drawings without any inventive effort. In the drawings: Figure 1 This is a schematic diagram illustrating an exemplary system architecture as shown in an exemplary embodiment of this application; Figure 2 An exemplary embodiment of this application illustrates a flowchart of a method for detecting the flow state of molten iron in a blast furnace; Figure 3 This is a schematic diagram of a blast furnace molten iron flow state detection device, as shown in an exemplary embodiment of this application; Figure 4 This is a schematic diagram of the structure of a computer system for an electronic device, as illustrated in an exemplary embodiment of this application. Detailed Implementation

[0018] The embodiments of this application will be described below with reference to the accompanying drawings and specific examples. Those skilled in the art can easily understand other advantages and effects of this application from the content disclosed in this specification. This application can also be implemented or applied through other different specific embodiments, and various details in this specification can also be modified or changed based on different viewpoints and applications without departing from the spirit of this application. It should be understood that the preferred embodiments are only for illustrating this application and are not intended to limit the scope of protection of this application.

[0019] It should be noted that the illustrations provided in the following embodiments are only schematic representations of the basic concept of this application. Therefore, the illustrations only show the components related to this application and are not drawn according to the number, shape and size of the components in actual implementation. In actual implementation, the form, quantity and proportion of each component can be arbitrarily changed, and the layout of the components may also be more complex.

[0020] In the following description, numerous details are explored to provide a more thorough explanation of embodiments of the present application. However, it will be apparent to those skilled in the art that embodiments of the present application may be practiced without these specific details. In other embodiments, well-known structures and devices are shown in block diagram form rather than in detail to avoid obscuring embodiments of the present application.

[0021] The term "and / or" used in this application describes the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A existing alone, A and B existing simultaneously, or B existing alone. The character " / " generally indicates that the related objects before and after it are in an "or" relationship.

[0022] During the implementation of related technologies, the extraction of molten iron flow boundaries is prone to discontinuity and inaccuracy due to the combined effects of complex environmental factors such as high-temperature radiation, smoke and dust interference, and light fluctuations. At the same time, the accuracy of capturing changes in molten iron flow morphology is insufficient, and the temporal dynamic characteristics cannot be effectively characterized. This significantly restricts the reliability and accuracy of molten iron flow state identification. In particular, the fracture problem in boundary extraction directly weakens the ability to represent the flow morphology, while the lack of temporal evolution rules further limits the real-time response capability of state identification. This has an adverse impact on the safety and stability of the iron tapping process and the continuity of the production process.

[0023] Based on the aforementioned technical problems and characteristics, this application proposes the following technical solution: By acquiring images of molten iron flow during blast furnace tapping and performing foreground region identification and image feature extraction, boundary extraction is performed on the foreground region image to obtain the initial upper and lower boundaries of the molten iron flow, and curve fitting is performed on each boundary to obtain the upper boundary fitting curve and the lower boundary fitting curve. Based on the initial upper boundary, initial lower boundary, upper boundary fitting curve, and lower boundary fitting curve, the boundary morphological features and temporal dynamic features of the molten iron flow are extracted and fused with the features of the molten iron flow image to obtain the current state information based on the fused features of the molten iron flow. By integrating foreground region identification, boundary extraction and fitting, and multi-dimensional feature extraction and fusion, the influence of high-temperature radiation and dust interference on boundary extraction is effectively overcome, ensuring the continuity and accuracy of boundary extraction. At the same time, by extracting temporal dynamic features and fusing image features, the evolution law of the flow in the time dimension is fully captured, thereby significantly improving the accuracy and reliability of molten iron flow state identification and providing technical support for the safe and stable operation of the blast furnace tapping process.

[0024] Please see Figure 1 , Figure 1 This is a schematic diagram illustrating an exemplary system architecture as shown in an exemplary embodiment of this application.

[0025] Reference Figure 1 As shown, the system architecture may include an image acquisition device 110 and a computer device 120. The computer device 120 acquires images of molten iron flow during blast furnace tapping via the image acquisition device 110, performs foreground region identification on the molten iron flow images to obtain foreground region images, extracts image features from the molten iron flow images to obtain molten iron flow image features, extracts the boundaries of the molten iron flow stream based on the foreground region images to obtain the initial upper and lower boundaries of the molten iron flow stream, and performs curve fitting on the initial upper and lower boundaries to obtain corresponding upper and lower boundary fitting curves. Based on the initial upper and lower boundaries, the upper and lower boundary fitting curves, the boundary morphology features and temporal dynamic features of the molten iron flow stream are extracted, and the molten iron flow image features are fused with the boundary morphology features and temporal dynamic features to obtain molten iron flow fusion features. Based on the molten iron flow fusion features, state recognition is performed to obtain the current state information of the molten iron flow stream. The aforementioned image acquisition device 110 is used to acquire and temporarily store images of molten iron flow during the tapping process, and transmit them to the computer device 120. The image acquisition device 110 includes, but is not limited to, industrial cameras, high-speed cameras, etc., and may further include an infrared thermal imager or be equipped with infrared thermal imaging function. The aforementioned computer device 120 refers to a computing power support terminal device used to carry out the program implementation environment for the blast furnace molten iron flow state detection method, including but not limited to microcomputers, tablet devices, industrial control computers, server clusters, and cloud servers.

[0026] This technical solution can be further discussed and analyzed in multiple steps; for specific steps, please refer to [link / reference needed]. Figure 2 , Figure 2 This is a flowchart illustrating an exemplary embodiment of a blast furnace molten iron flow state detection method. This method can be executed in implementation environments supported by various operating systems, and no specific limitation is made to the implementation environment herein. (Refer to...) Figure 2 As shown, the flowchart of the blast furnace molten iron flow state detection method includes at least steps S210 to S240, which are described in detail below: In step S210, an image of molten iron flow from the blast furnace is acquired, and foreground region recognition is performed on the molten iron flow image to obtain a foreground region image. Additionally, image feature extraction is performed on the molten iron flow image to obtain molten iron flow image features.

[0027] In one embodiment of this application, the above-mentioned image of molten iron flow during blast furnace tapping is obtained by capturing short-wave infrared images of the molten iron flow using a high-speed camera. In some specific implementations, 0.75-1.1 is used. Imaging of molten iron flow was performed in the spectral bands to avoid imaging glare caused by complex lighting conditions and ambient light reflection. Camera calibration was performed to address the distortion of the molten iron flow caused by the camera's shooting angle, using a pinhole model for derivation. Equation (1) in, For the camera intrinsic parameter matrix, It is an extrinsic parameter matrix. Here, s represents the image pixel coordinates, and s represents the actual coordinates. l Let (X, Y, Z) represent the brightness, and (X, Y, Z) represent the actual three-dimensional coordinates of the blast furnace taphole. Furthermore, to improve the visual perception consistency between the taphole and the molten iron flow in the image, gamma correction is also performed on the image.

[0028] In one embodiment of this application, the distance between each pixel to be processed in the molten iron flow image and each sample pixel in the corresponding preset background sample set is calculated. The preset background sample set is constructed by collecting the neighboring pixels of each pixel position in multiple consecutive frames of images. The preset background sample set includes multiple sample pixels. Based on the distance between the pixel to be processed and each sample pixel, the pixel category of the pixel to be processed is determined. The pixel category includes foreground region pixels and background pixels. The pixels to be processed that are classified as foreground region pixels are extracted to obtain an initial foreground image. The initial foreground image is then binarized to obtain a binarized foreground mask. All connected regions are detected in the binarized foreground mask, and the connected region with the largest area is selected to generate the foreground region image.

[0029] The preset background sample set can be constructed by maintaining historical frame pixel data using a sliding window mechanism, which enhances the background model's resistance to transient interference through the continuity of the time dimension; pixel category determination is an adaptive classification process based on pixel-level distance measurement, including ensuring accurate distinction between foreground and background through dynamic threshold determination strategy or probability model distribution calculation; binarized foreground mask refers to converting foreground pixel information into an intermediate representation of a binary image, simplifying the image structure and highlighting the foreground contour based on an adaptive threshold segmentation algorithm; the above-mentioned connected region detection refers to identifying continuous pixel regions in the image through region growing algorithms or morphological operations.

[0030] In one embodiment of this application, a preset background sample set is dynamically constructed, enabling the background model to respond to changes in neighboring pixels at each pixel location in multiple consecutive frames of images. This quantifies the similarity between the pixel to be processed and the background. Based on this distance metric, pixel category determination is performed. Subsequently, foreground pixels are extracted and binarized to form a clear foreground mask structure. Finally, connected component analysis is used to focus on the connected component with the largest area, ensuring that the foreground region image fully reflects the main shape of the molten iron flow. From pixel-level distance calculation to region-level shape extraction, the continuity and accuracy issues of foreground region extraction in complex environments are solved.

[0031] In the specific implementation of this application, since only the molten iron flow is a dynamic moving object in the tapping field during the tapping process, while the background wall, main iron trough, and other interfering factors are static features, the frame difference method for visual background extraction is used for foreground region recognition. First, the background model is initialized, and each pixel in the image is processed. of Samples are taken from adjacent frames to form a preset background sample set. For example, the pixels to be processed With any of the preset background sample sets The distance between each sample pixel is less than the pixel distance threshold. If the condition is met, the pixel to be processed is determined to be background; otherwise, it is considered foreground. This can be represented as: Equation (2) in, The pixel to be processed in the image is represented by 1, where 1 indicates that the pixel is in the foreground region and 0 indicates that the pixel is in the background region. The function is used to calculate the distance between the pixel to be processed and the sample pixels in a preset background sample set. These are sample pixels. If the current pixel is determined to be background, then... The probability is to write the current value into the preset background sample set. The original sample is then randomly replaced. Simultaneously, a neighboring pixel position is randomly selected from the neighborhood. The location of the neighboring pixels is also written into the preset background sample set. In this process, we can enhance our ability to perceive spatial diffusion.

[0032] In one embodiment of this application, an image of molten iron flow is input into a pre-trained structural feature extraction network to extract structural features from the image of molten iron flow, thereby obtaining image structural features, and the image structural features are determined as molten iron flow image features, which include texture features, shape features, and contour space features.

[0033] Specifically, the pre-trained structural feature extraction network can be trained using the MobileNetV2 convolutional neural network model. The aforementioned structural features are used to reflect the intrinsic properties of geometric shape and topological relationship, and can capture the physical characteristics of the molten iron surface. Specifically, the image structural features can be represented as multi-dimensional feature vectors or feature maps, where texture features are the subtle fluctuations and particle distribution patterns on the molten iron surface, shape features are the overall geometric shape of the flow stream, and contour space features are the spatial distribution characteristics of the boundary.

[0034] In step S220, the boundary of the molten iron stream is extracted based on the foreground region image to obtain the initial upper boundary and initial lower boundary of the molten iron stream. Curve fitting is then performed on the initial upper boundary and initial lower boundary to obtain the corresponding upper boundary fitting curve and lower boundary fitting curve.

[0035] In one embodiment of this application, the boundary of the foreground region image is determined as the initial contour curve of the active contour model, and an energy function is constructed. The energy function includes an external energy term and an internal energy term. The external energy term is used to drive the contour curve to move towards the boundary of the molten iron flow, and the internal energy term is used to constrain the continuity and smoothness of the contour curve. The initial contour curve is iteratively optimized with the goal of minimizing the energy function, and the initial upper boundary and initial lower boundary of the molten iron flow are obtained when the energy function is minimized.

[0036] Among them, the active contour model is a boundary detection mechanism that dynamically adjusts the contour position based on image features. It can be calculated based on the Snake model. The energy function is used to characterize the degree of matching between the contour and image features. It can be expressed based on the gradient magnitude as the input source of the external energy term and the contour curvature as the constraint factor of the internal energy term.

[0037] In one embodiment of this application, the boundary of the foreground region image is used as the initial contour curve. This fully utilizes the characteristic that the foreground region image has already removed background noise, allowing the contour starting point to focus on the real molten iron flow region, avoiding convergence deviation caused by starting from a noisy region. Secondly, an energy function containing external and internal energy terms is constructed. The external energy term drives the contour to move towards the molten iron flow boundary based on image gradient information, dynamically responding to boundary blurring changes caused by high-temperature radiation, and guiding the contour to accurately fit the actual boundary. The internal energy term effectively suppresses boundary breakage and jaggedness caused by illumination fluctuations by constraining the continuity and smoothness of the contour, ensuring the integrity of boundary extraction. Finally, the initial contour curve is iteratively optimized to minimize the energy function. By gradually adjusting the contour position to balance the external driving force and the internal constraint force, the contour dynamically adapts to changes in the flow morphology in complex environments such as smoke and dust interference. The initial upper and lower boundaries are obtained when the energy function is minimized, providing a high-precision foundation for subsequent boundary morphology feature extraction.

[0038] In a specific embodiment of this application, an active contour model is used for boundary extraction. The contour is first initialized. The boundary of the foreground region of the molten iron flow is set. The energy function of the active contour consists of external energy and internal energy, and the goal is to minimize the total energy by optimizing the contour. The external energy, defined based on the image gradient, reflects the edge strength and can be characterized as: Equation (3) The internal energy, determined by the smoothness of the contour, is used to limit the deformation of the contour and can be characterized as follows: Equation (4) in, It is the direction of the normal. It is the profile curvature. These are the weighting coefficients. The profile is optimized using a variational method. To obtain the minimum total energy and the upper and lower boundaries of the molten iron flow.

[0039] In one embodiment of this application, multiple discrete feature points of the upper boundary are collected from the initial upper boundary, and multiple discrete feature points of the lower boundary are collected from the initial lower boundary. Boundary weight coefficients are assigned to each discrete feature point of the upper boundary and each discrete feature point of the lower boundary. The boundary weight coefficients are inversely proportional to the Euclidean distance from the discrete feature point of the upper boundary or the discrete feature point of the lower boundary to the iron outlet position. Based on the weighted least squares fitting algorithm, parabolic trajectories are fitted to the discrete feature points of the upper boundary and the discrete feature points of the lower boundary with boundary weight coefficients to obtain the corresponding upper boundary fitting curve and lower boundary fitting curve.

[0040] Among them, the boundary weight coefficients are dynamically adjusted according to the spatial location of discrete feature points, and can be adjusted based on the continuous function relationship constructed by the inverse proportional relationship of the Euclidean distance from the discrete feature points to the iron outlet; the above-mentioned weighted least squares fitting algorithm is implemented by iteratively calculating and minimizing the weighted residual sum of squares; the parabolic trajectory fitting is based on polynomial regression analysis or least squares parameters to fit the discrete point set into a quadratic curve model.

[0041] Specifically, this application first extracts discrete feature points from the initial boundary as basic data input, and then dynamically assigns boundary weight coefficients based on the Euclidean distance from the point to the tap hole, so that the stable region near the tap hole receives higher weights, while the region far away receives lower weights due to the reduced reliability caused by environmental interference. On this basis, the weighted least squares fitting algorithm is used to integrate the weight coefficients into the parabolic trajectory fitting process, so that the generation of the fitting curve depends first on the boundary information of the high reliability region, avoiding excessive distortion of the curve shape by the far-end anomaly points. The final upper boundary fitting curve and lower boundary fitting curve not only maintain the smoothness of the boundary description, but also conform to the physical motion characteristics of the molten iron flow in the gravitational field.

[0042] In a specific embodiment of this application, the feature point sets of the upper and lower boundaries of the molten iron flow are obtained respectively using the active contour method. , ,in , During the jetting process, the jet stream is subjected to the combined effects of gravity and air resistance, forming a parabolic jet trajectory. Furthermore, the further the jet stream is from the taphole, the greater the disturbance and the higher the randomness of its motion. Therefore, a distance-weighted partial least squares boundary fitting method is used for curve fitting. During the fitting process, the weight of each boundary pixel is the coordinate of the point to be fitted. To the initial position of the taphole The reciprocal of the Euclidean distance can be represented as: Equation (5) The weighted sum of squared residuals minimized by the molten iron flow boundary curve can be characterized as follows: Equation (6) in, The equation for the parabolic trajectory is given, and the fitted curve at the upper boundary after fitting is as follows: Lower boundary fitting curve .

[0043] In step S230, based on the initial upper boundary, initial lower boundary, upper boundary fitting curve, and lower boundary fitting curve, the boundary morphology features and temporal dynamic features of the molten iron flow are extracted, and the molten iron flow image features are fused with the boundary morphology features and temporal dynamic features to obtain the molten iron flow fused features.

[0044] In one embodiment of this application, the ideal jet area is determined by calculating the area of ​​the region formed by the upper boundary fitting curve and the lower boundary fitting curve. It can be characterized as: Equation (7) The jet divergence coefficient is obtained by integrating and averaging the difference in the ordinates of the fitted curves at the upper and lower boundaries with the ratio of the jet width at the taphole. It can be characterized as: Equation (8) in, The lateral distance of the jet stream.

[0045] The flow stream boundary fluctuation coefficient is determined by calculating the average deviation between the initial upper and lower boundaries on the same horizontal axis and the corresponding fitted curve on the vertical axis. It can be characterized as follows: Equation (9) in, The boundary fluctuation coefficient of the flow stream. and They represent The ordinate values ​​of the initial upper and lower boundaries of the flow stream.

[0046] The rate of change of boundary displacement is determined by the change of the ordinate value per unit time when the initial upper and lower boundaries have the same abscissa, and can be characterized as follows:

[0047]

[0048] Equation (10) in, The rate of change of displacement at the upper boundary. The rate of change of the lower boundary displacement. The rate of change of the width displacement of the flow stream. and The x-coordinates are respectively Located in time The ordinates of the initial upper and lower boundaries at time points.

[0049] Based on the initial upper and lower boundaries, the standard deviation of the ordinate is calculated when the x-axis is the same, and the boundary fluctuation amplitude is obtained, which can be characterized as:

[0050]

[0051]

[0052] Equation (11) in, This represents the upper boundary fluctuation range; This represents the lower boundary fluctuation range; This represents the maximum instantaneous width of the stream. This represents the minimum instantaneous width of the stream.

[0053] The ideal jet area, jet divergence coefficient, and jet boundary fluctuation coefficient are defined as boundary morphology characteristics, and the boundary displacement change rate and boundary fluctuation amplitude are defined as time-series dynamic characteristics.

[0054] Among them, the ideal jet area is used to describe the characteristics of the taphole diameter and initial jet velocity in the current jet process. As the tapping process proceeds, the jet area of ​​the molten iron shows an increasing trend. The divergence coefficient describes the width change of the jet stream from the initial jet process at the taphole to the end, which can reflect the slag-iron mixing ratio of the molten iron stream and the situation of gas mixing into the stream stream. The stream boundary fluctuation coefficient is used to evaluate the stability of the stream boundary and filter high-frequency noise interference. The boundary displacement change rate describes the temporal fluctuation of the molten iron stream boundary morphology. The boundary fluctuation amplitude describes the temporal dynamic characteristics of the molten iron stream, such as instantaneous interruption and splashing.

[0055] In one embodiment of this application, the smoothness of the upper and lower boundary fitting curves is used to replace the original boundary for calculating key spatial parameters, effectively suppressing local noise caused by high-temperature radiation and smoke interference. Simultaneously, time-series analysis is performed on the initial upper and lower boundaries at the same abscissa position, ensuring spatial consistency and avoiding dynamic feature distortion caused by jet position shifts. The calculation of the ideal jet area utilizes the continuity of the fitting curve to provide a stable morphological benchmark. The jet divergence coefficient is normalized to eliminate the influence of operating condition differences. The jet boundary fluctuation coefficient quantifies the degree of fluctuation by combining the deviation between the original boundary and the fitting curve. The boundary displacement change rate and boundary fluctuation amplitude capture temporal dynamic characteristics from the perspectives of change rate and dispersion, respectively. This comprehensive approach enhances the anti-interference capability of morphological features in the spatial dimension and improves the real-time representation accuracy of dynamic features in the temporal dimension, providing highly robust feature inputs for state recognition.

[0056] In one embodiment of this application, feature fusion includes integrating molten iron flow image features with boundary-related features, and feature fusion operations can be performed based on feature vector concatenation or linear combination.

[0057] In step S240, state identification is performed based on the fusion characteristics of the molten iron flow to obtain the current state information of the molten iron flow stream.

[0058] In one embodiment of this application, the fusion characteristics of molten iron flow are input into a preset state recognition model, and a time-series correlation analysis is performed on the fusion characteristics of molten iron flow based on the long short-term memory network in the preset state recognition model to obtain the time-series correlation analysis results. Based on the time-series correlation analysis results, the probability distribution of the molten iron flow stream in each state category is determined, and the state category with the highest probability distribution is determined as the current state information of the molten iron flow stream. The current state information includes normal, interrupted flow, splashing, and jittering.

[0059] The preset state recognition model is a machine learning model architecture used to classify the state of molten iron flow. Its long short-term memory network is a variant of the recurrent neural network used to process sequential data. It includes unit structures such as input gate, forget gate, output gate, candidate cell state, cell state update, and hidden state output, which can effectively capture the feature evolution pattern over a long period of time. The above time-series correlation analysis results reflect the changes of molten iron flow characteristics over time, which can be realized based on the hidden state representation of the feature vector sequence. The probability distribution of the state category refers to the quantitative representation of the possibility of each state, which can be realized by using the normalized vector output by the softmax function. Its purpose is to provide a confidence assessment of state recognition.

[0060] In some other specific embodiments of this application, the weight parameters of the network can also be automatically updated based on the Adam optimizer.

[0061] In one embodiment of this application, by inputting the fusion features of molten iron flow into a preset state recognition model, the Long Short-Term Memory Network can serialize the fusion features at consecutive time points, use its gating mechanism to filter key historical information and suppress irrelevant noise, and establish a dynamic correlation model of molten iron flow features in the time dimension. Furthermore, the time-series correlation analysis results are transformed into probability distributions of each state category, and the final state category is determined by comparing the probability values. By utilizing the continuous characteristics of the evolution of the molten iron flow state, the system can identify the gradual change process of abnormal states such as flow interruption or splashing, avoiding the risk of misjudgment caused by relying solely on single-frame image feature analysis.

[0062] This application provides a method, apparatus, equipment, and storage medium for detecting the state of molten iron flow in a blast furnace. It acquires images of the molten iron flow during blast furnace tapping, identifies the foreground region, and extracts image features. Based on the foreground region image, it extracts the boundary of the molten iron flow stream, obtaining its initial upper and lower boundaries. Curve fitting is then performed on these boundaries to obtain corresponding upper and lower boundary fitting curves. Based on the initial upper and lower boundaries, the upper and lower boundary fitting curves, the boundary morphology and temporal dynamic features of the molten iron flow stream are extracted and fused with the molten iron flow image features. The current state information is obtained through state recognition based on these fused features. This application overcomes the interference of high-temperature radiation, smoke and dust obstruction, and light fluctuations on boundary extraction by integrating foreground region identification, boundary morphology features, and temporal dynamic features through multi-dimensional analysis. This achieves accurate perception of the molten iron flow state, improves the accuracy and robustness of molten iron flow boundary recognition, and enhances the reliability of molten iron flow stream state recognition.

[0063] The following describes an embodiment of the apparatus described in this application, which can be used to execute the blast furnace molten iron flow state detection method described above in this application. For details not disclosed in the apparatus embodiments of this application, please refer to the embodiments of the blast furnace molten iron flow state detection method described above in this application.

[0064] Figure 3 This is a schematic diagram illustrating an exemplary embodiment of a blast furnace molten iron flow state detection device. This device can be applied to... Figure 1 The method described is implemented in a device that has the necessary conditions for execution. This embodiment does not impose specific limitations on the devices to which the device is applicable.

[0065] like Figure 3 As shown, the exemplary blast furnace molten iron flow state detection device includes: an image acquisition and processing module 301, a boundary feature extraction module 302, and a flow state recognition module 303.

[0066] The image acquisition and processing module 301 is used to acquire images of molten iron flow during blast furnace tapping, identify the foreground region of the molten iron flow image to obtain a foreground region image, and extract image features from the molten iron flow image to obtain molten iron flow image features. The boundary feature extraction module 302 is used to extract the boundary of the molten iron flow stream based on the foreground region image to obtain the initial upper boundary and initial lower boundary of the molten iron flow stream, and to perform curve fitting on the initial upper boundary and initial lower boundary respectively to obtain the corresponding upper boundary fitting curve and lower boundary fitting curve. Based on the initial upper boundary, initial lower boundary, upper boundary fitting curve and lower boundary fitting curve, the boundary morphology features and temporal dynamic features of the molten iron flow stream are extracted, and the molten iron flow image features are fused with the boundary morphology features and temporal dynamic features to obtain molten iron flow fusion features. The stream state recognition module 303 is used to perform state recognition based on the molten iron flow fusion features to obtain the current state information of the molten iron flow stream.

[0067] Embodiments of this application also provide an electronic device, including: one or more processors; and a storage device for storing one or more programs, which, when executed by one or more processors, enable the electronic device to implement the blast furnace molten iron flow state detection method provided in the above embodiments.

[0068] Figure 4 This is a schematic diagram illustrating the structure of a computer system for an electronic device, as shown in an exemplary embodiment of this application. It should be noted that... Figure 4 The computer system 400 of the electronic device shown is merely an example and should not impose any limitation on the functionality and scope of use of the embodiments of this application.

[0069] like Figure 4 As shown, the computer system 400 includes a Central Processing Unit (CPU) 401, which can perform various appropriate actions and processes, such as executing the methods described in the above embodiments, based on programs stored in Read-Only Memory (ROM) 402 or programs loaded from storage into Random Access Memory (RAM) 403. The RAM 403 also stores various programs and data required for system operation. The CPU 401, ROM 402, and RAM 403 are interconnected via a bus. An I / O interface 405 is also connected to the bus 404, where the I / O interface 405 refers to an input / output interface.

[0070] The following components are connected to I / O interface 405: an input section 406 including a keyboard, mouse, etc.; an output section 407 including a cathode ray tube (CRT), liquid crystal display (LCD), etc., and speakers, etc.; a storage section 408 including a hard disk, etc.; and a communication section 409 including a network interface card such as a LAN (Local Area Network) card, modem, etc. The communication section performs communication processing via a network such as the Internet. A drive is also connected to I / O interface 405 as needed. Removable media 411, such as a disk, optical disk, magneto-optical disk, semiconductor memory, etc., are installed on drive 410 as needed so that computer programs read from them can be installed into storage section 408 as needed.

[0071] Specifically, according to embodiments of this application, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, embodiments of this application include a computer program product comprising a computer program carried on a computer-readable medium, the computer program including a computer program for performing the methods shown in the flowcharts. In such embodiments, the computer program can be downloaded and installed from a network via communication section 409, and / or installed from removable medium 411. When the computer program is executed by central processing unit (CPU) 401, it performs various functions defined in the system of this application.

[0072] It should be noted that the computer-readable medium shown in the embodiments of this application can be a computer-readable signal medium or a computer-readable storage medium, or any combination of the two. A computer-readable storage medium can be, for example, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples of a computer-readable storage medium may include, but are not limited to: an electrical connection having one or more wires, a portable computer disk, a hard disk, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM), flash memory, optical fiber, portable compact disc read-only memory (CD-ROM), optical storage device, magnetic storage device, or any suitable combination thereof. In this application, a computer-readable signal medium may include a data signal propagated in baseband or as part of a carrier wave, carrying a computer-readable computer program. Such propagated data signals can take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. Computer-readable signal media can also be any computer-readable medium other than computer-readable storage media, which can send, propagate, or transmit a program for use by or in connection with an instruction execution system, apparatus, or device. The computer program contained on the computer-readable medium can be transmitted using any suitable medium, including but not limited to wireless, wired, etc., or any suitable combination thereof.

[0073] 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. Each block in a flowchart or block diagram may represent a module, segment, or portion of code, which contains 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 a block diagram or flowchart, and combinations of blocks in a block diagram or flowchart, may 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.

[0074] In the corresponding figures of the above embodiments, connecting lines can represent the connection relationship between various components, indicating more constitutive signal paths and / or one or more ends of some lines having arrows to indicate the main information flow direction. Connecting lines are an identifier and are not a limitation on the scheme itself, but rather, using these lines in conjunction with one or more exemplary embodiments helps to more easily connect circuits or logic units. Any signal represented (determined by design requirements or preferences) can actually include one or more signals that can be transmitted in any direction and can be implemented in any suitable type of signal scheme.

[0075] The units described in the embodiments of this application can be implemented in software or hardware, and the described units can also be located in a processor. The names of these units do not necessarily limit the specific unit itself.

[0076] Another aspect of this application provides a computer-readable storage medium having a computer program stored thereon that, when executed by a processor, implements the method as described above. This computer-readable storage medium may be included in the electronic device described in the above embodiments, or it may exist independently and not assembled into the electronic device.

[0077] Embodiments of this application also provide a computer program product, including a computer program that, when executed by a processor, implements the blast furnace molten iron flow state detection method as described in any of the above embodiments.

[0078] It should be noted that although several modules or units for the device used to perform actions have been mentioned in the detailed description above, this division is not mandatory. In fact, according to the embodiments of this application, the features and functions of two or more modules or units described above can be embodied in one module or unit. Conversely, the features and functions of one module or unit described above can be further divided and embodied by multiple modules or units.

[0079] Through the above description of the embodiments, those skilled in the art will readily understand that the exemplary embodiments described herein can be implemented by software or by combining software with necessary hardware. Therefore, the technical solutions according to the embodiments of this application can be embodied in the form of a software product, which can be stored in a non-volatile storage medium (such as a CD-ROM, USB flash drive, external hard drive, etc.) or on a network, including several instructions to cause a computing device (such as a personal computer, server, touch terminal, or network device, etc.) to execute the method according to the embodiments of this application.

[0080] This application can be used in a wide range of general-purpose or special-purpose computing system environments or configurations. For example: personal computers, server computers, handheld or portable devices, tablet devices, multiprocessor systems, microprocessor-based systems, set-top boxes, programmable consumer electronics devices, network PCs, minicomputers, mainframe computers, distributed computing environments including any of the above systems or devices, etc.

[0081] Other embodiments of this application will readily occur to those skilled in the art upon consideration of the specification and practice of the embodiments disclosed herein. This application is intended to cover any variations, uses, or adaptations of this application that follow the general principles of this application and include common knowledge or customary techniques in the art not disclosed herein.

[0082] The above embodiments are merely illustrative of the principles and effects of this application and are not intended to limit this application. Any person skilled in the art can modify or alter the above embodiments without departing from the spirit and scope of this invention. Therefore, all equivalent modifications or alterations made by those skilled in the art without departing from the spirit and technical concept disclosed in this application should still be covered by the claims of this application.

Claims

1. A method of detecting the flow state of molten iron in a blast furnace, characterized by, The blast furnace molten iron flow state detection method comprises: obtaining a molten iron flow image of blast furnace tapping, and performing foreground region recognition on the molten iron flow image to obtain a foreground region image, and performing image feature extraction on the molten iron flow image to obtain molten iron flow image features; based on the foreground region image, extracting the boundary of the molten iron flow stream to obtain the initial upper boundary and the initial lower boundary of the molten iron flow stream, and respectively performing curve fitting on the initial upper boundary and the initial lower boundary to obtain the corresponding upper boundary fitting curve and lower boundary fitting curve; According to the initial upper boundary, the initial lower boundary, the upper boundary fitting curve and the lower boundary fitting curve, the boundary shape feature and the time sequence dynamic feature of the molten iron flow stream are extracted, and the molten iron flow image features and the boundary shape feature and the time sequence dynamic feature are fused to obtain the molten iron flow fusion features; based on the molten iron flow fusion features, performing state recognition to obtain the current state information of the molten iron flow stream.

2. The method of claim 1, wherein The foreground region recognition on the molten iron flow image obtains a foreground region image, which comprises: Calculate the distance between each pixel point to be processed in the molten iron flow image and each sample pixel point in the corresponding preset background sample set, the preset background sample set is constructed by collecting the neighborhood pixel points of each pixel point position in a plurality of continuous image frames, and the preset background sample set comprises a plurality of sample pixel points; According to the distance between the to-be-processed pixel point and each sample pixel point, determine the pixel category of the to-be-processed pixel point, the pixel category includes foreground region pixel and background pixel; extracting the to-be-processed pixel points with the pixel category of the foreground region pixel to obtain an initial foreground image, and performing binaryzation processing on the initial foreground image to obtain a binary foreground mask; Detect all connected regions in the binary foreground mask, and select the largest connected domain to generate a foreground region image.

3. The method of claim 1, wherein Based on the foreground region image, the boundary of the molten iron flow stream is extracted to obtain the initial upper boundary and the initial lower boundary of the molten iron flow stream, which comprises: Determine the boundary of the foreground region image as the initial contour curve of the active contour model, and construct an energy function, the energy function comprises an external energy term and an internal energy term, the external energy term is used to drive the contour curve to move to the molten iron flow boundary, and the internal energy term is used to constrain the continuity and smoothness of the contour curve; Iterative optimization is performed on the initial contour curve with the goal of minimizing the energy function, and the initial upper boundary and the initial lower boundary of the molten iron flow stream are obtained when the energy function is minimized.

4. The method of claim 3, wherein Respectively performing curve fitting on the initial upper boundary and the initial lower boundary to obtain the corresponding upper boundary fitting curve and lower boundary fitting curve comprises: Collecting a plurality of upper boundary discrete feature points from the initial upper boundary, collecting a plurality of lower boundary discrete feature points from the initial lower boundary, and respectively assigning a boundary weight coefficient to each upper boundary discrete feature point and each lower boundary discrete feature point, the boundary weight coefficient is inversely proportional to the Euclidean distance from the upper boundary discrete feature point or the lower boundary discrete feature point to the tapping hole position; The parabolic trajectory fitting is performed on the upper boundary discrete feature points and the lower boundary discrete feature points with the boundary weight coefficient based on a weighted least square fitting algorithm, so as to obtain the upper boundary fitting curve and the lower boundary fitting curve.

5. The method of claim 1, wherein The boundary shape features and the time sequence dynamic features of the molten iron flow are extracted according to the initial upper boundary, the initial lower boundary, the upper boundary fitting curve and the lower boundary fitting curve, including: The ideal jet area is determined by calculating the area formed by the upper boundary fitting curve and the lower boundary fitting curve; The flow divergence coefficient is obtained by integrating and averaging the ratio of the difference between the longitudinal coordinates of the upper boundary fitting curve and the lower boundary fitting curve and the jet width at the tapping hole; The flow boundary fluctuation coefficient is determined by calculating the average deviation of the longitudinal coordinates of the initial upper boundary and the initial lower boundary at the same horizontal coordinate and the corresponding fitting curve; The boundary displacement change rate is determined based on the change amount of the longitudinal coordinate value of the initial upper boundary and the initial lower boundary at the same horizontal coordinate within a unit time; The boundary fluctuation amplitude is obtained by calculating the standard deviation of the longitudinal coordinates of the initial upper boundary and the initial lower boundary at the same horizontal coordinate. The ideal jet area, the flow divergence coefficient and the flow boundary fluctuation coefficient are determined as the boundary shape features, and the boundary displacement change rate and the boundary fluctuation amplitude are determined as the time sequence dynamic features.

6. The method of claim 1-5, wherein The image feature extraction is performed on the molten iron flow image to obtain the molten iron flow image features, including: The structure features in the molten iron flow image are extracted by inputting the molten iron flow image into a pre-trained structure feature extraction network to obtain the image structure features, and the image structure features are determined as the molten iron flow image features, including the texture features, the shape features and the contour space features.

7. The method of claim 1-5, wherein The state recognition is performed based on the molten iron flow fusion features to obtain the current state information of the molten iron flow stream, including: The molten iron flow fusion features are input into a preset state recognition model, and the time sequence correlation analysis is performed on the molten iron flow fusion features based on the long short-term memory network in the preset state recognition model to obtain the time sequence correlation analysis result; The probability distribution of the molten iron flow stream in each state category is determined based on the time sequence correlation analysis result, and the state category with the highest probability distribution is determined as the current state information of the molten iron flow stream, including the normal, the flow interruption, the spattering and the shaking.

8. A molten iron flow state detecting device for a blast furnace, characterized by comprising: The blast furnace molten iron flow state detection device includes: An image acquisition and processing module is configured to acquire a molten iron flow image of a blast furnace tapping, recognize a foreground region of the molten iron flow image to obtain a foreground region image, and extract image features of the molten iron flow image. The boundary feature extraction module is configured to perform boundary extraction on the molten iron flow based on the foreground region image to obtain an initial upper boundary and an initial lower boundary of the molten iron flow, and perform curve fitting on the initial upper boundary and the initial lower boundary respectively to obtain a corresponding upper boundary fitting curve and a lower boundary fitting curve; the boundary shape feature and the time sequence dynamic feature of the molten iron flow are extracted according to the initial upper boundary, the initial lower boundary, the upper boundary fitting curve and the lower boundary fitting curve, and the molten iron flow image feature is fused with the boundary shape feature and the time sequence dynamic feature to obtain a molten iron flow fusion feature; The flow state recognition module is configured to perform state recognition based on the molten iron flow fusion feature to obtain current state information of the molten iron flow.

9. An electronic device, comprising: The computer program is stored on the computer readable medium, and is configured to enable the computer to execute the blast furnace molten iron flow state detection method according to any one of claims 1-7.

10. A computer-readable storage medium, characterized in that, The computer program is stored on the computer readable medium, and is configured to enable the computer to execute the blast furnace molten iron flow state detection method according to any one of claims 1-7.

Images