Dust removal control method and system for taphole of blast furnace

Abstract

The invention provides a blast furnace tap hole dust removal control method and system, relates to the field of blast furnace dust removal, and aims to solve the problem that smoke dust control is difficult to meet requirements due to the fact that only tapping hole molten iron discharge is considered in current air volume adjustment, and smoke dust concentration and distribution conditions can be more accurately judged by comprehensively considering operation machine and molten iron factors. The rotating speed of the fan can be adjusted in time at the moment that a large amount of smoke dust is generated by the operation machine, dynamic adjustment is performed based on the actual smoke dust condition and the state of the operation machine, unnecessary high-air-volume operation is avoided, smoke dust diffusion is effectively controlled, and therefore more accurate control over the dust removal fan is achieved.

Description

Technical Field

[0001] The present invention relates to the field of blast furnace dust removal, and in particular to a blast furnace taphole dust removal control method and system. Background Art

[0002] During the blast furnace ironmaking process, large amounts of dust-laden flue gas are released when iron is tapped from the taphole. Traditional dust removal fans, often operating at a fixed frequency or with simple timed adjustments, have significant drawbacks. Specifically, when the taphole is opened by the tapping machine, molten iron splashes everywhere, causing a sharp increase in dust levels. After the taphole is opened, the molten iron gushes out in a jet-like stream, also splashing around. Once the taphole is shielded by the molten iron hood, the splashing of molten iron at the taphole ceases, and dust emissions stabilize. After tapping is complete, the mud gun seals the taphole, stopping the outflow of molten iron and eliminating dust generation from the outflow. When dust emissions are high, failing to increase the fan frequency in a timely manner will result in ineffective dust removal, negatively impacting the working environment and steel quality. However, continuing to operate the fan at a high frequency when dust emissions are low results in unnecessary energy waste.

[0003] A Chinese patent (publication number CN 117450096 A, publication date 2024.01.26) discloses a blast furnace iron-making yard dust removal fan control method and device, which includes information acquisition, dust removal mapping and control modules. The information acquisition module obtains the amount of iron and temperature information by image acquisition and infrared thermal imaging; the dust removal mapping gives the dust removal wind force according to a preset model, controls the dust removal fan wind force based on the amount of iron and temperature of the molten iron discharged from the iron-making yard, builds a model based on the relationship between the two and the smoke concentration, and determines the dust removal wind force based on real-time collected data, mainly for energy saving and adapting to the dynamic changes of smoke. Although it applies image acquisition, it only considers the smoke generated during the discharge of the molten iron from the iron-making yard itself and its impact on the smoke. However, the operation of the operating machinery at the iron-making yard will also affect the generation and diffusion of smoke at the iron-making yard. Considering the molten iron at the iron-making yard alone will lead to deviations in the judgment of the smoke emission concentration, which in turn makes it difficult for the air volume adjustment to meet the requirements of the on-site smoke concentration and distribution, affecting the production environment of the iron-making yard. Summary of the Invention

[0004] The purpose of the present invention is to address the defects of the prior art and provide a blast furnace taphole dust removal control method and system. By comprehensively considering the factors of the operating machine and molten iron, the smoke concentration and distribution can be judged more accurately. At the moment when the operating machine generates a large amount of smoke, the fan speed can be adjusted in time, and dynamic adjustment can be made based on the actual smoke situation and the state of the operating machine, avoiding unnecessary high air volume operation, effectively controlling the diffusion of smoke, and thus achieving more precise control of the dust removal fan.

[0005] The first object of the present invention is to provide a blast furnace taphole dust removal control method, which adopts the following scheme:

[0006] include:

[0007] Collect image data of the taphole operating machine and the environment around the taphole, and extract image features of the working state of the taphole operating machine and the environment around the taphole;

[0008] Determine corresponding dust removal fan control information based on image features and a preset dust removal fan control model, where the preset dust removal fan control model is a correspondence between image features and dust removal fan control information;

[0009] The operating state of the dust removal fan is controlled according to the dust removal fan control information.

[0010] Furthermore, the working status of the taphole operation machine includes the taphole machine working status, the molten iron cover placement status and the mud gun working status.

[0011] Furthermore, the image data of the environment surrounding the tap hole is an image of smoke and dust generated during tap hole operation.

[0012] Furthermore, the determination of the preset dust removal fan control model includes:

[0013] Determine a first correspondence between an image feature of a working state of a taphole operating machine, an image feature of an environment surrounding the taphole, and smoke concentration;

[0014] Determine a second corresponding relationship between smoke dust concentration and dust removal fan control information;

[0015] A preset dust removal fan control model is obtained according to the first corresponding relationship and the second corresponding relationship.

[0016] Furthermore, obtaining a preset dust removal fan control model according to the first corresponding relationship and the second corresponding relationship includes:

[0017] Obtaining an initial model according to the first corresponding relationship and the second corresponding relationship;

[0018] Use a smoke detector to detect the smoke information after dust removal, and determine whether the initial model meets the requirements based on the smoke information;

[0019] If so, the initial model is used as the preset dust removal fan control model;

[0020] If not, adjust the dust removal fan control information and re-determine whether the initial model meets the requirements based on the smoke information obtained by detection.

[0021] Furthermore, the extraction of image features of the working state of the taphole operation machine and image features of the taphole surrounding environment includes:

[0022] Preprocess the image data;

[0023] Use edge detection algorithms to extract edge features of the taphole operating machinery and the surrounding environment, obtain the outline of the molten iron and the boundary information of the smoke, and determine the flow state of the molten iron and the diffusion range of the smoke;

[0024] Extract the texture features of smoke and dust to determine the concentration and distribution uniformity of smoke and dust;

[0025] Based on the flow state of molten iron, the diffusion range of smoke and dust, the concentration and distribution uniformity of smoke and dust are marked separately to match the corresponding relationship of the dust removal fan control information.

[0026] Furthermore, image acquisition parameters are configured according to the environmental conditions of the taphole, and the image acquisition parameters include resolution, frame rate and protection level.

[0027] Furthermore, the dust removal fan control information includes the dust removal fan operating frequency.

[0028] A second object of the present invention is to provide a blast furnace taphole dust removal control system as described in the first object, comprising:

[0029] a feature acquisition module configured to: collect image data of the taphole operating machine and the environment around the taphole, and extract image features of the working state of the taphole operating machine and the environment around the taphole;

[0030] A dust removal mapping module is configured to: determine corresponding dust removal fan control information based on image features and a preset dust removal fan control model, wherein the preset dust removal fan control model is a correspondence between image features and dust removal fan control information;

[0031] The dust removal control module is configured to control the operating state of the dust removal fan according to the dust removal fan control information.

[0032] Furthermore, the working status of the taphole operation machine includes the taphole machine working status, the molten iron cover placement status and the mud gun working status.

[0033] Compared with the prior art, the present invention has the following advantages and positive effects:

[0034] In response to the problem that the current air volume regulation only considers the discharge of molten iron from the taphole, resulting in smoke and dust control being unable to meet demand, by comprehensively considering the operating machine and molten iron factors, the smoke and dust concentration and distribution can be judged more accurately. At the moment when the operating machine generates a large amount of smoke and dust, the fan speed can be adjusted in time, and dynamic adjustments can be made based on the actual smoke and dust conditions and the operating machine status, avoiding unnecessary high air volume operation and effectively controlling smoke and dust diffusion, thereby achieving more precise control of the dust removal fan.

[0035] By collecting image data of the taphole operating machine, the collected image data is preprocessed, feature extracted and other steps to identify the status of the operating machine. Then, based on the identified operating machine status, combined with relevant algorithms and models, the smoke emission concentration is predicted. By analyzing the working conditions of the operating machine, such as the operating status of the equipment, the material processing process, etc., the amount of smoke that may be generated is inferred, thereby realizing the prediction of the smoke emission concentration and timely adjusting the fan operating status to meet the dust removal needs. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] The accompanying drawings, which constitute a part of the present invention, are used to provide a further understanding of the present invention. The exemplary embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute improper limitations on the present invention.

[0037] Figure 1 Schematic diagram of the architecture of the blast furnace taphole dust removal control method in Examples 1 and 2 of the present invention.

[0038] Figure 2 Schematic diagram of the flow chart of the blast furnace taphole dust removal control method in Examples 1 and 2 of the present invention.

[0039] In the figure, 1. Iron outlet; 2. High-definition industrial camera; 3. Signal converter; 4. Fiber optic transceiver; 5. Fiber optic transceiver; 6. Signal converter; 7. Computer host; 8. Display screen; 9. Controller; 10. Frequency converter; 11. Fan. DETAILED DESCRIPTION

[0040] Example 1

[0041] In a typical embodiment of the present invention, Figure 1-Figure 2 As shown, a method for controlling dust removal at the taphole of a blast furnace is given.

[0042] The existing technology only considers the smoke generated during the discharge of molten iron from the taphole, ignoring the impact of the operation of the taphole's operating machinery on the generation and diffusion of smoke. Due to the inaccurate judgment of smoke concentration by the existing technology, the air volume adjustment is difficult to meet the needs of on-site smoke concentration and distribution. Based on this, this embodiment can more comprehensively consider the factors affecting smoke concentration by collecting and analyzing the image features of the operating machinery, avoiding the deviation in smoke emission concentration judgment caused by the sole consideration of molten iron factors; and establishes a dust removal fan control model based on image features, which can accurately adjust the dust removal fan's air volume according to the actual smoke and operating conditions, so that it can better adapt to the dynamic changes of smoke on site and improve the production environment for tapping iron at the taphole.

[0043] like Figure 1 and Figure 2 As shown, the blast furnace taphole dust removal control method includes:

[0044] Collect image data of the taphole operating machine and the environment around the taphole, and extract image features of the working state of the taphole operating machine and the environment around the taphole;

[0045] Determine corresponding dust removal fan control information based on image features and a preset dust removal fan control model, where the preset dust removal fan control model is a correspondence between image features and dust removal fan control information;

[0046] The operating state of the dust removal fan is controlled according to the dust removal fan control information.

[0047] According to the above control method, in this embodiment, the control process of the above control method is implemented through multiple modules.

[0048] Specifically, the image acquisition module is used to capture images of the working status of the blast furnace taphole operating machine; the image recognition module analyzes and recognizes the captured images; the intelligent frequency modulation decision module analyzes and makes decisions based on the data provided by the image recognition module; the fan frequency modulation control module controls the frequency of the dust removal fan according to the frequency instructions output by the intelligent frequency modulation decision module; and the human-computer interaction module provides the user with the system's operating interface and status display.

[0049] The image acquisition module consists of a high-definition industrial camera 2, a mounting bracket, and data transmission lines. The camera is positioned appropriately near the blast furnace taphole to capture real-time images of the taphole's operating machinery and the surrounding environment. The camera is equipped with an automatic cleaning device to ensure image clarity.

[0050] The machines operating the blast furnace taphole primarily consist of the taphole opening machine, molten iron cover, and mud gun. When the taphole opening machine is in operation, the powerful mechanical force instantly disrupts the previously enclosed environment. The internal high pressure drives the airflow at high speed, stirring up a large amount of dust and forming smoke. During operation, if the equipment is poorly sealed or vibrates, the surrounding air is disturbed and drawn in, increasing the chance of dust becoming suspended. The duration and frequency of the machine's operation affect the amount of smoke. Prolonged, high-frequency operation continuously disturbs the environment around the taphole, leading to the accumulation of smoke.

[0051] Before the molten iron cover is placed, the molten iron is directly exposed without any cover, and the high-temperature molten iron evaporates, reacts chemically, and exchanges heat violently with the surrounding air to produce a large amount of smoke and dust. During the placement of the molten iron cover, the docking of the molten iron cover and the iron outlet requires precise operation. A slight deviation will cause a gap, and the high-temperature gas will leak with smoke and dust. The material and structure of the molten iron cover affect the dispersion of smoke and dust. If the thermal insulation performance is poor and the sealing structure is unreasonable, it will be difficult to effectively prevent the spread of smoke and dust, causing some smoke and dust to overflow.

[0052] When the mud gun is working to seal the taphole, the pressure and speed of the mud ejected from the mud gun cause strong disturbances in the surrounding air, raising dust to form smoke; the quality of the mud affects the sealing effect. If the quality is poor and the sealing is not tight or cracks and holes appear, the high-temperature molten iron and gas remaining in the taphole will seep out and interact with the outside air, continuously generating smoke; the maintenance condition of the mud gun is related to the working stability. Failure or performance degradation will cause abnormal operation, such as uneven impact force and prolonged sealing time, which will increase the probability and amount of smoke generation.

[0053] In this embodiment, the high-definition industrial camera 2 has a high resolution (1080p and above) and a high frame rate (more than 30 frames per second), and can clearly capture the details of the taphole 1 operating machine under different lighting conditions. The lens of the high-definition industrial camera 2 needs to have a suitable focal length and viewing angle to cover the key areas where the blast furnace taphole operating machine works. Generally, a fixed-focus lens is used, and a suitable focal length (such as 5-50mm) is selected according to the monitoring distance and range. The high-definition industrial camera 2 needs to be equipped with a protective housing and a mounting bracket. The protective housing can protect the camera from harsh environmental factors such as dust and high temperature. The mounting bracket should be able to flexibly adjust the angle and height of the camera to ensure the best shooting angle. For example, a rotatable and retractable metal bracket can be used, and the material can be stainless steel or aluminum alloy. In addition, in areas with darker light, auxiliary lighting equipment such as LED light strips or infrared lamps can be provided. LED light strips can provide stable visible light illumination, and infrared lamps can be used to achieve night vision functions in low light or no light conditions to ensure the continuity of image acquisition.

[0054] The image recognition module primarily consists of an image processor, image recognition algorithm software, and storage devices. The image processor is responsible for image preprocessing, such as denoising and enhancement, to improve image quality. The image recognition algorithm software uses the YOLOv8 algorithm to extract features from the image and identify the operating status of the taphole machinery, such as the tapping machine, the placement of the molten iron cover, and the working status of the mud gun, which are closely related to the amount of smoke and dust generated. The storage device is used to store the trained image recognition model and historical image data.

[0055] The image recognition module primarily includes a computer host 7. Computer host 7 is equipped with a dedicated graphics processing unit (GPU) or a high-performance central processing unit (CPU). GPUs have powerful parallel computing capabilities, making them suitable for processing large amounts of image data and accelerating the execution of image recognition algorithms. Certain NVIDIA GPU models feature a large number of CUDA cores, enabling efficient image processing. Computer host 7 also requires image recognition algorithm software, developed based on machine learning or deep learning algorithms. Machine learning algorithms may include support vector machines and random forests; deep learning algorithms often use convolutional neural networks (CNNs), such as AlexNet, VGG, and ResNet. These algorithm software must run on a server or high-performance computer and must include model training, parameter adjustment, and real-time image recognition capabilities. The storage device in computer host 7 is a large-capacity hard drive (such as a mechanical hard drive or solid-state drive) for storing training image data, trained model parameters, and intermediate calculation results. Furthermore, RAM (memory RAM) may be required to temporarily store the image data being processed. The RAM capacity should be sufficiently large (16GB or above) to ensure the efficient execution of the image recognition algorithm.

[0056] The intelligent frequency modulation decision module includes a data analysis processor, intelligent algorithm software and a database. The data analysis processor processes image features. The intelligent algorithm software determines the optimal operating frequency of the dust removal fan based on pre-established models and rules. The database is used to store historical data and decision results to optimize and improve the system.

[0057] The intelligent frequency regulation decision module is mainly a computer host 7. The computer host 7 has a data analysis processor with high-performance computing capabilities for processing large amounts of image feature data. The processor should have a multi-core architecture (quad-core, octa-core, etc.) and be able to quickly perform data analysis and algorithm operations. The computer host 7 is installed with intelligent algorithm software, which can be used to establish intelligent algorithms based on historical data and experimental results, such as fuzzy logic algorithms, neural network algorithms, etc. These algorithm software uses the working status data of the iron outlet operation machine as input and outputs the optimal frequency value of the dust removal fan. The software should have self-learning and self-adaptation capabilities and be able to continuously optimize the algorithm parameters according to the system operation conditions. The computer host 7 is installed with database software for storing historical image data, fan operating frequency data and decision results. The database can be a relational database (such as MySQL, Oracle, etc.) or a non-relational database (such as MongoDB, etc.), with large-capacity storage and fast query functions to support long-term storage and real-time retrieval of data.

[0058] The fan frequency control module consists of a variable frequency drive, a controller and a communication interface. The variable frequency drive adjusts the fan speed, the controller is responsible for receiving instructions and controlling the operation of the variable frequency drive, and the communication interface realizes information exchange with other modules to ensure the coordinated operation of the system.

[0059] The fan frequency regulation control module includes a controller 9, a frequency converter 10 and a fan 11. The frequency converter 10 receives the frequency control signal from the intelligent frequency regulation decision module and converts it into the voltage and frequency to drive the fan 11 motor, thereby changing the speed of the dust removal fan 11. The frequency converter 10 should have a wide frequency speed regulation range (such as 0-50Hz or wider), high-precision frequency control (frequency resolution can reach 0.01Hz or higher) and good overload protection capabilities. The controller 9 is used to coordinate the communication and control logic between the frequency converter 10 and other modules (computer host 7). It can be a programmable logic controller (PLC) or a dedicated microcontroller. The controller 9 receives the instructions of the intelligent frequency regulation decision module and passes them to the frequency converter 10, while monitoring the operating status of the frequency converter 10 and the fan 11, such as current, voltage, speed, etc., to ensure the safe and stable operation of the system.

[0060] The human-machine interaction module includes a display screen 8, an operation panel and an alarm device. The display screen displays the working status of the iron mouth operation machine, the operating frequency of the fan and other information in real time. The operation panel allows the user to perform parameter settings, manual control and other operations. When an abnormal situation occurs in the system, such as image recognition failure, fan failure, etc., the alarm device will emit an audible and visual alarm to remind the user to deal with it in time.

[0061] The human-computer interaction module includes a display screen 8, an operation panel, a keyboard, and a mouse. The display screen 8 can be an industrial-grade liquid crystal display (LCD) or a light-emitting diode display (LED), and its size is determined according to the operating distance and visibility requirements (such as 7 inches, 10 inches, etc.). The display screen should have high clarity, high contrast, and a good viewing angle to clearly display various system parameters, images, and alarm information. The operation panel includes input devices such as buttons and a touch screen. Buttons can be used for emergency operations, parameter adjustments, and other functions, while the touch screen provides a more intuitive and convenient operation method. Users can start, stop, and set parameters of the system through touch operations.

[0062] The communication module primarily includes a signal converter 3, a signal converter 6, a fiber optic transceiver 4, a fiber optic transceiver 5, optical fibers, and network cables. First, determine the number of devices that will be connected in the system. These include a high-definition industrial camera 1, an image processor, the server in the intelligent frequency modulation decision module, devices in the fan frequency modulation control module (such as variable frequency drives and controllers), and devices in the human-computer interaction module (such as display screen 8 and host computer). Generally, a certain amount of port redundancy should be reserved to accommodate future equipment expansion needs. For devices such as cameras, servers, and controllers, RJ45 ports provide stable wired network connections. Fiber optic ports are particularly useful if some devices in the system are remote or require extremely high interference resistance for data transmission. These devices can be connected to the switch's fiber optic ports using fiber optic patch cables, enabling high-speed, stable, and long-distance data transmission. In the complex electromagnetic environment of the taphole, fiber optic connections can effectively reduce the impact of electromagnetic interference on data transmission. For the high-definition camera in the image acquisition module, if high resolution (such as 4K) and high frame rate are used for image acquisition, a gigabit switch is recommended to ensure timely and smooth transmission of image data. If a large amount of image data needs to be transmitted simultaneously in the system, or if the system may be upgraded to applications with higher bandwidth requirements in the future (such as adding more high-definition cameras or performing real-time video analysis), a 10G switch can be considered.

[0063] The implementation of the blast furnace taphole dust removal control method includes the following steps:

[0064] (1) Installation of image acquisition equipment: Install high-definition cameras at key locations of the blast furnace taphole to ensure that they can fully cover the taphole opening machine, molten iron cover, mud gun, taphole and other related areas to obtain image data of the working status of the blast furnace taphole operating machines.

[0065] (2) Image recognition: De-noising is performed on the collected images of the taphole machine to remove interference factors in the image.

[0066] (3) Intelligent frequency adjustment decision: Through experiments and statistical analysis of historical data, a mapping relationship between image features and the optimal frequency of the dust removal fan is established, and this mapping relationship is stored in the control system in the form of a database or rule table. When making real-time decisions, the real-time data obtained by the integrated image recognition module is used to query the mapping relationship table to determine the optimal frequency of the dust removal fan under the current circumstances. Considering the stability of the system and the continuity of the dust removal effect, a certain frequency adjustment step and buffer zone can be set to avoid frequent and large fluctuations in the fan frequency.

[0067] (4) Fan frequency control: Based on the frequency value determined by the intelligent frequency modulation decision module, the control system outputs the corresponding control signal to the variable frequency drive of the dust removal fan to adjust the fan frequency. After the fan frequency is adjusted, the dust removal effect is continuously monitored. Based on the feedback of the dust removal effect, the mapping relationship between image recognition and fan frequency is optimized and adjusted to improve the accuracy and efficiency of the entire system.

[0068] Among them, in the image recognition model training and classification during image recognition, deep learning is adopted, and convolutional neural networks (CNN) such as ResNet, VGG and other network structures are selected. The YOLOv8 algorithm is used to train a large amount of labeled image data and classify and recognize the collected real-time images.

[0069] The YOLOv8 algorithm's network architecture consists of a backbone network, a neck network, and a head network. The backbone network is primarily used to extract image features, the neck network is primarily used to fuse features at different levels, and the head network is used for final object detection and classification.

[0070] During the training phase, the YOLOv8 algorithm mainly goes through four stages: data preparation and labeling, loss function definition, forward propagation, backpropagation and parameter update.

[0071] During the data preparation and labeling phase, the YOLOv8 algorithm collected a large amount of image data of blast furnace taphole operating machines and annotated each image. The annotated dataset was divided into training, validation, and test sets.

[0072] The YOLOv8 algorithm optimizes the model through classification loss, positioning loss, and target confidence loss during the loss function definition stage.

[0073] In the forward propagation stage of the YOLOv8 algorithm, the image first passes through the backbone network for feature extraction, then passes through the neck network for feature fusion, and finally generates the prediction box, category probability and confidence score in the head network.

[0074] In the backpropagation and parameter update phase of the YOLOv8 algorithm, after calculating the value of the loss function based on the prediction results and the true annotations, the gradient is calculated through the backpropagation algorithm.

[0075] By comprehensively considering the operating machine and molten iron, we can more accurately determine the smoke concentration and distribution, thereby achieving more precise control of the dust removal fan. For example, at the moment when the operating machine generates a large amount of smoke, the fan speed can be adjusted in time to effectively control the smoke spread.

[0076] Combine Figure 2 The blast furnace taphole dust removal control method specifically includes the following steps:

[0077] 1. System design and planning phase, including the following steps:

[0078] S1. Communicate with blast furnace taphole operators, environmental protection personnel, and other relevant personnel to clarify the system's functional requirements. Also, understand the actual operating conditions, such as the frequency of blast furnace tapping and the range of molten iron flow, to provide a basis for subsequent design.

[0079] S2. Select an appropriate image acquisition device (industrial camera) based on the environmental conditions at the blast furnace taphole (such as temperature, humidity, space, and lighting). Consider parameters such as the camera's resolution, frame rate, and protection level.

[0080] S3. Determine the models and installation locations of other auxiliary sensors (such as temperature sensors and smoke concentration sensors). Auxiliary sensors are used to obtain more environmental parameters and verify the image data.

[0081] S4. Plan the control interface of the dust removal fan and select a suitable intelligent controller (such as PLC, industrial computer, etc.) to ensure that it can receive control signals and accurately adjust the fan's speed, air volume and other parameters.

[0082] S5. Design the image recognition algorithm architecture and decide whether to use traditional image processing algorithms or deep learning algorithms. If using deep learning, select an appropriate network model (such as a convolutional neural network (CNN)) and plan the model's training strategy.

[0083] S6. Design the intelligent control algorithm architecture, such as parameter settings and rule formulation for fuzzy control or PID control algorithms. Also, plan the data communication module to ensure accurate transmission of image and sensor data between different devices and timely delivery of control commands to the dust removal blower.

[0084] 2. Data collection phase includes the following steps:

[0085] U1. Install and debug the image acquisition equipment to ensure that the operating machinery and the smoke generated in the blast furnace taphole area can be clearly captured. Adjust the camera's angle, focal length, aperture, and other parameters to ensure that the image contains sufficient information.

[0086] U2. Set an appropriate image acquisition frequency. According to the speed of iron tapping in the blast furnace and the dynamic characteristics of smoke and dust generation, the acquisition frequency should not be too low to avoid missing key information, but it should not be too high to cause excessive data volume.

[0087] U3. Install and calibrate other auxiliary sensors, such as a temperature sensor near the taphole to measure the molten iron temperature and a smoke concentration sensor in the ventilation duct to monitor the smoke concentration in real time.

[0088] U4. Synchronously collect auxiliary sensor data and image data, and ensure that different types of data correspond in time through timestamps or other synchronization mechanisms, facilitating subsequent data fusion and analysis.

[0089] 3. Data processing and model building phase, including the following steps:

[0090] Extracting the image features of the working state of the taphole machine and the image features of the taphole surrounding environment includes:

[0091] Preprocess the image data;

[0092] Use edge detection algorithms to extract edge features of the taphole operating machinery and the surrounding environment, obtain the outline of the molten iron and the boundary information of the smoke, and determine the flow state of the molten iron and the diffusion range of the smoke;

[0093] Extract the texture features of smoke and dust to determine the concentration and distribution uniformity of smoke and dust;

[0094] Based on the flow state of molten iron, the diffusion range of smoke and dust, the concentration and distribution uniformity of smoke and dust are marked separately to match the corresponding relationship of the dust removal fan control information.

[0095] The determination of the preset dust removal fan control model includes:

[0096] Determine a first correspondence between an image feature of a working state of a taphole operating machine, an image feature of an environment surrounding the taphole, and smoke concentration;

[0097] Determine a second corresponding relationship between smoke dust concentration and dust removal fan control information;

[0098] A preset dust removal fan control model is obtained according to the first corresponding relationship and the second corresponding relationship.

[0099] The preset dust removal fan control model obtained according to the first corresponding relationship and the second corresponding relationship includes:

[0100] Obtaining an initial model according to the first corresponding relationship and the second corresponding relationship;

[0101] Use a smoke detector to detect the smoke information after dust removal, and determine whether the initial model meets the requirements based on the smoke information;

[0102] If so, the initial model is used as the preset dust removal fan control model;

[0103] If not, adjust the dust removal fan control information and re-determine whether the initial model meets the requirements based on the smoke information obtained by detection.

[0104] Specifically:

[0105] X1. Grayscale the captured image, converting the color image to grayscale. This reduces the amount of data and simplifies subsequent calculations. Filter the image to remove noise. Select the appropriate filtering method based on the noise type, such as Gaussian filtering for Gaussian noise or median filtering for salt and pepper noise. Gaussian filtering smoothes the image by taking a weighted average of each pixel and its neighborhood. Median filtering replaces the center pixel's value with the median of its neighborhood.

[0106] X2. Use edge detection algorithms (such as the Canny operator) to extract edge features of the taphole machinery and smoke. The Canny operator identifies edges by finding the maximum gradient in the image. This can reveal information such as the outline of the molten iron and the boundaries of the smoke. These edge features help determine the flow state of the molten iron and the diffusion range of the smoke.

[0107] X3. Extract texture features and analyze the smoke texture using a gray-level co-occurrence matrix (GLCM). The GLCM counts the frequency of pixel pairs with different grayscale values ​​in an image and describes the texture by calculating parameters such as contrast, energy, and entropy, thereby determining the concentration and uniformity of the smoke distribution.

[0108] X4. Mark the molten iron flow state, smoke diffusion range, smoke concentration, and distribution uniformity separately to match the corresponding relationship of the dust removal fan control information.

[0109] The extracted features are labeled into different categories, such as "high smoke concentration", "medium smoke concentration", "low smoke concentration" and so on.

[0110] X5. Divide the dataset into training, validation, and test sets, generally in a ratio of 7:2:1. Use the training set to train the selected deep learning model (such as a CNN). Use the backpropagation algorithm to continuously adjust the network weights so that the model can accurately classify different categories. During training, use the validation set to monitor model performance and prevent overfitting. Finally, use the test set to evaluate the final model results.

[0111] X6. Use fuzzy control to determine the range of the fuzzy set. For smoke concentration, the fuzzy set can be defined as "low," "medium," and "high." For dust removal fan speed, the fuzzy set can be defined as "slow," "medium," and "fast."

[0112] X7 determine the working state of the taphole machine image characteristics, the taphole surroundings image characteristics and the first correspondence between smoke concentration;

[0113] Determine a second corresponding relationship between smoke dust concentration and dust removal fan control information;

[0114] A preset dust removal fan control model is obtained according to the first corresponding relationship and the second corresponding relationship, and the model has fuzzy rules.

[0115] Establish fuzzy rules, such as "If the smoke concentration is high, the dust removal fan speed is fast." The specific content of these rules is determined through expert experience or experimental data.

[0116] 4. System integration and debugging phase includes the following steps:

[0117] Y1. Integrate the image recognition software and intelligent control software into the selected controller (such as a PLC or industrial computer), ensuring that the software can correctly read the data from the hardware devices (image acquisition device, auxiliary sensors) and can send control instructions to the dust removal blower.

[0118] Y2. Establish communication links and transmit data between different devices through wired (such as Ethernet, RS485, etc.) or wireless (such as ZigBee, Wi-Fi, etc.) communication methods.

[0119] Y3. Perform offline debugging. First, use a pre-collected data set to test the operation of the entire system and check the accuracy of image recognition and the rationality of intelligent control. Check whether the image classification model can correctly identify images of different smoke concentrations and whether the intelligent control model can output reasonable fan control instructions based on the recognition results.

[0120] Y4. Conduct online debugging and testing in an actual blast furnace tapping environment. Observe the system's operation under actual operating conditions and promptly adjust any issues that arise. If image recognition deviates under strong light, adjust the camera parameters or optimize the image recognition algorithm. If the dust removal fan control is inaccurate, adjust the parameters of the intelligent control model.

[0121] 5. System optimization and maintenance phase, including the following steps:

[0122] U1. Optimize the image recognition and intelligent control algorithms based on data and feedback from the system's operation. Regularly update the weights of the image classification model to adapt to changes in the operating status of the operating machines during the blast furnace tapping process.

[0123] U2. Optimize the performance of hardware equipment, adjust the parameters of image acquisition equipment to obtain better image quality, maintain and upgrade the dust removal fan to improve its working efficiency.

[0124] U3. Establish a system maintenance plan and regularly check the operating status of hardware equipment, including image acquisition equipment, sensors, and dust removal fans. Clean camera lenses, check sensor accuracy, and replace worn parts of dust removal fans.

[0125] U4. Back up system data and parameters, including image datasets, model parameters, and control rules. This allows for rapid system recovery in the event of a system failure or upgrade. System software is also updated and protected against virus infection and malicious attacks.

[0126] Example 2

[0127] In another typical embodiment of the present invention, Figure 1-Figure 2 As shown in the figure, a blast furnace taphole dust removal control system is given.

[0128] A blast furnace taphole dust removal control system, comprising:

[0129] a feature acquisition module configured to: collect image data of the taphole operating machine and the environment around the taphole, and extract image features of the working state of the taphole operating machine and the environment around the taphole;

[0130] A dust removal mapping module is configured to: determine corresponding dust removal fan control information based on image features and a preset dust removal fan control model, wherein the preset dust removal fan control model is a correspondence between image features and dust removal fan control information;

[0131] The dust removal control module is configured to control the operating state of the dust removal fan according to the dust removal fan control information.

[0132] The working status of the taphole operation machine includes the working status of the taphole machine, the placement status of the molten iron cover and the working status of the mud gun.

[0133] The working process of the blast furnace taphole dust removal control system is similar to the blast furnace taphole dust removal control method in Example 1, and will not be described in detail here.

[0134] The foregoing description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. Those skilled in the art will readily appreciate that various modifications and variations of the present invention are possible. Any modifications, equivalent substitutions, or improvements made within the spirit and principles of the present invention are intended to be within the scope of protection of the present invention.

Claims

1. A blast furnace tap hole dust removal control method, characterized in that: include: Collect image data of the taphole operation machine and the environment around the taphole, and extract image features of the working state of the taphole operation machine and the environment around the taphole; Determine corresponding dust removal fan control information based on the image features and a preset dust removal fan control model, where the preset dust removal fan control model is a correspondence between the image features and the dust removal fan control information; The operating state of the dust removal fan is controlled according to the dust removal fan control information.

2. The blast furnace tap hole dust removal control method according to claim 1, characterized in that: The working state of the taphole operation machine includes the working state of the taphole opening machine, the placing state of the molten iron cover and the working state of the mud gun.

3. The blast furnace tap hole dust removal control method according to claim 1 or 2, characterized in that: The image data of the environment around the tap hole is an image of smoke and dust generated during the tap hole operation.

4. The blast furnace tap hole dust removal control method according to claim 1, characterized in that: The determination of the preset dust removal fan control model includes: Determine a first correspondence between the image features of the working state of the tap hole operating machine, the image features of the environment around the tap hole, and the smoke concentration; Determine a second corresponding relationship between smoke concentration and dust removal fan control information; A preset dust removal fan control model is obtained according to the first corresponding relationship and the second corresponding relationship.

5. The blast furnace tap hole dust removal control method according to claim 4, characterized in that: The preset dust removal fan control model is obtained according to the first corresponding relationship and the second corresponding relationship, including: An initial model is obtained according to the first corresponding relationship and the second corresponding relationship; The smoke information after dust removal is detected by a smoke detector, and whether the initial model meets the requirements is determined based on the smoke information; If so, the initial model is used as the preset dust removal fan control model; If not, adjust the dust removal fan control information and re-determine whether the initial model meets the requirements based on the smoke information obtained by the detection.

6. The blast furnace tap hole dust removal control method according to claim 1, characterized in that: The image features of the working state of the tap hole operation machine and the image features of the surrounding environment of the tap hole are extracted as follows: Preprocess the image data; The edge detection algorithm is used to extract the edge features of the taphole operating machine and the environment around the taphole, obtain the outline of the molten iron and the boundary information of the smoke, and determine the flow state of the molten iron and the diffusion range of the smoke; Extract the texture features of smoke and dust to determine the concentration and uniformity of distribution of smoke and dust; Based on the flow state of molten iron, the diffusion range of smoke, the concentration of smoke and the uniformity of distribution, they are marked separately to match the corresponding relationship of the dust removal fan control information.

7. The blast furnace tap hole dust removal control method according to claim 1, characterized in that: Configure image acquisition parameters according to the environmental conditions of the iron outlet. The image acquisition parameters include resolution, frame rate and protection level.

8. The blast furnace tap hole dust removal control method according to claim 1, characterized in that: The dust removal fan control information includes the dust removal fan operating frequency.

9. A blast furnace tap hole dust removal control system, characterized in that: include: The feature acquisition module is configured to: collect image data of the tap hole operation machine and the environment around the tap hole, and extract image features of the working state of the tap hole operation machine and the image features of the environment around the tap hole; A dust removal mapping module is configured to: determine corresponding dust removal fan control information based on image features and a preset dust removal fan control model, wherein the preset dust removal fan control model is a correspondence between image features and dust removal fan control information; The dust removal control module is configured to control the operating state of the dust removal fan according to the dust removal fan control information.

10. The blast furnace tap hole dust removal control system according to claim 9, characterized in that: The working state of the taphole operation machine includes the working state of the taphole opening machine, the placing state of the molten iron cover and the working state of the mud gun.

Images