A System and Method for Determining the End Point of Copper Converter Smelting Based on Flame Image Feature Recognition
By designing a copper converter blowing endpoint determination system based on flame image feature recognition, the stability and adaptability of image acquisition and feature extraction were achieved, solving the problems of unstable judgment and high misjudgment rate in the existing technology, and improving the automation and accuracy of the copper smelting process.
Patent Information
- Application Number
- CN202611002922.5
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-07-07
- Publication Date
- 2026-11-10
AI Technical Summary
Existing flame image endpoint determination systems have shortcomings in image acquisition stability, feature extraction robustness, and model adaptability, resulting in unstable judgment results and a high misjudgment rate, making it difficult to meet the needs of modern copper smelting full-process automation and intelligent control.
A copper converter blowing endpoint determination system based on flame image feature recognition was designed, including image acquisition, preprocessing, feature extraction and endpoint determination modules. Combined with a central control module, the image acquisition frequency, preprocessing algorithm and feature extraction threshold can be adjusted in real time. Through multi-dimensional feature analysis and closed-loop feedback optimization, the adaptability and consistency of the determination process are improved.
Stable acquisition and feature extraction of flame images were achieved, improving the accuracy and consistency of endpoint determination, reducing reliance on operator experience, and enhancing the automation level of the copper converter blowing process.
Smart Images

Figure CN123071699A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of copper smelting technology, and more specifically, to a system and method for determining the endpoint of copper converter blowing based on flame image feature recognition. Background Technology
[0002] Flame image feature recognition technology plays a crucial role in determining the endpoint of copper converter blowing. Its main function is to achieve real-time, non-contact monitoring of the blowing process by analyzing the visual characteristics of the flame at the furnace mouth, providing a basis for smelting process optimization and precise endpoint control. Existing technologies typically analyze changes in flame image features such as color, brightness, shape, and texture, and use empirical models or machine learning algorithms to determine the blowing stage. However, this process involves the coordination of multiple stages, including image acquisition quality, preprocessing effectiveness, and feature extraction accuracy, resulting in a long technical chain and placing high demands on the real-time performance and reliability of the results.
[0003] Currently, existing flame image judgment systems generally employ fixed or semi-fixed parameter settings. Image acquisition frequency, preprocessing algorithms, and feature extraction thresholds rely primarily on manual experience for pre-setting, making it difficult to adapt to different furnace conditions, raw material variations, and environmental interference. During image acquisition and processing, fluctuations in illumination, smoke interference, and camera shake can degrade image quality, ultimately leading to misjudgments or delays in endpoint determination. Furthermore, existing systems typically lack online analysis and parameter self-adjustment mechanisms based on real-time judgment results. The determination of the blowing endpoint often relies on manual verification or delayed analysis, resulting in untimely adjustments and inconsistent judgment standards between furnace runs.
[0004] It is evident that existing flame image endpoint determination technologies still have shortcomings in image acquisition stability, feature extraction robustness, model adaptability, and process closed-loop optimization, making it difficult to guarantee the accuracy of the determination results and the timeliness of process control. This not only increases the reliance on operator experience and introduces subjective errors but also limits the in-depth application of this technology in the automation and intelligent control of the entire modern copper smelting process. Therefore, there is an urgent need to provide a flame image blowing endpoint determination system that can achieve intelligent adjustment of image acquisition, highly robust feature extraction, online adaptive model analysis, and integrate real-time judgment feedback and parameter closed-loop optimization, in order to solve the problems of unstable judgment and high misjudgment rate in existing technologies. Summary of the Invention
[0005] In view of this, the present invention proposes a copper converter blowing endpoint determination system and method based on flame image feature recognition, aiming to solve the problems of lack of robustness and adaptability in the current judgment methods in terms of image acquisition condition control, feature extraction and processing and endpoint determination feedback adjustment, which leads to unstable judgment process and poor consistency of endpoint determination results.
[0006] This invention proposes a copper converter blowing endpoint determination system based on flame image feature recognition, comprising: The image acquisition module includes a camera device and a light source auxiliary device for real-time acquisition of images of the flame at the mouth of a copper converter; An image preprocessing module, connected to the image acquisition module, includes an image processor for denoising and enhancing the acquired flame images; The feature extraction module, connected to the image preprocessing module, includes a feature analyzer for extracting color distribution, brightness gradient, shape contour and texture features from flame images. The endpoint determination module is connected to the feature extraction module and is used to calculate the blowing state parameters based on the extracted flame image features. The blowing state parameters include flame stability, color saturation, brightness peak and shape variation rate. The central control module is electrically connected to the image acquisition module, image preprocessing module, feature extraction module, and endpoint determination module. It is used to determine whether the blowing endpoint has been reached based on the flame stability calculated by the endpoint determination module and the preset stability threshold. If the endpoint has not been reached, it will adjust one or more of the following parameters for the next blowing cycle: image acquisition frequency, preprocessing algorithm, or feature extraction threshold, based on the deviation of flame stability and the combined analysis results of other blowing state parameters.
[0007] Furthermore, the central control module includes: The data acquisition unit is used to acquire the values of all refining state parameters calculated by the endpoint judgment module in real time; The model comparison unit is used to compare the flame stability obtained by the data acquisition unit with the first preset stability and the second preset stability, and generate the endpoint determination result. The parameter adjustment unit is used to optimize and adjust the enhancement algorithm of the image preprocessing module in the next blowing cycle based on the color saturation detection value when the model comparison unit determines that the blowing endpoint has been reached, or to correct and adjust the process parameters of the corresponding module based on the endpoint determination result when the blowing endpoint has not been reached. The model comparison unit determines that the copper converter blowing endpoint has been reached under the first determination condition; determines that the copper converter blowing endpoint needs to be corrected at the first level under the second determination condition; and determines that the copper converter blowing endpoint has not been reached under the third determination condition, and determines the reason for not reaching the endpoint based on the data of brightness peak and shape variation rate, including: the first determination condition is that the flame stability is greater than or equal to the first preset stability. The second determination condition is that the flame stability is less than the first preset stability and greater than or equal to the second preset stability. The third criterion is that the flame stability is less than the second preset stability.
[0008] Furthermore, when the parameter adjustment unit determines the optimized adjustment method for the enhancement algorithm in the next blowing cycle based on the color saturation calculated by the endpoint judgment module, it includes: When the color saturation is less than the preset lower limit of saturation, the central control module selects the first algorithm enhancement coefficient ε1 to adjust the enhancement algorithm to the corresponding value; When the color saturation is greater than the preset saturation upper limit, the central control module uses the first algorithm reduction coefficient ε2 to adjust the enhancement algorithm to the corresponding value; When the color saturation is between the preset lower limit and the preset upper limit, the current enhancement algorithm remains unchanged.
[0009] Furthermore, when a first-level correction is required, the parameter adjustment unit records the difference between the first preset stability and the measured flame stability as the first-level stability difference, and determines the adjustment method for the acquisition frequency of the image acquisition module in the next blowing cycle based on the first-level stability difference, including: When the first-level stability difference is greater than or equal to the preset first-level difference threshold, the parameter adjustment unit uses the first frequency adjustment coefficient φ1 to linearly increase the acquisition frequency to the corresponding target frequency value. When the first-level stability difference is less than the preset first-level difference threshold, the parameter adjustment unit selects the second frequency adjustment coefficient φ2 to nonlinearly increase the acquisition frequency to the corresponding target frequency value.
[0010] Furthermore, after determining the target frequency value for the frequency to be collected in the next blowing cycle, the parameter adjustment unit enters the secondary judgment process. The difference between the target frequency value and the safe frequency threshold of the camera device is recorded as the secondary frequency difference. When determining the adjustment method for subsequent parameters based on the secondary frequency difference, the process includes: When the secondary frequency difference is greater than or equal to zero, the parameter adjustment unit determines to adjust the sampling frequency of the next blowing cycle to the target frequency value. When the secondary frequency difference is less than zero, the parameter adjustment unit determines to keep the acquisition frequency unchanged and enables the preprocessing correction program, selecting the first preprocessing adjustment coefficient ν1 to adjust the denoising algorithm of the image preprocessing module to the corresponding value.
[0011] Furthermore, the parameter adjustment unit determines when the copper converter blowing endpoint has not been reached based on the brightness peak and shape variation rate calculated by the endpoint judgment module, including: When the peak brightness is greater than or equal to the preset peak brightness limit and the absolute value of the shape variation rate is less than the preset variation rate threshold, the parameter adjustment unit determines that the feature extraction is incomplete and adjusts the extraction threshold of the feature extraction module for the next blowing cycle to the corresponding value according to the specific value of the peak brightness. When the peak brightness is less than the preset peak brightness limit and the absolute value of the shape variation rate is less than the preset variation rate threshold, the parameter adjustment unit determines that the image acquisition is incomplete and adjusts the light source intensity of the image acquisition module for the next blowing cycle to the corresponding value according to the specific value of the shape variation rate. When the peak brightness is greater than or equal to the preset peak brightness limit and the absolute value of the shape variation rate is greater than or equal to the preset variation rate threshold, the parameter adjustment unit determines that the preprocessing algorithm is unbalanced and corrects the preset preprocessing algorithm of the next blowing cycle to the corresponding value according to the coupling relationship between the peak brightness and the shape variation rate.
[0012] Furthermore, when the parameter adjustment unit determines the adjustment method for extracting the threshold in the next blowing cycle based on the calculated brightness peak value, it includes: When the ratio of the peak brightness to the preset peak brightness limit is greater than or equal to the first preset ratio, the parameter adjustment unit selects the first threshold coefficient κ1 to adjust the extraction threshold to the corresponding value. When the ratio of the peak brightness to the preset peak brightness limit is less than the first preset ratio, the parameter adjustment unit selects the second threshold coefficient κ2 to adjust the extraction threshold to the corresponding value.
[0013] Furthermore, when the parameter adjustment unit determines the adjustment method for the light source intensity in the next blowing cycle based on the difference between the calculated shape variation rate and the preset variation rate threshold, it includes: When the difference is greater than or equal to the preset variation rate difference, the parameter adjustment unit selects the first intensity coefficient σ1 to adjust the light source intensity to the corresponding value. When the difference is less than the preset variation rate difference, the parameter adjustment unit selects the second intensity coefficient σ2 to adjust the light source intensity to the corresponding value.
[0014] Furthermore, when the parameter adjustment unit modifies the current preprocessing algorithm, it includes: The parameter adjustment unit establishes a multivariate nonlinear regression model of the peak brightness and shape variation rate with the preprocessing algorithm, and uses the calculated peak brightness and shape variation rate as input variables to determine the target correction algorithm. The parameter adjustment unit is also used to record the difference vector between the target correction algorithm and the current blowing cycle algorithm as the algorithm correction vector, and to determine the correction method according to the modulus of the algorithm correction vector. When the modulus is greater than or equal to the preset modulus threshold, the parameter adjustment unit selects the first correction matrix N1 to correct the current algorithm. When the modulus is less than the preset modulus threshold, the parameter adjustment unit uses the second correction matrix N2 to correct the current algorithm.
[0015] Compared with the prior art, the beneficial effects of the present invention are as follows: By setting up an image acquisition module, stable acquisition and controllable illumination of furnace flame images were achieved, providing a high-quality and consistent visual signal source for subsequent feature analysis. Secondly, the synergy between the image preprocessing module and the feature extraction module effectively eliminated environmental noise interference and enhanced key flame features, avoiding image quality fluctuations and feature distortion problems caused by traditional fixed algorithms, thereby improving the repeatability and accuracy of flame feature extraction. Furthermore, in the feature extraction module, the use of a multi-dimensional feature analyzer comprehensively captured the flame's color, brightness, shape, and texture information, ensuring the integrity and representativeness of the feature data. Simultaneously, by setting up an endpoint judgment module, key blowing state parameters such as flame stability, color saturation, peak brightness, and shape variation rate can be quickly calculated based on the extracted multi-dimensional features, achieving automated and quantitative analysis from image features to process state. Finally, the central control module can compare the calculated blowing state parameters with preset standards in real time. When the judgment result does not reach the endpoint, it can automatically adjust key judgment conditions such as the image acquisition frequency, preprocessing algorithm, or feature extraction threshold for the next blowing cycle based on the degree of flame stability deviation and other parameter combinations. This mechanism establishes a closed-loop control path from image acquisition and feature analysis to condition optimization, ensuring the consistency between the adaptive optimization of the judgment process and the endpoint determination result, which helps to improve the stability and reliability of the endpoint determination of copper converter blowing.
[0016] On the other hand, this application also provides a method for determining the endpoint of copper converter blowing based on flame image feature recognition, including: Obtain the blowing state parameters of the copper converter, including flame stability, color saturation, peak brightness, and shape variation rate. The calculation of flame stability and the preset stability threshold determine whether the copper converter blowing endpoint has been reached. If the endpoint has not been reached, the image acquisition frequency, preprocessing algorithm, or feature extraction threshold of the next blowing cycle are adjusted based on the deviation of flame stability and the combined analysis results of other blowing state parameters. Attached Figure Description
[0017] Various other advantages and benefits will become apparent to those skilled in the art upon reading the following detailed description of preferred embodiments. The accompanying drawings are for illustrative purposes only and are not intended to limit the invention. Furthermore, the same reference numerals denote the same parts throughout the drawings. In the drawings: Figure 1 This is a functional block diagram of a copper converter blowing endpoint determination system based on flame image feature recognition provided in an embodiment of the present invention; Figure 2This is a flowchart illustrating a copper converter blowing endpoint determination system based on flame image feature recognition, provided in an embodiment of the present invention. Figure 3 The flowchart is provided for an embodiment of the present invention to illustrate a method for determining the end point of copper converter blowing based on flame image feature recognition. Detailed Implementation
[0018] Exemplary embodiments of the present disclosure will now be described in more detail with reference to the accompanying drawings. While exemplary embodiments of the present disclosure are shown in the drawings, it should be understood that the present disclosure may be implemented in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided to enable a more thorough understanding of the present disclosure and to fully convey the scope of the disclosure to those skilled in the art. It should be noted that, unless otherwise specified, embodiments and features in the embodiments of the present invention can be combined with each other. The present invention will now be described in detail with reference to the accompanying drawings and embodiments.
[0019] like Figures 1-2 As shown in some embodiments of this application, this embodiment provides a copper converter blowing endpoint determination system based on flame image feature recognition, including: an image acquisition module, an image preprocessing module, a feature extraction module, an endpoint determination module, and a central control module.
[0020] Specifically, the image acquisition module includes a camera device and a light source auxiliary device for real-time acquisition of flame images at the copper converter mouth; the image preprocessing module is connected to the image acquisition module and includes an image processor for denoising and enhancing the acquired flame images; the feature extraction module is connected to the image preprocessing module and includes a feature analyzer for extracting color distribution, brightness gradient, shape contour, and texture features from the flame images; the endpoint determination module is connected to the feature extraction module and is used to calculate the blowing state parameters based on the extracted flame image features, wherein the blowing state parameters include flame stability, color saturation, brightness peak, and shape variation rate; the central control module is electrically connected to the image acquisition module, the image preprocessing module, the feature extraction module, and the endpoint determination module, and is used to determine whether the blowing endpoint has been reached based on the flame stability calculated by the endpoint determination module and a preset stability threshold, and if the endpoint has not been reached, to adjust one or more of the image acquisition frequency, preprocessing algorithm, or feature extraction threshold for the next blowing cycle based on the deviation of flame stability and the combined analysis results of other blowing state parameters.
[0021] Understandably, through modular design, intelligent judgment is achieved throughout the entire process, from image acquisition, preprocessing, feature analysis to endpoint determination and result feedback. In the image acquisition module, auxiliary lighting equipment provides stable illumination, and a camera captures real-time images of the furnace flame sequence, providing a high-quality and consistent visual signal source for analysis. In the image preprocessing module, an image processor performs noise reduction and contrast enhancement on the original images, effectively eliminating interference from smoke, uneven lighting, and other factors, providing clear and standardized image data for subsequent feature extraction. The feature extraction module utilizes a feature analyzer to extract key visual features such as color, brightness, shape, and texture from the preprocessed images in multiple dimensions, comprehensively describing the dynamic and static characteristics of the flame. Based on the extracted multi-dimensional features, the endpoint determination module calculates quantitative blowing state parameters such as flame stability, color saturation, peak brightness, and shape variation rate, providing an objective basis for endpoint determination. Finally, the central control module compares the calculated flame stability with the preset threshold. When the judgment result does not reach the endpoint, it can automatically adjust the key judgment conditions such as the image acquisition frequency, preprocessing algorithm or feature extraction threshold of the next blowing cycle based on the combination analysis of the degree of deviation of flame stability and other state parameters, so as to realize the closed-loop optimization of the judgment process and improve the accuracy and consistency of the endpoint judgment.
[0022] In a specific embodiment of this application, the above steps are implemented as follows: A copper converter blowing endpoint determination system based on flame image feature recognition includes an image acquisition module, an image preprocessing module, a feature extraction module, an endpoint determination module, and a central control module. The image acquisition module includes a high dynamic range industrial camera and a near-infrared auxiliary light source, used to acquire furnace mouth flame images in real time at a frequency of 25 frames per second. The image preprocessing module uses an image processor to perform median filtering and histogram equalization on the original image. The feature extraction module uses an OpenCV-based feature analyzer to extract the HSV color histogram, brightness gradient histogram, contour shape descriptor, and LBP texture features of the flame region. Based on the extracted features, the endpoint determination module calculates flame stability, color saturation, brightness peak, and shape variation rate using a support vector machine (SVM) model. The central control module compares the calculated flame stability with preset thresholds, such as the first preset stability of 0.92 and the second preset stability of 0.85. When the result does not reach the endpoint, it can automatically adjust the image acquisition frequency, preprocessing enhancement intensity or feature extraction threshold for the next blowing cycle based on the degree of deviation and other parameters, such as brightness peak and shape variation rate, to achieve closed-loop feedback optimization.
[0023] The above scenarios are merely preferred embodiments of the present invention and are not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.
[0024] Specifically, the central control module includes: a data acquisition unit for real-time acquisition of all blowing state parameters calculated by the endpoint judgment module; a model comparison unit for comparing the flame stability acquired by the data acquisition unit with the first preset stability and the second preset stability, and generating an endpoint judgment result; and a parameter adjustment unit for optimizing and adjusting the enhancement algorithm of the image preprocessing module in the next blowing cycle based on the color saturation detection value when the model comparison unit determines that the blowing endpoint has been reached, or for correcting and adjusting the process parameters of the corresponding module based on the endpoint judgment result when the blowing endpoint has not been reached. The model comparison unit determines that the copper converter blowing endpoint has been reached under the first judgment condition; determines that the copper converter blowing endpoint requires a first-level correction under the second judgment condition; and determines that the copper converter blowing endpoint has not been reached under the third judgment condition, and determines the reason for not reaching the endpoint based on the brightness peak and shape variation rate data, including: the first judgment condition is that the flame stability is greater than or equal to the first preset stability; the second judgment condition is that the flame stability is less than the first preset stability but greater than or equal to the second preset stability; and the third judgment condition is that the flame stability is less than the second preset stability.
[0025] Understandably, the central control module intelligently controls the flame image endpoint determination process in a closed-loop manner. This module includes a data acquisition unit, a model comparison unit, and a parameter adjustment unit. The data acquisition unit acquires all refining state parameters calculated by the endpoint determination module in real time, such as flame stability, color saturation, peak brightness, and shape variation rate, providing basic data for subsequent judgment and adjustment. The model comparison unit compares the flame stability acquired by the data acquisition unit with a first preset stability and a second preset stability, generating an endpoint determination result based on preset judgment conditions. Specifically, if the flame stability is greater than or equal to the first preset stability, the refining endpoint is determined to have been reached; if the flame stability is between the first and second preset stability, a first-level correction is required; if the flame stability is lower than the second preset stability, the endpoint is determined not to have been reached, and the specific reason for not reaching the endpoint is confirmed by combining the peak brightness and shape variation rate data. Finally, the parameter adjustment unit performs corresponding adjustments based on the judgment results generated by the model comparison unit: when the blowing end point is reached, the enhancement algorithm of the image preprocessing module for the next blowing cycle can be optimized based on color saturation to further improve image quality and feature stability; when correction is needed or the end point is not reached, the judgment parameters of the corresponding module, such as image acquisition frequency, preprocessing algorithm, feature extraction threshold or light source intensity, can be corrected and adjusted to achieve closed-loop optimization control of the judgment process.
[0026] Specifically, when the parameter adjustment unit determines the optimization adjustment method of the enhancement algorithm for the next blowing cycle based on the color saturation calculated by the endpoint judgment module, it includes: when the color saturation is less than the preset lower limit of saturation, the central control module selects the first algorithm enhancement coefficient ε1 to adjust the enhancement algorithm to the corresponding value; when the color saturation is greater than the preset upper limit of saturation, the central control module selects the first algorithm reduction coefficient ε2 to adjust the enhancement algorithm to the corresponding value; when the color saturation is between the preset lower limit of saturation and the preset upper limit of saturation, the current enhancement algorithm remains unchanged.
[0027] Understandably, the parameter adjustment unit dynamically optimizes the image preprocessing stage during the judgment process. Based on the color saturation calculated in real-time by the endpoint judgment module, the parameter adjustment unit adjusts the image enhancement algorithm for the next blowing cycle, thereby optimizing the image preprocessing effect to more accurately represent the flame color characteristics. Specifically, when the color saturation is below a preset lower limit, it indicates that the current enhancement algorithm may not adequately highlight the flame's color characteristics. The central control module will select the first algorithm enhancement coefficient ε1 to increase the intensity of the enhancement algorithm to the corresponding optimized value. When the color saturation is above a preset upper limit, to avoid over-enhancement leading to color distortion or noise amplification, the central control module will select the first algorithm reduction coefficient ε2 to reduce the intensity of the enhancement algorithm to the corresponding optimized value. When the color saturation is between the preset lower and upper limits, the current enhancement algorithm remains unchanged, thus maintaining the stability of the preprocessing effect.
[0028] Specifically, when a first-level correction is required, the parameter adjustment unit records the difference between the first preset stability and the measured flame stability as the first-level stability difference, and determines the adjustment method for the acquisition frequency of the image acquisition module in the next blowing cycle based on the first-level stability difference. This includes: when the first-level stability difference is greater than or equal to the preset first-level difference threshold, the parameter adjustment unit uses the first frequency adjustment coefficient φ1 to linearly increase the acquisition frequency to the corresponding target frequency value; when the first-level stability difference is less than the preset first-level difference threshold, the parameter adjustment unit uses the second frequency adjustment coefficient φ2 to non-linearly increase the acquisition frequency to the corresponding target frequency value.
[0029] Specifically, after determining the target frequency value for the sampling frequency in the next blowing cycle, the parameter adjustment unit enters the secondary judgment process. The difference between the target frequency value and the safe frequency threshold of the camera device is recorded as the secondary frequency difference. When determining the adjustment method of subsequent parameters based on the secondary frequency difference, the parameter adjustment unit determines that: when the secondary frequency difference is greater than or equal to zero, the parameter adjustment unit determines to adjust the sampling frequency of the next blowing cycle to the target frequency value; when the secondary frequency difference is less than zero, the parameter adjustment unit determines to keep the sampling frequency unchanged and activates the preprocessing correction program, using the first preprocessing adjustment coefficient ν1 to adjust the denoising algorithm of the image preprocessing module to the corresponding value.
[0030] Understandably, the parameter adjustment unit enables intelligent, closed-loop control of the key condition—image acquisition frequency—during the judgment process. When a first-level correction is required, the parameter adjustment unit first calculates the difference between the first preset stability and the measured flame stability, recording this as the first-level stability difference. This difference reflects the degree of deviation between the judgment result and the target value, serving as a crucial basis for determining the direction of frequency optimization for the next cycle. The parameter adjustment unit selects different frequency adjustment strategies based on the first-level stability difference: when the difference is greater than or equal to a preset threshold, indicating a large deviation, the parameter adjustment unit uses a first frequency adjustment coefficient φ1 to linearly increase the acquisition frequency to the calculated target frequency value in order to obtain more continuous flame dynamic information for correction; when the difference is less than the preset threshold, indicating a small deviation, the parameter adjustment unit uses a second frequency adjustment coefficient φ2 to non-linearly increase the acquisition frequency to the target frequency value, achieving finer adjustment and avoiding data redundancy or processing burden due to excessive frequency changes. After determining the target frequency value, the parameter adjustment unit further enters the second-level judgment process to ensure the safety of the adjustment operation. This process compares the target frequency value with the safe frequency threshold of the camera device's hardware performance and calculates a secondary frequency difference. If the secondary frequency difference is greater than or equal to zero, it indicates that the target frequency is within the equipment's safe range, and the acquisition frequency for the next blowing cycle is adjusted to the target value. If the secondary frequency difference is less than zero, it indicates that the target frequency may exceed the equipment's limits. For equipment protection, the parameter adjustment unit keeps the acquisition frequency unchanged and simultaneously activates the preprocessing correction program. The strength of the denoising algorithm in the image preprocessing module is adjusted through the first preprocessing adjustment coefficient ν1 to compensate for the information loss that may be caused by frequency limitations by improving the quality of a single frame image, thereby achieving multi-dimensional parameter collaborative optimization.
[0031] In a specific embodiment of this application, the above steps are implemented as follows: A parameter adjustment unit is set in the central control module to intelligently control the acquisition frequency of the image acquisition module and the denoising algorithm of the image preprocessing module when the judgment result requires first-level correction. During a blowing process, the endpoint judgment module calculates the current flame stability as 0.87, the first preset stability as 0.92, and the second preset stability as 0.85, thus determining that first-level correction is required. The parameter adjustment unit first records the difference (0.05) between the first preset stability and the measured stability as the first-level stability difference. If the first-level stability difference is greater than or equal to the preset first-level difference threshold (e.g., 0.03), the first frequency adjustment coefficient φ1 is selected to linearly increase the acquisition frequency to the calculated target frequency value (e.g., from 25Hz to 30Hz); if the first-level stability difference is less than the preset first-level difference threshold, the second frequency adjustment coefficient φ2 is selected to non-linearly increase the acquisition frequency to the corresponding target frequency value (e.g., from 25Hz to 27Hz). After determining the target frequency value, the parameter adjustment unit enters the secondary judgment process, comparing the target frequency value with the camera's safe frequency threshold (e.g., 35Hz) and calculating the secondary frequency difference. If the secondary frequency difference is greater than or equal to zero, the acquisition frequency for the next blowing cycle is adjusted to the target frequency value; if the secondary frequency difference is less than zero, the acquisition frequency remains unchanged, and the preprocessing correction program is activated. The median filter kernel size of the image preprocessing module is adjusted to the corresponding value (e.g., increased from 3x3 to 5x5) through the first preprocessing adjustment coefficient ν1, thereby optimizing the judgment conditions while ensuring equipment safety.
[0032] As can be seen from the above embodiments, the present invention utilizes a parameter adjustment unit combined with a first-level stability difference judgment and a second-level safety frequency judgment to achieve intelligent closed-loop adjustment of the acquisition frequency and preprocessing algorithm, so that the key parameters in the flame image endpoint judgment process are dynamically optimized, ensuring the accuracy and stability of the judgment result, and improving the safety and reliability of the judgment process.
[0033] The above scenarios are merely preferred embodiments of the present invention and are not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.
[0034] Specifically, the parameter adjustment unit determines when the copper converter blowing endpoint has not been reached based on the brightness peak and shape variation rate calculated by the endpoint judgment module. This includes: when the brightness peak is greater than or equal to the preset brightness peak upper limit and the absolute value of the shape variation rate is less than the preset variation rate threshold, the parameter adjustment unit determines the cause as incomplete feature extraction and adjusts the extraction threshold of the feature extraction module for the next blowing cycle to the corresponding value based on the specific value of the brightness peak; when the brightness peak is less than the preset brightness peak upper limit and the absolute value of the shape variation rate is less than the preset variation rate threshold, the parameter adjustment unit determines the cause as incomplete image acquisition and adjusts the light source intensity of the image acquisition module for the next blowing cycle to the corresponding value based on the specific value of the shape variation rate; when the brightness peak is greater than or equal to the preset brightness peak upper limit and the absolute value of the shape variation rate is greater than or equal to the preset variation rate threshold, the parameter adjustment unit determines the cause as an unbalanced preprocessing algorithm and corrects the preset preprocessing algorithm for the next blowing cycle to the corresponding value based on the coupling relationship between the brightness peak and the shape variation rate.
[0035] Specifically, when the parameter adjustment unit determines the adjustment method for the extraction threshold in the next blowing cycle based on the calculated brightness peak value, it includes: when the ratio of the brightness peak value to the preset brightness peak value upper limit is greater than or equal to the first preset ratio, the parameter adjustment unit selects the first threshold coefficient κ1 to adjust the extraction threshold to the corresponding value; when the ratio of the brightness peak value to the preset brightness peak value upper limit is less than the first preset ratio, the parameter adjustment unit selects the second threshold coefficient κ2 to adjust the extraction threshold to the corresponding value.
[0036] Specifically, when the parameter adjustment unit determines the adjustment method for the light source intensity in the next blowing cycle based on the difference between the calculated shape variation rate and the preset variation rate threshold, the following steps are taken: when the difference is greater than or equal to the preset variation rate difference, the parameter adjustment unit selects the first intensity coefficient σ1 to adjust the light source intensity to the corresponding value; when the difference is less than the preset variation rate difference, the parameter adjustment unit selects the second intensity coefficient σ2 to adjust the light source intensity to the corresponding value.
[0037] Specifically, when the parameter adjustment unit corrects the current preprocessing algorithm, it includes: establishing a multivariate nonlinear regression model of the brightness peak and shape variation rate with the preprocessing algorithm, and using the calculated brightness peak and shape variation rate as input variables to determine the target correction algorithm; the parameter adjustment unit is also used to record the difference vector between the target correction algorithm and the current blowing cycle algorithm as the algorithm correction vector, and to determine the correction method according to the modulus of the algorithm correction vector, wherein: when the modulus is greater than or equal to the preset modulus threshold, the parameter adjustment unit selects the first correction matrix N1 to correct the current algorithm; when the modulus is less than the preset modulus threshold, the parameter adjustment unit selects the second correction matrix N2 to correct the current algorithm.
[0038] Understandably, the parameter adjustment unit enables intelligent closed-loop optimization and control of image acquisition quality, feature extraction accuracy, and preprocessing effect during the judgment process. When the brightness peak and shape variation rate calculated by the endpoint judgment module indicate that the blowing endpoint has not been reached, the parameter adjustment unit determines the cause of the anomaly based on different combinations of indicators and selects the corresponding adjustment strategy accordingly. Specifically, when the brightness peak is too high and the shape variation rate is too low, the main cause is determined to be an improper setting of the feature extraction threshold, resulting in the neglect of some weak shape change features. The parameter adjustment unit adjusts the feature extraction threshold for the next blowing cycle based on the specific value of the brightness peak. When the brightness peak is normal but the shape variation rate is too low, the main cause is determined to be insufficient light source intensity during image acquisition, resulting in an overall dark image and blurred outlines. The parameter adjustment unit enhances the auxiliary light source intensity for the next blowing cycle based on the specific value of the shape variation rate. When the brightness peak is too high and the shape variation rate is also abnormal, the main cause is determined to be an imbalance in the parameters of the preprocessing algorithm (such as enhancement or denoising), which fails to bring the image to the optimal analysis state. The parameter adjustment unit then corrects the preprocessing algorithm parameters based on the coupling relationship between the two. Secondly, regarding the feature extraction threshold adjustment, the parameter adjustment unit calculates the ratio of the peak brightness to a preset upper limit and selects different threshold coefficients (κ1 or κ2) to adjust the extraction threshold to the corresponding value, thereby ensuring the integrity of feature extraction. Regarding the light source intensity adjustment, the parameter adjustment unit selects different intensity coefficients (σ1 or σ2) to adjust the light source intensity to the corresponding value based on the difference between the shape variation rate and the preset threshold, thereby improving image acquisition quality. Finally, in terms of preprocessing algorithm correction, the parameter adjustment unit establishes a multivariate nonlinear regression model between the peak brightness and shape variation rate and the preprocessing algorithm parameters, using the calculated values as input to determine the target correction algorithm parameter set. Subsequently, the difference vector between the target algorithm parameter set and the current algorithm parameter set is recorded as the algorithm correction vector, and the correction method is determined according to its magnitude: when the magnitude is large, the first correction matrix N1 is used for a larger correction; when the magnitude is small, the second correction matrix N2 is used for fine-tuning. This method achieves precise and adaptive control of key image processing parameters.
[0039] In a specific embodiment of this application, the above steps are implemented as follows: When the endpoint determination module calculates the blowing state parameters, and the brightness peak and shape variation rate indicate that the endpoint has not been reached, the parameter adjustment unit in the central control module determines the cause of the anomaly based on different combinations of indicators, and selects the corresponding adjustment strategy accordingly. For example: 1. If the measured brightness peak is 235 (the preset upper limit is 220), and the shape variation rate is 0.10 (the preset threshold is 0.15, and the absolute value is less than the threshold), the system determines that the cause is incomplete feature extraction. The parameter adjustment unit calculates the ratio of the brightness peak to the upper limit (approximately 1.068). If this ratio is greater than or equal to the first preset ratio (e.g., 1.05), the first threshold coefficient κ1 is used to reduce the gradient threshold of the shape contour extraction from 50 to 40; if it is less than the first preset ratio, the coefficient κ2 is used to fine-tune it to 45. 2. If the measured brightness peak is 200 (below the upper limit), and the shape variation rate is 0.12 (still less than the threshold), the system determines that the cause is incomplete image acquisition. The parameter adjustment unit calculates the difference between the shape variation rate and the threshold (-0.03). If the absolute value of this difference is greater than or equal to the preset variation rate difference (e.g., 0.02), the first intensity coefficient σ1 is used to increase the auxiliary light source power by 15%; if it is less than this difference, the σ2 coefficient is used to increase it by 8%. 3. If the measured brightness peak is 230 (higher than the upper limit) and the shape variation rate is 0.18 (higher than the threshold), the system determines the cause to be an imbalance in the preprocessing algorithm. The parameter adjustment unit inputs the brightness peak and shape variation rate into the pre-trained regression model to obtain the target correction algorithm parameter set (e.g., adjusting the contrast enhancement coefficient from 1.2 to 1.1, and the Gaussian filter σ value from 1.5 to 1.8). The algorithm correction vector magnitude is calculated. If the magnitude is greater than or equal to the preset magnitude threshold, the first correction matrix N1 is used for correction; if it is less than the threshold, the second correction matrix N2 is used for fine-tuning.
[0040] Through the aforementioned dynamic closed-loop adjustment using multiple indicators and strategies, this embodiment achieves precise and coordinated control of image acquisition, preprocessing, and feature extraction during the flame image endpoint determination process. This significantly improves the accuracy and robustness of endpoint determination and reduces fluctuations in determination between furnace cycles. Simultaneously, this method can respond to determination results in real time, enabling automatic optimization and intelligent control of analysis conditions, reducing reliance on operator experience, and enhancing the automation level of the smelting process.
[0041] The above scenarios are merely preferred embodiments of the present invention and are not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.
[0042] In the above embodiments, by setting up an image acquisition module, stable acquisition of furnace flame images is achieved, providing a high-quality input source for visual analysis. Through the collaboration of the image preprocessing module and the feature extraction module, image quality is effectively improved and key features are comprehensively extracted, ensuring the accuracy and robustness of feature analysis. The endpoint judgment module realizes the quantitative conversion from image features to blowing state parameters, providing an objective basis for endpoint decision-making. Finally, the central control module compares the judgment result with preset standards and intelligently adjusts front-end parameters when the endpoint is not reached, establishing a closed-loop path of "acquisition-analysis-judgment-optimization." This ensures the adaptive optimization and continuous improvement of the entire judgment system, effectively enhancing the stability, accuracy, and reliability of copper converter blowing endpoint judgment.
[0043] In another preferred embodiment based on the above embodiments, such as Figure 3 As shown, this embodiment provides a method for determining the endpoint of copper converter blowing based on flame image feature recognition, including: Step S100: Obtain the refining state parameters of the copper converter, wherein the refining state parameters include flame stability, color saturation, brightness peak and shape variation rate. Step S200: Determine whether the copper converter blowing endpoint has been reached based on the calculated flame stability and the preset stability threshold. If the endpoint has not been reached, adjust one or more of the following parameters for the next blowing cycle: image acquisition frequency, preprocessing algorithm, or feature extraction threshold, based on the deviation of flame stability and the combined analysis results of other blowing state parameters.
[0044] It is understood that the copper converter blowing endpoint determination system and method based on flame image feature recognition in the above embodiments of the present invention have the same beneficial effects, and will not be described again.
[0045] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program goods. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program goods embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.
[0046] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program goods according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.
[0047] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.
[0048] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.
[0049] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit it. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that modifications or equivalent substitutions can still be made to the specific implementation of the present invention. Any modifications or equivalent substitutions that do not depart from the spirit and scope of the present invention should be covered within the scope of protection of the claims of the present invention.
Claims
1. A copper converter blowing endpoint determination system based on flame image feature recognition, characterized in that, include: The image acquisition module includes a camera device and a light source auxiliary device for real-time acquisition of images of the flame at the mouth of a copper converter; An image preprocessing module, connected to the image acquisition module, includes an image processor for denoising and enhancing the acquired flame images; The feature extraction module, connected to the image preprocessing module, includes a feature analyzer for extracting color distribution, brightness gradient, shape contour and texture features from flame images. The endpoint determination module is connected to the feature extraction module and is used to calculate the blowing state parameters based on the extracted flame image features. The blowing state parameters include flame stability, color saturation, brightness peak and shape variation rate. The central control module is connected to the image acquisition module, image preprocessing module, feature extraction module, and endpoint determination module. It is used to determine whether the blowing endpoint has been reached based on the flame stability calculated by the endpoint determination module and the preset stability threshold. If the endpoint has not been reached, it will adjust one or more of the following parameters for the next blowing cycle: image acquisition frequency, preprocessing algorithm, or feature extraction threshold, based on the deviation of flame stability and the combined analysis results of other blowing state parameters.
2. The copper converter blowing endpoint determination system based on flame image feature recognition as described in claim 1, characterized in that, The central control module includes: The data acquisition unit is used to acquire the values of all refining state parameters calculated by the endpoint judgment module in real time; The model comparison unit is used to compare the flame stability obtained by the data acquisition unit with the first preset stability and the second preset stability, and generate the endpoint determination result. The parameter adjustment unit is used to optimize and adjust the enhancement algorithm of the image preprocessing module in the next blowing cycle based on the color saturation detection value when the model comparison unit determines that the blowing endpoint has been reached, or to correct and adjust the process parameters of the corresponding module based on the endpoint determination result when the blowing endpoint has not been reached. The model comparison unit determines that the copper converter blowing endpoint has been reached under the first determination condition; determines that the copper converter blowing endpoint needs to be corrected at the first level under the second determination condition; and determines that the copper converter blowing endpoint has not been reached under the third determination condition, and determines the reason for not reaching the endpoint based on the data of brightness peak and shape variation rate, including: the first determination condition is that the flame stability is greater than or equal to the first preset stability. The second determination condition is that the flame stability is less than the first preset stability and greater than or equal to the second preset stability. The third criterion is that the flame stability is less than the second preset stability.
3. The copper converter blowing endpoint determination system based on flame image feature recognition as described in claim 2, characterized in that, When the parameter adjustment unit determines the optimized adjustment method for the enhancement algorithm in the next blowing cycle based on the color saturation calculated by the endpoint judgment module, it includes: When the color saturation is less than the preset lower limit of saturation, the central control module selects the first algorithm enhancement coefficient ε1 to adjust the enhancement algorithm to the corresponding value; When the color saturation is greater than the preset saturation upper limit, the central control module uses the first algorithm reduction coefficient ε2 to adjust the enhancement algorithm to the corresponding value; When the color saturation is between the preset lower limit and the preset upper limit, the current enhancement algorithm remains unchanged.
4. The copper converter blowing endpoint determination system based on flame image feature recognition as described in claim 3, characterized in that, When a first-level correction is required, the parameter adjustment unit records the difference between the first preset stability and the measured flame stability as the first-level stability difference, and determines the adjustment method for the acquisition frequency of the image acquisition module in the next blowing cycle based on the first-level stability difference, including: When the first-level stability difference is greater than or equal to the preset first-level difference threshold, the parameter adjustment unit uses the first frequency adjustment coefficient φ1 to linearly increase the acquisition frequency to the corresponding target frequency value. When the first-level stability difference is less than the preset first-level difference threshold, the parameter adjustment unit selects the second frequency adjustment coefficient φ2 to nonlinearly increase the acquisition frequency to the corresponding target frequency value.
5. The copper converter blowing endpoint determination system based on flame image feature recognition as described in claim 4, characterized in that, After determining the target frequency value for the frequency to be collected in the next blowing cycle, the parameter adjustment unit enters the secondary judgment process. The difference between the target frequency value and the safe frequency threshold of the camera device is recorded as the secondary frequency difference. When determining the adjustment method for subsequent parameters based on the secondary frequency difference, the following steps are taken: When the secondary frequency difference is greater than or equal to zero, the parameter adjustment unit determines to adjust the sampling frequency of the next blowing cycle to the target frequency value. When the secondary frequency difference is less than zero, the parameter adjustment unit determines to keep the acquisition frequency unchanged and enables the preprocessing correction program, selecting the first preprocessing adjustment coefficient ν1 to adjust the denoising algorithm of the image preprocessing module to the corresponding value.
6. The copper converter blowing endpoint determination system based on flame image feature recognition as described in claim 5, characterized in that, The parameter adjustment unit determines when the copper converter blowing endpoint has not been reached based on the brightness peak and shape variation rate calculated by the endpoint judgment module, including: When the peak brightness is greater than or equal to the preset peak brightness limit and the absolute value of the shape variation rate is less than the preset variation rate threshold, the parameter adjustment unit determines that the feature extraction is incomplete and adjusts the extraction threshold of the feature extraction module for the next blowing cycle to the corresponding value according to the specific value of the peak brightness. When the peak brightness is less than the preset peak brightness limit and the absolute value of the shape variation rate is less than the preset variation rate threshold, the parameter adjustment unit determines that the image acquisition is incomplete and adjusts the light source intensity of the image acquisition module for the next blowing cycle to the corresponding value according to the specific value of the shape variation rate. When the peak brightness is greater than or equal to the preset peak brightness limit and the absolute value of the shape variation rate is greater than or equal to the preset variation rate threshold, the parameter adjustment unit determines that the preprocessing algorithm is unbalanced and corrects the preset preprocessing algorithm of the next blowing cycle to the corresponding value according to the coupling relationship between the peak brightness and the shape variation rate.
7. The copper converter blowing endpoint determination system based on flame image feature recognition as described in claim 6, characterized in that, When the parameter adjustment unit determines the adjustment method for extracting the threshold in the next blowing cycle based on the calculated peak brightness, it includes: When the ratio of the peak brightness to the preset peak brightness limit is greater than or equal to the first preset ratio, the parameter adjustment unit selects the first threshold coefficient κ1 to adjust the extraction threshold to the corresponding value. When the ratio of the peak brightness to the preset peak brightness limit is less than the first preset ratio, the parameter adjustment unit selects the second threshold coefficient κ2 to adjust the extraction threshold to the corresponding value.
8. The copper converter blowing endpoint determination system based on flame image feature recognition as described in claim 7, characterized in that, When the parameter adjustment unit determines the adjustment method for the light source intensity in the next blowing cycle based on the difference between the calculated shape variation rate and the preset variation rate threshold, it includes: When the difference is greater than or equal to the preset variation rate difference, the parameter adjustment unit selects the first intensity coefficient σ1 to adjust the light source intensity to the corresponding value. When the difference is less than the preset variation rate difference, the parameter adjustment unit selects the second intensity coefficient σ2 to adjust the light source intensity to the corresponding value.
9. The copper converter blowing endpoint determination system based on flame image feature recognition as described in claim 8, characterized in that, When the parameter adjustment unit modifies the current preprocessing algorithm, it includes: The parameter adjustment unit establishes a multivariate nonlinear regression model of the peak brightness and shape variation rate with the preprocessing algorithm, and uses the calculated peak brightness and shape variation rate as input variables to determine the target correction algorithm; The parameter adjustment unit is also used to record the difference vector between the target correction algorithm and the current blowing cycle algorithm as the algorithm correction vector, and to determine the correction method according to the modulus of the algorithm correction vector. When the modulus is greater than or equal to the preset modulus threshold, the parameter adjustment unit selects the first correction matrix N1 to correct the current algorithm. When the modulus is less than the preset modulus threshold, the parameter adjustment unit uses the second correction matrix N2 to correct the current algorithm.
10. A method for determining the endpoint of copper converter blowing based on flame image feature recognition, applicable to the copper converter blowing endpoint determination system based on flame image feature recognition as described in any one of claims 1-9, characterized in that, include: Obtain the blowing state parameters of the copper converter, including flame stability, color saturation, peak brightness, and shape variation rate. The calculation of flame stability and the preset stability threshold determine whether the copper converter blowing endpoint has been reached. If the endpoint has not been reached, the image acquisition frequency, preprocessing algorithm, or feature extraction threshold of the next blowing cycle are adjusted based on the deviation of flame stability and the combined analysis results of other blowing state parameters.
