Converter unmanned deslagging monitoring method and device based on artificial intelligence

Abstract

The invention provides a converter unmanned deslagging monitoring method and device based on artificial intelligence. The method comprises the following steps: acquiring historical image data of a converter deslagging process; performing labeling and feature extraction on the historical image data; establishing a database based on the extracted features, and training a slag tapping monitoring model by using the database, so that the slag tapping monitoring model learns, identifies and judges normal slag tapping, abnormal slag tapping, starting of converter steel slag outflow, slag tapping ending, the position relationship between the normal converter and the slag tank and the position relationship between the abnormal converter and the slag tank; and deploying the trained slag tapping monitoring model in an actual production environment, monitoring the unmanned slag tapping process of the converter in real time, and providing a slag tapping process and an alarm prompt. According to the invention, through intelligent identification of the AI model, a more comprehensive safety monitoring means is provided, and intelligent monitoring is realized; in addition, whether the position relation between the converter and the slag ladle is normal or not can be recognized and judged, and reliable safety guarantee is provided for unmanned operation.

Description

Technical Field

[0001] The present invention relates to the field of artificial intelligence and steelmaking technology, and in particular to an artificial intelligence-based converter unmanned slag monitoring method and device. Background Art

[0002] At present, the unmanned slag discharge of the converter is implemented by laser positioning to confirm that the converter slag tank car is in the right position, and by infrared photography to confirm that the converter slag falls into the slag tank under the furnace. At present, in the field of converter slag discharge monitoring, CN 118308561 A discloses a converter slag discharge control method, which mainly determines the acquisition of real-time image data containing the slag discharge port of the target converter and the slag tank on the target slag car based on real-time image data; determines whether there is a target slag flow at the slag discharge port based on real-time image data; after the target slag flow exists at the slag discharge port, determines the actual clearance height in the slag tank based on real-time image data; determines whether the target converter stops discharging slag into the slag tank based on the actual clearance height. The invention identifies the clearance of the slag tank, stops slag discharge in time when the remaining space of the slag tank is insufficient, reduces the probability of slag overflow in the slag tank, and thus reduces the probability of splashing in the slag tank, thereby improving the safety of the slag discharge process.

[0003] However, the prior art still has the following defects:

[0004] 1. Single monitoring method: It mainly relies on laser positioning and infrared camera. When any of the systems fails or is interfered with (such as laser positioning obstruction or infrared camera failure), there is a lack of backup monitoring methods.

[0005] 2. Functional limitations: It can only measure the clearance of the slag pot and confirm the slag flow position. It cannot determine whether the slag discharge is normal, cannot identify abnormal slag discharge, and cannot fully monitor the relative position relationship between the converter and the slag pot.

[0006] 3. Low level of intelligence: It mainly relies on simple image measurement and position positioning, lacks intelligent recognition and judgment capabilities, is unable to learn and accumulate experience, and cannot adapt to complex and changing on-site environments.

[0007] 4. Insufficient safety: Unable to identify and warn of abnormal situations in a timely manner, lack of comprehensive monitoring of the slag discharge process, and unable to provide effective safety protection in the event of equipment failure. Summary of the invention

[0008] In view of this, the present invention provides an artificial intelligence-based converter unmanned slag discharge monitoring method and device to solve at least one of the above-mentioned problems.

[0009] In order to achieve the above object, the present invention adopts the following scheme:

[0010] According to a first aspect of the present invention, there is provided a method for monitoring unmanned slag tapping of a converter based on artificial intelligence, the method comprising: acquiring historical image data of a slag tapping process of a converter, the historical image data comprising normal slag tapping image data and abnormal slag tapping image data; annotating and extracting features from the historical image data, the features comprising slag flow morphology features, slag position features, relative position features between a converter and a slag pot, a slag tapping start time and a slag tapping end time; establishing a database based on the extracted features, and using the database to train a slag tapping monitoring model, so that the slag tapping monitoring model learns to identify and judge normal slag tapping, abnormal slag tapping, the start of converter slag flow, the end of slag tapping, the normal position relationship between the converter and the slag pot, and the abnormal position relationship between the converter and the slag pot; deploying the trained slag tapping monitoring model into an actual production environment, monitoring the unmanned slag tapping process of the converter in real time, and providing the slag tapping process and alarm prompts.

[0011] As an embodiment of the present invention, the prompts of the slag discharge process include: a slag discharge start stage and a slag discharge end stage; the alarm prompts include: a slag overflow risk alarm.

[0012] As an embodiment of the present invention, the above method also includes: based on the normal slag discharge image data in the historical image data, using a single-classification support vector machine algorithm to train an abnormality recognition model to identify abnormal slag discharge conditions that do not appear in the abnormal slag discharge image data.

[0013] As an embodiment of the present invention, the above-mentioned single-classification support vector machine algorithm adopts a radial basis kernel function, and optimizes the kernel parameter γ and the penalty parameter ν by a grid search or cross-validation method to determine the optimal anomaly recognition model.

[0014] As an embodiment of the present invention, the above method also includes: obtaining historical slag discharge data of the converter, the historical slag discharge data including slag discharge time, slag discharge amount, converter tilting angle, slag pot type, and slag pot initial weight; performing data cleaning and preprocessing on the historical slag discharge data; using the historical slag discharge data that has undergone data cleaning and preprocessing to train a slag discharge time and slag discharge amount prediction model; deploying the slag discharge time and slag discharge amount prediction model in an actual production environment, and optimizing production scheduling in combination with the slag discharge monitoring model.

[0015] As an embodiment of the present invention, the above-mentioned slag discharge time and slag discharge amount prediction model combined with the slag discharge monitoring model optimizes production scheduling, including: in the process of the slag discharge monitoring model monitoring the slag discharge process, if the slag discharge amount predicted by the slag discharge time and slag discharge amount prediction model exceeds a preset percentage of the slag pot capacity, the slag discharge monitoring model issues an early warning to prompt the operator to replace the slag pot or adjust the slag discharge speed to avoid slag pot overflow; if the slag discharge time predicted by the slag discharge time and slag discharge amount prediction model exceeds a preset threshold, the slag discharge monitoring model issues an early warning to prompt the operator to adjust the production scheduling plan; the slag discharge time and slag discharge amount prediction model dynamically adjusts the positional relationship between the two according to the predicted slag discharge amount and slag discharge time, combined with the relative position of the converter and the slag pot monitored in real time by the slag discharge monitoring model, to ensure that the steel slag falls into the slag pot to avoid abnormal slag discharge; according to the slag discharge time and slag discharge amount prediction model, according to the predicted slag discharge amount and slag discharge time, arrange the appropriate number and type of slag pot trucks to the designated location in advance to avoid the slag pot truck waiting time too long or insufficient resources, thereby improving production efficiency.

[0016] According to a second aspect of the present invention, there is provided an unmanned slag tapping monitoring device for a converter based on artificial intelligence, the device comprising: acquiring historical image data of a slag tapping process of a converter, the historical image data comprising normal slag tapping image data and abnormal slag tapping image data; annotating and extracting features of the historical image data, the features comprising slag flow morphology features, slag position features, converter and slag pot relative position features, slag tapping start time and slag tapping end time; establishing a database based on the extracted features, and using the database to train a slag tapping monitoring model, so that the slag tapping monitoring model learns to identify and judge normal slag tapping, abnormal slag tapping, the start of converter slag flow, the end of slag tapping, the normal position relationship between the converter and the slag pot, and the abnormal position relationship between the converter and the slag pot; deploying the trained slag tapping monitoring model into an actual production environment, real-time monitoring of the unmanned slag tapping process of the converter is performed, and the slag tapping process and alarm prompts are provided.

[0017] As an embodiment of the present invention, the prompts of the slag discharge process include: a slag discharge start stage and a slag discharge end stage; the alarm prompts include: a slag overflow risk alarm.

[0018] As an embodiment of the present invention, the above-mentioned device also includes: an abnormality recognition unit, which is used to train an abnormality recognition model based on the normal slag discharge image data in the historical image data using a single-classification support vector machine algorithm to identify abnormal slag discharge conditions that do not appear in the abnormal slag discharge image data.

[0019] As an embodiment of the present invention, the above-mentioned single-classification support vector machine algorithm adopts a radial basis kernel function, and optimizes the kernel parameter γ and the penalty parameter ν by a grid search or cross-validation method to determine the optimal anomaly recognition model.

[0020] As an embodiment of the present invention, the above-mentioned device also includes: a slag data acquisition unit, used to acquire historical slag data of the converter slag, the historical slag data including slag time, slag amount, converter tilting angle, slag pot type, and slag pot initial weight; a preprocessing unit, used to perform data cleaning and preprocessing on the historical slag data; a prediction model training unit, used to train a slag time and slag amount prediction model using the historical slag data that has been cleaned and preprocessed; a scheduling optimization unit, used to deploy the slag time and slag amount prediction model into an actual production environment, and optimize production scheduling in combination with the slag monitoring model.

[0021] As an embodiment of the present invention, the scheduling optimization unit optimizes production scheduling using the slag discharge time and slag discharge amount prediction model in combination with the slag discharge monitoring model, including: in the process of the slag discharge monitoring model monitoring the slag discharge process, if the slag discharge amount predicted by the slag discharge time and slag discharge amount prediction model exceeds a preset percentage of the slag pot capacity, the slag discharge monitoring model issues an early warning to prompt the operator to replace the slag pot or adjust the slag discharge speed to avoid slag pot overflow; if the slag discharge time predicted by the slag discharge time and slag discharge amount prediction model exceeds a preset threshold, the slag discharge monitoring model issues an early warning to prompt the operator to adjust the production scheduling plan; the slag discharge time and slag discharge amount prediction model dynamically adjusts the positional relationship between the converter and the slag pot according to the predicted slag discharge amount and slag discharge time, combined with the relative position of the converter and the slag pot monitored in real time by the slag discharge monitoring model, to ensure that the steel slag falls into the slag pot to avoid abnormal slag discharge; according to the slag discharge time and slag discharge amount prediction model, according to the predicted slag discharge amount and slag discharge time, arrange the appropriate number and type of slag pot trucks to the designated location in advance to avoid the slag pot truck waiting time too long or insufficient resources, thereby improving production efficiency.

[0022] According to a third aspect of the present invention, there is provided an electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the steps of the above method when executing the computer program.

[0023] According to a fourth aspect of the present invention, there is provided a computer-readable storage medium having a computer program stored thereon, wherein the computer program implements the steps of the above method when executed by a processor.

[0024] According to a fifth aspect of the present invention, there is provided a computer program product, comprising a computer program / instructions, which implement the steps of the above method when executed by a processor.

[0025] It can be seen from the above technical scheme that the artificial intelligence-based converter unmanned slag discharge monitoring method and device provided by the present invention, through the intelligent recognition of the AI ​​model, provides a more comprehensive safety monitoring means, realizes intelligent monitoring, and breaks through the limitation of the prior art that can only perform simple image measurement and position positioning; in addition, through AI model training, the system of the present invention has similar manual recognition and judgment capabilities, and can intelligently identify normal / abnormal slag discharge, the beginning of converter slag flow and the end of slag discharge, etc., and can also identify and judge whether the positional relationship between the converter and the slag pot is normal, and can promptly discover and warn of abnormal situations, providing reliable safety guarantees for unmanned operation. BRIEF DESCRIPTION OF THE DRAWINGS

[0026] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the prior art descriptions. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work. In the drawings:

[0027] Figure 1 It is a flow chart of an artificial intelligence-based converter unmanned slag monitoring method provided in an embodiment of the present application;

[0028] Figure 2 It is a flow chart of an artificial intelligence-based converter unmanned slag monitoring method provided by another embodiment of the present application;

[0029] Figure 3 It is a flow chart of an artificial intelligence-based converter unmanned slag monitoring method provided by another embodiment of the present application;

[0030] Figure 4 It is a flow chart of optimizing production scheduling by combining a slag discharge monitoring model provided in an embodiment of the present application with a slag discharge time and slag discharge amount prediction model;

[0031] Figure 5 It is a structural schematic diagram of an unmanned converter slag monitoring device based on artificial intelligence provided in an embodiment of the present application;

[0032] Figure 6 It is a schematic diagram of an electronic device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0033] To make the purpose, technical solution and advantages of the embodiments of the present invention more clear, the embodiments of the present invention are further described in detail below in conjunction with the accompanying drawings. Here, the exemplary embodiments of the present invention and their descriptions are used to explain the present invention, but are not intended to limit the present invention.

[0034] like Figure 1 The figure shows a flow chart of an artificial intelligence-based converter unmanned slag monitoring method provided in an embodiment of the present application, the method comprising the following steps:

[0035] Step S101: acquiring historical image data of a converter slag tapping process, wherein the historical image data includes normal slag tapping image data and abnormal slag tapping image data.

[0036] The historical image data here mainly use a large number of videos of the converter slag discharge process recorded by high-definition cameras, and then extract image data from the video. The historical image data collected here should include two categories: normal slag discharge image data and abnormal slag discharge image data. The normal slag discharge image data includes the image of the converter slag being completely mixed into the slag pot, only a small amount of converter slag splashing in the air, the converter and the slag pot maintaining the correct relative position relationship, and the slag flow morphology being stable and the landing point being accurate; the abnormal slag discharge image data includes the image of abnormal conditions such as the slag falling outside the slag pot, the incorrect relative position relationship between the converter and the slag pot, the abnormal slag flow morphology, and the slag pot overflowing.

[0037] In order to ensure the training effect of the subsequent slag monitoring model, this step requires the collection of sufficient data samples, covering all possible normal and abnormal situations. At the same time, the collected image data is stored according to normal and abnormal classification for subsequent annotation and feature extraction.

[0038] Step S102: annotating and extracting features from the historical image data, wherein the features include slag flow morphology features, slag position features, relative position features between the converter and the slag pot, slag tapping start time and slag tapping end time.

[0039] Here, the annotation of historical image data can be done manually or by a semi-automatic annotation method that uses a pre-trained model for preliminary annotation and then manual correction. Because there is currently no mature artificial intelligence application in the field of converter unmanned slag monitoring, and for things like "normal slag" and "abnormal slag", certain professional knowledge and judgment are required. Therefore, using manual annotation or manual assisted annotation to annotate training data is a reliable method to ensure the quality of annotation. In addition to the labels for "normal slag discharge" and "abnormal slag discharge", you can also mark: "Labels related to slag flow morphology characteristics", such as whether the slag flow is stable, whether there is splashing in the slag flow, and the thickness of the slag flow; "Labels related to slag position characteristics", such as the location of the slag landing point (whether it falls into the slag pot), the range of slag splashing, and the height of the slag liquid level in the slag pot; "Labels related to the relative position characteristics of the converter and the slag pot", such as the horizontal distance between the converter and the slag pot, the vertical height difference between the converter and the slag pot, the tilting angle of the converter, and whether the position of the slag pot is correct; "Labels related to time nodes", such as the start time of slag discharge, the end time of slag discharge, and the duration of slag discharge; "Labels related to slag pot status", such as whether there is a risk of overflow in the slag pot and the remaining capacity status of the slag pot.

[0040] Correspondingly, the feature extraction in this step refers to extracting the following features from the annotated historical image data:

[0041] "Slag flow morphological characteristics": the continuity and stability of the slag flow, the shape and trajectory of the slag flow, the degree of splashing of the slag flow, and the changes in the thickness of the slag flow.

[0042] "Slag position characteristics": the coordinate position of the slag landing point, the deviation distance between the slag and the center of the slag pot, the spatial distribution range of the slag splashing, and the accumulation height of the slag in the slag pot.

[0043] "Relative position characteristics of the converter and slag pot": the horizontal distance between the converter slag outlet and the center of the slag pot, the vertical height difference between the converter slag outlet and the upper edge of the slag pot, the tilting angle parameters of the converter, and the spatial position coordinates of the slag pot.

[0044] "Time characteristics": the time characteristics when the slag starts to flow out, the time characteristics when the slag discharge ends, and the duration characteristics of the entire slag discharge process.

[0045] The extraction of these features is crucial for the subsequent training of the slag monitoring model, because they constitute the basic data for model recognition and judgment. Through these features, the slag monitoring model can learn the difference between normal and abnormal slag, as well as the correctness of the position relationship between the converter and the slag pot, thereby realizing accurate monitoring and early warning functions in practical applications.

[0046] Step S103: A database is established based on the extracted features, and a slag monitoring model is trained using the database, so that the slag monitoring model learns to identify and judge normal slag, abnormal slag, the beginning of converter slag outflow, the end of slag outflow, the normal positional relationship between the converter and the slag pot, and the abnormal positional relationship between the converter and the slag pot.

[0047] The extracted feature data and corresponding image labels are stored in the database to form a training data set, and then a suitable AI model, such as a convolutional neural network (CNN), a recurrent neural network (RNN), or a Transformer, is selected for training the slag monitoring model. During the model training process, the performance of the model is continuously optimized by adjusting model parameters, increasing training data, etc., to improve the accuracy of identifying abnormal situations.

[0048] Through the above training, the slag monitoring model of the present application can learn the relationship between different features and different slag conditions. In addition to giving the judgment of "normal slag discharge / abnormal slag discharge", it can also give various situations corresponding to normal slag discharge or abnormal slag discharge. For example, when the slag flow splashes to a large extent and over a wide range, it can correspond to "abnormal slag discharge-serious splashing"; when the tilting angle of the converter is too large or too small, it can correspond to "abnormal slag discharge-abnormal inclination angle"; when the slag pot is not aligned with the slag discharge port of the converter, it can correspond to "abnormal slag discharge-position deviation"; when the slag flow is intermittent or unstable, it can correspond to "abnormal slag discharge-abnormal slag flow"; when the liquid level of the slag pot is about to reach the upper limit, it can correspond to "abnormal slag discharge-overflow risk"; when the horizontal distance or vertical height difference between the converter and the slag pot exceeds the safety range, it can correspond to "abnormal slag discharge-abnormal spacing" and so on.

[0049] Step S104: deploy the trained slag discharge monitoring model into the actual production environment, monitor the unmanned slag discharge process of the converter in real time and provide slag discharge process and alarm prompts.

[0050] The trained slag monitoring model is deployed to the hardware equipment in the actual production environment, such as industrial computers or embedded devices. At the same time, the image data of the converter slag process is collected in real time using a high-definition camera and input into the deployed slag monitoring model for analysis. Finally, the key parameters of the slag process, such as slag state parameters, position relationship parameters, slag pot state parameters, etc., can be displayed in real time on the system interface to facilitate operator monitoring.

[0051] Preferably, the prompts of the slag discharge process include: the slag discharge start stage and the slag discharge end stage; and the alarm prompts include: slag overflow risk alarm. When the slag discharge monitoring model detects an abnormal situation, it immediately sends out an audible and visual alarm signal, and displays information such as the abnormal type, hazard level, and handling suggestions to remind the operator to take timely measures. For example, if the slag pot position deviation is detected, an alarm of "abnormal slag pot position" is issued, and it is recommended to "correct the slag pot position immediately."

[0052] It can be seen from the above technical scheme that the artificial intelligence-based converter unmanned slag discharge monitoring method provided by the present invention provides a more comprehensive safety monitoring means through the intelligent recognition of the AI ​​model, realizes intelligent monitoring, and breaks through the limitation of the prior art that can only perform simple image measurement and position positioning; in addition, through AI model training, the system of the present invention has similar manual recognition and judgment capabilities, and can intelligently identify various states such as normal / abnormal slag discharge, the beginning of converter slag flow and the end of slag discharge, and can also identify and judge whether the positional relationship between the converter and the slag pot is normal, and can promptly discover and warn of abnormal situations, providing reliable safety guarantees for unmanned operation.

[0053] The above-mentioned converter unmanned slag discharge monitoring method can detect abnormal situations in the historical image data during the slag discharge process when the amount of training data is sufficient. However, when it is difficult to collect enough abnormal image data samples, the above-mentioned slag discharge monitoring model may miss the situation. In addition, since the converter unmanned slag discharge process itself has a wide variety of abnormal situations, in order to increase the reliability of the final monitoring results, such as Figure 2 As shown, the artificial intelligence-based converter unmanned slag monitoring method of the present application may include the following steps:

[0054] Step S201: Acquire historical image data of a converter slag tapping process, wherein the historical image data includes normal slag tapping image data and abnormal slag tapping image data.

[0055] Step S202: annotating and extracting features from the historical image data, wherein the features include slag flow morphology features, slag position features, relative position features between the converter and the slag pot, slag tapping start time and slag tapping end time.

[0056] Step S203: A database is established based on the extracted features, and a slag monitoring model is trained using the database, so that the slag monitoring model learns to identify and judge normal slag, abnormal slag, the beginning of converter slag outflow, the end of slag discharge, the normal positional relationship between the converter and the slag pot, and the abnormal positional relationship between the converter and the slag pot.

[0057] The above steps S201-S203 are the same as steps S101-S103 in the aforementioned embodiment, and will not be further described here.

[0058] Step S204: Based on the normal slag discharge image data in the historical image data, a single-class support vector machine algorithm (One-Class SVM) is used to train an abnormality recognition model to identify abnormal slag discharge situations that do not appear in the abnormal slag discharge image data.

[0059] This step is performed simultaneously with S202 and S203. The difference from step S202 and step S203 is that this step only uses the normal slag discharge image data in the historical image data. The purpose of this step is to identify abnormal slag discharge situations that have not appeared in the abnormal slag discharge image data.

[0060] Single-class support vector machine is an anomaly detection algorithm. It does not identify anomalies by learning abnormal samples, but by learning the characteristics of normal slag data to build a model that describes the normal slag state. This model can be regarded as a boundary surrounding normal data points. During the training process, the single-class support vector machine algorithm will try to find a hyperplane that includes as many normal slag data points as possible and minimizes the area outside the hyperplane. This hyperplane defines the boundary of "normal" slag. When new slag data arrives, if the data point falls within the hyperplane (that is, within the boundary), it is considered to be normal slag; if the data point falls outside the hyperplane, it is considered to be abnormal slag, even if this abnormal situation has never appeared in the training data. Therefore, the single-class support vector machine can be trained to identify abnormal situations based only on normal slag data.

[0061] Further preferably, the above-mentioned single-classification support vector machine algorithm adopts a radial basis kernel function, and optimizes the kernel parameter γ and the penalty parameter ν by a grid search or cross-validation method to determine the optimal anomaly recognition model.

[0062] Single-class support vector machines can use different kernel functions to map data to high-dimensional space. The RBF kernel function is a commonly used kernel function that can effectively process nonlinearly separable data. The reason for selecting the RBF kernel function in this embodiment is that it has good generalization ability and flexibility and can adapt to various data distributions. The formula of the RBF kernel function is as follows:

[0063] K(x,y)=exp(-γ*||xy||2)

[0064] Where x and y are two data points, ||xy||2 is the square of their Euclidean distance, and γ is the kernel parameter that controls the width of the RBF kernel function.

[0065] Kernel parameter γ (Gamma): The γ value determines the influence range of the RBF kernel function. A smaller γ value indicates a larger influence range, and the model will be smoother and prone to underfitting; a larger γ value indicates a smaller influence range, and the model will be more complex and prone to overfitting. Therefore, choosing an appropriate γ value is crucial.

[0066] Penalty parameter ν (Nu): ν controls the sensitivity of the model to outliers. It indicates the upper limit of the proportion of training data that is allowed to be considered as outliers, and also indicates the lower limit of the proportion of support vectors. A smaller ν value means that the model is more sensitive to outliers and is prone to misclassifying normal data points as abnormal; a larger ν value means that the model is more tolerant to outliers and is prone to misclassifying abnormal data points as normal.

[0067] The grid search in this step is a hyperparameter optimization method that finds the best parameter value by trying different parameter combinations. For OC-SVM, you can use grid search to try different γ and ν values ​​and select the best combination based on the performance of the model.

[0068] Cross-validation is a model evaluation method that divides the data set into multiple subsets, uses one of the subsets as the test set and the remaining subsets as the training set in turn, and the final performance indicator is the average of multiple test results. Cross-validation can more accurately evaluate the generalization ability of the model and avoid overfitting.

[0069] Specifically, the above optimization process is as follows:

[0070] 1) Select RBF kernel function: First, determine to use RBF kernel function as the kernel function of One-Class SVM.

[0071] 2) Define parameter range: Define a reasonable search range for γ and ν. For example, you can set the range of γ to [0.001, 1000] and the range of ν to [0.01, 0.5].

[0072] 3) Grid search or cross validation: Use grid search or cross validation methods to try different combinations of γ and ν within the defined parameter range. For each combination, train an OC-SVM model and evaluate its performance using cross validation.

[0073] 4) Selecting the best parameters: Select the best performing values ​​of γ and ν and use these values ​​to train the final anomaly recognition model.

[0074] Through this optimization method, the optimal kernel parameter γ and penalty parameter ν can be found, thereby improving the accuracy and generalization ability of the One-ClassSVM model, enabling it to more effectively identify abnormal slag discharge situations that do not appear in the abnormal slag discharge image data.

[0075] Step S205: deploy the trained slag discharge monitoring model and the abnormality recognition model into the actual production environment, perform real-time monitoring on the unmanned slag discharge process, and provide slag discharge process and alarm prompts.

[0076] Different from the above step S204, this step uses both the slag discharge monitoring model and the abnormality recognition model to monitor the unmanned slag discharge process in real time and provide slag discharge process and alarm prompts. This involves the comprehensive use of two models. In this embodiment, the master-slave collaboration mode can be used for monitoring:

[0077] Main model (slag monitoring model): This model is trained based on normal and abnormal slag data and is mainly responsible for identifying known normal and abnormal slag conditions. It can classify the current slag condition and determine whether an alarm is needed based on the preset threshold. For example, if the deviation of the slag pot position exceeds the threshold, an alarm of "abnormal slag - abnormal slag pot position" is issued.

[0078] Auxiliary model (abnormal identification model - OC-SVM): This model is trained only on normal slag discharge data and is mainly responsible for identifying unknown abnormal situations, that is, those abnormal situations that have not appeared in the training data. It learns the normal slag discharge pattern and determines whether the current slag discharge situation deviates from the normal pattern. If the degree of deviation exceeds a certain threshold, an "unknown abnormality" alarm is issued.

[0079] Collaborative work: The two models analyze the real-time camera images simultaneously and perform collaborative analysis according to the following three situations:

[0080] Case 1: The main model identifies a known anomaly. If the main model identifies a known anomaly, such as "abnormal slag discharge - abnormal slag pot position", it will directly issue a corresponding alarm and take corresponding measures according to the preset strategy, such as stopping the slag discharge operation. At this time, the output of the auxiliary model can be used as a reference, but it does not play a decisive role.

[0081] Case 2: The primary model does not identify an abnormality, but the auxiliary model does. If the primary model believes that the current slag discharge is normal, but the auxiliary model identifies a large deviation from the normal pattern, an "unknown abnormality" alarm is issued to prompt the operator to pay attention and conduct further inspections. This can serve as an additional safety guarantee to prevent unknown risks.

[0082] Case 3: Both models do not identify anomalies. If both models believe that the current slag discharge is normal, the slag discharge operation will continue.

[0083] From the above, it can be seen that this collaborative working mode can combine the advantages of the two models to improve the reliability and safety of the system. The main model is responsible for identifying known anomalies, and the auxiliary model is responsible for identifying unknown anomalies, thereby covering a more comprehensive range of abnormal situations, realizing comprehensive monitoring of the unmanned slag discharge process, and ultimately promoting the popularization and application of unmanned slag discharge technology for converters.

[0084] The above embodiments discuss the method of using the slag monitoring model and the abnormality recognition model for monitoring. However, in order to enable the monitoring method to help optimize production scheduling, ensure the rational allocation of resources, reduce waiting time, and improve overall production efficiency. The present application embodiment further considers the solution of multi-model fusion, combining the slag monitoring model with the slag time and slag amount prediction model. Specifically, Figure 3 As shown, the unmanned converter slag monitoring method based on artificial intelligence in this embodiment may include the following steps:

[0085] Step S301: acquiring historical image data of a converter slag tapping process, wherein the historical image data includes normal slag tapping image data and abnormal slag tapping image data.

[0086] Step S302: annotating and extracting features from the historical image data, wherein the features include slag flow morphology features, slag position features, relative position features between the converter and the slag pot, slag tapping start time and slag tapping end time.

[0087] Step S303: A database is established based on the extracted features, and a slag monitoring model is trained using the database, so that the slag monitoring model learns to identify and judge normal slag, abnormal slag, the beginning of converter slag outflow, the end of slag discharge, the normal positional relationship between the converter and the slag pot, and the abnormal positional relationship between the converter and the slag pot.

[0088] The above steps S301-S303 are the same as steps S101-S103 in the above embodiment, and will not be further described here.

[0089] Step S304: Acquire historical slag tapping data of the converter, wherein the historical slag tapping data includes slag tapping time, slag tapping amount, converter tilting angle, slag pot type, and slag pot initial weight.

[0090] These historical data are not the image data in the above step S301, but a series of parameter data in the system. By learning these data, the prediction model can infer the slag discharge time and slag discharge amount of this slag discharge based on these parameter data.

[0091] Step S305: cleaning and preprocessing the historical slag discharge data.

[0092] The data cleaning in this step includes: removing outliers and erroneous data, processing missing values, and eliminating unreasonable data records; while data preprocessing includes: data format unification, normalization, feature extraction and conversion, etc.

[0093] Step S306: using the historical slag discharge data that has undergone data cleaning and preprocessing to train a slag discharge time and slag discharge amount prediction model.

[0094] In this embodiment, random forest regression is selected for training. On the one hand, it can handle multiple input features (slag discharge time, converter tilting angle, slag pot type, etc.) and is insensitive to abnormal values. On the other hand, it can also handle nonlinear relationships and evaluate feature importance. For predicting continuous values ​​such as slag discharge, it can complete the task well.

[0095] During model training, the historical slag discharge data is divided into data sets: the data set is divided into training set, validation set and test set. Then the training set is used to train the slag discharge time and slag discharge amount prediction model, and the validation set is used to tune the model and select the best model parameters and hyperparameters.

[0096] Step S307: deploy the trained slag discharge monitoring model into the actual production environment, monitor the unmanned slag discharge process of the converter in real time, and provide slag discharge process and alarm prompts.

[0097] This step is the same as the above step S104 and will not be described in detail here.

[0098] Step S308: deploying the slag discharge time and slag discharge amount prediction model into the actual production environment, and optimizing production scheduling in combination with the slag discharge monitoring model.

[0099] and Figure 1 The difference from the corresponding embodiment is that the slag monitoring model in this embodiment can not only independently complete the real-time monitoring of the unmanned slag discharging process of the converter and provide slag discharging process and alarm prompts, but also optimize production scheduling in combination with the slag discharging time and slag discharging amount prediction model.

[0100] Preferably, Figure 4 As shown, this step may further include the following sub-steps:

[0101] Step S3081: During the process of the slag discharge monitoring model monitoring the slag discharge process, if the slag discharge amount predicted by the slag discharge time and slag discharge amount prediction model exceeds the preset percentage of the slag pot capacity, the slag discharge monitoring model will issue an early warning to prompt the operator to replace the slag pot or adjust the slag discharge speed to avoid slag pot overflow; if the slag discharge time predicted by the slag discharge time and slag discharge amount prediction model exceeds the preset threshold, the slag discharge monitoring model will issue an early warning to prompt the operator to adjust the production scheduling plan.

[0102] Step S3082: According to the predicted slag discharge amount and slag discharge time based on the slag discharge time and slag discharge prediction model, combined with the relative position of the converter and the slag pot monitored in real time by the slag discharge monitoring model, the positional relationship between the two is dynamically adjusted to ensure that the slag falls into the slag pot and avoid abnormal slag discharge.

[0103] Step S3083: According to the slag discharge time and slag discharge amount prediction model and the predicted slag discharge amount and slag discharge time, arrange the appropriate number and type of slag tank trucks to the designated location in advance to avoid long waiting time or insufficient resources of the slag tank trucks, thereby improving production efficiency.

[0104] As can be seen from the above, the artificial intelligence-based converter unmanned slag discharge monitoring method provided in this embodiment optimizes the production scheduling of the converter unmanned slag discharge process by combining the slag discharge monitoring model and the slag discharge time and slag discharge amount prediction model, and brings the following beneficial effects: through the combination of real-time monitoring and prediction models, the system can issue an early warning when the slag discharge amount is close to the slag pot capacity or the slag discharge time exceeds the preset threshold. This early warning mechanism can effectively avoid the safety hazards caused by slag pot overflow and improper production scheduling. The slag discharge time and slag discharge amount information provided by the prediction model makes the production scheduling more accurate and efficient. By dynamically adjusting the positional relationship between the converter and the slag pot, it is ensured that the steel slag falls accurately into the slag pot and the occurrence of abnormal slag discharge is reduced. By arranging the appropriate number and type of slag pot trucks to the designated location in advance, the waiting time and resource shortage of the slag pot trucks are reduced, thereby improving the overall production efficiency. The system can automatically monitor and adjust the slag discharge process, reduce the dependence on manual operation, and improve the degree of unmanned operation.

[0105] In general, this embodiment not only improves the safety and efficiency of the converter slag discharge process through the application of AI technology, but also provides a more intelligent and reliable solution for unmanned operation.

[0106] like Figure 5 The structure diagram of an artificial intelligence-based converter unmanned slag monitoring device provided in an embodiment of the present application is shown, and the device includes: a historical data acquisition unit 510, a feature extraction unit 520, a model training unit 530 and a model deployment unit 540, which are sequentially connected. Among them:

[0107] The historical data acquisition unit 510 is used to acquire historical image data of the converter slag tapping process, wherein the historical image data includes normal slag tapping image data and abnormal slag tapping image data.

[0108] The feature extraction unit 520 is used to annotate and extract features from the historical image data, wherein the features include slag flow morphology features, slag position features, relative position features between the converter and the slag pot, slag tapping start time and slag tapping end time.

[0109] The model training unit 530 is used to establish a database based on the extracted features, and use the database to train a slag monitoring model, so that the slag monitoring model learns to identify and judge normal slag, abnormal slag, the beginning of the flow of converter slag, the end of slag, the normal position relationship between the converter and the slag pot, and the abnormal position relationship between the converter and the slag pot.

[0110] The model deployment unit 540 is used to deploy the trained slag discharge monitoring model into the actual production environment, to monitor the unmanned slag discharge process of the converter in real time and to provide slag discharge process and alarm prompts.

[0111] Preferably, the prompts of the slag discharge process include: a slag discharge start stage and a slag discharge end stage; and the warning prompts include: a slag overflow risk warning.

[0112] Preferably, the above-mentioned device also includes: an abnormality recognition unit, which is used to train an abnormality recognition model based on the normal slag discharge image data in the historical image data using a single-classification support vector machine algorithm to identify abnormal slag discharge conditions that do not appear in the abnormal slag discharge image data.

[0113] Preferably, the above-mentioned single-classification support vector machine algorithm adopts a radial basis kernel function, and optimizes the kernel parameter γ and the penalty parameter ν by a grid search or cross-validation method to determine the optimal anomaly recognition model.

[0114] Preferably, the above-mentioned device also includes: a slag data acquisition unit, used to acquire historical slag data of the converter slag, the historical slag data including slag time, slag amount, converter tilting angle, slag pot type, and slag pot initial weight; a preprocessing unit, used to perform data cleaning and preprocessing on the historical slag data; a prediction model training unit, used to train a slag time and slag amount prediction model using the historical slag data that has been cleaned and preprocessed; a scheduling optimization unit, used to deploy the slag time and slag amount prediction model in an actual production environment, and optimize production scheduling in combination with the slag monitoring model.

[0115] Preferably, the above-mentioned scheduling optimization unit optimizes production scheduling by using the slag discharge time and slag discharge amount prediction model in combination with the slag discharge monitoring model, including: in the process of the slag discharge monitoring model monitoring the slag discharge process, if the slag discharge amount predicted by the slag discharge time and slag discharge amount prediction model exceeds a preset percentage of the slag pot capacity, the slag discharge monitoring model issues an early warning to prompt the operator to replace the slag pot or adjust the slag discharge speed to avoid slag pot overflow; if the slag discharge time predicted by the slag discharge time and slag discharge amount prediction model exceeds a preset threshold, the slag discharge monitoring model issues an early warning to prompt the operator to adjust the production scheduling plan; the slag discharge time and slag discharge amount prediction model dynamically adjusts the positional relationship between the two according to the predicted slag discharge amount and slag discharge time, combined with the relative position of the converter and the slag pot monitored in real time by the slag discharge monitoring model, to ensure that the steel slag falls into the slag pot and avoid abnormal slag discharge; according to the slag discharge time and slag discharge amount prediction model, according to the predicted slag discharge amount and slag discharge time, arrange the appropriate number and type of slag pot trucks to the designated location in advance to avoid the slag pot truck waiting time too long or insufficient resources, thereby improving production efficiency.

[0116] The detailed description of each of the above units can be found in the corresponding description in the aforementioned method embodiment, which will not be further elaborated here.

[0117] It can be seen from the above technical scheme that the artificial intelligence-based converter unmanned slag discharge monitoring device provided by the present invention provides a more comprehensive safety monitoring means through the intelligent recognition of the AI ​​model, realizes intelligent monitoring, and breaks through the limitation of the prior art that can only perform simple image measurement and position positioning; in addition, through AI model training, the system of the present invention has similar manual recognition and judgment capabilities, and can intelligently identify normal / abnormal slag discharge, the beginning of converter slag flow and the end of slag discharge, etc., and can also identify and judge whether the positional relationship between the converter and the slag pot is normal, and can promptly discover and warn of abnormal situations, providing reliable safety guarantees for unmanned operation.

[0118] Figure 6 is a schematic diagram of an electronic device provided by an embodiment of the present invention. Figure 6 The electronic device shown is a general data processing device, which includes a general computer hardware structure, which at least includes a processor 801 and a memory 802. The processor 801 and the memory 802 are connected via a bus 803. The memory 802 is suitable for storing one or more instructions or programs executable by the processor 801. The one or more instructions or programs are executed by the processor 801 to implement the steps in the above-mentioned artificial intelligence-based converter unmanned slag monitoring method.

[0119] The processor 801 may be an independent microprocessor or a collection of one or more microprocessors. Thus, the processor 801 executes the commands stored in the memory 802 to execute the method flow of the embodiment of the present invention as described above to realize the processing of data and the control of other devices. The bus 803 connects the above-mentioned multiple components together, and at the same time connects the above-mentioned components to the display controller 804 and the display device and the input / output (I / O) device 805. The input / output (I / O) device 805 may be a mouse, a keyboard, a modem, a network interface, a touch input device, a somatosensory input device, a printer, and other devices known in the art. Typically, the input / output (I / O) device 805 is connected to the system via an input / output (I / O) controller 806.

[0120] The memory 802 may store software components such as an operating system, a communication module, an interaction module, and an application program. Each of the modules and applications described above corresponds to a set of executable program instructions that implement one or more functions and methods described in the embodiments of the invention.

[0121] An embodiment of the present invention also provides a computer-readable storage medium having a computer program stored thereon. When the computer program is executed by a processor, the steps of the above-mentioned artificial intelligence-based converter unmanned slag monitoring method are implemented.

[0122] An embodiment of the present invention also provides a computer program product, including a computer program / instruction, which, when executed by a processor, implements the steps of the above-mentioned artificial intelligence-based converter unmanned slag monitoring method.

[0123] To sum up, the artificial intelligence-based converter unmanned slag discharge monitoring method and device provided by the present invention provide a more comprehensive safety monitoring means through the intelligent recognition of AI models, realize intelligent monitoring, and break through the limitation of the prior art that can only perform simple image measurement and position positioning; in addition, through AI model training, the system of the present invention has similar manual recognition and judgment capabilities, and can intelligently identify normal / abnormal slag discharge, the beginning of converter slag flow and the end of slag discharge, etc., and can also identify and judge whether the positional relationship between the converter and the slag pot is normal, and can promptly discover and warn of abnormal situations, providing reliable safety guarantees for unmanned operation.

[0124] The preferred embodiments of the present invention have been described above with reference to the accompanying drawings. Many features and advantages of these embodiments are clear from this detailed description, and therefore the claims are intended to cover all such features and advantages of these embodiments that fall within their true spirit and scope. In addition, since many modifications and changes are readily apparent to those skilled in the art, the embodiments of the present invention are not intended to be limited to the precise structure and operation illustrated and described, but all suitable modifications and equivalents that fall within their scope may be included.

[0125] Those skilled in the art will appreciate that embodiments of the present invention may be provided as methods, systems, or computer program products. Therefore, the present invention may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, the present invention may take the form of a computer program product implemented 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.

[0126] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0127] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0128] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0129] The specific embodiments described above further illustrate the objectives, technical solutions and beneficial effects of the present invention in detail. It should be understood that the above description is only a specific embodiment of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. An unmanned converter slag monitoring method based on artificial intelligence, characterized in that: The method comprises: Acquire historical image data of the converter slag tapping process, wherein the historical image data includes normal slag tapping image data and abnormal slag tapping image data; Annotating and extracting features from the historical image data, wherein the features include slag flow morphology features, slag position features, relative position features between the converter and the slag pot, slag discharge start time and slag discharge end time; A database is established based on the extracted features, and a slag discharge monitoring model is trained using the database, so that the slag discharge monitoring model learns to identify and judge normal slag discharge, abnormal slag discharge, the beginning of converter slag outflow, the end of slag discharge, the normal position relationship between the converter and the slag pot, and the abnormal position relationship between the converter and the slag pot; The trained slag discharge monitoring model is deployed in the actual production environment to monitor the unmanned slag discharge process of the converter in real time and provide slag discharge process and alarm prompts.

2. The unmanned converter slag monitoring method based on artificial intelligence according to claim 1, characterized in that: The prompts of the slag discharge process include: a slag discharge start stage and a slag discharge end stage; the alarm prompts include: a slag overflow risk alarm.

3. The unmanned converter slag discharge monitoring method based on artificial intelligence according to claim 1, characterized in that: The method further includes: based on the normal slag discharge image data in the historical image data, using a single-classification support vector machine algorithm to train an abnormality recognition model to recognize abnormal slag discharge situations that do not appear in the abnormal slag discharge image data.

4. The unmanned converter slag monitoring method based on artificial intelligence as claimed in claim 3 is characterized in that: The single-class support vector machine algorithm adopts a radial basis kernel function and optimizes the kernel parameter γ and the penalty parameter ν by a grid search or cross-validation method to determine the optimal anomaly recognition model.

5. The unmanned converter slag monitoring method based on artificial intelligence according to claim 1, characterized in that: The method further comprises: Acquire historical slag tapping data of the converter, wherein the historical slag tapping data includes slag tapping time, slag tapping amount, converter tilting angle, slag pot type, and slag pot initial weight; Performing data cleaning and preprocessing on the historical slag discharge data; Using the historical slag discharge data that has been cleaned and preprocessed to train a slag discharge time and slag discharge amount prediction model; The slag discharge time and slag discharge amount prediction model is deployed in the actual production environment, and the production scheduling is optimized in combination with the slag discharge monitoring model.

6. The unmanned converter slag discharge monitoring method based on artificial intelligence according to claim 5, characterized in that: The slag discharge time and slag discharge amount prediction model combined with the slag discharge monitoring model optimizes production scheduling including: In the process of the slag discharge monitoring model monitoring the slag discharge process, if the slag discharge amount predicted by the slag discharge time and slag discharge amount prediction model exceeds the preset percentage of the slag pot capacity, the slag discharge monitoring model issues an early warning to prompt the operator to replace the slag pot or adjust the slag discharge speed to avoid slag pot overflow; if the slag discharge time predicted by the slag discharge time and slag discharge amount prediction model exceeds the preset threshold, the slag discharge monitoring model issues an early warning to prompt the operator to adjust the production scheduling plan; According to the predicted slag discharge amount and slag discharge time based on the slag discharge time and slag discharge amount prediction model, combined with the relative position of the converter and the slag pot monitored in real time by the slag discharge monitoring model, the positional relationship between the two is dynamically adjusted to ensure that the steel slag falls into the slag pot to avoid abnormal slag discharge; According to the slag discharge time and slag discharge amount prediction model and the predicted slag discharge amount and slag discharge time, a suitable number and type of slag tank trucks are arranged in advance to the designated location to avoid long waiting time or insufficient resources of the slag tank trucks, thereby improving production efficiency.

7. An unmanned converter slag monitoring device based on artificial intelligence, characterized in that: The device comprises: A historical data acquisition unit, used to acquire historical image data of the converter slag tapping process, wherein the historical image data includes normal slag tapping image data and abnormal slag tapping image data; A feature extraction unit, used for annotating and extracting features from the historical image data, wherein the features include slag flow morphology features, slag position features, relative position features between the converter and the slag pot, slag discharge start time and slag discharge end time; A model training unit, for establishing a database based on the extracted features, and using the database to train a slag discharge monitoring model, so that the slag discharge monitoring model learns to identify and judge normal slag discharge, abnormal slag discharge, the beginning of converter slag outflow, the end of slag discharge, the normal positional relationship between the converter and the slag pot, and the abnormal positional relationship between the converter and the slag pot; The model deployment unit is used to deploy the trained slag discharge monitoring model into the actual production environment, to monitor the unmanned slag discharge process of the converter in real time and to provide slag discharge process and alarm prompts.

8. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 6 are implemented.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 6 are implemented.

10. A computer program product comprising a computer program / instructions, characterized in that When the computer program / instructions are executed by a processor, the steps of the method according to any one of claims 1 to 6 are implemented.

Images