Slagging-off path planning method, control method, apparatus, system, device, and medium

Abstract

Disclosed are a slagging-off path planning method, a control method, an apparatus, a system, a device, and a medium. The slagging-off path planning method comprises: acquiring a distribution condition of molten iron and iron slag in a ladle; on the basis of the distribution condition of the molten iron and the iron slag, carrying out area division on an area waiting for slagging-off, and calculating a slagging-off weight of each area; and calculating all possible slagging-off paths on the basis of the slagging-off weight of each area, and selecting an optimal slagging-off path as an automatic slagging-off path. The control method comprises: acquiring slagging-off input data of a ladle, separately carrying out linear prediction and nonlinear prediction on the basis of the slagging-off input data to obtain linear slagging-off prediction data and nonlinear slagging-off prediction data, determining, by using the linear slagging-off prediction data and the nonlinear slagging-off prediction data, target prediction data corresponding to the slagging-off input data, and thus carrying out slagging-off control on the ladle on the basis of the target prediction data.

Description

Slag removal path planning method, control method, device, system, equipment and medium

[0001] Citation of Related Applications

[0002] This disclosure claims all rights and interests in the invention patent application with application number 202410236111.6 filed with the State Intellectual Property Office of the People's Republic of China on March 1, 2024, entitled "A method, system and device for automatic slag removal path planning" and the invention patent application with application number 202410035215.0 filed with the State Intellectual Property Office of the People's Republic of China on January 9, 2024, entitled "Method, device, electronic equipment and storage medium for slag removal control of ladle", and incorporates the entire contents thereof into this document by reference.

[0003] field

[0004] The present disclosure relates to the field of control technology, and in particular to a slag removal path planning method, control method, device, system, equipment and medium.

[0005] background

[0006] Slag skimming is a specialized process used to remove desulfurized slag from the molten iron ladle after desulfurization. This process requires manual operation or control of a slag rake to remove the slag from the ladle's surface. The most common methods for this process involve manually planning the slag removal path or following a fixed path. Automated slag removal path control is rarely used, utilizing manual control and program-planned fixed paths. This process requires workers to observe the distribution of slag in the hot iron ladle to operate the slag rake, posing significant safety risks and increasing labor costs for steel companies.

[0007] During the specific slag removal operation, the ladle containing molten steel needs to be tilted at a certain angle so that the liquid level and the edge of the ladle are roughly even. The slag removal operation begins after the ladle is tilted. During the slag removal operation, different slag removal data are required, such as the tilting angle, slag removal time, or number of slag removals. In existing technical solutions, the determination of slag removal data is mainly based on manual judgment and determination. That is, the operator sets the slag removal data based on operational experience to perform the slag removal operation. However, this manual determination of slag removal data relies on the operator's experience and judgment, requiring the operator to spend time observing the status of each ladle during the slag removal process, which affects efficiency and is affected by human subjectivity, reducing the accuracy of slag removal control.

[0008] Overview

[0009] The present disclosure provides a slag removal path planning method, a control method, an apparatus, a system, equipment and a medium.

[0010] According to one aspect of the present disclosure, a method for automatic slag removal path planning is provided, comprising the following steps:

[0011] S1: Obtain the distribution of molten iron and slag in the ladle;

[0012] S2: Divide the slag removal area into regions according to the distribution of molten iron and slag, and calculate the slag removal weight of each region;

[0013] S3: Calculate all possible scraping paths according to the scraping weights of each area, and select the optimal scraping path as the automatic scraping path.

[0014] In some embodiments, S1 further includes S0 before S0:

[0015] According to the steel type and the temperature of the molten iron entering the ladle, the fixed route of the coarse slag removal model is called, and the slag removal path is output to the slag removal machine control part to perform coarse slag removal on the molten iron ladle.

[0016] In certain embodiments, the specific manner of dividing the area to be scraped is as follows:

[0017] When the system is first run on the obtained image of the distribution of molten iron and iron slag in the ladle, the edge of the ladle is identified by the image algorithm in the system imaging screen, and the internal and external areas of the ladle are distinguished according to the edge of the ladle. The calculation area in the internal area of ​​the ladle is divided into small rectangles with specific side lengths.

[0018] In some embodiments, calculating the scraping weight of each region includes the following steps:

[0019] S201: establishing a rule correspondence table between the number of iron slag points and the average brightness of the iron slag;

[0020] S202: Calculate the number of iron slag points and the average brightness of the iron slag in each calculation area, and obtain the initial weight value according to the rule correspondence table;

[0021] S203: Calculating a weight change value based on the location of the calculation area and the initial weight value;

[0022] S204: The slag removal weight of the area calculated during the last slag removal is used as a reference, and the slag removal weight of this area is obtained according to the weight change value.

[0023] In some embodiments, the method of selecting the optimal scraping path includes:

[0024] Traverse all possible paths and select the path with the highest average weight as the automatic scraping path;

[0025] In some embodiments, the method of selecting the optimal scraping path further includes:

[0026] Select the weight sum of two adjacent areas for comparison, take the two largest areas as the base points, expand upward, downward and to the origin, and take the path with the highest average weight.

[0027] In some embodiments, the method of selecting the optimal scraping path further includes:

[0028] Take the path with the highest average weight as the first backup scraping path, select the weight sum of two adjacent areas for comparison, take the two largest areas as base points, expand upward, downward and to the origin, take the path with the highest average weight as the second backup path, repeat S1-S3 to obtain multiple groups of first backup scraping paths and second backup scraping paths, match the first backup scraping paths and the second backup scraping paths for familiarity, and select the first backup scraping path or the second backup scraping path with the highest familiarity as the optimal scraping path.

[0029] According to another aspect of the present disclosure, there is provided an automatic slag removal path planning system, comprising:

[0030] The first module is used to obtain the distribution of molten iron and slag in the ladle;

[0031] The second module is used to divide the slag removal area according to the distribution of molten iron and slag, and calculate the slag removal weight of each area;

[0032] The third module is used to calculate all possible scraping paths according to the scraping weights of each area, and select the optimal scraping path as the automatic scraping path.

[0033] In some embodiments, the automatic slag removal path planning system also includes a coarse slag removal module: for calling the coarse slag removal model fixed route according to the type of steel and the temperature of the molten iron entering the ladle, outputting the slag removal path to the slag removal machine control part, and performing coarse slag removal on the molten iron ladle.

[0034] According to another aspect of the present disclosure, an automatic slag scraping path planning device is provided, comprising an image acquisition device and any of the above-mentioned automatic slag scraping path planning systems, wherein the distribution of molten iron and slag in the ladle obtained by the first module in the automatic slag scraping path planning system is collected through the image acquisition device.

[0035] According to another aspect of the present disclosure, a method for controlling slag removal of a ladle is provided, comprising:

[0036] Get the slag removal input data of the ladle;

[0037] Based on the scraping input data, linear prediction and nonlinear prediction are performed respectively to obtain linear scraping prediction data and nonlinear scraping prediction data;

[0038] Using the linear slag removal prediction data and the nonlinear slag removal prediction data, determining target prediction data corresponding to the slag removal input data;

[0039] The ladle is subjected to slag removal control according to the target prediction data.

[0040] In certain embodiments, performing linear prediction based on the scraping input data to obtain linear scraping prediction data includes:

[0041] Get the preset linear regression parameters;

[0042] Performing linear prediction using the linear regression parameters and the scraping input data to obtain a first scraping start angle parameter, a first scraping end angle parameter, a first scraping time parameter, and a first scraping number parameter;

[0043] The first slag scraping start angle parameter, the first slag scraping end angle parameter, the first slag scraping time parameter, and the first slag scraping number parameter are determined as the linear slag scraping prediction data.

[0044] In certain embodiments, performing nonlinear prediction based on the slag removal input data to obtain nonlinear slag removal prediction data includes:

[0045] Inputting the scraping input data into a pre-trained nonlinear prediction model for prediction to obtain an output result of the nonlinear prediction model;

[0046] Extracting a second slag scraping start angle parameter, a second slag scraping end angle parameter, a second slag scraping time parameter, and a second slag scraping number parameter from the output result;

[0047] The second slag scraping start angle parameter, the second slag scraping end angle parameter, the second slag scraping time parameter, and the second slag scraping number parameter are determined as the nonlinear slag scraping prediction data.

[0048] In certain embodiments, the determining target prediction data corresponding to the slag scraping input data using the linear slag scraping prediction data and the nonlinear slag scraping prediction data includes:

[0049] Calculating using the linear slag removal prediction data and the nonlinear slag removal prediction data to obtain an average value of the slag removal prediction data;

[0050] The target prediction data is determined based on the average value of the slagging prediction data.

[0051] In certain embodiments, performing slag removal control on the ladle based on the target prediction data includes:

[0052] generating control instructions for the slag removal equipment according to the target prediction data;

[0053] Based on the control instruction, the ladle is deslagging by a deslagging device, and real-time deslagging parameters of the deslagging device are obtained;

[0054] When the real-time slag scraping parameters meet the end conditions corresponding to the target prediction parameters, slag scraping end information is generated.

[0055] In some embodiments, the target prediction data includes a target scraping end angle parameter, a target scraping time parameter, and a target scraping number parameter, and the real-time scraping parameters include a real-time scraping angle parameter, a real-time scraping time parameter, and a real-time scraping number parameter;

[0056] After obtaining the real-time scraping parameters of the scraping equipment, the method further includes:

[0057] When the real-time slag scraping angle parameter reaches the target slag scraping end angle parameter and the real-time slag scraping time parameter reaches the target slag scraping time parameter, determining that the real-time slag scraping parameter meets the end condition corresponding to the target prediction parameter; or

[0058] When the real-time slag scraping angle parameter reaches the target slag scraping end angle parameter and the real-time slag scraping number parameter reaches the target slag scraping number parameter, determining that the real-time slag scraping parameter meets the end condition corresponding to the target prediction parameter; or

[0059] When the real-time slag scraping number parameter reaches the target slag scraping number parameter and the real-time slag scraping time parameter reaches the target slag scraping time parameter, it is determined that the real-time slag scraping parameter meets the end condition corresponding to the target prediction parameter.

[0060] In certain embodiments, after performing slag removal control on the ladle according to the target prediction data, the method further comprises:

[0061] Based on the deslagging completion information, obtaining deslagging completion parameters of the ladle;

[0062] When the scraping end parameter satisfies a preset condition, the scraping end parameter and the scraping input data are used to update a training sample set of a nonlinear prediction model, and the number of updates of the training sample set is determined;

[0063] When the update number reaches a preset update number threshold, the nonlinear prediction model is retrained using sample data in the training sample set, where the sample data includes the slag scraping end parameter and the slag scraping input data.

[0064] According to another aspect of the present disclosure, a slag removal control device for a ladle is provided, comprising:

[0065] Acquisition module, used to obtain the slag input data of the ladle;

[0066] A prediction module is used to perform linear prediction and nonlinear prediction based on the scraping input data to obtain linear scraping prediction data and nonlinear scraping prediction data;

[0067] A determination module, configured to determine target prediction data corresponding to the slag removal input data using the linear slag removal prediction data and the nonlinear slag removal prediction data;

[0068] A control module is used to control the slag removal of the ladle based on the target prediction data.

[0069] According to another aspect of the present disclosure, there is also provided an electronic device, comprising a processor, a communication interface, a memory, and a communication bus, wherein the processor, the communication interface, and the memory communicate with each other via the communication bus;

[0070] Memory for storing computer programs;

[0071] The processor is used to implement the above-mentioned ladle slag removal control method when executing the program stored in the memory.

[0072] According to another aspect of the present disclosure, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the ladle slag removal control method as described in any embodiment of the first aspect is implemented.

[0073] In certain embodiments, the above-mentioned slag removal path planning method can shorten the slag removal time, break away from the fully automatic algorithm of manual control, and can significantly shorten the slag removal process time. Improve the cleaning rate. The iron slag is initially removed by fine slag removal, and then the iron slag is specifically tracked by fine slag removal. The cleaning rate of this method far exceeds that of traditional manual slag removal and fixed path planning slag removal. Reduce labor costs. The operation of slag removal path planning can be implemented by cameras and computer software, reducing the labor costs of production enterprises. High precision. The distribution of iron slag is accurately analyzed through image processing, and the path is planned according to the distribution of iron slag, which can achieve high precision and is targeted at removing iron slag from the liquid surface of molten steel. Strong variability. The slag removal path can be changed in real time according to the changes in the distribution of iron slag on the liquid surface to achieve a higher cleaning rate.

[0074] In certain embodiments, by obtaining the slag removal input data of the ladle, linear prediction and nonlinear prediction are performed based on the slag removal input data to obtain linear slag removal prediction data and nonlinear slag removal prediction data. The linear slag removal prediction data and the nonlinear slag removal prediction data are used to determine the target prediction data corresponding to the slag removal input data, so that the slag removal of the ladle can be controlled based on the target prediction data, without the need to manually determine the slag removal data during the slag removal process, thereby improving the accuracy and efficiency of slag removal.

[0075] BRIEF DESCRIPTION OF THE DRAWINGS

[0076] The drawings described herein are used to provide further understanding of the present disclosure and constitute a part of the present disclosure. The exemplary embodiments of the present disclosure and their descriptions are used to explain the present disclosure and do not constitute improper limitations on the present disclosure.

[0077] In order to more clearly illustrate the technical solutions of the embodiments of the present disclosure, the following briefly introduces the drawings required for describing the embodiments. Obviously, for ordinary technicians in this field, other drawings can be obtained based on these drawings without any creative work.

[0078] FIG1 is a flow chart of an automatic slag removal path planning method provided by the present disclosure;

[0079] FIG2 is a schematic diagram of a coarse scraping path provided by the present disclosure;

[0080] FIG3 is a diagram of the slag removal area provided by the present disclosure;

[0081] FIG4 is a schematic flow diagram of a ladle slag removal control method provided by the present disclosure;

[0082] FIG5 is a schematic structural diagram of a slag skimming control device for a ladle provided by the present disclosure;

[0083] FIG6 is a schematic structural diagram of an electronic device provided by the present disclosure.

[0084] Details

[0085] The embodiments of the present disclosure are described in detail below with reference to the accompanying drawings. However, the present disclosure can be implemented in many different ways as defined and covered by the claims.

[0086] To make the objectives, technical solutions, and advantages of the embodiments of the present disclosure more clear, the technical solutions in the embodiments of the present disclosure will be clearly and completely described below in conjunction with the accompanying drawings. Obviously, the described embodiments are part of the embodiments of the present disclosure, not all of the embodiments. All other embodiments obtained by ordinary technicians in this field based on the embodiments of the present disclosure without making any creative efforts shall fall within the scope of protection of the present disclosure.

[0087] The present disclosure first provides an automatic scraping path planning method, as shown in FIG1 , comprising the following steps:

[0088] S0: According to the steel type and the temperature of the molten iron entering the ladle, the fixed route of the rough slag removal model is called, the slag removal path is output to the slag removal machine control part, and the molten iron ladle is roughly slag removed.

[0089] A coarse slag removal model is selected based on the steel grade and inlet temperature, and a coarse slag removal path is output. Figure 2 shows the coarse slag removal model within the normal slag removal temperature range for a certain steel grade. At this point, the desulfurization process in the molten iron ladle has just concluded, and desulfurization slag covers the molten iron surface. Therefore, fine slag removal path planning is ineffective, and the slag removal path is output according to the fixed slag removal model. The steel grade and inlet temperature are input externally via Ethernet communication. Alternative coarse slag removal path planning models can be selected based on actual needs.

[0090] S1: Obtain the distribution of molten iron and slag in the ladle.

[0091] The camera captures images of the molten steel surface in the molten iron ladle, providing material for image processing. After rough slag removal, a high-precision camera captures real-time image information from the ladle interior. Figure 3 shows the distribution of slag on the molten iron surface after rough slag removal for a certain steel grade.

[0092] S2: Divide the slag removal area into regions according to the distribution of molten iron and slag, and calculate the slag removal weight of each region.

[0093] The possible slag removal range is divided and numbered. A small rectangle with a specific side length is used as the analysis and calculation area, expanding upward, downward, left, and right to the edge of the ladle and assigning a number. The slag content within the image area is analyzed and the iron slag content is calculated. The weight change value is then calculated based on the location of the area. Finally, the final area weight is calculated by referring to the previous slag removal path.

[0094] Calculating the scraping weight of each area includes the following steps:

[0095] S201: Establishing a rule correspondence table between the number of iron slag points and the average brightness of the iron slag.

[0096] The iron slag content is equal to the iron slag points. Each iron slag point corresponds to a brightness value. The average brightness of all iron slag points is the iron slag brightness average. The regular correspondence table of the iron slag points P and the iron slag brightness average E is shown in Table 1 below:

[0097] Table 1

[0098] Explanation of the symbol suffixes in the above table:

[0099] HS: negative biased; MS: negative biased; A: intermediate value; MB: positive biased; HB: positive biased;

[0100] S202: Calculate the number of iron slag points and the average brightness of the iron slag in each calculation area, and obtain the initial weight value according to the rule correspondence table.

[0101] Assuming that the number of iron slag points in this area is positively small (P_MB) and the average brightness of the iron slag is negatively small (E_MS), the initial weight value of this area can be found to be W2.4 according to the rule correspondence table.

[0102] S203: Calculate the weight change value based on the location of the calculation area and the initial weight value.

[0103] Combined with the location of the area, the weight change value is calculated. For example, according to the rule correspondence table, the initial value of the area weight is W2.4. The area is on the edge and the area coefficient is 0.6, then the weight change value is 0.6*W2.4.

[0104] S203: The slag removal weight of the area calculated during the last slag removal is used as a reference, and the slag removal weight of this area is obtained according to the weight change value.

[0105] The weight of the area passed by the last scraping needs to be reduced appropriately. Different proportions C are taken according to the size of the area covered and the time lag. The final weight value is 0.6*C*W2.4

[0106] S3: Calculate all possible slag removal paths according to the final slag removal weights of each area, and select the optimal slag removal path as the automatic slag removal path.

[0107] There are three ways to select the optimal scraping path, including:

[0108] Static: All possible paths are calculated each time, and the path with the highest average weight is selected;

[0109] Dynamic: Compare the weights of two adjacent areas, take the two largest areas as the base points, expand upward, downward and to the origin, and take the path with the highest average weight.

[0110] Merge: Analyze single areas and merged areas separately, and take the path that basically overlaps among the top 5 paths as the final path.

[0111] The final path is displayed and marked on the image, as shown in Figure 3, and compared with human judgment to determine whether it is appropriate.

[0112] Based on the above method, the present disclosure also provides an automatic slag removal path planning system, which includes:

[0113] The first module is used to obtain the distribution of molten iron and slag in the ladle;

[0114] The second module is used to divide the slag removal area according to the distribution of molten iron and slag, and calculate the slag removal weight of each area;

[0115] The third module is used to calculate all possible scraping paths according to the scraping weights of each area, and select the optimal scraping path as the automatic scraping path.

[0116] Preferably, the automatic slag removal path planning system also includes a coarse slag removal module: used to call the coarse slag removal model fixed route according to the steel type and the molten iron inlet temperature, output the slag removal path to the slag removal machine control part, and perform coarse slag removal on the molten iron ladle.

[0117] Relying on the above system, the present disclosure also provides an automatic slag scraping path planning device, including an image acquisition device and any of the above automatic slag scraping path planning systems. The distribution of molten iron and slag in the ladle obtained by the first module in the automatic slag scraping path planning system is collected through the image acquisition device.

[0118] The specific operation process of slag skimming is to tilt the ladle to a preset angle and then skim off the impurities on the surface of the molten metal in the ladle. Among them, different slag skimming methods need to be adopted in different situations. For example, when the liquid level of the ladle is different, the tilting angle required for slag skimming will be different, or when the slag thickness in the ladle is different, the number of slag skimming times or the time required will also be different. Currently, there are also semi-automatic slag skimming methods in automatic slag skimming. For example, after the ladle is tilted to a certain angle automatically, the angle is adjusted manually, and after the slag skimming is completed, it is still necessary to manually observe the liquid level with the naked eye to manually end the slag skimming. Manual auxiliary operation is required to effectively complete the slag skimming operation, and manual auxiliary operation will make the slag skimming efficiency low.

[0119] In order to solve the problem in the above-mentioned prior art that manual assisted operation will result in low slag removal efficiency, the embodiment of the present invention obtains the slag removal input data of the ladle, and performs linear prediction and nonlinear prediction based on the slag removal input data to obtain linear slag removal prediction data and nonlinear slag removal prediction data. The linear slag removal prediction data and the nonlinear slag removal prediction data are used to determine the target prediction data corresponding to the slag removal input data, so that the slag removal of the ladle can be controlled according to the target prediction data, without the need to manually determine the slag removal data during the slag removal process, thereby improving the accuracy and efficiency of slag removal.

[0120] FIG4 is a schematic flow chart of a ladle slag removal control method provided in an embodiment of the present disclosure.

[0121] As shown in FIG4 , the present disclosure discloses an embodiment, which provides a method for controlling slag removal of a ladle, comprising:

[0122] Step S110: Obtaining the ladle's slag removal input data.

[0123] Among them, the ladle may refer to the ladle that currently needs to be deslagging, the ladle represents a container for holding molten metal, and the deslagging input data represents the status information of the current ladle, which may include parameter information such as the ladle age, arrival tare weight, arrival net weight, liquid level, arrival slag thickness, and exit slag thickness of the ladle; wherein the method of obtaining the deslagging input data of the ladle may be through user input, sensor collection, etc.

[0124] Step S120: Based on the scraping input data, linear prediction and nonlinear prediction are performed respectively to obtain linear scraping prediction data and nonlinear scraping prediction data.

[0125] Specifically, after obtaining the slag scraping input data, this embodiment can perform linear prediction and nonlinear prediction based on the slag scraping input data, so as to perform real-time linear prediction and nonlinear prediction on the slag scraping process of the ladle according to the slag scraping input data, and obtain linear slag scraping prediction data and nonlinear slag scraping prediction data corresponding to the slag scraping input data; wherein, linear prediction means linear logical prediction of the slag scraping process based on the slag scraping input data, for example, prediction can be performed using linear equations, coefficient relationships, etc., and this embodiment does not impose specific restrictions on this. The linear slag scraping prediction data means the data obtained after linear prediction of the slag scraping process based on the slag scraping input data; nonlinear prediction means nonlinear logical prediction of the slag scraping process based on the slag scraping input data, for example, prediction can be performed using random forest regression, gradient boosting regression, BP neural network model, XGBoost algorithm, etc., and this embodiment does not impose specific restrictions on this. The linear slag scraping prediction data means the data obtained after linear prediction of the slag scraping process based on the slag scraping input data.

[0126] It should be noted that the linear slag removal prediction data and the nonlinear slag removal prediction data may contain one or more data parameters, and the types of data parameters contained in the linear slag removal prediction data and the nonlinear slag removal prediction data may be consistent. For example, the linear slag removal prediction data and the nonlinear slag removal prediction data may both contain slag removal start angle parameters, slag removal end angle parameters, slag removal time parameters, and slag removal number parameters, etc.

[0127] Step S130: using the linear slag removal prediction data and the nonlinear slag removal prediction data to determine target prediction data corresponding to the slag removal input data.

[0128] Specifically, after obtaining the linear slag removal prediction data and the nonlinear slag removal prediction data, the linear slag removal prediction data and the nonlinear slag removal prediction data can be used to perform preset calculation processing to determine the target prediction data corresponding to the slag removal input data; wherein the preset calculation processing can be to calculate the average value or variance value of various data parameters in the linear slag removal prediction data and the nonlinear slag removal prediction data, so that the data parameter set after calculation processing can be used as the target prediction data.

[0129] Since the ladle deslagging input data represents the current state of the ladle, and different deslagging methods are required for different ladle states, there is no direct and accurate linear relationship between them. In other words, if a single linear calculation or linear prediction is used, it is impossible to obtain predicted data that matches the deslagging input data. In addition, since the accuracy of nonlinear prediction is high, and the training samples and data accuracy requirements are high, if a single nonlinear calculation or linear prediction is used, it is also impossible to accurately and effectively obtain predicted data that is very consistent with the deslagging input data.

[0130] In addition, nonlinear prediction using linear prediction and machine learning has some advantages:

[0131] (1) Improved robustness and accuracy: Combining linear and nonlinear models can make up for the shortcomings of each model. Linear models may not be able to capture complex nonlinear relationships in the data, but they perform well for certain linear relationships; nonlinear models may be more suitable for capturing complex patterns and features, providing more accurate predictions.

[0132] (2) Flexibility and interpretability: We can take advantage of the interpretability and simplicity of linear models while combining them with the flexibility of nonlinear models. This allows us to maintain the interpretability of the model to a certain extent while using nonlinear models to capture more complex data relationships.

[0133] (3) Reduce the risk of overfitting: Linear models are relatively simple and less likely to perform well on training data but poorly on test data due to overfitting. At the same time, nonlinear models can better adapt to the complexity of data and help improve the generalization ability of the overall model.

[0134] (4) Adaptability to specific situations: In some specific situations, data may have both linear and nonlinear characteristics. In this case, using both linear and nonlinear predictions can better adapt to the diversity and complexity of the data.

[0135] Therefore, this embodiment combines the two means of linear prediction and nonlinear prediction. By adopting linear prediction and nonlinear prediction, two kinds of prediction data, linear slag removal prediction data and nonlinear slag removal prediction data, are obtained. Then, after calculating and processing the two prediction data, the linear slag removal prediction data and the nonlinear slag removal prediction data, the target prediction data is obtained. At this time, the target prediction data is a combination of the linear slag removal prediction data and the nonlinear slag removal prediction data, thereby improving the fit between the target prediction data and the slag removal input data.

[0136] Step S140: performing slag removal control on the ladle according to the target prediction data.

[0137] Specifically, after determining the target prediction data, the ladle can be controlled for slag removal based on the target prediction data. Specifically, a control method can be to generate control instructions based on the target prediction data, and the control instructions are used to control the slag removal equipment to perform the slag removal operation on the ladle. In this embodiment, the slag removal control process does not require manual determination of slag removal parameters. Instead, the target prediction data is determined through linear and nonlinear predictions. This solves the problem that the manual determination of slag removal parameters in the existing related art inevitably takes a long time and has a high probability of error, that is, the problem of low accuracy and efficiency in manually determining slag removal parameters. This effectively improves the accuracy and efficiency of slag removal.

[0138] In an optional embodiment of the present disclosure, step S110 obtains the slag skimming input data of the ladle, which may specifically include the following sub-steps: obtaining the ladle age parameter, arrival tare weight parameter, arrival net weight parameter, liquid level parameter, arrival slag thickness parameter and exit slag thickness parameter of the ladle; using the ladle age parameter, arrival tare weight parameter, arrival net weight parameter, liquid level parameter, arrival slag thickness parameter and exit slag thickness parameter to generate the slag skimming input data.

[0139] Specifically, after determining the ladle that needs to be deslagging, data parameters such as the ladle age parameter, arrival tare weight parameter, arrival net weight parameter, liquid level height parameter, arrival slag thickness parameter and exit slag thickness parameter of the ladle can be collected. The specific collection method can be through user input, sensor collection, database acquisition, etc., which is not specifically limited in this embodiment.

[0140] In an optional embodiment of the present disclosure, step S120 includes performing linear prediction based on the scraping input data to obtain linear scraping prediction data, which may specifically include the following sub-steps: obtaining preset linear regression parameters; using the linear regression parameters and the scraping input data to perform linear prediction to obtain a first scraping start angle parameter, a first scraping end angle parameter, a first scraping time parameter and a first scraping number parameter; and determining the first scraping start angle parameter, the first scraping end angle parameter, the first scraping time parameter and the first scraping number parameter as the linear scraping prediction data.

[0141] Specifically, in the process of linear prediction based on slag scraping input data, preset linear regression parameters can be obtained, wherein the linear round parameters can represent the preset parameters for linear prediction; thereby linear prediction is performed using the linear regression parameters and the slag scraping input data, wherein the linear prediction can be a linear calculation performed in combination with the linear regression parameters and the slag scraping input data, and then the first slag scraping start angle parameter, the first slag scraping end angle parameter, the first slag scraping time parameter and the first slag scraping number parameter are obtained based on the calculation results, wherein the first slag scraping start angle parameter represents the tilting angle required for the ladle to start slag scraping, the first slag scraping end angle parameter represents the tilting angle when the ladle ends slag scraping, the first slag scraping time parameter represents the time for the ladle to scrape the slag, and the first slag scraping number parameter represents the number of times the ladle scrapes the slag; the first slag scraping start angle parameter, the first slag scraping end angle parameter, the first slag scraping time parameter and the first slag scraping number parameter can be determined as linear slag scraping prediction data.

[0142] In a specific embodiment, when the linear prediction is a linear regression equation, the scraping input data includes the bag age (x1), the arrival tare weight (x2), the arrival net weight (x3), the liquid level (x4), the arrival slag thickness (x5), and the discharge slag thickness (x6). The preset linear regression parameters are obtained, including the calculation coefficients a0~a4, b0~b7, c0~c6 and d0~d6. The process of linear prediction using the linear regression parameters and the scraping input data is as follows: the first scraping start angle parameter (y1) is used to construct a linear regression equation: y1=a0+a1x1+a2x2+a3x3+a4x4; since there is also a significant relationship between the first scraping start angle parameter and the first scraping end angle parameter, the first scraping start angle parameter (y1) can be used when constructing the linear regression equation for the first scraping end angle parameter (y2). ) is also used as the independent variable: y2=b0+b1x1+b2x2+b3x3+b4x4+b5x5+b6x6+b7y1; a linear regression equation is constructed for the first slag scraping time parameter (y3): y3=c0+c1x1+c2x2+c3x3+c4x4+c5x5+c6x6; a linear equation is constructed for the first slag scraping times parameter (y4): y4=d0+d1x1+d2x2+d3x3+d4x4+d5x5+d6x6; thus, the first slag scraping start angle parameter, the first slag scraping end angle parameter, the first slag scraping time parameter and the first slag scraping times parameter are calculated according to the above linear regression equation, and then the first slag scraping start angle parameter, the first slag scraping end angle parameter, the first slag scraping time parameter and the first slag scraping times parameter can be determined as linear slag scraping prediction data.

[0143] In an optional embodiment of the present disclosure, step S120 includes performing nonlinear prediction based on the slag scraping input data to obtain nonlinear slag scraping prediction data, which may specifically include the following sub-steps: inputting the slag scraping input data into a pre-trained nonlinear prediction model for prediction to obtain the output result of the nonlinear prediction model; extracting the second slag scraping start angle parameter, the second slag scraping end angle parameter, the second slag scraping time parameter and the second slag scraping number parameter from the output result; and determining the second slag scraping start angle parameter, the second slag scraping end angle parameter, the second slag scraping time parameter and the second slag scraping number parameter as the nonlinear slag scraping prediction data.

[0144] Specifically, in the process of linear prediction based on scraping input data, the scraping input data can be input into a pre-trained nonlinear prediction model for prediction, and the model calculation is performed through the nonlinear prediction model to obtain the output result of the nonlinear prediction model, wherein the nonlinear prediction model can be Lasso regression, Ridge regression, L1 regularization & L2 regularization, elastic network regression, Bayesian ridge regression, Huber regression, KNN, SVM, decision tree regression, random forest regression, gradient boosting regression, BP neural network model, XGBoost algorithm and other models; and then the second scraping start angle parameter, the second scraping end angle parameter, the second scraping time parameter and the first scraping time parameter can be extracted from the output result. The second slag removal number parameter is used, and the second slag removal start angle parameter, the second slag removal end angle parameter, the second slag removal time parameter and the second slag removal number parameter are determined as nonlinear slag removal prediction data; wherein, the second slag removal start angle parameter, the second slag removal end angle parameter, the second slag removal time parameter and the second slag removal number parameter are similar to the aforementioned first slag removal start angle parameter, the first slag removal end angle parameter, the first slag removal time parameter and the first slag removal number parameter, the second slag removal start angle parameter indicates the tilting angle required for the ladle to start slag removal, the second slag removal end angle parameter indicates the tilting angle when the ladle ends slag removal, the second slag removal time parameter indicates the time for ladle slag removal, and the second slag removal number parameter indicates the number of times the ladle slags are removed.

[0145] In specific implementations, historical data on ladle deslagging operations is obtained as a training sample set. The historical data can include historical deslagging input data and historical deslagging result data. The training sample set is used to train a preset model to obtain a nonlinear prediction model. The input layer of the nonlinear prediction model is the ladle age (x1), arrival tare weight (x2), arrival net weight (x3), liquid level (x4), arrival slag thickness (x5), and exit slag thickness (x6). The output layer of the nonlinear prediction model is the second deslagging start angle parameter (y1), the second deslagging end angle parameter (y2), the second deslagging duration parameter (y3), and the second deslagging number parameter (y4).

[0146] In an optional embodiment of the present disclosure, step S130 uses linear slag removal prediction data and nonlinear slag removal prediction data to determine the target prediction data corresponding to the slag removal input data, which may specifically include the following sub-steps: using linear slag removal prediction data and nonlinear slag removal prediction data to perform calculations to obtain the average value of the slag removal prediction data; and determining the target prediction data based on the average value of the slag removal prediction data.

[0147] Specifically, in the process of determining the target prediction data corresponding to the slag scraping input data, the first slag scraping start angle parameter, the first slag scraping end angle parameter, the first slag scraping time parameter and the first slag scraping number parameter can be extracted from the linear slag scraping prediction data, and the second slag scraping start angle parameter, the second slag scraping end angle parameter, the second slag scraping time parameter and the second slag scraping number parameter can be extracted from the nonlinear slag scraping prediction data; thereby, the target slag scraping end angle parameter is generated based on the first slag scraping end angle parameter and the second slag scraping end angle parameter, the target slag scraping start angle parameter is generated based on the first slag scraping start angle parameter and the second slag scraping start angle parameter, the target slag scraping time parameter is generated based on the first slag scraping time parameter and the second slag scraping number parameter, and the target slag scraping number parameter is generated based on the first slag scraping number parameter and the second slag scraping number parameter. In the above process, the average value can be calculated, that is, the target slag scraping end angle parameter represents the average value of the first slag scraping end angle parameter and the second slag scraping end angle parameter, the target slag scraping start angle parameter represents the average value of the first slag scraping start angle parameter and the second slag scraping start angle parameter, and the target slag scraping time parameter represents the average value of the first slag scraping start angle parameter and the second slag scraping start angle parameter. The parameter represents the average value of the first slag scraping time parameter and the second slag scraping time parameter, and the target slag scraping number parameter represents the average value of the first slag scraping number parameter and the second slag scraping number parameter; that is, the average value of the slag scraping prediction data may include the target slag scraping end angle parameter, the target slag scraping start angle parameter, the target slag scraping time parameter and the target slag scraping number parameter; more specifically, in the process of calculating the average value of the slag scraping prediction data, two average value calculations may be included. The first average value calculation is as described above. After obtaining the target slag scraping end angle parameter, the target slag scraping start angle parameter, the target slag scraping time parameter and the target slag scraping number parameter, the parameter data of which the average value of the slag scraping prediction data deviates from the target slag scraping end angle parameter, the target slag scraping start angle parameter, the target slag scraping time parameter and the target slag scraping number parameter by more than one standard deviation may be removed, and the remaining parameter data may be subjected to a second average value calculation. The target slag scraping end angle parameter, the target slag scraping start angle parameter, the target slag scraping time parameter and the target slag scraping number parameter obtained by this average value calculation are used as the average value of the slag scraping prediction data, and the target prediction data may be determined by the average value of the slag scraping prediction data.

[0148] In addition, the use of linear and nonlinear scraping prediction data to calculate the average value of the scraping prediction data in this embodiment is only an example of feasibility. In specific implementations, other stacking models can also be adopted. The stacking model is to train on multiple models and then use another model (called a meta-model) to combine the prediction results of these models. The linear model can be used as one of the basic models, and its prediction results can be combined with other more complex nonlinear models.

[0149] It should be noted that due to the instability of a single nonlinear prediction model, in this embodiment, the slag removal input data can be input into at least two pre-trained nonlinear prediction models for prediction, thereby obtaining at least two nonlinear slag removal prediction data. In other words, the aforementioned target prediction data determination process can be calculated based on at least one linear slag removal prediction data and at least two nonlinear slag removal prediction data to obtain the average slag removal prediction data. This improves the accuracy of the final slag removal prediction data average.

[0150] Furthermore, there are the following reasons and advantages for using two or more machine learning-based nonlinear prediction methods to jointly predict the results:

[0151] (1) Improved robustness and accuracy: Different machine learning models may have different strengths in capturing data patterns and features. By combining the prediction results of multiple models, the bias of individual models can be reduced, thereby improving the accuracy of the overall prediction. Even if one model performs poorly in certain situations, the prediction results of other models can compensate for this deficiency and improve the robustness of the overall prediction.

[0152] (2) Reduce the risk of overfitting: Using multiple different types of models for prediction can reduce the risk of overfitting. If a model is too sensitive to the training data, it may lead to overfitting. By combining multiple models, the overfitting of a single model to specific data can be reduced, and the generalization ability of the model can be improved.

[0153] (3) Providing diverse perspectives: Different machine learning models may use different feature representations and learning methods, thus providing diverse perspectives on the data. This diversity helps capture different patterns and structures in the data, providing more comprehensive predictions.

[0154] (4) Addressing model bias: Different models may have different biases in their data, and combining multiple models can compensate for the bias of a single model to a certain extent. This can improve the robustness of the model and reduce prediction errors caused by the limitations of a single model.

[0155] In an optional embodiment of the present disclosure, step S140 controls the slag removal of the ladle based on the target prediction data, which may specifically include the following sub-steps: generating a control instruction for the slag removal equipment based on the target prediction data; based on the control instruction, performing slag removal processing on the ladle through the slag removal equipment, and obtaining the real-time slag removal parameters of the slag removal equipment; generating slag removal end information when the real-time slag removal parameters meet the end conditions corresponding to the target prediction parameters.

[0156] Specifically, in the present embodiment, during the process of controlling the slag removal of the ladle, a control instruction of the slag removal device can be generated based on the target prediction data, wherein the slag removal device represents a device for performing slag removal processing on the ladle, and the control instruction represents an instruction for controlling the slag removal device; the process of generating the control can be to extract the target slag removal start angle parameter, the target slag removal end angle parameter, the target slag removal time parameter and the target slag removal number parameter from the target prediction data, and generate the control instruction based on the target slag removal start angle parameter, the target slag removal end angle parameter, the target slag removal time parameter and the target slag removal number parameter; thereby, the slag removal processing of the ladle can be performed by the slag removal device based on the control instruction, and the real-time slag removal parameters of the slag removal device can be obtained in real time, and the real-time slag removal parameters represent the current state of the slag removal device, such as the real-time slag removal angle parameter, the real-time slag removal time parameter and the real-time slag removal time parameter. long parameter and real-time slag scraping times parameter, the real-time slag scraping angle parameter represents the current tilting angle of the slag scraping ladle, the real-time slag scraping time parameter represents the current slag scraping time, and the real-time slag scraping times parameter represents the current slag scraping times; and judge whether the real-time slag scraping parameters meet the end conditions corresponding to the target prediction parameters, wherein the end conditions are the conditions for stopping the slag scraping operation, and the end conditions can be determined based on the target prediction parameters; when the real-time slag scraping parameters meet the end conditions corresponding to the target prediction parameters, slag scraping end information is generated, indicating that the conditions for stopping the slag scraping operation are currently met. At this time, the output of the control instruction can be stopped, or the stop instruction can be output to control the slag scraping equipment to stop slag scraping, thereby generating slag scraping end information, wherein the slag scraping end information represents the state of the ladle at the end of slag scraping, that is, the real-time slag scraping parameters collected at the end of slag scraping.

[0157] In an optional embodiment of the present disclosure, the target prediction data includes a target slag scraping end angle parameter, a target slag scraping time parameter and a target slag scraping number parameter, and the real-time slag scraping parameter includes a real-time slag scraping angle parameter, a real-time slag scraping time parameter and a real-time slag scraping number parameter; after obtaining the real-time slag scraping parameters of the slag scraping equipment, the following sub-steps may also be included: when the real-time slag scraping angle parameter reaches the target slag scraping end angle parameter and the real-time slag scraping time parameter reaches the target slag scraping time parameter, determining that the real-time slag scraping parameter meets the end condition corresponding to the target prediction parameter; or, when the real-time slag scraping angle parameter reaches the target slag scraping end angle parameter and the real-time slag scraping number parameter reaches the target slag scraping number parameter, determining that the real-time slag scraping parameter meets the end condition corresponding to the target prediction parameter; or, when the real-time slag scraping number parameter reaches the target slag scraping number parameter and the real-time slag scraping time parameter reaches the target slag scraping time parameter, determining that the real-time slag scraping parameter meets the end condition corresponding to the target prediction parameter.

[0158] In this embodiment, after extracting the target scraping end angle parameter, the target scraping time parameter and the target scraping number parameter from the target prediction data, and extracting the real-time scraping angle parameter, the real-time scraping time parameter and the real-time scraping number parameter from the real-time scraping parameters, it can be judged whether the real-time scraping angle parameter reaches the target scraping end angle parameter, whether the real-time scraping time parameter reaches the target scraping time parameter and whether the real-time scraping number parameter reaches the target scraping number parameter. If at least two of the above judgment results are reached, it can be determined that the real-time scraping parameters meet the end conditions corresponding to the target prediction parameters; that is, in the real When the real-time slag scraping angle parameter reaches the target slag scraping end angle parameter and the real-time slag scraping time parameter reaches the target slag scraping time parameter, it is determined that the real-time slag scraping parameter meets the end condition corresponding to the target prediction parameter; or, when the real-time slag scraping angle parameter reaches the target slag scraping end angle parameter and the real-time slag scraping number parameter reaches the target slag scraping number parameter, it is determined that the real-time slag scraping parameter meets the end condition corresponding to the target prediction parameter; or, when the real-time slag scraping number parameter reaches the target slag scraping number parameter and the real-time slag scraping time parameter reaches the target slag scraping time parameter, it can be determined that the real-time slag scraping parameter meets the end condition corresponding to the target prediction parameter.

[0159] In an optional embodiment of the present disclosure, after S140 controls the slag removal of the ladle according to the target prediction data, it may also include the following sub-steps: based on the slag removal end information, obtaining the slag removal end parameters of the ladle; when the slag removal end parameters meet the preset conditions, using the slag removal end parameters and the slag removal input data to update the training sample set of the nonlinear prediction model, and determining the number of updates to the training sample set; when the update number reaches the preset update number threshold, using the sample data in the training sample set to retrain the nonlinear prediction model, the sample data includes the slag removal end parameters and the slag removal input data.

[0160] Specifically, after the ladle is controlled to be deslagging, deslagging completion information can be obtained, and the deslagging completion parameter of the ladle can be obtained based on the deslagging completion information, the deslagging completion parameter indicating the state of the ladle when deslagging is completed; and it is judged whether the deslagging completion parameter meets a preset condition, the preset condition indicating the preset state of the ladle after the deslagging is completed, for example, when the deslagging completion parameter includes an impurity residue parameter and a splashing liquid parameter, the preset condition may include an impurity residue parameter threshold and a splashing liquid parameter threshold, and when the impurity residue parameter is lower than the impurity residue parameter threshold and the splashing liquid parameter is lower than the splashing liquid parameter threshold, it can be determined that the deslagging completion parameter meets the preset condition; when the deslagging completion parameter meets the preset condition, the deslagging completion parameter and the deslagging input data can be used to update the training sample set of the nonlinear prediction model, The specific updating method is that the scraping end parameters and the scraping input data can replace the training samples with the earliest timestamp in the training sample set, so as to keep the number of training samples in the training sample set unchanged; and the update number of the training sample set can be determined, and the update number represents the number of the scraping end parameters and the scraping input data updated to the end of the training sample set, and then judge whether the update number reaches the preset update number threshold. When the update number reaches the preset update number threshold, the sample data in the training sample set is used to retrain the nonlinear prediction model, and the sample data includes the scraping end parameters and the scraping input data; so that the nonlinear prediction model in this embodiment can continue to use the updated training sample set to retrain the nonlinear prediction model, and the update number threshold can be used to control the frequency of retraining the nonlinear prediction model.

[0161] It should be noted that when the number of updates reaches the preset update number threshold, the linear regression parameters and their corresponding linear equations in the aforementioned embodiment can also be reconstructed based on the sample data in the training sample set, so as to use the reconstructed linear regression parameters and linear equations to perform linear prediction on the corresponding scraping input data.

[0162] As shown in FIG5 , the present disclosure further discloses an embodiment, which provides a slag removal control device for a ladle, comprising:

[0163] An acquisition module 210 is used to obtain the slag removal input data of the ladle;

[0164] Prediction module 220, for performing linear prediction and nonlinear prediction based on the scraping input data, to obtain linear scraping prediction data and nonlinear scraping prediction data;

[0165] A determination module 230 is configured to determine target prediction data corresponding to the slag removal input data using the linear slag removal prediction data and the nonlinear slag removal prediction data;

[0166] The control module 240 is used to control the slag removal of the ladle according to the target prediction data.

[0167] In one embodiment, the prediction module 220 may include:

[0168] A first acquisition unit, configured to acquire preset linear regression parameters;

[0169] The first prediction unit is used to perform linear prediction using linear regression parameters and slag scraping input data to obtain a first slag scraping start angle parameter, a first slag scraping end angle parameter, a first slag scraping time parameter, and a first slag scraping number parameter;

[0170] The first determining unit is configured to determine a first slag scraping start angle parameter, a first slag scraping end angle parameter, a first slag scraping time parameter, and a first slag scraping number parameter as linear slag scraping prediction data.

[0171] In one embodiment, the prediction module 220 may include:

[0172] The second prediction unit is used to input the scraping input data into a pre-trained nonlinear prediction model for prediction, and obtain the output result of the nonlinear prediction model;

[0173] The first extraction unit is used to extract the second slag scraping start angle parameter, the second slag scraping end angle parameter, the second slag scraping time parameter and the second slag scraping number parameter from the output result;

[0174] The second determining unit is configured to determine the second slag scraping start angle parameter, the second slag scraping end angle parameter, the second slag scraping time parameter, and the second slag scraping number parameter as nonlinear slag scraping prediction data.

[0175] In one embodiment, the determination module 230 may include:

[0176] The first calculation unit is used to calculate using the linear slag removal prediction data and the nonlinear slag removal prediction data to obtain an average value of the slag removal prediction data;

[0177] The third determining unit is configured to determine target prediction data based on an average value of the slag removal prediction data.

[0178] In one embodiment, the control module 240 may include:

[0179] The first generating unit is used to generate control instructions for the slag scraping equipment according to the target prediction data;

[0180] An acquisition unit is used to perform slag removal processing on the ladle through the slag removal equipment based on the control instruction and obtain the real-time slag removal parameters of the slag removal equipment;

[0181] The second generating unit is configured to generate slag scraping end information when the real-time slag scraping parameters meet the end conditions corresponding to the target prediction parameters.

[0182] In one embodiment, the target prediction data includes a target scraping end angle parameter, a target scraping time parameter, and a target scraping number parameter, and the real-time scraping parameters include a real-time scraping angle parameter, a real-time scraping time parameter, and a real-time scraping number parameter; the control module 240 may further include:

[0183] The fourth determining unit is configured to determine that the real-time slag scraping parameter meets the end condition corresponding to the target prediction parameter when the real-time slag scraping angle parameter reaches the target slag scraping end angle parameter and the real-time slag scraping time parameter reaches the target slag scraping time parameter; or

[0184] A fifth determining unit is configured to determine that the real-time slag scraping parameter meets the end condition corresponding to the target prediction parameter when the real-time slag scraping angle parameter reaches the target slag scraping end angle parameter and the real-time slag scraping number parameter reaches the target slag scraping number parameter; or

[0185] The sixth determining unit is configured to determine whether the real-time slag scraping number parameter meets the end condition corresponding to the target prediction parameter when the real-time slag scraping number parameter reaches the target slag scraping number parameter and the real-time slag scraping time parameter reaches the target slag scraping time parameter.

[0186] In one embodiment, the apparatus may further include:

[0187] End module, used to obtain the slag removal end parameters of the ladle based on the slag removal end information;

[0188] An updating module is used to update the training sample set of the nonlinear prediction model using the slag removal end parameter and the slag removal input data when the slag removal end parameter meets the preset conditions, and to determine the update quantity of the training sample set;

[0189] The training module is used to retrain the nonlinear prediction model using sample data in the training sample set when the update number reaches a preset update number threshold, and the sample data includes scraping end parameters and scraping input data.

[0190] The implementation process of the functions and effects of each module in the above-mentioned device is specifically described in the implementation process of the corresponding steps in the above-mentioned method, and will not be repeated here.

[0191] As shown in FIG6 , an embodiment of the present disclosure provides an electronic device, including a processor 310, a communication interface 320, a memory 330, and a communication bus 340, wherein the processor 310, the communication interface 320, and the memory 330 communicate with each other via the communication bus 340.

[0192] Memory 330, for storing computer programs;

[0193] In one embodiment of the present disclosure, the processor 310 is used to implement the slag removal control method of the ladle provided by any one of the aforementioned method embodiments when executing the program stored on the memory 330, by obtaining the slag removal input data of the ladle, and performing linear prediction and nonlinear prediction based on the slag removal input data, respectively, to obtain linear slag removal prediction data and nonlinear slag removal prediction data, so as to determine the target prediction data corresponding to the slag removal input data by using the linear slag removal prediction data and the nonlinear slag removal prediction data, so that the slag removal of the ladle can be controlled according to the target prediction data, without the need to manually determine the slag removal data during the slag removal process, thereby improving the accuracy and efficiency of the slag removal.

[0194] An embodiment of the present disclosure also provides a computer-readable storage medium having a computer program stored thereon. When the computer program is executed by a processor, the ladle slag removal control method provided in any one of the aforementioned method embodiments is implemented, by obtaining the slag removal input data of the ladle, and performing linear prediction and nonlinear prediction based on the slag removal input data, respectively, to obtain linear slag removal prediction data and nonlinear slag removal prediction data. The linear slag removal prediction data and the nonlinear slag removal prediction data are used to determine target prediction data corresponding to the slag removal input data, so that the ladle can be slag removed based on the target prediction data, without the need for manual determination of the slag removal data during the slag removal process, thereby improving the accuracy and efficiency of slag removal.

[0195] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the terms "comprises," "comprising," or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, article, or device comprising a series of elements includes not only those elements, but also other elements not explicitly listed, or elements inherent to such process, method, article, or device. In the absence of further limitations, an element defined by the phrase "comprising a ..." does not exclude the presence of other identical elements in the process, method, article, or device comprising the element.

[0196] The foregoing description of specific embodiments of the present disclosure is provided herein. Other embodiments are within the scope of the appended claims. In some cases, the actions or steps recited in the claims may be performed in an order different from that described in the embodiments and still achieve the desired results. Furthermore, the processes depicted in the accompanying drawings do not necessarily require the specific order shown or the sequential order shown to achieve the desired results. In certain embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0197] The foregoing description is intended only to provide specific embodiments of the present disclosure, intended to enable those skilled in the art to understand and implement the present disclosure. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present disclosure. Therefore, the present disclosure is not to be limited to the embodiments shown herein, but is to be construed in the broadest manner consistent with the principles and novel features claimed herein.

Claims

1. An automatic slag skimming path planning method, wherein, It includes the following steps: S1: Obtain the distribution of molten iron and slag in the ladle. S2: Divide the slag-skimming area according to the distribution of molten iron and slag, and calculate the slag-skimming weight of each area. S3: Calculate all possible slag-skimming paths according to the slag-skimming weights of each area, and select the optimal slag-skimming path as the automatic slag-skimming path.

2. The automatic slag skimming path planning method according to claim 1, wherein, Before S1, there is also S0, and S0: According to the steel grade type and the molten iron tapping temperature, call the fixed route of the rough slag-skimming model, output the slag-skimming path to the slag-skimming machine control part, and perform rough slag-skimming on the ladle.

3. The automatic slag skimming path planning method according to claim 1, wherein, The specific method for dividing the slag-skimming area in S2 is: Identify the ladle edge through image algorithm for the obtained image of the distribution of molten iron and slag in the ladle, distinguish the internal area and the external area of the ladle according to the ladle edge, and divide and calculate the area in the internal area of the ladle according to a preset shape.

4. The automatic slag skimming path planning method according to claim 1, wherein, Calculating the slag-skimming weight of each area includes the following steps: S201: Establish a rule correspondence table between the number of slag points and the average brightness of slag. S202: Calculate the number of slag points and the average brightness of slag in each calculation area, and obtain the initial weight value according to the rule correspondence table. S203: Combine the position where the calculation area is located and the initial weight value to obtain the weight change value. S204: Take the slag-skimming weight of the calculation area during the previous slag-skimming as the benchmark, and obtain the slag-skimming weight of the current area according to the weight change value.

5. The automatic slag skimming path planning method according to claim 1, wherein, The methods for selecting the optimal slag-skimming path include: Traverse all possible paths that can pass through, and take the path with the highest average weight as the automatic slag-skimming path.

6. The automatic slag skimming path planning method according to claim 1, wherein The methods for selecting the optimal slag-skimming path also include: Select the sum of the weights of two adjacent areas for comparison, take the two largest areas as the base points, expand upward, downward, and to the origin, and take the path with the highest average weight.

7. The automatic slag skimming path planning method according to claim 4, wherein, The methods for selecting the optimal slag-skimming path also include: Take the path with the highest average weight as the first standby slag-skimming path; select the sum of the weights of two adjacent areas for comparison, take the two largest areas as the base points, expand upward, downward, and to the origin, and take the path with the highest average weight as the second standby path. Repeat S1 - S3 to obtain multiple groups of the first standby slag-skimming path and the second standby slag-skimming path, perform similarity matching on the first standby slag-skimming path and the second standby slag-skimming path, and select the first standby slag-skimming path or the second standby slag-skimming path with the highest similarity as the optimal slag-skimming path.

8. An automatic slag skimming path planning system, wherein, It includes: The first module: used to obtain the distribution of molten iron and slag in the ladle. The second module: used to divide the slag-skimming area according to the distribution of molten iron and slag, and calculate the slag-skimming weight of each area. The third module: used to calculate all possible slag-skimming paths according to the slag-skimming weights of each area, and select the optimal slag-skimming path as the automatic slag-skimming path.

9. The automatic slag skimming path planning system according to claim 8, wherein, The automatic slag-skimming path planning system also includes a rough slag-skimming module: used to call the fixed route of the rough slag-skimming model according to the steel grade type and the molten iron tapping temperature, output the slag-skimming path to the slag-skimming machine control part, and perform rough slag-skimming on the ladle.

10. An automatic slag skimming path planning system, wherein, It includes an image acquisition device and the automatic slag skimming path planning system as described in claim 8 or 9. The distribution of molten iron and slag in the ladle obtained by the first module in the automatic slag skimming path planning system is collected by the image acquisition device.

11. A slag skimming control method, wherein, It includes: Obtaining the slag skimming input data of the container for holding molten metal; Based on the slag skimming input data, performing linear prediction and non-linear prediction respectively to obtain linear slag skimming prediction data and non-linear slag skimming prediction data; Using the linear slag skimming prediction data and the non-linear slag skimming prediction data to determine the target prediction data corresponding to the slag skimming input data; Controlling the slag skimming of the container according to the target prediction data.

12. The slag skimming control method according to claim 11, wherein, Performing linear prediction based on the slag skimming input data to obtain linear slag skimming prediction data, including: Obtaining preset linear regression parameters; Performing linear prediction using the linear regression parameters and the slag skimming input data to obtain a first slag skimming start angle parameter, a first slag skimming end angle parameter, a first slag skimming duration parameter, and a first slag skimming number parameter; Determining the first slag skimming start angle parameter, the first slag skimming end angle parameter, the first slag skimming duration parameter, and the first slag skimming number parameter as the linear slag skimming prediction data.

13. The slag skimming control method according to claim 11, wherein, Performing non-linear prediction based on the slag skimming input data to obtain non-linear slag skimming prediction data, including: Inputting the slag skimming input data into a pre-trained non-linear prediction model for prediction to obtain the output result of the non-linear prediction model; Extracting a second slag skimming start angle parameter, a second slag skimming end angle parameter, a second slag skimming duration parameter, and a second slag skimming number parameter from the output result; Determining the second slag skimming start angle parameter, the second slag skimming end angle parameter, the second slag skimming duration parameter, and the second slag skimming number parameter as the non-linear slag skimming prediction data.

14. The slag skimming control method according to claim 11, wherein, The step of using the linear slag skimming prediction data and the non-linear slag skimming prediction data to determine the target prediction data corresponding to the slag skimming input data includes: Calculating using the linear slag skimming prediction data and the non-linear slag skimming prediction data to obtain the average value of the slag skimming prediction data; Determining the target prediction data based on the average value of the slag skimming prediction data.

15. The slag skimming control method according to claim 11, wherein, The step of controlling the slag skimming of the container according to the target prediction data includes: Generating a control instruction for the slag skimming device according to the target prediction data; Based on the control instruction, performing slag skimming on the container by the slag skimming device and obtaining the real-time slag skimming parameters of the slag skimming device; Generating slag skimming end information when the real-time slag skimming parameters meet the end condition corresponding to the target prediction parameters.

16. The slag skimming control method according to claim 15, wherein, The target prediction data includes a target slag skimming end angle parameter, a target slag skimming duration parameter, and a target slag skimming number parameter, and the real-time slag skimming parameters include a real-time slag skimming angle parameter, a real-time slag skimming duration parameter, and a real-time slag skimming number parameter; After obtaining the real-time slag skimming parameters of the slag skimming device, it further includes: When the real-time slag skimming angle parameter reaches the target slag skimming end angle parameter and the real-time slag skimming duration parameter reaches the target slag skimming duration parameter, it is determined that the real-time slag skimming parameter meets the end condition corresponding to the target prediction parameter; or, When the real-time slag skimming angle parameter reaches the target slag skimming end angle parameter and the real-time slag skimming times parameter reaches the target slag skimming times parameter, it is determined that the real-time slag skimming parameter meets the end condition corresponding to the target prediction parameter; or, When the real-time slag skimming times parameter reaches the target slag skimming times parameter and the real-time slag skimming duration parameter reaches the target slag skimming duration parameter, it is determined that the real-time slag skimming parameter meets the end condition corresponding to the target prediction parameter.

17. The slag skimming control method according to claim 15, wherein, After controlling the slag skimming of the container according to the target prediction data, it further includes: Based on the slag skimming end information, obtaining the slag skimming end parameter of the container; When the slag skimming end parameter meets the preset condition, using the slag skimming end parameter and the slag skimming input data to update the training sample set of the non-linear prediction model, and determining the update quantity of the training sample set; When the update quantity reaches the preset update quantity threshold, retraining the non-linear prediction model with the sample data in the training sample set, where the sample data includes the slag skimming end parameter and the slag skimming input data.

18. A slag skimming control device, wherein, It includes: An acquisition module, configured to acquire the slag skimming input data of a container filled with molten metal; A prediction module, configured to respectively perform linear prediction and non-linear prediction based on the slag skimming input data to obtain linear slag skimming prediction data and non-linear slag skimming prediction data; A determination module, configured to use the linear slag skimming prediction data and the non-linear slag skimming prediction data to determine the target prediction data corresponding to the slag skimming input data; A control module, configured to control the slag skimming of the container according to the target prediction data.

19. An electronic device, wherein, It includes a processor, a communication interface, a memory and a communication bus, wherein the processor, the communication interface and the memory complete communication with each other through the communication bus; The memory is used for storing a computer program; The processor, when executing the program stored on the memory, implements the slag skimming control method according to any one of claims 11-17.

20. A computer-readable storage medium having a computer program stored thereon, wherein, The computer program, when executed by the processor, implements the slag skimming control method according to any one of claims 11-17.

Images