Sintered ore production method, device and system, electronic equipment and storage medium

Abstract

The invention provides a sinter production method, device and system, electronic equipment and a storage medium, and the method comprises the steps: inputting the to-be-predicted sintering production information into a pre-trained sinter metallurgical performance prediction model, and obtaining the predicted sinter metallurgical performance data; and determining a data difference between the predicted sintered ore metallurgical performance data and the target sintered ore metallurgical performance data, and if the data difference exceeds a preset difference interval, generating a sintered ore production early warning message, according to the method and the device, the production proportion information of the sinter to be adopted currently in the sintering production information to be predicted is adjusted through the prompt of the sinter production early warning message, the metallurgical performance of the sinter is predicted, and when the sintering production information of the sinter to be produced in real time is adopted, the metallurgical performance of the sinter can be predicted in real time, so that the production efficiency of the sinter is improved. And further, the situation that the process requirement is not met can be adjusted in time, the stability of the performance index of the sintered ore obtained through production can be improved, and smooth smelting of a blast furnace is guaranteed.

Description

Technical Field

[0001] The embodiments of the present application relate to the field of metallurgical engineering and intelligent manufacturing technology, and in particular to a sintered ore production method, device, system, electronic equipment and storage medium. Background Art

[0002] Sintered ore is one of the important raw materials for blast furnace smelting, and stable sintered ore performance indicators are the prerequisite for smooth smelting in blast furnaces. The detection of the metallurgical properties of sintered ore has been one of the difficult problems in the metallurgical industry for many years. This is due to the small amount of data, unstable data, and few important influencing factors in its metallurgical performance indicators. Based on prior knowledge of the process, it is known that the metallurgical properties of sintered ore are highly correlated with the basic properties of mineral powder. However, due to the wide variety of mineral powders used and the expensive testing and analysis of the basic properties of mineral powders, the basic properties data of mineral powders are seriously missing. In addition, the sintering process is a smelting process of uneven duration. It is difficult for traditional process equipment to grasp the data correlation between the input parameters and output parameters of sintering, and it is impossible to accurately and real-time estimate the metallurgical properties of sintered ore, resulting in the performance indicators of the sintered ore produced being difficult to maintain stability. Summary of the invention

[0003] The embodiments of the present application provide a sintered ore production method, device, system, electronic device and storage medium to solve the technical problem in the related art that the metallurgical properties of the sintered ore cannot be accurately and real-time estimated, resulting in difficulty in maintaining stable performance indicators of the sintered ore produced.

[0004] A sintered ore production method provided in an embodiment of the present application includes: obtaining sintered production information to be predicted, wherein the sintered production information to be predicted includes sintered production parameters to be predicted and sintered ore production ratio information currently to be adopted; inputting the sintered production information to be predicted into a pre-trained sintered ore metallurgical property prediction model to obtain predicted sintered ore metallurgical property data, wherein the pre-trained sintered ore metallurgical property prediction model is trained based on historical sintered production information and historical sintered ore metallurgical property test data; determining the data difference between the predicted sintered ore metallurgical property data and the target sintered ore metallurgical property data; if the data difference exceeds a preset difference interval, generating a sintered ore production early warning message, so as to prompt the current sintered ore production ratio information to be adjusted through the sintered ore production early warning message.

[0005] In an embodiment of the present application, the sintering production information to be predicted is input into a pre-trained sintered ore metallurgical property prediction model to obtain predicted sintered ore metallurgical property data, including: inputting the sintering production information to be predicted into a first prediction sub-model to obtain a first prediction sub-result; inputting the sintering ingredient single ratio to be predicted in the sintering production information to be predicted into a second prediction sub-model to obtain a second prediction sub-result; and determining the predicted sintered ore metallurgical property data based on the first prediction sub-result and the second prediction sub-result; wherein the sintered ore metallurgical property prediction model includes a first prediction sub-model and a second prediction sub-model, and the sintering production parameters to be predicted include the sintering ingredient single ratio to be predicted, the calculated value of the sintered ore mixture to be predicted, the fuel and flux data of the sintering process to be predicted, and the sintered ore production operation parameters to be predicted.

[0006] In an embodiment of the present application, if the data difference exceeds a preset difference range, the method further includes: inputting the sintering production parameters to be predicted and the adjusted sintering ore production ratio information currently to be adopted into a pre-trained sintering ore metallurgical property prediction model to obtain new predicted sintering ore metallurgical property data, until the data difference between the new predicted sintering ore metallurgical property data and the preset sintering ore metallurgical property data falls into the preset difference range; determining the final adjusted sintering ore production ratio information currently to be adopted as the target sintering ore production ratio information; and producing sintering ore based on the target sintering ore production ratio information and the sintering production parameters to be predicted.

[0007] In the first embodiment of the present application, the training method of the sintered ore metallurgical property prediction model includes: obtaining multiple training data, the training data including historical sintering production information and historical sintered ore metallurgical property inspection data of the sintered ore produced based on the historical sintering production information, the historical sintering production information including historical sintered ore mixture calculation value, historical sintered ore production operation parameters, historical sintering ingredient single ratio, historical sintering process fuel flux data and historical sintered ore production ratio information; inputting the historical sintering production information into the first prediction sub-model based on the residual neural network for data fitting to obtain data fitting results; inputting the historical sintering ingredient single ratio in the historical sintering production information into the first prediction sub-model based on the residual neural network to obtain data fitting results. The second prediction sub-model is input to perform convolution kernel processing and activation function processing on the historical sintering ingredient single ratio in sequence to obtain processed data; the processed data is multiplied by the data fitting result, and the training sintered ore metallurgical property data is determined based on the multiplication result; the loss function is determined according to the training sintered ore metallurgical property data of all training data and the corresponding historical sintered ore metallurgical property inspection data; the model parameters of the first prediction sub-model and the model parameters of the second prediction sub-model are updated based on the loss function until convergence to obtain a pre-trained sintered ore metallurgical property prediction model; wherein the sintered ore metallurgical property prediction model includes a first prediction sub-model and a second prediction sub-model.

[0008] In an embodiment of the present application, the method for obtaining the training data includes: obtaining historical sintered ore mixture calculation values, historical sintered ore production operation parameters, historical sintering batching ratios, historical sintering process fuel and flux data, historical sintered ore production ratio information and historical sintered ore metallurgical performance test data of multiple batches of sintered ore; storing the historical sintered ore mixture calculation values ​​and historical sintering process fuel and flux data in a first message queue according to the collected production parameter change time; storing the historical sintering batching ratio in a second message queue according to the batching execution time; storing the historical sintering ore production operation parameters in a third message queue according to the real-time sintering production time; storing the historical sintered ore metallurgical performance test data in a fourth message queue according to the test time; determining the production parameter change time of each batch of sintered ore as the time identifier of the corresponding batch of sintered ore; associating the historical sintering batching ratio with the historical sintered ore production operation parameters based on the batching execution time and the real-time sintering production time; and based on each batch of sintered ore The historical sintered ore mixture calculation value, the historical sintering process fuel and flux data and the historical sintered ore metallurgical property test data are associated based on the actual offline time and the test time; based on the time mark and the batching execution time, the historical sintered ore mixture calculation value, the historical sintering process fuel and flux data are associated with the historical sintering batching ratio and the historical sintered ore production operation parameters, so as to achieve the association between the historical sintered ore mixture calculation value, the historical sintered ore production operation parameters, the historical sintering batching ratio, the historical sintering process fuel and flux data, the historical sintering batching ratio and the historical sintered ore metallurgical property test data of each batch of sintered ore, and obtain multiple initial data sets; preliminary data cleaning is performed on the initial data sets of multiple batches according to the production abnormality period of the sintered ore and the time mark; the noise data is screened out from the initial data set after the preliminary data cleaning according to the Laida criterion to obtain the screened initial data set; the screened initial data set is determined as the training data to obtain multiple training data.

[0009] In an embodiment of the present application, after the sintering production information to be predicted is input into a pre-trained sintered ore metallurgical property prediction model to obtain predicted sintered ore metallurgical property data, the method also includes: obtaining historical predicted sintered ore metallurgical property data output by the pre-trained sintered ore metallurgical property prediction model, and real sintered ore metallurgical property data corresponding to each historical predicted sintered ore metallurgical property data; sorting the predicted sintered ore metallurgical property data and the historical predicted sintered ore metallurgical property data according to data time stamps to generate a predicted numerical curve time series diagram; sorting the real sintered ore metallurgical property data according to data time stamps to generate a real numerical curve time series diagram; and displaying the predicted numerical curve time series diagram and the real numerical curve time series diagram.

[0010] The embodiment of the present application also provides a sintered ore production device, which includes: an acquisition module, used to acquire sintered production information to be predicted, wherein the sintered production information to be predicted includes sintered production parameters to be predicted and sintered ore production ratio information currently to be adopted; a prediction module, used to input the sintered production information to be predicted into a pre-trained sintered ore metallurgical property prediction model to obtain predicted sintered ore metallurgical property data, wherein the pre-trained sintered ore metallurgical property prediction model is trained based on historical sintered production information and historical sintered ore metallurgical property inspection data; a difference determination module, used to determine the data difference between the predicted sintered ore metallurgical property data and the target sintered ore metallurgical property data; an adjustment module, used to generate a sintered ore production warning message if the data difference exceeds a preset difference interval, so as to prompt the adjustment of the sintered ore production ratio information currently to be adopted through the sintered ore production warning message.

[0011] The embodiment of the present application also provides a sinter production system, the system comprising: a data acquisition module, for collecting sinter production information to be predicted, historical sinter production information, and historical sinter metallurgical performance test data of sinter produced based on the historical sinter production information, the historical sinter production information including historical sinter mixture calculation values ​​of multiple batches of sinter, historical sinter production operation parameters, historical sintering batch ratios, historical sinter process fuel and flux data, historical sinter production ratio information, the sinter production information to be predicted includes the sinter production parameters to be predicted and the sinter production ratio information currently to be adopted; a material tracking module, for collecting production parameter change time, batch execution time, sinter real-time production time, inspection and testing time and actual offline time; a server, the server comprising a data processing module, a data association module, a storage module, a training module, a test module and an online evaluation module, wherein the data association module is used to determine the production parameter change time of each batch of sintered ore as the time mark of the corresponding batch of sintered ore; based on the batching execution time and the real-time sintering production time, the historical sintering batching ratio is associated with the historical sintered ore production operation parameters; based on the actual offline time and inspection and testing time of each batch of sintered ore, the historical sintered ore mixture calculation value, the historical sintering process fuel and flux data are associated with the historical sintered ore metallurgical performance inspection and testing data; based on the time mark and the batching ratio During execution, the historical sinter mixture calculation value, the historical sintering process fuel and flux data are associated with the historical sintering batch ratio and the historical sinter production operation parameters, so as to realize the association between the historical sinter mixture calculation value, the historical sinter production operation parameters, the historical sintering batch ratio, the historical sinter process fuel and flux data, the historical sinter production ratio and the historical sinter metallurgical performance test data of each batch of sinter, and obtain multiple initial data sets; the data processing module is used to perform preliminary data cleaning on the initial data sets of multiple batches according to the production abnormality period of the sinter and the time mark; the noise data is screened out according to the Laida criterion for the initial data set after the preliminary data cleaning, Obtaining a set of initial data after screening; determining the set of initial data after screening as the training data to obtain a plurality of training data; the storage module is used to store the plurality of initial data sets; the training module is used to obtain a plurality of training data, the training data including historical sintering production information and historical sintered ore metallurgical performance test data of the sintered ore produced based on the sintering production information, the historical sintering production information including historical sintered ore mixture calculation values, historical sintered ore production operation parameters, historical sintering ingredient single ratios, historical sintering process fuel flux data and historical sintered ore production ratio information; inputting the historical sintering production information into a first prediction sub-model based on a residual neural network for data fitting to obtain a data fitting result;Input the historical sintering ingredient list ratio in the historical sintering production information into the second prediction sub-model, so as to perform convolution kernel processing and activation function processing on the historical sintering ingredient list ratio in sequence to obtain processed data; multiply the processed data with the data fitting result, and determine the training sintered ore metallurgical property data based on the multiplication result; determine the loss function according to the training sintered ore metallurgical property data of all training data and the corresponding historical sintered ore metallurgical property inspection data; update the model parameters of the first prediction sub-model and the model parameters of the second prediction sub-model based on the loss function until convergence, and obtain a pre-trained sintered ore metallurgical property prediction model; wherein the sintered ore metallurgical property prediction model includes the first prediction sub-model and the second prediction sub-model type; the test module is used to input the sintering production information to be predicted into the pre-trained sintering ore metallurgical performance prediction model to obtain the predicted sintering ore metallurgical performance data; the online evaluation module is used to determine the data difference between the predicted sintering ore metallurgical performance data and the target sintering ore metallurgical performance data; if the data difference exceeds the preset difference interval, a sintering ore production warning message is generated to prompt the current sintering ore production ratio information to be adjusted through the sintering ore production warning message; the front-end display module is used to input the predicted sintering ore metallurgical performance data and history into the pre-trained sintering ore metallurgical performance prediction model to obtain the historical predicted sintering ore metallurgical performance data, sort them according to the data time mark, and display the time series data. ;

[0012] An embodiment of the present application also provides an electronic device, including a processor, a memory and a communication bus; the communication bus is used to connect the processor and the memory; the processor is used to execute a computer program stored in the memory to implement a method provided in any one of the above embodiments.

[0013] An embodiment of the present application further provides a computer-readable storage medium on which a computer program is stored, wherein the computer program is used to enable a computer to execute a method provided in any of the above embodiments.

[0014] Beneficial effects of the embodiments of the present application: The embodiments of the present application propose a sintered ore production method, device, system, electronic device and storage medium. The method obtains predicted sintered ore metallurgical property data by inputting the sintering production information to be predicted into a pre-trained sintered ore metallurgical property prediction model, and then determines the data difference between the predicted sintered ore metallurgical property data and the target sintered ore metallurgical property data. If the data difference exceeds the preset difference interval, a sintered ore production early warning message is generated, so as to prompt the current sintered ore production ratio information to be adopted in the predicted sintering production information to be adjusted through the sintered ore production early warning message, so as to realize the prediction of the metallurgical properties of the sintered ore. When the sintering production information to be predicted is the sintering production information of the sintered ore to be produced in real time, the metallurgical properties of the sintered ore can be estimated in real time, and then the situation that does not meet the process requirements can be adjusted in time, so as to improve the stability of the performance indicators of the sintered ore produced and ensure the smooth smelting of the blast furnace. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] Figure 1 A schematic diagram of a process for producing sintered ore provided in an embodiment of the present application;

[0016] Figure 2 A specific schematic diagram of a sintered ore production method provided in an embodiment of the present application;

[0017] Figure 3 It is a structural schematic diagram of a sintered ore production device provided in Example 1 of the present application;

[0018] Figure 4 It is a structural schematic diagram of a sintered ore production system provided in Example 1 of the present application;

[0019] Figure 5 It is a structural schematic diagram of an electronic device provided in one embodiment of the present application. DETAILED DESCRIPTION

[0020] The following is an explanation of the implementation of the embodiments of the present application by specific examples, and those skilled in the art can easily understand other advantages and effects of the embodiments of the present application from the contents disclosed in this specification. The embodiments of the present application can also be implemented or applied through other different specific implementations, and the details in this specification can also be modified or changed in various ways based on different viewpoints and applications without departing from the spirit of the embodiments of the present application. It should be noted that the following embodiments and features in the embodiments can be combined with each other without conflict.

[0021] It should be noted that the illustrations provided in the following embodiments are only schematic illustrations of the basic concept of the embodiments of the present application. Therefore, the drawings only show components related to the embodiments of the present application rather than being drawn according to the number, shape and size of components in actual implementation. In actual implementation, the type, quantity and proportion of each component may be changed at will, and the component layout may also be more complicated.

[0022] In the following description, a large number of details are discussed to provide a more thorough explanation of the embodiments of the present application. However, it is obvious to those skilled in the art that the embodiments of the present application can be implemented without these specific details. In other embodiments, well-known structures and devices are shown in the form of block diagrams rather than in detail to avoid making the embodiments of the present application difficult to understand.

[0023] The following is an exemplary description of some technical terms involved in the embodiments of the present application to facilitate understanding of the embodiments of the present application.

[0024] Sintering is the process of mixing various powdered iron-containing raw materials with appropriate amounts of fuel and flux, adding appropriate amounts of water, mixing and pelletizing, and then subjecting the materials to a series of physical and chemical changes on sintering equipment to bond the mineral powder particles into blocks.

[0025] The quality indicators for evaluating sintered ore (metallurgical performance data of sintered ore) mainly include: chemical composition and its stability, particle size composition and screening index, drum strength, falling strength, low-temperature reduction pulverization, reducibility, soft melting property, etc.

[0026] See also Figure 1 , Figure 1 A schematic diagram of a sintered ore production method provided in an embodiment of the present application, with reference to Figure 1 The sintered ore production method comprises the following steps:

[0027] Step S110: Obtaining sintering production information to be predicted.

[0028] Among them, the sintering production information to be predicted includes the sintering production parameters to be predicted and the current sintering ore production ratio information to be adopted. The sintering production parameters to be predicted include the single ratio of the sintering ingredients to be predicted, the calculated value of the sintering ore mixture to be predicted, the fuel and flux data of the sintering process to be predicted, and the sintering ore production operation parameters to be predicted. The calculated value of the sintering ore mixture to be predicted can be understood as the calculated value obtained by calculating the ratio of each part of the mixture of the mixed ore, fuel, and solvent in the raw materials of the sintering ore to be produced, which is a percentage value.

[0029] The sintering production information to be predicted may be the information related to the raw materials and production line of the sintered ore to be produced. In the related art, the time information corresponding to the sintering production information is often missing. As an example, the sintering production information to be predicted can be obtained by adding corresponding equipment to the production line to monitor the relevant data, and then the time information corresponding to the sintering production information can be obtained. The time information can facilitate the subsequent time series display of the prediction results.

[0030] Step S120: inputting the sintering production information to be predicted into a pre-trained sintered ore metallurgical property prediction model to obtain predicted sintered ore metallurgical property data.

[0031] Among them, the pre-trained sintered ore metallurgical property prediction model is trained based on historical sintering production information and historical sintered ore metallurgical property test data.

[0032] As an example, the sintered ore metallurgical properties prediction model is implemented based on the residual neural network.

[0033] In one embodiment, the sintering production information to be predicted is input into a pre-trained sintered ore metallurgical property prediction model to obtain predicted sintered ore metallurgical property data, including: inputting the sintering production information to be predicted into a first prediction sub-model to obtain a first prediction sub-result; inputting the sintering ingredient list ratio to be predicted in the sintering production information to be predicted into a second prediction sub-model to obtain a second prediction sub-result; determining the predicted sintered ore metallurgical property data based on the first prediction sub-result and the second prediction sub-result; wherein the sintered ore metallurgical property prediction model includes a first prediction sub-model and a second prediction sub-model, wherein the sintering production parameters to be predicted include the sintering ingredient list ratio to be predicted, the calculated value of the sintered ore mixture to be predicted, the fuel and flux data of the sintering process to be predicted, the sintering ore production operation parameters to be predicted, and the sintering ore production ratio information to be predicted.

[0034] As an example, the sinter metallurgical property prediction model is implemented based on residual neural network. The sinter metallurgical property prediction model includes two parts, a first prediction sub-model and a second prediction sub-model. The first prediction sub-model uses multiple residual neural network blocks as basic model blocks to process the predicted sintering production information. The second prediction sub-model includes a convolution module and an activation module, which is used to perform convolution and activation processing on the predicted sintering ingredient single ratio, and then jointly determine the final prediction result with the first prediction sub-result obtained by the first prediction sub-model, that is, predict the sinter metallurgical property data.

[0035] Since the physicochemical properties of sintering ingredient ore powder are seriously missing, and the physicochemical properties of sintering ingredient ore powder are highly correlated with metallurgical properties. Based on the prior knowledge of the process that the physicochemical properties of sintering ingredient ore powder are somewhat correlated with the historical sintering ingredient single ratio, two prediction sub-models are designed to fit the metallurgical properties test data of sintered ore. The information fitted by the model containing residuals is comprehensive information based on the missing physicochemical properties of the ore powder, and the second prediction sub-result related to the sintering ingredient single ratio output by the second prediction sub-model and the information fitted by the first prediction sub-result output by the first prediction sub-model are more relevant to the metallurgical performance target, and the prediction effect is better.

[0036] In this embodiment, the metallurgical performance data of the sintered ore can be represented by the drum index indicator. Of course, those skilled in the art can also use other indicators to predict the metallurgical performance of the sintered ore by the method provided in this embodiment as needed.

[0037] In one embodiment, a training method for a sintered ore metallurgical property prediction model comprises: obtaining a plurality of training data, the training data comprising historical sintering production information and historical sintered ore metallurgical property test data of sintered ore produced based on the historical sintering production information, the historical sintering production information comprising historical sintered ore mixture calculation values, historical sintered ore production operation parameters, historical sintering batching ratios, historical sintering process fuel and flux data, and historical sintered ore production ratio information; inputting the historical sintering production information into a first prediction sub-model based on a residual neural network for data fitting to obtain a data fitting result; inputting the historical sintering batching ratios in the historical sintering production information into a first prediction sub-model based on a residual neural network for data fitting to obtain a data fitting result; The second prediction sub-model is input to perform convolution kernel processing and activation function processing on the historical sintering ingredient single ratio in sequence to obtain processed data; the processed data is multiplied by the data fitting result, and the training sintered ore metallurgical property data is determined based on the multiplication result; the loss function is determined according to the training sintered ore metallurgical property data of all training data and the corresponding historical sintered ore metallurgical property test data; the model parameters of the first prediction sub-model and the model parameters of the second prediction sub-model are updated based on the loss function until convergence to obtain a pre-trained sintered ore metallurgical property prediction model; wherein the sintered ore metallurgical property prediction model includes the first prediction sub-model and the second prediction sub-model.

[0038] It can be understood that the input data of the first prediction sub-model in the sintered ore metallurgical properties prediction model is the historical sintering production information (historical sintered ore mixture calculation value, historical sintered ore production operation parameters, historical sintering ingredient single ratio, historical sintering process fuel and flux data and historical sintered ore production ratio information), and the input data of the second prediction sub-model in the sintered ore metallurgical properties prediction model is the historical sintering ingredient single ratio. Then, the data fitting result obtained by the first prediction sub-model is supplemented with missing information based on the characteristics obtained from the historical sintering ingredient single ratio, so as to make the output result of the model more reliable. The first prediction sub-model is based on the residual neural network block and is designed to obtain the correlation between the metallurgical properties of sintered ore and the sintering production parameters and proportions (sintering production information). The second prediction sub-model performs convolution kernel processing on the single proportion of sintering ingredients separately, and then multiplies the data fitting result obtained by the first prediction sub-model after processing with the activation function. The loss function can be determined by MSE (mean-square error) and the specific type of loss function can be selected by technical personnel in this field as needed. The above is only an example.

[0039] The above method can be used to train a sintered ore metallurgical property prediction model that can obtain output data (sintered ore metallurgical properties) based on input data (sintered production information), so that the real-time sintering production information to be predicted can be used to predict the sintered ore metallurgical property data through the trained sintered ore metallurgical property prediction model.

[0040] In one embodiment, the method for obtaining training data includes: obtaining historical sintered ore mixture calculation values, historical sintered ore production operation parameters, historical sintering batching ratios, historical sintering process fuel and flux data, historical sintered ore production ratio information and historical sintered ore metallurgical performance test data of multiple batches of sintered ore; storing the historical sintered ore mixture calculation values ​​and historical sintering process fuel and flux data in a first message queue according to the collected production parameter change time; storing the historical sintering batching ratios in a second message queue according to the batching execution time; storing the historical sintering ore production operation parameters in a third message queue according to the real-time sintering production time; storing the historical sintered ore metallurgical performance test data in a fourth message queue according to the test time; determining the production parameter change time of each batch of sintered ore as the time identifier of the corresponding batch of sintered ore; associating the historical sintering batching ratio with the historical sintered ore production operation parameters based on the batching execution time and the real-time sintering production time; based on each batch of sintered ore The actual offline time and inspection time of the sintered ore are used to associate the historical sintered ore mixture calculation value, the historical sintering process fuel and flux data with the historical sintered ore metallurgical performance inspection data; based on the time mark and the batching sheet execution time, the historical sintered ore mixture calculation value, the historical sintering process fuel and flux data are associated with the historical sintering batching sheet ratio and the historical sintered ore production operation parameters, so as to achieve the association between the historical sintered ore mixture calculation value, the historical sintered ore production operation parameters, the historical sintering batching sheet ratio, the historical sintering process fuel and flux data, the historical sintering batching sheet ratio and the historical sintered ore metallurgical performance inspection data of each batch of sintered ore, and obtain multiple initial data sets; preliminary data cleaning is performed on the initial data sets of multiple batches according to the production abnormality period and time mark of the sintered ore; the noise data is screened out from the initial data set after the preliminary data cleaning according to the Laida criterion to obtain the screened initial data set; the screened initial data set is determined as training data to obtain multiple training data.

[0041] It can be understood that for the production of sintered ore, the corresponding equipment can be added to the production line to realize the collection of historical sintering production information corresponding to multiple batches of sintered ore that have been produced. The historical sintering production information includes at least three parts of information, mixed ore composition information (historical sintered ore mixture calculation value, historical sintering batching ratio, historical sintering process fuel flux data), operating parameters in the sintering process (historical sintered ore production operating parameters) and mix ratio information (historical sintered ore production ratio information). In addition, it is also necessary to collect the historical sintered ore metallurgical performance test data of these batches of sintered ore. The above three parts of data can be divided into four categories based on their data properties, and are stored in 4 message queues respectively. In order to more conveniently associate these scattered collected data into a group of training data, it is necessary to associate them according to the timestamps corresponding to each data (production parameter change time, batching execution time, sintering real-time production time, and inspection time). Of course, these timestamps often have a certain time difference due to factors such as the time-consuming production process. Those skilled in the art can determine the time interval relationship between the timestamps based on actual production practices, and then realize the generation of the initial data set.

[0042] As an example, the abnormal production period of sintered ore may be the equipment downtime period, the equipment abnormal operation period, or other time periods when the production equipment cannot operate normally. By excluding the data corresponding to these abnormal production time periods through the time dimension, the reliability of the training data itself can be improved, thereby improving the reliability of the model trained based on the training data.

[0043] The Laida criterion is a statistical method for identifying and eliminating gross errors in data. This criterion assumes that the data contains only random errors and obtains the standard deviation by calculation. Then, based on a specific probability interval, it is determined which errors exceed the normal range, thus identifying them as gross errors, and eliminating the corresponding data points. As an example, the 3-Sigma algorithm can be used to screen out noise data from the initial data set after preliminary data cleaning to avoid excessive data noise and further improve the accuracy of the data. The specific data cleaning method can also use other methods known to those skilled in the art, which is only an example here.

[0044] In one embodiment, after the sintering production information to be predicted is input into the pre-trained sinter metallurgical property prediction model to obtain the predicted sinter metallurgical property data, the method further includes: obtaining the historical predicted sinter metallurgical property data output by the pre-trained sinter metallurgical property prediction model, and the real sinter metallurgical property data corresponding to each historical predicted sinter metallurgical property data; sorting the predicted sinter metallurgical property data and the historical predicted sinter metallurgical property data according to the data time mark to generate a predicted numerical curve time series diagram; sorting the real sinter metallurgical property data according to the data time mark to generate a real numerical curve time series diagram; displaying the predicted numerical curve time series diagram and the real numerical curve time series diagram. For example, the predicted numerical curve time series diagram and the real numerical curve time series diagram can be displayed in the same coordinate system for representation, so that the real inspection and testing data of the sinter ore that has been produced can be compared with the predicted inspection and testing data, and the predicted sinter metallurgical property data of the sinter ore that has not yet been produced can be displayed in real time, which is convenient for relevant users to view and makes the display of the results more intuitive.

[0045] It can be understood that the predicted sintered ore metallurgical performance data of multiple batches of sintered ore produced by one or more production lines can be sorted according to the production time sequence, or the time sequence input into the prediction model, etc., to obtain multiple sintered ore metallurgical performance data with a time series relationship, and then these sintered ore metallurgical performance data are generated into a graph, such as a line graph or other schematic diagrams pre-set by technicians in this field, and then the time series data obtained after sorting are displayed, so that relevant personnel can intuitively know the changes in the sintered ore metallurgical performance data of different batches of sintered ore. As an example, the historical predicted sintered ore metallurgical performance data of the batches of sintered ore that have been produced can also be distinguished from the current predicted sintered ore metallurgical performance data by using different line types, different colors, etc., so as to know which are the predicted results of the sintered ore that have been produced and which are the predicted results that have not yet been produced.

[0046] Step S130: determining the data difference between the predicted sintered ore metallurgical property data and the target sintered ore metallurgical property data.

[0047] The target sintered ore metallurgical performance data can be set by those skilled in the art as required, and are not limited here. The data difference can be expressed by the difference between the data, or by the difference amplitude (percentage) between the data. For example, when the data difference is expressed by the difference amplitude, the following formula can be referred to:

[0048] A=(BC) / C Formula (1),

[0049] Among them, A is the data difference, B is the predicted sintered ore metallurgical properties data, and C is the target sintered ore metallurgical properties data.

[0050] It should be noted that the data difference may also be represented by other methods known to those skilled in the art, and the above are only some examples.

[0051] Step S140: if the data difference exceeds the preset difference range, a sintered ore production warning message is generated to prompt the user to adjust the current sintered ore production ratio information to be adopted.

[0052] The preset difference interval can be a data interval with a maximum value and a minimum value set by technicians in this field based on the calculation method of data difference. In very strict scenarios, it can also be a preset difference threshold with the maximum value and the minimum value equal. If the data difference exceeds the tolerable interval, a sinter production warning message is generated to remind relevant personnel to adjust the current sinter production ratio information to avoid the sinter produced not meeting the requirements.

[0053] The sinter production warning message can be used to prompt relevant technical personnel in a manner known to those skilled in the art, such as system messages, sound and light warnings, emails, etc. As an example, when displaying the predicted sinter metallurgical performance data in real time, the data point can also be highlighted (such as red, bold, etc.) to provide a prompt.

[0054] This method can promptly identify ratio problems and inform relevant personnel during actual production.

[0055] In one embodiment, if the data difference exceeds a preset difference range, the method further includes: inputting the sintering production parameters to be predicted and the adjusted sintering ore production ratio information currently to be adopted into a pre-trained sintering ore metallurgical property prediction model to obtain new predicted sintering ore metallurgical property data, until the data difference between the new predicted sintering ore metallurgical property data and the preset sintering ore metallurgical property data falls into the preset difference range; determining the final adjusted sintering ore production ratio information currently to be adopted as the target sintering ore production ratio information; and producing sintering ore based on the target sintering ore production ratio information and the sintering production parameters to be predicted.

[0056] Of course, if it is found that the sintered ore produced by the current sintering production information may not meet the process requirements, then the metallurgical properties of the sintered ore finally produced can be adjusted by adjusting the current sintering ore production ratio information, and finally adjusted to a satisfactory state, that is, the data difference between the new predicted sintered ore metallurgical property data and the preset sintered ore metallurgical property data falls into the preset difference range. At this time, the adjusted current sintering ore production ratio information is used for production to ensure that the sintered ore actually produced meets the process requirements.

[0057] The sintered ore production method provided in the above embodiment obtains predicted sintered ore metallurgical property data by inputting the sintered production information to be predicted into a pre-trained sintered ore metallurgical property prediction model, and then determines the data difference between the predicted sintered ore metallurgical property data and the target sintered ore metallurgical property data. If the data difference exceeds the preset difference interval, a sintered ore production early warning message is generated, so as to prompt the current sintered ore production ratio information to be adopted in the predicted sintered production information to be adjusted through the sintered ore production early warning message, thereby realizing the prediction of the metallurgical properties of the sintered ore. When the sintered production information to be predicted is the sintering production information of the sintered ore to be produced in real time, the metallurgical properties of the sintered ore can be estimated in real time, and then the situation that does not meet the process requirements can be adjusted in time, thereby improving the stability of the performance indicators of the produced sintered ore and ensuring the smooth smelting of the blast furnace.

[0058] Optionally, through a pre-trained sinter metallurgical property prediction model, data association between sintering input parameters and output parameters can be achieved, so as to accurately predict the sinter metallurgical properties and control the stability and quality of the sinter raw materials.

[0059] The following is an exemplary description of the sintered ore production method provided above using a specific embodiment. Figure 2 , Figure 2 A specific flow chart of the sintered ore production method provided in one embodiment of the present application is shown in FIG. Figure 2 As shown, the process includes:

[0060] Step S210, collecting sintering production related data.

[0061] As an example, the sintering production related data includes but is not limited to the sintering ingredient single ratio, sintering ore mixture calculation value, sintering ore production operation parameters, sintering ore production ratio information, sintering process fuel flux data and sintering ore metallurgical performance test data. The sintering production related data can be collected by adding relevant material tracking collection modules and data collection modules.

[0062] The four categories of data are collected, namely, sintering material tracking data (calculated value of sintering ore mixture, fuel flux data of sintering process), sintering batching ratio, sintering production data (sintering ore production ratio information, sintering ore production operation parameters), and sintering metallurgical performance test data. The sintering material tracking data is stored in the message queue according to the time of parameter change, the sintering batching ratio data is stored in the message queue according to the batching execution time, the sintering production data is stored in the message queue according to the real-time production time of sintering, and the sintering metallurgical performance data is stored in the message queue according to the test time. The above-mentioned message queue can be a message queue corresponding to each type of data, or multiple types of data can correspond to the same message queue. The specific data can be selected by technicians in this field as needed.

[0063] Step S220, storing sintering production related data.

[0064] The collected historical sintering production related data are stored in the database.

[0065] Then, the data batches are associated for the input data in the message queue. The material tracking data based on equipment feedback has two important time parameters: production parameter change time and expected offline time (actual offline time). The production parameter change time is used as the timestamp for each batch of sintered ore. For the same batch of sintered materials, the sintering production parameters are associated according to the batching execution time in the sintered mineral material tracking data, and the inspection and testing data are associated according to the expected offline time of the finished product in the sintered mineral material tracking data. The completely processed sintering production-related data is stored in the target message queue for subsequent use.

[0066] Step S230: data preprocessing.

[0067] This process mainly includes data cleaning and batch association processing. Common data anomalies such as shutdown and equipment monitoring anomalies in production are cleaned, and batch association processing is performed on time-delayed data based on production material tracking equipment, and production parameters (sintering ingredient single ratio, sintering ore mixture calculation value, sintering ore production operation parameters, sintering ore production ratio information, sintering process fuel and flux data) are spliced ​​with the final sintering metallurgical performance inspection information.

[0068] An example method of data preprocessing is to clean the obvious equipment abnormality and shutdown data in the sintering production related data, and use the data processing 3-Sigma algorithm to preliminarily screen the data to avoid excessive data noise.

[0069] Step S240, establishing a prediction model for metallurgical properties of sintered ore based on residual neural network.

[0070] For example, by incorporating sinter ratio information, a sinter metallurgical property prediction model based on residual neural network can be built. The data set obtained in the above steps is divided into training data set and test data set according to the time sequence and proportion of parameter changes. The data set is divided into independent variables (sintering production information) and dependent variables (sinter metallurgical property test data).

[0071] In order to solve the problem of missing data in the basic property library of mineral powder due to its high price, based on the prior knowledge that the mineral powder in the ingredient list has similar basic properties, a deep learning algorithm based on residual neural network is proposed to fit the sintering ingredient list ratio and sintering production parameters (calculated value of sintered ore mixture, sintered ore production operation parameters, sintering process fuel and flux data, sintered ore production ratio information), aiming to generate intermediate characteristic indicators with high correlation with metallurgical performance targets, that is, to evaluate the performance of the mixed ore.

[0072] There are two paths in total for building the neural network model (metallurgical performance prediction model for ore formation). One is to obtain the correlation between sintering metallurgical performance and sintering production parameters and proportion (sintering ingredient single proportion), using the residual neural network block as the basic model block; the other path is to obtain the intermediate high correlation characteristic index, and the proportion data (sintering ingredient single proportion) is processed by convolution kernel alone, and then multiplied with the fitting result of another data after processing with the activation function. The common MSE is used as the loss function. The model after the training set converges is saved.

[0073] As an example, the method also includes putting all the test set data into the offline model (the trained ore metallurgical properties prediction model) to obtain the predicted values ​​of the dependent variables, comparing all the predicted values ​​of the dependent variables in the test set (the predicted sintered ore metallurgical properties test data) with the actual values ​​of the dependent variables (the actual sintered ore metallurgical properties test data), calculating the regression interval value hit rate and trend hit rate, and evaluating the effect of the trained ore metallurgical properties prediction model.

[0074] Step S250, data display.

[0075] For example, the real-time sintering production information and the corresponding inspection information of a batch of materials can be displayed in batches throughout the entire process from mixing and proportioning to sintering inspection, so as to monitor the metallurgical performance indicators of sintered ore in real time.

[0076] Based on the prediction results (predicted sintered ore metallurgical performance data) of the trained ore metallurgical performance prediction model obtained in step S204, the real-time prediction results are stored, and the real-time true and predicted value curve time series diagram is drawn for real-time monitoring. In addition to the target data and predicted values, the corresponding sintering production and batching parameters of the samples after batch association are supported to assist in controlling production information.

[0077] Taking the prediction of sinter metallurgical performance data as the prediction of drum index as an example, as an example, the data points with a data difference of more than 2% in the actual and predicted value curves can be highlighted. In addition, when the data difference exceeds the preset difference range, the predicted sinter metallurgical performance data can be highlighted.

[0078] The above embodiment provides a method for predicting the metallurgical properties of sintered ore based on residual neural network, by collecting the sintered mixed ore composition information (sintering ingredient single ratio, sintered ore mixture calculated value, sintering process fuel flux data), the operating parameters in the sintering process (sintered ore production operating parameters) and the mix ratio information three kinds of sintering production information (sintered ore production ratio information) as input data; at the same time, the ore powder ore phase information (sintering ingredient single ratio) is integrated as input; the drum index test value (sintered ore metallurgical performance data) of the metallurgical properties of the sintered ore is obtained, combined with the historical sintered ore information, the sintered ore process material tracking information (time change), to establish a deep learning network model, to construct a sintered ore metallurgical performance prediction model based on residual neural network, to predict the metallurgical properties of the sintered ore, such as the prediction of the drum index, and to display the time series data, to draw a time series data curve to display the production stability of the sintered ore and the future change trend, to carry out real-time monitoring of the metallurgical performance quality of the sintered ore and to observe the future trend changes. The input of the sinter metallurgical properties prediction model based on the residual neural network is combined with the material tracking information fed back by the hardware equipment. The impact of the time difference between the input data and the output results of each batch of sinter ore on the model is minimized by adding equipment, and accurate time series association is performed. In the form of residual neural networks, it is expected to learn the metallurgical properties and their important characteristic information, reduce the impact of the lack of basic characteristics of ore powder on the prediction of metallurgical properties, and realize online prediction of the metallurgical properties of sinter ore. During the determination of the sinter ore ratio (currently it is planned to use the sinter ore production ratio information), the sinter ore ratio can be tested and guided, so as to avoid the problem of unstable metallurgical performance quality caused by manual experience, and the quality of the sinter ore does not meet the standards, affecting the subsequent blast furnace smelting process.

[0079] In the process of establishing and training the sinter metallurgical performance prediction model based on residual neural network, by collecting sintering production data and material tracking data, the whole sintering process data from batching to inspection and testing is associated based on process rules, and the associated data is used to build and train the offline model. After the newly input samples are associated with data, they are put into prediction testing. The metallurgical performance index data under the current batching list (sintering batching list ratio) is estimated in real time and monitored to provide reference for operators, effectively identify abnormal deviations in metallurgical performance quality, and avoid quality problems of sintered ore products caused by manual experience. The degree of automation is high and the evaluation accuracy is high.

[0080] See also Figure 3 , Figure 3 Schematic diagram of a sintered ore production device provided in Example 1 of the present application. Figure 3 As shown, a sintered ore production device 300 is provided, including: an acquisition module 301, used to obtain sintering production information to be predicted, the sintering production information to be predicted includes sintering production parameters to be predicted and sintering ore production ratio information currently to be adopted; a prediction module 302, used to input the sintering production information to be predicted into a pre-trained sintering ore metallurgical property prediction model to obtain predicted sintering ore metallurgical property data, the pre-trained sintering ore metallurgical property prediction model is trained based on historical sintering production information and historical sintering ore metallurgical property test data; a difference determination module 303, used to determine the data difference between the predicted sintering ore metallurgical property data and the target sintering ore metallurgical property data; an adjustment module 304, used to generate a sintering ore production warning message if the data difference exceeds a preset difference interval, so as to prompt the current sintering ore production ratio information to be adopted to be adjusted through the sintering ore production warning message.

[0081] In one embodiment, the adjustment module is also used to input the sintering production parameters to be predicted and the adjusted sintering ore production ratio information currently to be adopted into a pre-trained sintering ore metallurgical property prediction model if the data difference exceeds a preset difference interval, to obtain new predicted sintering ore metallurgical property data, until the data difference between the new predicted sintering ore metallurgical property data and the preset sintering ore metallurgical property data falls into the preset difference interval; determine the final adjusted sintering ore production ratio information currently to be adopted as the target sintering ore production ratio information; and produce sintering ore based on the target sintering ore production ratio information and the sintering production parameters to be predicted.

[0082] In one embodiment, the device also includes a display module, which is used to input the sintering production information to be predicted into a pre-trained sintered ore metallurgical property prediction model to obtain the predicted sintered ore metallurgical property data, and then obtain the historical predicted sintered ore metallurgical property data output by the pre-trained sintered ore metallurgical property prediction model, as well as the real sintered ore metallurgical property data corresponding to each historical predicted sintered ore metallurgical property data; sort the predicted sintered ore metallurgical property data and the historical predicted sintered ore metallurgical property data according to the data time stamp to generate a predicted numerical curve time series diagram; sort the real sintered ore metallurgical property data according to the data time stamp to generate a real numerical curve time series diagram; and display the predicted numerical curve time series diagram and the real numerical curve time series diagram.

[0083] The specific definition of the sintered ore production device can be found in the definition of the sintered ore production method above, which will not be repeated here. Each module in the above-mentioned sintered ore production device can be implemented in whole or in part by software, hardware and a combination thereof. The above-mentioned modules can be embedded in or independent of the processor in the computer device in the form of hardware, or can be stored in the memory of the computer device in the form of software, so that the processor can call and execute the operations corresponding to the above modules.

[0084] See also Figure 4 , Figure 4 Schematic diagram of a sintered ore production system provided in Example 1 of the present application. Figure 4As shown, a sinter production system 400 is provided, including: a data acquisition module 410, which is used to collect sinter production information to be predicted, historical sinter production information, and historical sinter metallurgical performance test data of sinter produced based on the historical sinter production information, the historical sinter production information includes historical sinter mixture calculation values ​​of multiple batches of sinter, historical sinter production operation parameters, historical sintering batching ratios, historical sinter process fuel flux data, and historical sinter production ratio information, and the sinter production information to be predicted includes the sinter production parameters to be predicted and the sinter production ratio information currently to be adopted; a material tracking module 420, which is used to collect production parameter change time, batching execution time, sinter real-time production time, and material tracking information; The server 430 includes a data processing module 431, a data association module 432, a storage module 433, a training module 434, a test module 435 and an online evaluation module 436, wherein the data association module 432 is used to determine the production parameter change time of each batch of sintered ore as the time mark of the corresponding batch of sintered ore; based on the batching execution time and the real-time sintering production time, the historical sintering batching ratio is associated with the historical sintered ore production operation parameters; based on the actual offline time and inspection time of each batch of sintered ore, the historical sintered ore mixture calculation value, the historical sintering process fuel flux data and the historical sintered ore metallurgical performance inspection data are associated; based The historical sintered ore mixture calculation value, the historical sintering process fuel and flux data, the historical sintering batching ratio, and the historical sintered ore production operation parameters are associated with the time mark and the batching execution time, so as to realize the association between the historical sintered ore mixture calculation value, the historical sintered ore production operation parameters, the historical sintering batching ratio, the historical sintering process fuel and flux data, the historical sintered ore production ratio and the historical sintered ore metallurgical performance inspection data of each batch of sintered ore, and obtain multiple initial data sets; the data processing module 431 is used to perform preliminary data cleaning on the initial data sets of multiple batches according to the production abnormality period and time mark of the sintered ore; the initial data set after the preliminary data cleaning is cleaned according to the Laida criterion The data are screened out to obtain an initial data set after screening; the initial data set after screening is determined as training data to obtain multiple training data; the storage module 433 is used to store multiple initial data sets; the training module 434 is used to obtain multiple training data, the training data includes historical sintering production information and historical sintered ore metallurgical performance test data of the sintered ore produced based on the sintering production information, the historical sintering production information includes historical sintered ore mixture calculation value, historical sintered ore production operation parameters, historical sintering ingredient single ratio, historical sintering process fuel flux data and historical sintered ore production ratio information; the historical sintering production information is input into the first prediction sub-model based on the residual neural network for data fitting to obtain the data fitting result;Input the historical sintering ingredient list ratio in the historical sintering production information into the second prediction sub-model, so as to perform convolution kernel processing and activation function processing on the historical sintering ingredient list ratio in sequence to obtain processed data; multiply the processed data with the data fitting result, and determine the training sintered ore metallurgical property data based on the multiplication result; determine the loss function according to the training sintered ore metallurgical property data of all training data and the corresponding historical sintered ore metallurgical property inspection data; update the model parameters of the first prediction sub-model and the model parameters of the second prediction sub-model based on the loss function until convergence, and obtain the pre-trained sintered ore metallurgical property prediction model; wherein the sintered ore metallurgical property prediction model includes the first prediction sub-model and the second prediction sub-model; test model Block 435 is used to input the sintering production information to be predicted into the pre-trained sintering ore metallurgical performance prediction model to obtain the predicted sintering ore metallurgical performance data; the online evaluation module 436 is used to determine the data difference between the predicted sintering ore metallurgical performance data and the target sintering ore metallurgical performance data; if the data difference exceeds the preset difference range, a sintering ore production warning message is generated to prompt the current sintering ore production ratio information to be adjusted through the sintering ore production warning message; the front-end display module 440 is used to input the predicted sintering ore metallurgical performance data and history into the pre-trained sintering ore metallurgical performance prediction model to obtain the historical predicted sintering ore metallurgical performance data, sort them according to the data time mark, and display the time series data. ;

[0085] As an example, the data acquisition module includes multiple data sensing devices.

[0086] As an example, the material tracking module can also characterize the actual offline time by collecting the estimated offline time of the sintered ore finished product (estimated offline time).

[0087] As an example, the pre-trained sinter metallurgical property prediction model can be saved offline and used as needed.

[0088] For the specific definition of the sintered ore production system, please refer to the definition of the sintered ore production method above, which will not be repeated here. Each module in the above-mentioned sintered ore production system can be implemented in whole or in part by software, hardware and a combination thereof. The above-mentioned modules can be embedded in or independent of the processor in the computer device in the form of hardware, or can be stored in the memory of the computer device in the form of software, so that the processor can call and execute the operations corresponding to the above modules.

[0089] See also Figure 5The embodiment of the present application also provides an electronic device 1000, including a processor 1001, a memory 1002 and a communication bus 1003; the communication bus 1003 is used to connect the processor 1001 and the memory 1002; the processor 1001 is used to execute the computer program stored in the memory 1002 to implement one or more of the methods described in the above embodiments.

[0090] An embodiment of the present application further provides a computer-readable storage medium on which a computer program is stored, and the computer program is used to enable a computer to execute the method described in any one of the above embodiments.

[0091] The embodiment of the present application also provides a non-volatile readable storage medium, which stores one or more modules (programs). When the one or more modules are applied to a device, the device can execute instructions (instructions) of the steps included in the embodiment 1 of the embodiment of the present application.

[0092] It should be noted that the computer-readable medium mentioned above in the present disclosure may be a computer-readable signal medium or a computer-readable storage medium or any combination of the above two. The computer-readable storage medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared or semiconductor system, device or device, or any combination of the above. More specific examples of computer-readable storage media may include, but are not limited to: an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above. In the present disclosure, a computer-readable storage medium may be any tangible medium containing or storing a program that can be used by or in combination with an instruction execution system, device or device. In the present disclosure, a computer-readable signal medium may include a data signal propagated in a baseband or as part of a carrier wave, which carries a computer-readable program code. This propagated data signal may take a variety of forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination of the above. The computer readable signal medium may also be any computer readable medium other than a computer readable storage medium, which may send, propagate or transmit a program for use by or in conjunction with an instruction execution system, apparatus or device. The program code contained on the computer readable medium may be transmitted using any suitable medium, including but not limited to: wires, optical cables, RF (radio frequency), etc., or any suitable combination of the above.

[0093] The computer-readable medium may be included in the electronic device, or may exist independently without being incorporated into the electronic device.

[0094] Computer program code for performing the operations of the present disclosure may be written in one or more programming languages, or a combination thereof, including object-oriented programming languages, such as Java, Smalltalk, C++, and conventional procedural programming languages, such as "C" or similar programming languages. The program code may be executed entirely on the user's computer, partially on the user's computer, as a separate software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving a remote computer, the remote computer may be connected to the user's computer through any type of network, including a local area network (LAN) or a wide area network (WAN), or may be connected to an external computer (e.g., through the Internet using an Internet service provider).

[0095] The flowcharts and block diagrams in the accompanying drawings illustrate the possible implementation architecture, functions and operations of the methods and computer program products according to various embodiments of the present disclosure. In this regard, each box in the flowchart or block diagram can represent a module, a program segment or a part of a code, and the module, a program segment or a part of the code contains one or more executable instructions for implementing the specified logical functions. It should also be noted that in some alternative implementations, the functions marked in the box can also occur in an order different from that marked in the accompanying drawings. For example, two boxes represented in succession can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each box in the block diagram and / or flowchart, and the combination of boxes in the block diagram and / or flowchart, can be implemented with a dedicated hardware-based system that performs a specified function or operation, or can be implemented with a combination of dedicated hardware and computer instructions.

[0096] The above embodiments are merely illustrative of the principles and effects of the embodiments of the present application, and are not intended to limit the embodiments of the present application. Anyone familiar with the technology may modify or change the above embodiments without violating the spirit and scope of the embodiments of the present application. Therefore, all equivalent modifications or changes made by those with ordinary knowledge in the technical field without departing from the spirit and technical ideas disclosed in the embodiments of the present application shall still be covered by the claims of the embodiments of the present application.

Claims

1. A method for producing sintered ore, characterized in that: The method comprises: Acquiring the sintering production information to be predicted, wherein the sintering production information to be predicted includes the sintering production parameters to be predicted and the sintering ore production ratio information to be currently adopted; Inputting the sintering production information to be predicted into a pre-trained sintered ore metallurgical property prediction model to obtain predicted sintered ore metallurgical property data, wherein the pre-trained sintered ore metallurgical property prediction model is trained based on historical sintering production information and historical sintered ore metallurgical property test data; Determining the data difference between the predicted sintered ore metallurgical property data and the target sintered ore metallurgical property data; If the data difference exceeds a preset difference range, a sintered ore production warning message is generated to prompt the user to adjust the sintered ore production ratio information currently to be adopted through the sintered ore production warning message.

2. The method for producing sintered ore according to claim 1, characterized in that: The sintering production information to be predicted is input into a pre-trained sintered ore metallurgical property prediction model to obtain predicted sintered ore metallurgical property data, including: Inputting the sintering production information to be predicted into the first prediction sub-model to obtain a first prediction sub-result; Inputting the single ratio of the sintering ingredients to be predicted in the sintering production information to be predicted into the second prediction sub-model to obtain a second prediction sub-result; Determine predicted sinter metallurgical property data based on the first prediction sub-result and the second prediction sub-result; Among them, the sintered ore metallurgical property prediction model includes a first prediction sub-model and a second prediction sub-model, and the sintering production parameters to be predicted include the single ratio of sintering ingredients to be predicted, the calculated value of the sintered ore mixture to be predicted, the fuel and flux data of the sintering process to be predicted, and the sintering ore production operation parameters to be predicted.

3. The method for producing sintered ore according to claim 1, characterized in that: If the data difference exceeds a preset difference range, the method further includes: Inputting the sintering production parameters to be predicted and the adjusted current sintering ore production ratio information to be adopted into a pre-trained sintering ore metallurgical property prediction model to obtain new predicted sintering ore metallurgical property data, until the data difference between the new predicted sintering ore metallurgical property data and the preset sintering ore metallurgical property data falls into the preset difference range; Determine the final adjusted sinter production ratio information currently to be adopted as the target sinter production ratio information; Sintered ore is produced based on the target sintered ore production ratio information and the sintering production parameters to be predicted.

4. The method for producing sintered ore according to claim 1, characterized in that: The training method of the sintered ore metallurgical property prediction model includes: Acquire a plurality of training data, wherein the training data includes historical sintering production information and historical sintered ore metallurgical property test data of sintered ore produced based on the historical sintering production information, wherein the historical sintering production information includes historical sintered ore mixture calculation value, historical sintered ore production operation parameters, historical sintering ingredient single ratio, historical sintering process fuel flux data and historical sintered ore production ratio information; Inputting the historical sintering production information into a first prediction sub-model based on a residual neural network to perform data fitting to obtain a data fitting result; Inputting the historical sintering batch proportion in the historical sintering production information into the second prediction sub-model, so as to sequentially perform convolution kernel processing and activation function processing on the historical sintering batch proportion to obtain processed data; multiplying the processed data with the data fitting result, and determining the metallurgical property data of the training sintered ore based on the multiplication result; Determine the loss function based on the training sinter metallurgical property data of all training data and the corresponding historical sinter metallurgical property test data; Based on the loss function, the model parameters of the first prediction sub-model and the model parameters of the second prediction sub-model are updated until convergence, so as to obtain a pre-trained sinter metallurgical property prediction model; Wherein, the sintered ore metallurgical property prediction model includes a first prediction sub-model and a second prediction sub-model.

5. The method for producing sintered ore according to claim 4, characterized in that: The method for obtaining the training data includes: Obtain historical sinter mixture calculation values, historical sinter production operation parameters, historical sintering batch ratios, historical sintering process fuel and flux data, historical sinter production ratio information and historical sinter metallurgical performance test data of multiple batches of sinter; The historical sintered ore mixture calculation value and the historical sintering process fuel and flux data are stored in the first message queue according to the collected production parameter change time; Storing the historical sintering batching ratios in the second message queue according to the batching execution time; Storing the historical sintering ore production operation parameters in a third message queue according to the real-time sintering production time; Storing the historical sintered ore metallurgical property test data in the fourth message queue according to the test time; Determine the production parameter change time of each batch of sintered ore as the time mark of the corresponding batch of sintered ore; Associating the historical sintering batching ratio with the historical sintering ore production operation parameters based on the batching execution time and the real-time sintering production time; Based on the actual offline time and inspection time of each batch of sintered ore, the historical sintered ore mixture calculation value, the historical sintering process fuel and flux data and the historical sintered ore metallurgical performance inspection data are associated; Based on the time mark and the batching sheet execution time, the historical sintered ore mixture calculation value, the historical sintering process fuel and flux data are associated with the historical sintering batching sheet ratio, and the historical sintered ore production operation parameters, so as to achieve the association between the historical sintered ore mixture calculation value, the historical sintered ore production operation parameters, the historical sintering batching sheet ratio, the historical sintering process fuel and flux data, the historical sintering batching sheet ratio and the historical sintered ore metallurgical property inspection data of each batch of sintered ore, and obtain multiple initial data sets; Performing preliminary data cleaning on multiple batches of initial data sets according to the abnormal production period of the sintered ore and the time mark; The initial data set after preliminary data cleaning is filtered out according to the Laida criterion to obtain the filtered initial data set; The initial data set after screening is determined as the training data to obtain a plurality of training data.

6. The method for producing sintered ore according to any one of claims 1 to 5, characterized in that: After inputting the sintering production information to be predicted into a pre-trained sintered ore metallurgical property prediction model to obtain predicted sintered ore metallurgical property data, the method further includes: Obtaining historical predicted sintered ore metallurgical property data output by a pre-trained sintered ore metallurgical property prediction model, and actual sintered ore metallurgical property data corresponding to each historical predicted sintered ore metallurgical property data; The predicted sintered ore metallurgical property data and the historical predicted sintered ore metallurgical property data are sorted according to data time identifiers to generate a predicted value curve time series diagram; Sorting the real sintered ore metallurgical performance data according to the data time mark to generate a real numerical curve time series diagram; The predicted value curve timing diagram and the actual value curve timing diagram are displayed.

7. A sintered ore production device, characterized in that: The sintered ore production device comprises: An acquisition module is used to acquire the sintering production information to be predicted, wherein the sintering production information to be predicted includes the sintering production parameters to be predicted and the sintering ore production ratio information to be currently adopted; A prediction module, used for inputting the sintering production information to be predicted into a pre-trained sintered ore metallurgical property prediction model to obtain predicted sintered ore metallurgical property data, wherein the pre-trained sintered ore metallurgical property prediction model is trained based on historical sintering production information and historical sintered ore metallurgical property test data; A difference determination module, used to determine the data difference between the predicted sintered ore metallurgical property data and the target sintered ore metallurgical property data; The adjustment module is used to generate a sintered ore production warning message if the data difference exceeds a preset difference interval, so as to prompt the user to adjust the sintered ore production ratio information currently to be adopted through the sintered ore production warning message.

8. A sintered ore production system, characterized in that: The sintered ore production system comprises: A data acquisition module is used to collect the sintering production information to be predicted, the historical sintering production information, and the historical sintering ore metallurgical performance test data of the sintering ore produced based on the historical sintering production information, wherein the historical sintering production information includes the historical sintering ore mixture calculation values ​​of multiple batches of sintering ore, the historical sintering ore production operation parameters, the historical sintering ingredient single ratio, the historical sintering process fuel flux data, and the historical sintering ore production ratio information, and the sintering production information to be predicted includes the sintering production parameters to be predicted and the sintering ore production ratio information currently to be adopted; The material tracking module is used to collect the production parameter change time, batching list execution time, sintering real-time production time, inspection and testing time, and actual offline time; The server comprises a data processing module, a data association module, a storage module, a training module, a test module and an online evaluation module, wherein the data association module is used to determine the production parameter change time of each batch of sintered ore as the time mark of the corresponding batch of sintered ore; associate the historical sintering batching ratio with the historical sintered ore production operation parameters based on the batching execution time and the sintering real-time production time; associate the historical sintered ore mixture calculation value, the historical sintering process fuel and flux data with the historical sintered ore metallurgical performance test data based on the actual offline time and test time of each batch of sintered ore; associate the historical sintered ore mixture calculation value, the historical sintering process fuel and flux data with the historical sintered ore metallurgical performance test data based on the time mark and the batching execution time The calculated value, the fuel and flux data of the historical sintering process are associated with the historical sintering batch ratio and the historical sintering ore production operation parameters, so as to realize the association between the historical sintering ore mixture calculated value, the historical sintering ore production operation parameters, the historical sintering batch ratio, the historical sintering process fuel and flux data, the historical sintering ore production ratio and the historical sintering ore metallurgical performance test data of each batch of sintering ore, and obtain multiple initial data sets; the data processing module is used to perform preliminary data cleaning on the initial data sets of multiple batches according to the production abnormality period of the sintering ore and the time mark; the noise data is screened out from the initial data set after the preliminary data cleaning according to the Laida criterion to obtain the screened initial data set; The initial data set after screening is determined as the training data to obtain multiple training data; the storage module is used to store multiple initial data sets; the training module is used to obtain multiple training data, the training data includes historical sintering production information and historical sintered ore metallurgical performance test data of the sintered ore produced based on the sintering production information, the historical sintering production information includes historical sintered ore mixture calculation value, historical sintered ore production operation parameters, historical sintering ingredient single ratio, historical sintering process fuel flux data and historical sintered ore production ratio information; the historical sintering production information is input into the first prediction sub-model based on the residual neural network for data fitting to obtain data fitting results; the historical sintering production information is input into the first prediction sub-model based on the residual neural network ... The historical sintering ingredient list ratio in the information is input into the second prediction sub-model, so as to perform convolution kernel processing and activation function processing on the historical sintering ingredient list ratio in sequence to obtain processed data; the processed data is multiplied by the data fitting result, and the training sintered ore metallurgical property data is determined based on the multiplication result; the loss function is determined according to the training sintered ore metallurgical property data of all training data and the corresponding historical sintered ore metallurgical property inspection data; the model parameters of the first prediction sub-model and the model parameters of the second prediction sub-model are updated based on the loss function until convergence, so as to obtain a pre-trained sintered ore metallurgical property prediction model; wherein the sintered ore metallurgical property prediction model includes a first prediction sub-model and a second prediction sub-model;The test module is used to input the sintering production information to be predicted into a pre-trained sintering ore metallurgical property prediction model to obtain predicted sintering ore metallurgical property data; the online evaluation module is used to determine the data difference between the predicted sintering ore metallurgical property data and the target sintering ore metallurgical property data; if the data difference exceeds a preset difference range, a sintering ore production warning message is generated, so as to prompt the current sintering ore production ratio information to be adjusted through the sintering ore production warning message; The front-end display module is used to input the predicted sintered ore metallurgical performance data and history into a pre-trained sintered ore metallurgical performance prediction model to obtain the historical predicted sintered ore metallurgical performance data, sort them according to the data time mark, and display the time series data.

9. An electronic device, characterized in that: The system comprises a processor, a memory and a communication bus; the communication bus is used to connect the processor and the memory; the processor is used to execute a computer program stored in the memory to implement the method according to any one of claims 1 to 6.

10. A computer-readable storage medium, characterized in that: A computer program is stored thereon, and the computer program is used to make a computer execute the method according to any one of claims 1 to 6.

Images