Blast furnace condition identification and control method, system and equipment and storage medium

Abstract

The invention relates to the technical field of blast furnace ironmaking, and discloses a blast furnace condition identification and control method, system and equipment and a storage medium. The method comprises the following steps: acquiring blast furnace historical data, and performing data cleaning and preprocessing on the blast furnace historical data to obtain historical time sequence data; carrying out blast furnace adjustment capture and state and event identification on the historical time sequence data to obtain result time sequence data; constructing a deep learning model from a time sequence to a sequence; training a time sequence-to-sequence deep learning model based on the historical time sequence data and the result time sequence data to obtain a trained deep learning model; and inputting real-time acquired data into the trained deep learning model, and outputting a furnace condition recognition result and an operation suggestion result. According to the method, historical blast furnace data are collected, historical furnace condition changes are analyzed and captured, manual operation adjustment is carried out to learn manual operation experience, a trained deep learning model is obtained, and real-time collected data are input into the trained deep learning model to complete diagnosis and operation of the current furnace condition.

Description

Technical Field

[0001] The present invention relates to the technical field of blast furnace ironmaking, and in particular to a blast furnace condition identification and control method, system, equipment and storage medium. Background Art

[0002] At present, blast furnace production mainly relies on manual operation, but the experience of manual operation varies greatly, and there are problems such as manual judgment errors or irregular operations. It is urgent to realize real-time diagnosis and operation of blast furnace conditions, which is conducive to stable operation, so as to improve production efficiency and reduce energy consumption and emissions.

[0003] The factors for manually judging changes in blast furnace conditions and performing operations are basically obtained from monitoring data and material inspection data collected from various systems, equipment, modules, etc. of blast furnace production, based on comprehensive operating procedures, current operating policies, and operating experience formed by long-term operation learning. At present, various evaluation and analysis models can be established based on process mechanisms, data-driven methods, etc. to help understand blast furnace conditions, and logical process-type rules can be established based on experience to evaluate and judge various types of blast furnace conditions, but there is a lack of comprehensive methods that can combine blast furnace condition diagnosis and operation to learn or simulate manual operation. Summary of the invention

[0004] The main purpose of the present invention is to provide a blast furnace condition identification and control method, system, equipment and storage medium, aiming to solve at least one of the above-mentioned technical problems.

[0005] To achieve the above object, the present invention provides a blast furnace condition identification and control method, comprising:

[0006] Acquire historical data of the blast furnace, perform data cleaning and preprocessing on the historical data of the blast furnace, and obtain historical time series data;

[0007] Capturing blast furnace adjustment and identifying states and events on the historical time series data to obtain result time series data;

[0008] Build time series to sequence deep learning models;

[0009] Training the time series to sequence deep learning model based on the historical time series data and the result time series data to obtain a trained deep learning model;

[0010] The real-time collected data is input into the trained deep learning model, and the furnace condition identification results and operation suggestion results are output.

[0011] In some embodiments, the acquiring of blast furnace historical data, and performing data cleaning and preprocessing on the blast furnace historical data to obtain historical time series data include:

[0012] Select blast furnace monitoring parameters and inspection parameters for reference in blast furnace operation;

[0013] Collect historical data of different blast furnaces based on the blast furnace monitoring parameters and inspection parameters to obtain blast furnace historical data;

[0014] Perform at least one of abnormal monitoring and screening, noise filtering, missing data filling, abnormal filtering and data normalization operations on the blast furnace historical data to obtain historical time series data.

[0015] In some embodiments, the blast furnace monitoring parameters include blast furnace basic automation parameters and monitoring equipment parameters; wherein the blast furnace basic automation parameters include air supply, gas, charging, distribution, coal injection, water cooling and cooling wall temperature parameters, and the monitoring equipment parameters include tuyere imaging parameters and furnace top infrared imaging parameters;

[0016] The test parameters include material composition and performance parameters of coke, ore, coal powder and iron slag.

[0017] In some embodiments, the capturing of blast furnace adjustment and identification of status and events on the historical time series data to obtain result time series data includes:

[0018] Collect specific parameter data of air supply, material loading, coal injection or automation system according to the historical time series data to capture blast furnace adjustment;

[0019] Performing data analysis and logical judgment on the specific parameter data to obtain the changed adjustment parameter action amount;

[0020] Based on a preset data processing algorithm, a blast furnace process mechanism and an empirical model, the historical time series data is subjected to blast furnace status and event recognition to obtain status and event recognition results;

[0021] Result time series data is generated according to the changed adjustment parameter action amount and the state and event recognition results.

[0022] In some embodiments, the step of constructing a time series to sequence deep learning model includes:

[0023] Obtaining data length and data type according to the historical time series data and the result time series data;

[0024] A time series to sequence deep learning model is constructed based on the data length and the data type.

[0025] In some embodiments, the training of the time series to sequence deep learning model based on the historical time series data and the result time series data to obtain the trained deep learning model includes:

[0026] Obtaining integrated data according to the historical time series data and the result time series data;

[0027] extracting high-level abstract information from the integrated data;

[0028] Training the time series to sequence deep learning model according to the high-level abstract information in a self-supervised learning manner to obtain a pre-trained deep learning model;

[0029] The pre-trained deep learning model is fine-tuned based on the furnace condition diagnosis target and the operation adjustment target to obtain a trained deep learning model.

[0030] In some embodiments, fine-tuning the pre-trained deep learning model based on the furnace condition diagnosis target and the operation adjustment target to obtain the trained deep learning model includes:

[0031] According to the furnace condition diagnosis target and the operation adjustment target, a corresponding output node or network structure is added to the tail end of the pre-trained deep learning model to obtain an overall model;

[0032] The overall model is fine-tuned and trained using an artificially labeled dataset to obtain a trained deep learning model for furnace condition diagnosis and operation in blast furnace production.

[0033] In addition, to achieve the above-mentioned purpose, the present invention also proposes a blast furnace condition identification and control system, comprising:

[0034] A data acquisition and processing module is used to obtain the blast furnace historical data, perform data cleaning and preprocessing on the blast furnace historical data, and obtain historical time series data;

[0035] A data identification module is used to capture blast furnace adjustment and identify states and events of the historical time series data to obtain result time series data;

[0036] Model building module for building time series to sequence deep learning models;

[0037] A model training module, used for training the time series to sequence deep learning model based on the historical time series data and the result time series data to obtain a trained deep learning model;

[0038] The real-time processing module is used to input the real-time collected data into the trained deep learning model and output the furnace condition recognition results and operation suggestion results.

[0039] In addition, to achieve the above-mentioned purpose, the present invention also proposes an electronic device, which includes: a memory, a processor, and a blast furnace condition identification and control program stored in the memory and executable on the processor, wherein the blast furnace condition identification and control program is configured to implement the blast furnace condition identification and control method as described above.

[0040] In addition, to achieve the above-mentioned purpose, the present invention also proposes a storage medium, which stores a blast furnace condition identification and control program, and the blast furnace condition identification and control program is used to enable the processor to implement the blast furnace condition identification and control method described above when executing.

[0041] The present invention provides a blast furnace condition identification and control method, including: obtaining blast furnace historical data, performing data cleaning and preprocessing on the blast furnace historical data to obtain historical time series data; performing blast furnace adjustment capture and state and event recognition on the historical time series data to obtain result time series data; constructing a time series to sequence deep learning model; training the time series to sequence deep learning model based on the historical time series data and the result time series data to obtain a trained deep learning model; inputting real-time collected data into the trained deep learning model, and outputting furnace condition identification results and operation suggestion results. In the present invention, by collecting the historical time series data of the blast furnace, analyzing and capturing historical furnace condition changes and manual operation adjustments, in order to learn manual operation experience, a trained deep learning model is obtained, and the real-time collected data is input into the trained deep learning model to complete the diagnosis and operation of the current furnace condition, and the blast furnace condition changes are judged according to the monitoring data to perform targeted operations, thereby realizing auxiliary or closed-loop blast furnace production operations and achieving a stable and smooth effect. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] Figure 1 A schematic diagram of the structure of an electronic device in a hardware operating environment involved in an embodiment of the present invention;

[0043] Figure 2 A schematic flow chart of an embodiment of a method for identifying and controlling blast furnace conditions of the present invention;

[0044] Figure 3 It is a structural block diagram of an embodiment of a blast furnace condition identification and control system of the present invention;

[0045] Figure 4 It is a schematic diagram of the overall implementation of the system involved in the embodiment of the present invention.

[0046] The realization of the purpose, functional features and advantages of the present invention will be further explained in conjunction with embodiments and with reference to the accompanying drawings. DETAILED DESCRIPTION

[0047] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0048] It should be noted that all directional indications in the embodiments of the present invention (such as up, down, left, right, front, back, etc.) are only used to explain the relative position relationship, movement status, etc. between the components under a certain specific posture (as shown in the accompanying drawings). If the specific posture changes, the directional indication will also change accordingly.

[0049] In addition, the descriptions of "first", "second", etc. in the present invention are only used for descriptive purposes and cannot be understood as indicating or implying their relative importance or implicitly indicating the number of the indicated technical features. Therefore, the features defined as "first" and "second" may explicitly or implicitly include at least one of the features. In addition, the technical solutions between the various embodiments can be combined with each other, but it must be based on the ability of ordinary technicians in the field to implement. When the combination of technical solutions is contradictory or cannot be implemented, it should be deemed that such combination of technical solutions does not exist and is not within the scope of protection required by the present invention. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0050] Reference Figure 1 , Figure 1 The figure is a schematic diagram of the structure of an electronic device of the hardware operating environment involved in the embodiment of the present invention.

[0051] like Figure 1As shown, the electronic device may include: a processor 1001, such as a central processing unit (CPU), a communication bus 1002, a user interface 1003, a network interface 1004, and a memory 1005. Among them, the communication bus 1002 is used to realize the connection and communication between these components. The user interface 1003 may include a display screen (Display), an input unit such as a keyboard (Keyboard), and the optional user interface 1003 may also include a standard wired interface and a wireless interface. The network interface 1004 may optionally include a standard wired interface and a wireless interface (such as a wireless fidelity (Wireless-Fidelity, Wi-Fi) interface). The memory 1005 may be a high-speed random access memory (Random Access Memory, RAM memory) or a stable non-volatile memory (Non-Volatile Memory, NVM), such as a disk memory. The memory 1005 may also be a storage device independent of the aforementioned processor 1001.

[0052] Those skilled in the art will understand that Figure 1 The structure shown in the figure does not constitute a limitation on the electronic device, and may include more or less components than shown in the figure, or combine certain components, or arrange the components differently.

[0053] like Figure 1 As shown, the memory 1005 as a storage medium may include an operating system, a network communication module, a user interface module, and a blast furnace condition identification and control program.

[0054] exist Figure 1 In the electronic device shown, the network interface 1004 is mainly used for data communication with a network server; the user interface 1003 is mainly used for data interaction with a user; the processor 1001 and the memory 1005 in the electronic device of the present invention can be set in the electronic device, and the electronic device calls the blast furnace condition identification and control program stored in the memory 1005 through the processor 1001, and executes the blast furnace condition identification and control method provided in an embodiment of the present invention.

[0055] The present invention provides a blast furnace condition identification and control method, system, equipment and storage medium.

[0056] The embodiment of the present invention provides a blast furnace condition identification and control method, referring to Figure 2 , Figure 2 The figure is a flow chart of an embodiment of a method for identifying and controlling blast furnace conditions of the present invention.

[0057] like Figure 2 As shown, the blast furnace condition identification and control method comprises:

[0058] Step S100: acquiring blast furnace historical data, performing data cleaning and preprocessing on the blast furnace historical data, and obtaining historical time series data;

[0059] Step S200: Capturing blast furnace adjustment and identifying states and events on the historical time series data to obtain result time series data;

[0060] Step S300: constructing a time series to sequence deep learning model;

[0061] Step S400: training the time series to sequence deep learning model based on the historical time series data and the result time series data to obtain a trained deep learning model;

[0062] Step S500: input the real-time collected data into the trained deep learning model, and output the furnace condition recognition results and operation suggestion results.

[0063] It should be noted that the execution subject in this embodiment may be an electronic device, which may be a computer device with data processing functions, or other devices that can achieve the same or similar functions. This embodiment does not limit this. In this embodiment, a computer device is used as an example for explanation.

[0064] In one embodiment, historical data of a blast furnace is obtained, and the historical data of the blast furnace is cleaned and preprocessed to obtain historical time series data, including: selecting blast furnace monitoring parameters and inspection parameters for reference of blast furnace operation; collecting historical data of different blast furnaces based on the blast furnace monitoring parameters and inspection parameters to obtain blast furnace historical data; performing at least one of abnormal monitoring screening, noise filtering, missing data filling, abnormal filtering and data normalization operations on the blast furnace historical data to obtain historical time series data.

[0065] In one embodiment, the blast furnace monitoring parameters include basic blast furnace automation parameters and monitoring equipment parameters; wherein, the basic blast furnace automation parameters include air supply, gas, loading, distribution, coal injection, water cooling and cooling wall temperature parameters, and the monitoring equipment parameters include tuyere imaging parameters and furnace top infrared imaging parameters; the inspection parameters include material composition and performance parameters of coke, ore, coal powder and iron slag.

[0066] Specifically, historical data of blast furnaces are collected, and data cleaning and preprocessing are performed: in this embodiment, monitoring parameters and inspection parameters that are referenced by the foreman (such as blast furnace operator) for general operations are selected, and long-term historical data of different blast furnaces are collected based on the selected parameters, and the collected historical data are cleaned and preprocessed.

[0067] Exemplarily, the selected parameters include, but are not limited to, basic automation parameters of blast furnaces (e.g., air supply, gas, charging, material distribution, coal injection, water cooling, cooling wall temperature, etc.), monitoring equipment parameters (e.g., tuyere imaging, furnace top infrared imaging, etc.), and inspection and testing parameters (e.g., material composition and performance parameters of coke, ore, coal powder, iron slag, etc.). Thus, the long-term historical data collected based on the selected parameters include, but are not limited to, basic automation data of blast furnaces (e.g., time series data of parameters such as air supply, gas, charging, material distribution, coal injection, water cooling, cooling wall temperature, etc.), monitoring equipment data (e.g., unstructured data such as tuyere imaging, furnace top infrared imaging, etc.), and inspection and testing data (e.g., material composition and performance parameter data of coke, ore, coal powder, iron slag, etc.).

[0068] It is understandable that the historical operation data of the blast furnace is collected, including but not limited to temperature, pressure, gas composition, charge composition, operating parameters, etc., and this embodiment does not impose any limitation on this.

[0069] Exemplarily, data cleaning and preprocessing include abnormal monitoring screening, noise filtering and missing data filling for time series data, abnormal filtering and data normalization for video stream data, etc. Among them, the model of this embodiment can process input time series data and / or input video stream data.

[0070] It should be noted that data cleaning includes but is not limited to processing missing values, outliers and noise, etc. Preprocessing includes but is not limited to standardizing or normalizing data to make it suitable for input of subsequent machine learning models. This embodiment does not limit the specific operations of data cleaning and preprocessing.

[0071] In one embodiment, the historical time series data is subjected to blast furnace adjustment capture and state and event identification to obtain result time series data, including: collecting specific parameter data of air supply, charging, coal injection or automation system according to the historical time series data to capture blast furnace adjustment; performing data analysis and logical judgment on the specific parameter data to obtain the changed adjustment parameter action amount; based on a preset data processing algorithm, blast furnace process mechanism and empirical model, the historical time series data is subjected to blast furnace state and event identification to obtain state and event identification results; and generating result time series data according to the changed adjustment parameter action amount and the state and event identification results.

[0072] Specifically, adjustment capture and status and event identification: the various types of historical data after processing are subjected to operation adjustment capture and status and event identification, and after sorting and summarizing, the identification results of the time series data type (result time series data) are obtained.

[0073] Exemplarily, the operation adjustment capture is to obtain the amount of the adjustment parameter action that has changed by collecting specific parameter data, performing reasonable data analysis and logical judgment. The specific parameters include but are not limited to parameters such as air supply, charging and coal injection, or set parameters collected by the automated process system. Through adjustment capture, the amount of the adjustment parameter action that has changed in the actual production of the blast furnace is obtained through data analysis and logical judgment from historical data or real-time data.

[0074] Exemplarily, blast furnace state and event recognition is to identify and judge various blast furnace states and events through data analysis, according to certain data processing methods (preset data processing algorithms) and blast furnace process mechanisms and empirical models. Among them, the preset data processing algorithm can consider using time series analysis models, recurrent neural networks (RNN), long short-term memory networks (LSTM), gated recurrent units (GRU), Transformer-based models, support vector machines (SVM), random forests, gradient boosting trees (GBDT), etc.; blast furnace process mechanism models are based on classical mechanisms such as material balance, thermal balance, chemical balance, reaction rate and flow field heat and mass transfer, or various mathematical models established in combination with on-site statistical laws and process theories, which can be used to analyze and judge the current blast furnace status in various aspects, and help process guidance and control by outputting reasonable analysis and calculation data results.

[0075] Specifically, the blast furnace status includes but is not limited to high pressure difference, airflow distribution of tuyere, furnace body and furnace top, gas utilization level, iron-tapping status, furnace temperature trend, furnace hearth activity and soft melting zone area, etc. Blast furnace events include but are not limited to wind stop, wind resumption, slow wind, slag skin shedding, heat load fluctuation, furnace wall thickness, hot blast furnace replacement, gas composition analyzer purging and charge composition changes, etc.

[0076] It can be understood that, in one example, the blast furnace status and event identification is performed through the blast furnace historical data, and the following data analysis and machine learning steps can be adopted: collect the historical operation data of the blast furnace including temperature, pressure, gas composition, charge composition, operating parameters, etc., clean the data and standardize or normalize the data to make it suitable for subsequent machine learning models; select features related to the blast furnace status and events according to the mechanism and experience of blast furnace operation, and use statistical methods or machine learning techniques to extract time series features, frequency domain features, statistical features, etc.; select a suitable machine learning model according to the nature of the problem. For time series data, you can consider using time series analysis models, recurrent neural networks, long short-term memory networks, gated recurrent units or Transformer-based models. For classification problems, you can use support vector machines (SVM), random forests, gradient boosting trees (GBDT), etc.; use historical data to train the model, and divide the data set into training set, validation set and test set, adjust the model parameters, and use the validation set to evaluate the model performance to prevent overfitting; use the trained model to identify the status of the real-time data of the blast furnace, such as normal, abnormal, faulty and other states. According to the output of the model, the operation status of the blast furnace is monitored and warned in real time; the model is used to identify specific events, such as deterioration of furnace conditions, equipment failures, production anomalies, etc. For complex events, it may be necessary to build a multi-model system or use an ensemble method. The model is deployed in the production environment to achieve real-time blast furnace status monitoring and event identification, continuously monitor the performance of the model, and regularly use new data to update and optimize the model. Through this process, the historical data of the blast furnace can be used to identify the operation status of the blast furnace and possible events, thereby improving the production efficiency and safety level of the blast furnace.

[0077] In one embodiment, a time series to sequence deep learning model is constructed, including: obtaining data length and data type according to the historical time series data and the result time series data; and constructing a time series to sequence deep learning model based on the data length and the data type.

[0078] Specifically, a time series to sequence deep learning model is established to learn the characteristics of blast furnace data: the historical time series data after data cleaning and preprocessing and the resulting time series data obtained by blast furnace adjustment capture and status and event identification are organized, and according to the length and number of types of these data, a time series to sequence deep learning model with corresponding input sequence length and appropriate size architecture is designed to complete the establishment of the model.

[0079] Exemplarily, the time series to sequence deep learning model is a model that has undergone certain network structure design and hyperparameter training adjustment, inputs time series data that has been cleaned, preprocessed, and integrated, and can achieve end-to-end output of time series or timestamp prediction results. For example, a time series to sequence deep learning model for blast furnace condition diagnosis and operation can be designed using a Transformer architecture, or other types of time series to sequence deep learning models such as a time convolutional network TCN can be used, and this embodiment does not limit this.

[0080] In one embodiment, the time series to sequence deep learning model is trained based on the historical time series data and the result time series data to obtain a trained deep learning model, including: obtaining integrated data based on the historical time series data and the result time series data; extracting high-level abstract information from the integrated data; training the time series to sequence deep learning model based on the high-level abstract information based on a self-supervised learning method to obtain a pre-trained deep learning model; and fine-tuning the pre-trained deep learning model based on a furnace condition diagnosis target and an operation adjustment target to obtain a trained deep learning model.

[0081] Specifically, the self-supervised learning method is adopted to allow the established deep learning model (time series to sequence deep learning model) to extract high-level abstract information from the integrated data, learn a large number of hidden features of blast furnace data, and complete the pre-training of the established model. Among them, the integrated data is the historical time series data that has been cleaned and preprocessed and the result time series data that has been captured by blast furnace adjustment and recognized by status and events.

[0082] Exemplarily, the task methods of self-supervised learning include but are not limited to: using past sequences to predict future time series or specific timestamps, predicting invisible parts of masked time series, reconstructing data by minimizing errors, determining whether different variable sampling sequences are the same time window, determining whether different subsequences are the same variable, determining whether masked subsequences are true, etc.

[0083] In one embodiment, the pre-trained deep learning model is fine-tuned based on the furnace condition diagnosis target and the operation adjustment target to obtain a trained deep learning model, including: adding corresponding output nodes or network structures at the end of the pre-trained deep learning model according to the furnace condition diagnosis target and the operation adjustment target to obtain an overall model; fine-tuning the overall model using an artificially labeled data set to obtain a trained deep learning model for furnace condition diagnosis and operation in blast furnace production.

[0084] It should be noted that the furnace condition diagnosis target and the operation adjustment target are fine-tuned: for the furnace condition diagnosis target and the operation adjustment target, the corresponding output node or several layers of network structure are added to the tail end of the pre-trained model, and a small amount of manually labeled data sets are used to fine-tune the overall model to obtain a trained model that can be used for furnace condition diagnosis and operation in blast furnace production (trained deep learning model). Among them, operation adjustment refers to the numerical adjustment actions made to various adjustment parameters in blast furnace operation based on the current changes in the blast furnace operating state and the occurrence of special furnace conditions in order to stabilize production operations.

[0085] For example, the labeled data set can be obtained by manually finding typical target cases from historical data and labeling them. After marking the event periods for all time series data in the labeled data set, they are organized into a training data set. In this way, the model is trained based on the training data set, and the model can learn different parameter fluctuation characteristics that are potentially associated with furnace condition events. In one example, the labeled data set can be the time period when material slippage occurs based on the measurement data of the blast furnace top probe device.

[0086] In one embodiment, the real-time collected data is input into the trained deep learning model, and the furnace condition recognition results and operation suggestion results are output.

[0087] Specifically, real-time collected data is input, and furnace condition identification and operation suggestion results are given: the real-time collected data is cleaned and preprocessed as well as the adjustment capture and status and event identification processes in the above steps are performed, the collected real-time time series data is processed and integrated, and input into the trained model (deep learning model after training). After model calculation, the current furnace condition identification and operation suggestion results are obtained.

[0088] It is understandable that when the model is put into production, data and manual labeling results can be collected regularly, and the trained model (trained deep learning model) can be further trained and fine-tuned to make its output results more in line with production reality.

[0089] It should be noted that the above method of this embodiment can learn manual operation experience by collecting historical data, analyzing and capturing historical furnace condition changes and manual operation adjustments, and can complete the diagnosis and operation of the current furnace condition in real time based on the collected monitoring data, realize auxiliary or closed-loop blast furnace production operation, and achieve a stable and smooth effect.

[0090] The present embodiment provides a method for identifying and controlling a blast furnace condition, including: obtaining blast furnace historical data, performing data cleaning and preprocessing on the blast furnace historical data, and obtaining historical time series data; performing blast furnace adjustment capture and state and event recognition on the historical time series data, and obtaining result time series data; constructing a time series to sequence deep learning model; training the time series to sequence deep learning model based on the historical time series data and the result time series data, and obtaining a trained deep learning model; inputting real-time collected data into the trained deep learning model, and outputting furnace condition recognition results and operation suggestion results. In the present embodiment, by collecting the historical time series data of the blast furnace, analyzing and capturing the historical furnace condition changes and manual operation adjustments, in order to learn manual operation experience, a trained deep learning model is obtained, and the real-time collected data is input into the trained deep learning model to complete the diagnosis and operation of the current furnace condition, and the blast furnace condition changes are judged according to the monitoring data to perform targeted operations, thereby realizing auxiliary or closed-loop blast furnace production operations, and achieving a stable and smooth effect.

[0091] In addition, an embodiment of the present invention further proposes a storage medium, on which a blast furnace condition identification and control program is stored. When the blast furnace condition identification and control program is executed by a processor, the steps of the blast furnace condition identification and control method described above are implemented.

[0092] Reference Figure 3 , Figure 3 It is a structural block diagram of an embodiment of a blast furnace condition identification and control system of the present invention.

[0093] like Figure 3 As shown, the blast furnace condition identification and control system comprises:

[0094] The data acquisition and processing module 10 is used to obtain the blast furnace historical data, perform data cleaning and preprocessing on the blast furnace historical data, and obtain historical time series data;

[0095] A data identification module 20 is used to capture blast furnace adjustment and identify states and events of the historical time series data to obtain result time series data;

[0096] A model building module 30, for building a time series to sequence deep learning model;

[0097] A model training module 40 is used to train the time series to sequence deep learning model based on the historical time series data and the result time series data to obtain a trained deep learning model;

[0098] The real-time processing module 50 is used to input the real-time collected data into the trained deep learning model and output the furnace condition recognition results and operation suggestion results.

[0099] It is understandable that in the system, the data acquisition and processing module 10 may include a data acquisition and storage module and a data processing module. Figure 4 As shown, the data acquisition and storage module is used to collect the monitoring parameters and test parameter data required by the method of the present invention in real time, and store the original data and the data processed by other modules in the database. Exemplarily, the stored data formats include time series, relational, and video stream types.

[0100] Specifically, the data processing module is used to perform operation adjustment capture and state and event identification on various types of historical data after processing, and obtain the identification result of the time series data type after sorting and summarizing. Among them, operation adjustment capture is to collect specific parameter data, perform reasonable data analysis and logical judgment, and obtain the changed adjustment parameter action amount. Exemplarily, specific parameters include parameters such as air supply, feeding and coal injection, or set parameters collected by the automation process system. Blast furnace state and event identification is to identify and judge various blast furnace states and events through data analysis based on certain data processing methods, blast furnace process mechanisms and empirical models. Blast furnace states include but are not limited to high pressure difference, air flow distribution of tuyere, furnace body and furnace top, gas utilization level, iron tapping state, furnace temperature trend, furnace hearth activity, soft melting zone area, etc. Blast furnace events include but are not limited to wind stop, rewind, slow wind, slag skin falling off, heat load fluctuation, furnace wall thickness, hot blast furnace replacement, gas composition analyzer purging, charge composition changes, etc.

[0101] Specifically, the data identification module 20 may include a transfer capture module. Figure 4 As shown, the adjustment capture module is used to obtain the adjustment parameter action amount that has changed in the actual production of the blast furnace through data analysis and logical judgment from historical data or real-time data. Among them, the adjustment parameters include but are not limited to cold air flow, hot air pressure, hot air temperature, coal injection flow, furnace top pressure, oxygen enrichment flow, charging ratio and distribution matrix.

[0102] Specifically, the data identification module 20 may include a blast furnace process mechanism and an empirical model. The blast furnace process mechanism model is based on classical mechanisms such as material balance, thermal balance, chemical balance, reaction rate, and flow field heat and mass transfer, or various mathematical models established in combination with on-site statistical laws and process theories, which can be used to analyze and judge the current blast furnace status in various aspects, and help process guidance and control by outputting reasonable analysis and calculation data results. Among them, the blast furnace process mechanism model includes but is not limited to a furnace top airflow identification model, a tuyere airflow identification model, a furnace body airflow analysis model, a heat load calculation model, a slag skin shedding and furnace wall thickness identification model, a furnace temperature trend identification model, a furnace temperature field and a soft melting zone calculation model, a distribution simulation and a charge tracking model, etc.

[0103] Specifically, the model building module 30 may include a time series deep learning model. The time series deep learning model is a model that has undergone certain network structure design and hyperparameter training adjustment, inputs time series data that has been cleaned, preprocessed and integrated, and can achieve end-to-end output of time series or timestamp prediction results. Exemplarily, a time series deep learning model for blast furnace condition diagnosis and operation can be designed using a Transformer architecture.

[0104] Specifically, the system also includes a manual marking module, which can help blast furnace operators or managers quickly analyze historical data and mark data at the time or time period of target events, adjustment actions, status changes, etc. This module allows data marking work without the need for programming technology or professional software technology foundation.

[0105] The above-mentioned system in this embodiment can learn manual operation experience by collecting historical data, analyzing and capturing historical furnace condition changes and manual operation adjustments, and can complete the diagnosis and operation of the current furnace condition in real time based on the collected monitoring data, realize auxiliary or closed-loop blast furnace production operation, and achieve a stable and smooth effect.

[0106] It should be noted that the technical details that are not described in detail in the embodiment of the blast furnace condition identification and control system can be referred to the blast furnace condition identification and control method as described above provided in any embodiment of the present invention, and will not be repeated here.

[0107] It should be understood that the above is only an example and does not constitute any limitation on the technical solution of the present invention. In specific applications, technicians in this field can make settings as needed, and the present invention does not limit this.

[0108] It should be noted that the workflow described above is merely illustrative and does not limit the scope of protection of the present invention. In practical applications, technicians in this field can select part or all of them according to actual needs to achieve the purpose of the present embodiment, and no limitation is made here.

[0109] In addition, it should be noted that, in this article, the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, article or system including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or system. In the absence of further restrictions, an element defined by the sentence "comprises a ..." does not exclude the existence of other identical elements in the process, method, article or system including the element.

[0110] The serial numbers of the above embodiments of the present invention are only for description and do not represent the advantages or disadvantages of the embodiments.

[0111] Through the description of the above implementation methods, those skilled in the art can clearly understand that the above-mentioned embodiment methods can be implemented by means of software plus a necessary general hardware platform, and of course by hardware, but in many cases the former is a better implementation method. Based on such an understanding, the technical solution of the present invention, or the part that contributes to the prior art, can be embodied in the form of a software product, which is stored in a storage medium (such as a read-only memory (ROM) / RAM, a magnetic disk, or an optical disk), and includes a number of instructions for a terminal device (which can be a mobile phone, a computer, a server, or a network device, etc.) to execute the methods described in each embodiment of the present invention.

[0112] The above are only preferred embodiments of the present invention, and are not intended to limit the patent scope of the present invention. Any equivalent structure or equivalent process transformation made using the contents of the present invention specification and drawings, or directly or indirectly applied in other related technical fields, are also included in the patent protection scope of the present invention.

Claims

1. A blast furnace condition identification and control method, characterized in that: include: Acquire historical data of the blast furnace, perform data cleaning and preprocessing on the historical data of the blast furnace, and obtain historical time series data; Capturing blast furnace adjustment and identifying states and events on the historical time series data to obtain result time series data; Build time series to sequence deep learning models; Training the time series to sequence deep learning model based on the historical time series data and the result time series data to obtain a trained deep learning model; The real-time collected data is input into the trained deep learning model, and the furnace condition identification results and operation suggestion results are output.

2. The method according to claim 1, characterized in that The acquiring of blast furnace historical data, and performing data cleaning and preprocessing on the blast furnace historical data to obtain historical time series data include: Select blast furnace monitoring parameters and inspection parameters for reference in blast furnace operation; Collect historical data of different blast furnaces based on the blast furnace monitoring parameters and inspection parameters to obtain blast furnace historical data; Perform at least one of abnormal monitoring and screening, noise filtering, missing data filling, abnormal filtering and data normalization operations on the blast furnace historical data to obtain historical time series data.

3. The method according to claim 2, characterized in that The blast furnace monitoring parameters include blast furnace basic automation parameters and monitoring equipment parameters; wherein the blast furnace basic automation parameters include air supply, gas, charging, distribution, coal injection, water cooling and cooling wall temperature parameters, and the monitoring equipment parameters include tuyere imaging parameters and furnace top infrared imaging parameters; The test parameters include material composition and performance parameters of coke, ore, coal powder and iron slag.

4. The method according to claim 1, characterized in that The blast furnace adjustment capture and state and event identification are performed on the historical time series data to obtain result time series data, including: Collect specific parameter data of air supply, material loading, coal injection or automation system according to the historical time series data to capture blast furnace adjustment; Performing data analysis and logical judgment on the specific parameter data to obtain the changed adjustment parameter action amount; Based on a preset data processing algorithm, a blast furnace process mechanism and an empirical model, the historical time series data is subjected to blast furnace status and event recognition to obtain status and event recognition results; Result time series data is generated according to the changed adjustment parameter action amount and the state and event recognition results.

5. The method according to claim 1, characterized in that The construction of a time series to sequence deep learning model includes: Obtaining data length and data type according to the historical time series data and the result time series data; A time series to sequence deep learning model is constructed based on the data length and the data type.

6. The method according to claim 1, characterized in that The step of training the time series to sequence deep learning model based on the historical time series data and the result time series data to obtain a trained deep learning model includes: Obtaining integrated data according to the historical time series data and the result time series data; extracting high-level abstract information from the integrated data; Training the time series to sequence deep learning model according to the high-level abstract information in a self-supervised learning manner to obtain a pre-trained deep learning model; The pre-trained deep learning model is fine-tuned based on the furnace condition diagnosis target and the operation adjustment target to obtain a trained deep learning model.

7. The method according to claim 6, characterized in that The pre-trained deep learning model is fine-tuned based on the furnace condition diagnosis target and the operation adjustment target to obtain a trained deep learning model, including: According to the furnace condition diagnosis target and the operation adjustment target, a corresponding output node or network structure is added to the tail end of the pre-trained deep learning model to obtain an overall model; The overall model is fine-tuned and trained using an artificially labeled dataset to obtain a trained deep learning model for furnace condition diagnosis and operation in blast furnace production.

8. A blast furnace condition identification and control system, characterized in that: include: A data acquisition and processing module is used to obtain the blast furnace historical data, perform data cleaning and preprocessing on the blast furnace historical data, and obtain historical time series data; A data identification module is used to capture blast furnace adjustment and identify states and events of the historical time series data to obtain result time series data; Model building module for building time series to sequence deep learning models; A model training module, used for training the time series to sequence deep learning model based on the historical time series data and the result time series data to obtain a trained deep learning model; The real-time processing module is used to input the real-time collected data into the trained deep learning model and output the furnace condition recognition results and operation suggestion results.

9. An electronic device, characterized in that: The electronic device includes: a memory, a processor, and a blast furnace condition identification and control program stored in the memory and executable on the processor, wherein the blast furnace condition identification and control program is configured to implement a blast furnace condition identification and control method as described in any one of claims 1 to 7.

10. A storage medium, characterized in that: The storage medium stores a blast furnace condition identification and control program, and the blast furnace condition identification and control program is used to enable the processor to implement the blast furnace condition identification and control method as described in any one of claims 1 to 7 when executed.

Images