A fully intelligent online furnace start-up system for blast furnace oxygen lance

Abstract

The present application relates to the field of intelligent control, and specifically discloses a fully intelligent online furnace opening system for blast furnace oxygen lances, which obtains the gun body temperature and furnace temperature of the blast furnace oxygen lance collected in real time by a temperature sensor, and uses a data processing and analysis algorithm based on a deep learning neural network to perform a time series feature analysis on the gun body temperature and the furnace temperature, thereby intelligently obtaining an intelligent control instruction for the oxygen flow value at the current time point to increase, decrease or maintain according to the time series interactive response between the gun body temperature and the furnace temperature. In this way, the correlation between oxygen flow and temperature can be grasped more accurately, and intelligent oxygen flow control can be achieved, thereby effectively reducing energy waste, reducing dependence on operators, and improving the utilization efficiency of oxygen in the blast furnace ironmaking process, making the blast furnace ironmaking process more stable and efficient.

Description

Technical Field

[0001] The present application relates to the field of intelligent control, and more specifically, to a fully intelligent online furnace opening system for a blast furnace oxygen lance. Background Art

[0002] The blast furnace oxygen lance is a process equipment mainly used in the oxygen converter steelmaking process. Its function is to blow high-pressure and high-purity oxygen at supersonic speed into the top of the molten metal pool in the converter. It is also equipped with a high-pressure water cooling protection system. The oxygen lance can withstand extremely high temperatures and the erosion of molten metal, and is an indispensable equipment in the steelmaking process.

[0003] The opening of the blast furnace oxygen lance refers to the first injection of a certain amount of oxygen into the blast furnace during the blast furnace ironmaking process. This is an important step in blast furnace production, marking the official start of blast furnace production. The timing of opening the blast furnace oxygen lance depends on the specific conditions of the blast furnace, usually after the blast furnace reaches a certain temperature and the furnace condition is stable. Oxygen injection can speed up the blast furnace reaction, increase the output of the blast furnace, prevent the charge from agglomerating, and stabilize the blast furnace operation.

[0004] However, it is often difficult to accurately control the amount of oxygen in traditional blast furnace oxygen lances when the furnace is turned on, resulting in excessive or insufficient oxygen supply, resulting in uneven distribution of oxygen inside the blast furnace, thereby increasing energy consumption costs and affecting the effect of the combustion reaction and the stability of the temperature inside the furnace. In addition, the oxygen flow control of traditional oxygen lances usually requires experienced operators to adjust, which is complicated to operate and prone to misoperation, further affecting the efficiency of ironmaking.

[0005] Therefore, an optimized blast furnace oxygen lance fully intelligent online furnace start-up system is desired. Summary of the invention

[0006] In order to solve the above technical problems, the present application is proposed. The embodiment of the present application provides a fully intelligent online furnace opening system for a blast furnace oxygen lance, which obtains the gun body temperature and the furnace temperature of the blast furnace oxygen lance collected in real time by a temperature sensor, and uses a data processing and analysis algorithm based on a deep learning neural network to perform a time series feature analysis on the gun body temperature and the furnace temperature, thereby intelligently obtaining the intelligent control instructions for the oxygen flow value at the current time point to increase, decrease or maintain according to the time series interactive response between the gun body temperature and the furnace temperature. In this way, the correlation between the oxygen flow rate and the temperature can be grasped more accurately, and intelligent oxygen flow rate control can be achieved, thereby effectively reducing energy waste, reducing dependence on operators, and improving the utilization efficiency of oxygen in the blast furnace ironmaking process, making the blast furnace ironmaking process more stable and efficient.

[0007] According to one aspect of the present application, a blast furnace oxygen lance fully intelligent online furnace start-up system is provided, which includes:

[0008] The blast furnace oxygen lance temperature acquisition module is used to obtain the time series of the blast furnace oxygen lance body temperature and the time series of the furnace temperature collected by the temperature sensor;

[0009] A time dimension arrangement module, used for arranging the time series of the gun body temperature and the time series of the temperature inside the furnace according to the time dimension into a gun body temperature time series input vector and a furnace temperature time series input vector, and then performing a difference calculation between the characteristic values ​​of two adjacent positions to obtain a gun body temperature time series fluctuation input vector and a furnace temperature time series fluctuation input vector;

[0010] A gun body temperature multi-scale feature cascade module, used for cascading the gun body temperature time series input vector and the gun body temperature time series fluctuation input vector to obtain a gun body temperature multi-scale time series input vector;

[0011] A furnace temperature multi-scale feature cascade module, used for cascading the furnace temperature time series fluctuation input vector and the furnace temperature time series input vector to obtain a furnace temperature multi-scale time series input vector;

[0012] A time series correlation pattern feature extraction module is used to extract time series correlation pattern features from the multi-scale time series input vector of the gun body temperature and the multi-scale time series input vector of the temperature inside the furnace to obtain a multi-scale time series feature vector of the gun body temperature and a multi-scale time series feature vector of the temperature inside the furnace;

[0013] A feature interaction response module is used to process the gun body temperature multi-scale time series feature vector and the furnace temperature multi-scale time series feature vector using a gamma interaction response module to obtain a furnace temperature-gun body temperature multi-scale time series interaction response feature vector as a furnace temperature-gun body temperature multi-scale time series interaction response feature;

[0014] The control instruction generation module is used to obtain intelligent control instructions based on the multi-scale time-series interactive response characteristics of the furnace temperature and the gun body temperature.

[0015] Compared with the prior art, the present application provides a fully intelligent online furnace opening system for blast furnace oxygen lances, which obtains the gun body temperature and furnace temperature of the blast furnace oxygen lance collected in real time by a temperature sensor, and uses a data processing and analysis algorithm based on a deep learning neural network to perform a time series feature analysis on the gun body temperature and the furnace temperature, thereby intelligently obtaining an intelligent control instruction for the oxygen flow value at the current time point to increase, decrease or maintain according to the time series interactive response between the gun body temperature and the furnace temperature. In this way, the correlation between oxygen flow and temperature can be grasped more accurately, and intelligent oxygen flow control can be achieved, thereby effectively reducing energy waste, reducing dependence on operators, and improving the utilization efficiency of oxygen in the blast furnace ironmaking process, making the blast furnace ironmaking process more stable and efficient. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] By describing the embodiments of the present application in more detail in conjunction with the accompanying drawings, the above and other purposes, features and advantages of the present application will become more apparent. The accompanying drawings are used to provide a further understanding of the embodiments of the present application and constitute a part of the specification. Together with the embodiments of the present application, they are used to explain the present application and do not constitute a limitation of the present application. In the accompanying drawings, the same reference numerals generally represent the same components or steps.

[0017] Figure 1 It is a block diagram of a fully intelligent online furnace start-up system for blast furnace oxygen lances according to an embodiment of the present application.

[0018] Figure 2 This is a schematic diagram of the architecture of a fully intelligent online furnace start-up system for blast furnace oxygen lances according to an embodiment of the present application.

[0019] Figure 3 It is a block diagram of the time dimension arrangement module in the fully intelligent online furnace opening system of the blast furnace oxygen lance according to an embodiment of the present application.

[0020] Figure 4 It is a block diagram of the training module in the fully intelligent online furnace opening system of the blast furnace oxygen lance according to an embodiment of the present application. DETAILED DESCRIPTION

[0021] The embodiments of the present disclosure will be described in more detail below with reference to the accompanying drawings. Although certain embodiments of the present disclosure are shown in the accompanying drawings, it should be understood that the present disclosure can be implemented in various forms and should not be construed as being limited to the embodiments described herein, but rather these embodiments are provided for a more thorough and complete understanding of the present disclosure. It should be understood that the drawings and embodiments of the present disclosure are only for exemplary purposes and are not intended to limit the scope of protection of the present disclosure.

[0022] It should be understood that the various steps described in the method embodiments of the present disclosure may be performed in different orders and / or in parallel. In addition, the method embodiments may include additional steps and / or omit the steps shown. The scope of the present disclosure is not limited in this respect.

[0023] In the description of the embodiments of the present disclosure, the term "including" and similar terms should be understood as open inclusion, that is, "including but not limited to". The term "based on" should be understood as "based at least in part on". The term "one embodiment" or "the embodiment" should be understood as "at least one embodiment". The terms "first", "second", etc. may refer to different or the same objects. Other explicit and implicit definitions may also be included below.

[0024] It should be noted that the modifications of "one" and "plurality" mentioned in the present disclosure are illustrative rather than restrictive, and those skilled in the art should understand that unless otherwise clearly indicated in the context, it should be understood as "one or more".

[0025] The blast furnace oxygen lance is a process equipment mainly used in the oxygen converter steelmaking process. Its function is to blow high-pressure and high-purity oxygen at supersonic speed into the top of the molten metal pool in the converter. It is also equipped with a high-pressure water cooling protection system. The oxygen lance can withstand extremely high temperatures and the erosion of molten metal, and is an indispensable equipment in the steelmaking process.

[0026] The opening of the blast furnace oxygen lance refers to the first injection of a certain amount of oxygen into the blast furnace during the blast furnace ironmaking process. This is an important step in blast furnace production, marking the official start of blast furnace production. The timing of opening the blast furnace oxygen lance depends on the specific conditions of the blast furnace, usually after the blast furnace reaches a certain temperature and the furnace condition is stable. Oxygen injection can speed up the blast furnace reaction, increase the output of the blast furnace, prevent the charge from agglomerating, and stabilize the blast furnace operation.

[0027] However, it is often difficult to accurately control the amount of oxygen in traditional blast furnace oxygen lances when the furnace is turned on, resulting in excessive or insufficient oxygen supply, resulting in uneven distribution of oxygen inside the blast furnace, thereby increasing energy consumption costs and affecting the effect of the combustion reaction and the stability of the temperature inside the furnace. In addition, the oxygen flow control of traditional oxygen lances usually requires experienced operators to adjust, which is complicated to operate and prone to misoperation, further affecting the efficiency of ironmaking.

[0028] Therefore, in response to the above technical problems, the technical concept of the present application is to obtain the gun body temperature and furnace temperature of the blast furnace oxygen gun collected in real time by the temperature sensor, and use the data processing and analysis algorithm based on the deep learning neural network to perform time series feature analysis on the gun body temperature and the furnace temperature, so as to intelligently obtain the intelligent control instructions of the oxygen flow value at the current time point to increase, decrease or maintain according to the time series interactive response between the gun body temperature and the furnace temperature. In this way, the correlation between the oxygen flow rate and the temperature can be grasped more accurately, and intelligent oxygen flow rate control can be realized, thereby effectively reducing energy waste, reducing dependence on operators, and improving the utilization efficiency of oxygen in the blast furnace ironmaking process, making the blast furnace ironmaking process more stable and efficient.

[0029] Figure 1 It is a block diagram of a fully intelligent online furnace start-up system for blast furnace oxygen lances according to an embodiment of the present application. Figure 2 Schematic diagram of the architecture of the fully intelligent online furnace opening system for blast furnace oxygen lances according to an embodiment of the present application. Figure 1 and Figure 2As shown, according to the embodiment of the present application, the fully intelligent online furnace opening system 100 of the blast furnace oxygen lance includes: a blast furnace oxygen lance temperature acquisition module 110, which is used to obtain the time series of the lance body temperature of the blast furnace oxygen lance and the time series of the furnace temperature collected by the temperature sensor; a time dimension arrangement module 120, which is used to arrange the time series of the lance body temperature and the time series of the furnace temperature respectively according to the time dimension as a lance body temperature time series input vector and a furnace temperature time series input vector, and then respectively perform a difference calculation between the characteristic values ​​of two adjacent positions to obtain a lance body temperature time series fluctuation input vector and a furnace temperature time series fluctuation input vector; a lance body temperature multi-scale feature cascade module 130, which is used to cascade the lance body temperature time series input vector and the lance body temperature time series fluctuation input vector to obtain a lance body temperature multi-scale time series input vector; a furnace temperature multi-scale feature cascade module 140, which is used to The internal temperature time series fluctuation input vector and the furnace temperature time series input vector are cascaded to obtain the furnace temperature multi-scale time series input vector; the time series association pattern feature extraction module 150 is used to extract the time series association pattern features of the gun body temperature multi-scale time series input vector and the furnace temperature multi-scale time series input vector to obtain the gun body temperature multi-scale time series feature vector and the furnace temperature multi-scale time series feature vector; the feature interaction response module 160 is used to use the gamma interaction response module to process the gun body temperature multi-scale time series feature vector and the furnace temperature multi-scale time series feature vector to obtain the furnace temperature-gun body temperature multi-scale time series interaction response feature vector as the furnace temperature-gun body temperature multi-scale time series interaction response feature; and the control instruction generation module 170 is used to obtain the intelligent control instruction based on the furnace temperature-gun body temperature multi-scale time series interaction response feature.

[0030] In an embodiment of the present application, the blast furnace oxygen lance temperature acquisition module 110 is used to obtain the time series of the lance body temperature of the blast furnace oxygen lance collected by the temperature sensor and the time series of the temperature in the furnace. It should be understood that, considering that the lance body temperature refers to the temperature near the oxygen lance nozzle, it can reflect the combustion of oxygen and coal gas. The change in the lance body temperature can indicate the combustion state inside the blast furnace, which has an important guiding role in adjusting the oxygen flow rate, gas supply, etc. The temperature in the furnace refers to the temperature distribution of various parts inside the blast furnace, reflecting the combustion state of the charge and the progress of the ironmaking process. Based on this, in the technical solution of the present application, the time series of the lance body temperature of the blast furnace oxygen lance collected by the temperature sensor and the time series of the temperature in the furnace are obtained, and the lance body temperature and the temperature in the furnace are subjected to time series feature analysis and interactive correlation response, so that the oxygen flow rate can be adjusted in time, energy consumption can be reduced, and the stability and efficiency of the combustion reaction in the furnace can be ensured.

[0031] In the embodiment of the present application, the time dimension arrangement module 120 is used to arrange the time series of the gun body temperature and the time series of the furnace temperature according to the time dimension into a gun body temperature time series input vector and a furnace temperature time series input vector, respectively, and then perform a difference calculation between the eigenvalues ​​of two adjacent positions to obtain a gun body temperature time series fluctuation input vector and a furnace temperature time series fluctuation input vector. Figure 3 : is a block diagram of a time dimension arrangement module in a blast furnace oxygen lance fully intelligent online furnace opening system according to an embodiment of the present application. Specifically, in the embodiment of the present application, Figure 3 As shown, the time dimension arrangement module 120 includes: a time series arrangement unit 121, which is used to arrange the time series of the gun body temperature and the time series of the temperature inside the furnace according to the time dimension into the gun body temperature time series input vector and the furnace temperature time series input vector respectively; and a difference calculation unit 122, which is used to calculate the difference between the characteristic values ​​of each two adjacent positions in the gun body temperature time series input vector and the furnace temperature time series input vector to obtain the gun body temperature time series fluctuation input vector and the furnace temperature time series fluctuation input vector.

[0032] Specifically, the time series arrangement unit 121 is used to arrange the time series of the gun body temperature and the time series of the temperature in the furnace into the gun body temperature time series input vector and the furnace temperature time series input vector respectively according to the time dimension. Accordingly, considering that time is an important dimension in the gun body temperature and the temperature in the furnace data, and there is usually a time series dependency between the temperature data, that is, the temperature value at the previous moment may affect the temperature change at the subsequent moment. Therefore, in order to be able to retain the time series information of the gun body temperature and the temperature in the furnace, so as to accurately capture the law and trend of the gun body temperature and the temperature in the furnace changing with time, in the technical solution of the present application, the time series of the gun body temperature and the time series of the temperature in the furnace are arranged into the gun body temperature time series input vector and the furnace temperature time series input vector respectively according to the time dimension, so as to better analyze the time series change trend of the gun body temperature and the temperature in the furnace, so as to accurately control the oxygen flow rate subsequently.

[0033] Specifically, the difference calculation unit 122 is used to calculate the difference between the characteristic values ​​of each two adjacent positions in the gun body temperature time series input vector and the furnace temperature time series input vector to obtain the gun body temperature time series fluctuation input vector and the furnace temperature time series fluctuation input vector. It should be understood that considering that the gun body temperature and the furnace temperature change with the passage of time, that is, the gun body temperature and the furnace temperature have different time series characteristic information between two different time points. Therefore, in order to better understand the temperature changes of the gun body temperature and the furnace temperature between different time points, that is, the volatility of the temperature, so as to perform accurate analysis and processing later, in the technical solution of the present application, the difference between the characteristic values ​​of each two adjacent positions in the gun body temperature time series input vector and the furnace temperature time series input vector is calculated to obtain the gun body temperature time series fluctuation input vector and the furnace temperature time series fluctuation input vector. It should be understood that the difference reflects the temperature change trend between the gun body temperature and the temperature in the furnace at adjacent moments, such as the temperature change rate, periodic change trend, etc. By calculating the difference between the characteristic values ​​of each two adjacent positions, the fluctuation characteristics of the gun body temperature and the temperature in the furnace can be highlighted, such as whether the temperature is gradually increasing or decreasing, so as to better capture the temperature change trend and fluctuation over time.

[0034] In the embodiment of the present application, the gun body temperature multi-scale feature cascade module 130 is used to cascade the gun body temperature time series input vector and the gun body temperature time series fluctuation input vector to obtain the gun body temperature multi-scale time series input vector. Accordingly, considering that the gun body temperature time series input vector contains the original gun body temperature data information, and the gun body temperature time series fluctuation input vector contains the characteristics of temperature fluctuation changes. Therefore, in order to comprehensively consider the overall change trend and volatility of the gun body temperature, so that the model can better capture the time series feature information between the gun body temperature, in the technical solution of the present application, the gun body temperature time series input vector and the gun body temperature time series fluctuation input vector are cascaded to obtain the gun body temperature multi-scale time series input vector. In other words, by cascading gun body temperature input vectors of different scales, the gun body temperature time series feature expression capability can be enriched, so as to better capture the different time series temperature features and patterns in the gun body temperature time series data in the future, so as to more accurately capture the time series change law and fluctuation trend of the gun body temperature.

[0035] In an embodiment of the present application, the furnace temperature multi-scale feature cascade module 140 is used to cascade the furnace temperature time series fluctuation input vector and the furnace temperature time series input vector to obtain the furnace temperature multi-scale time series input vector. Similarly, considering that the furnace temperature time series input vector contains the original furnace temperature data information, and the furnace temperature time series fluctuation input vector contains the characteristics of the furnace temperature fluctuation. Therefore, in order to comprehensively consider the volatility and overall trend of the furnace temperature in the time dimension, so that the model can better understand the characteristics of the furnace temperature data, in the technical solution of the present application, the furnace temperature time series fluctuation input vector and the furnace temperature time series input vector are cascaded to obtain the furnace temperature multi-scale time series input vector.

[0036] In the embodiment of the present application, the time series association pattern feature extraction module 150 is used to extract the time series association pattern features of the multi-scale time series input vector of the gun body temperature and the multi-scale time series input vector of the temperature inside the furnace to obtain the multi-scale time series feature vector of the gun body temperature and the multi-scale time series feature vector of the temperature inside the furnace. Specifically, in the embodiment of the present application, the time series association pattern feature extraction module is used to: pass the multi-scale time series input vector of the gun body temperature and the multi-scale time series input vector of the temperature inside the furnace through a time series association pattern feature extractor based on deep convolution to obtain the multi-scale time series feature vector of the gun body temperature and the multi-scale time series feature vector of the temperature inside the furnace. In particular, the time series association pattern feature extractor based on deep convolution is a time series association pattern feature extractor based on a one-dimensional convolution layer. Accordingly, considering that both the multi-scale time series input vector of the gun body temperature and the multi-scale time series input vector of the temperature in the furnace have a temperature variation pattern and structure in the time series, and the one-dimensional convolution layer can effectively capture the time series pattern and correlation features in the data when processing time series data, it is helpful to extract important time series feature information in the time series data. Therefore, in the technical scheme of the present application, the multi-scale time series input vector of the gun body temperature and the multi-scale time series input vector of the temperature in the furnace are respectively captured and mined out of the feature information about the gun body temperature and the temperature in the furnace in the time dimension through a time series correlation pattern feature extractor based on a one-dimensional convolution layer, so as to obtain a multi-scale time series feature vector of the gun body temperature and a multi-scale time series feature vector of the temperature in the furnace with stronger time series feature representation.

[0037] In the embodiment of the present application, the characteristic interaction response module 160 is used to process the multi-scale time series feature vector of the gun body temperature and the multi-scale time series feature vector of the temperature in the furnace using the gamma interaction response module to obtain the furnace temperature-gun body temperature multi-scale time series interaction response feature vector as the furnace temperature-gun body temperature multi-scale time series interaction response feature. It should be understood that, considering that the multi-scale time series feature vector of the gun body temperature and the multi-scale time series feature vector of the temperature in the furnace have a mutual response relationship, for example, as the temperature in the furnace increases, the gun body temperature will also increase accordingly. Therefore, in order to better capture the interactive response relationship between the multi-scale time series feature of the gun body temperature and the multi-scale time series feature of the temperature in the furnace, and provide support for the subsequent oxygen flow control, in the technical solution of the present application, the multi-scale time series feature vector of the gun body temperature and the multi-scale time series feature vector of the temperature in the furnace are processed using the gamma interaction response module to obtain the furnace temperature-gun body temperature multi-scale time series interaction response feature vector. Specifically, the gamma interactive response module can help the model capture the nonlinear complex response association and interaction between the multi-scale time series characteristics of the gun body temperature and the multi-scale time series characteristics of the furnace temperature, and help to discover the temporal dynamic correlation and interaction relationship between the multi-scale time series characteristics of the gun body temperature and the multi-scale time series characteristics of the furnace temperature, thereby improving the expression and characterization capabilities of the furnace temperature-gun body temperature multi-scale time series interactive response feature vector.

[0038] Specifically, in the embodiment of the present application, the feature interactive response module is used to: process the multi-scale time series feature vector of the gun body temperature and the multi-scale time series feature vector of the furnace temperature using the following interactive response formula to obtain the furnace temperature-gun body temperature multi-scale time series interactive response feature vector; wherein the interactive response formula is:

[0039]

[0040] Among them, v 1i and v 2i are the eigenvalues ​​of the i-th position of the multi-scale time series feature vector of the gun body temperature and the multi-scale time series feature vector of the temperature in the furnace, respectively, i is the eigenvalue of the i-th position in the multi-scale time-series interactive response eigenvector of the internal temperature-gun body temperature, and A, B, C and D are the corresponding adjustment parameters.

[0041] In the embodiment of the present application, the control instruction generation module 170 is used to obtain intelligent control instructions based on the multi-scale temporal interactive response characteristics of the furnace temperature-gun body temperature. Specifically, in the embodiment of the present application, the control instruction generation module is used to: pass the multi-scale temporal interactive response feature vector of the furnace temperature-gun body temperature through the classifier-based oxygen flow intelligent controller to obtain intelligent control instructions, and the intelligent control instructions are used to indicate whether the oxygen flow value at the current time point should be increased, decreased or maintained. That is, the multi-scale temporal interactive response characteristics of the furnace temperature-gun body temperature obtained by gamma interactive response of the multi-scale temporal feature vector of the gun body temperature and the multi-scale temporal feature vector of the furnace temperature are classified and processed, so as to intelligently obtain the intelligent control instructions that the oxygen flow value at the current time point should be increased, decreased or maintained according to the temporal interactive response between the gun body temperature and the furnace temperature. In this way, the correlation between oxygen flow and temperature can be grasped more accurately, and intelligent oxygen flow control can be achieved, thereby effectively reducing energy waste, reducing dependence on operators, and improving the utilization efficiency of oxygen in the blast furnace ironmaking process, making the blast furnace ironmaking process more stable and efficient.

[0042] More specifically, in the embodiment of the present application, the control instruction generation module is used to: use the classifier-based oxygen flow intelligent controller to process the furnace temperature-gun body temperature multi-scale time series interactive response feature vector using the following classification formula to obtain the classification result;

[0043] Wherein, the formula is: O = softmax{(W n ,B n ):…:(W1,B1)|X}, where W1 to W n is the weight matrix, B1 to B n is a bias vector, X is the multi-scale time series interactive response feature vector of the furnace temperature-gun body temperature, softmax represents the softmax function, and O represents the classification result.

[0044] It is worth mentioning that those skilled in the art should be aware that before applying a deep neural network model for inference, the deep neural network model must be trained so that the deep neural network can implement specific functional capabilities.

[0045] Specifically, in the technical solution of the present application, the blast furnace oxygen lance fully intelligent online furnace opening system also includes a training module for training the temporal correlation pattern feature extractor based on the one-dimensional convolution layer, the gamma interactive response module and the classifier-based oxygen flow intelligent controller.

[0046] Figure 4: is a block diagram of a training module in a fully intelligent online furnace opening system for a blast furnace oxygen lance according to an embodiment of the present application. Specifically, in the embodiment of the present application, Figure 4As shown, the training module 200 includes: a training data acquisition unit 210, which is used to obtain the time series of the training gun body temperature and the time series of the training furnace temperature of the blast furnace oxygen gun collected by the temperature sensor, as well as the real value of the oxygen flow value at the current time point that should be increased, decreased or maintained; a training time dimension arrangement unit 220, which is used to arrange the time series of the training gun body temperature and the time series of the training furnace temperature into a training gun body temperature time series input vector and a training furnace temperature time series input vector according to the time dimension respectively; a training difference calculation unit 230, which is used to calculate the characteristic value of each adjacent two positions in the training gun body temperature time series input vector and the training furnace temperature time series input vector The difference between them is used to obtain the training gun body temperature time series fluctuation input vector and the training furnace temperature time series fluctuation input vector; the training gun body temperature cascade unit 240 is used to cascade the training gun body temperature time series input vector and the training gun body temperature time series fluctuation input vector to obtain the training gun body temperature multi-scale time series input vector; the training furnace temperature cascade unit 250 is used to cascade the training furnace temperature time series fluctuation input vector and the training furnace temperature time series input vector to obtain the training furnace temperature multi-scale time series input vector; the training data time series feature extraction unit 260 is used to cascade the training gun body temperature multi-scale time series input vector and the training furnace temperature multi-scale time series input vector through the training data time series feature extraction unit 260 based on A time series correlation pattern feature extractor of a one-dimensional convolutional layer is used to obtain a training gun body temperature multi-scale time series feature vector and a training furnace temperature multi-scale time series feature vector; a training time series interactive response unit 270 is used to use the gamma interactive response module to process the training gun body temperature multi-scale time series feature vector and the training furnace temperature multi-scale time series feature vector to obtain a training furnace temperature-gun body temperature multi-scale time series interactive response feature vector; a feature optimization unit 280 is used to optimize the training furnace temperature-gun body temperature multi-scale time series interactive response feature vector to obtain an optimized training furnace temperature-gun body temperature multi-scale time series interactive response feature vector; a classification loss calculation unit 290 is used to convert the optimized The multi-scale temporal interaction response feature vector of the furnace temperature-gun body temperature is trained through the classifier-based oxygen flow intelligent controller to obtain a classification loss function value; and a model training unit 300 is used to train the temporal association pattern feature extractor based on the one-dimensional convolution layer, the gamma interaction response module and the classifier-based oxygen flow intelligent controller based on the classification loss function value and through back propagation of gradient descent, wherein each time the training furnace temperature-gun body temperature multi-scale temporal interaction response feature vector is classified and iterated through the classifier-based oxygen flow intelligent controller, the training furnace temperature-gun body temperature multi-scale temporal interaction response feature vector is optimized.

[0047] Specifically, the model training unit 300 is used to train the temporal correlation pattern feature extractor based on the one-dimensional convolution layer, the gamma interactive response module and the oxygen flow intelligent controller based on the classifier based on the classification loss function value and through the back propagation of gradient descent, wherein each time the training furnace temperature-gun body temperature multi-scale temporal interactive response feature vector is classified and iterated by the classifier-based oxygen flow intelligent controller, the training furnace temperature-gun body temperature multi-scale temporal interactive response feature vector is optimized. It should be understood that in the technical solution of the present application, the training gun body temperature multi-scale temporal series feature vector represents the temporal correlation multi-scale features of the absolute value of the gun body temperature and the neighborhood fluctuation amount of the gun body temperature in the time domain receptive field determined based on the one-dimensional convolution layer. The training furnace temperature multi-scale temporal series feature vector represents the temporal correlation multi-scale features of the absolute value of the furnace temperature and the neighborhood fluctuation amount of the furnace temperature in the time domain receptive field determined based on the one-dimensional convolution layer. However, considering that the gun body temperature of the training blast furnace oxygen gun and the training furnace temperature have differences in the time series distribution in the data source domain, the training gun body temperature multi-scale time series feature vector and the training furnace temperature multi-scale time series feature vector have inconsistencies in the overall distribution of the feature vectors due to the differences in the time series distribution in the data source domain. In this way, when the training gun body temperature multi-scale time series feature vector and the training furnace temperature multi-scale time series feature vector are processed using the gamma interactive response module, the obtained training furnace temperature-gun body temperature multi-scale time series interactive response feature vector has a discretized local feature distribution, thereby affecting the convergence effect of the training furnace temperature-gun body temperature multi-scale time series interactive response feature vector to the class probability density space when it is classified by the classifier-based oxygen flow intelligent controller. Based on this, in the technical solution of the present application, each time the training furnace temperature-gun body temperature multi-scale time series interactive response feature vector is classified and iterated by the classifier-based oxygen flow intelligent controller, the training furnace temperature-gun body temperature multi-scale time series interactive response feature vector is optimized.

[0048] Specifically, in the embodiment of the present application, the model training unit is used to optimize the training furnace temperature-gun body temperature multi-scale time series interaction response feature vector using the following optimization formula to obtain the optimized training furnace temperature-gun body temperature multi-scale time series interaction response feature vector each time the training furnace temperature-gun body temperature multi-scale time series interaction response feature vector is classified and iterated by the classifier-based oxygen flow intelligent controller; wherein the optimization formula is:

[0049]

[0050]

[0051] Among them, v i and v j are the eigenvalues ​​of the i-th and j-th positions of the training furnace temperature-gun body temperature multi-scale time series interactive response eigenvector V, M μ (i, j) represents the eigenvalue of the (i, j)th position of the mean matrix of the training furnace temperature-gun body temperature multi-scale time series interactive response eigenvector, M σ (i, j) represents the eigenvalue of the (i, j)th position of the mean difference matrix of the training furnace temperature-gun body temperature multi-scale time series interaction response eigenvector, V represents the training furnace temperature-gun body temperature multi-scale time series interaction response eigenvector, V′ represents the optimized training furnace temperature-gun body temperature multi-scale time series interaction response eigenvector, represents matrix multiplication, Indicates addition by position.

[0052] That is, by introducing the local statistical information distribution of the training furnace temperature-gun body temperature multi-scale time series interactive response feature vector V as an external information source to perform feature vector retrieval enhancement, local overflow information distribution is avoided based on local statistical dense information structuring to cause the distribution illusion of the training furnace temperature-gun body temperature multi-scale time series interactive response feature vector V, thereby obtaining the information credible response reasoning based on the local distribution group dimension retention of the training furnace temperature-gun body temperature multi-scale time series interactive response feature vector V, so as to obtain the credible distribution response of the training furnace temperature-gun body temperature multi-scale time series interactive response feature vector V in the probability density space based on the discretized local feature distribution, thereby improving the convergence effect of the probability density space, so as to improve the training speed and the accuracy of the training results. In this way, the correlation between oxygen flow and temperature can be grasped more accurately, and intelligent oxygen flow control can be realized, thereby effectively reducing energy waste, reducing dependence on operators, and improving the utilization efficiency of oxygen in the blast furnace ironmaking process, making the blast furnace ironmaking process more stable and efficient.

[0053] In summary, the fully intelligent online furnace opening system 100 of the blast furnace oxygen lance based on the embodiment of the present application is explained, which obtains the gun body temperature and the furnace temperature of the blast furnace oxygen lance collected in real time by the temperature sensor, and uses the data processing and analysis algorithm based on the deep learning neural network to perform time series feature analysis on the gun body temperature and the furnace temperature, so as to intelligently obtain the intelligent control instructions of the oxygen flow value at the current time point that should be increased, decreased or maintained according to the time series interactive response between the gun body temperature and the furnace temperature. In this way, the correlation between the oxygen flow rate and the temperature can be grasped more accurately, and intelligent oxygen flow rate control can be realized, thereby effectively reducing energy waste, reducing dependence on operators, and improving the utilization efficiency of oxygen in the blast furnace ironmaking process, making the blast furnace ironmaking process more stable and efficient.

[0054] As described above, the blast furnace oxygen lance fully intelligent online furnace start-up system 100 according to the embodiment of the present application can be implemented in various wireless terminals, such as a server with a blast furnace oxygen lance fully intelligent online furnace start-up algorithm. In a possible implementation, the blast furnace oxygen lance fully intelligent online furnace start-up system 100 according to the embodiment of the present application can be integrated into the wireless terminal as a software module and / or a hardware module. For example, the blast furnace oxygen lance fully intelligent online furnace start-up system 100 can be a software module in the operating system of the wireless terminal, or it can be an application developed for the wireless terminal; of course, the blast furnace oxygen lance fully intelligent online furnace start-up system 100 can also be one of the many hardware modules of the wireless terminal.

[0055] Alternatively, in another example, the blast furnace oxygen lance fully intelligent online furnace start-up system 100 and the wireless terminal may also be separate devices, and the blast furnace oxygen lance fully intelligent online furnace start-up system 100 may be connected to the wireless terminal via a wired and / or wireless network, and transmit interactive information in accordance with an agreed data format.

[0056] The above descriptions of various implementations of the present disclosure are exemplary and not exhaustive. The disclosure is not limited to the disclosed implementations, and many modifications and variations are obvious to those of ordinary skill in the art without departing from the scope and spirit of the described implementations. The terms used herein are selected to best explain the principles of the implementations, practical applications, or improvements to the technology in the market, or to enable other persons of ordinary skill in the art to understand the various practical methods disclosed herein.

Claims

1. A fully intelligent online furnace opening system for blast furnace oxygen lance, characterized in that: include: The blast furnace oxygen lance temperature acquisition module is used to obtain the time series of the blast furnace oxygen lance body temperature and the time series of the furnace temperature collected by the temperature sensor; A time dimension arrangement module, used for arranging the time series of the gun body temperature and the time series of the temperature inside the furnace according to the time dimension into a gun body temperature time series input vector and a furnace temperature time series input vector, and then performing a difference calculation between the characteristic values ​​of two adjacent positions to obtain a gun body temperature time series fluctuation input vector and a furnace temperature time series fluctuation input vector; A gun body temperature multi-scale feature cascade module, used for cascading the gun body temperature time series input vector and the gun body temperature time series fluctuation input vector to obtain a gun body temperature multi-scale time series input vector; A furnace temperature multi-scale feature cascade module, used for cascading the furnace temperature time series fluctuation input vector and the furnace temperature time series input vector to obtain a furnace temperature multi-scale time series input vector; A time series correlation pattern feature extraction module is used to extract time series correlation pattern features from the multi-scale time series input vector of the gun body temperature and the multi-scale time series input vector of the temperature inside the furnace to obtain a multi-scale time series feature vector of the gun body temperature and a multi-scale time series feature vector of the temperature inside the furnace; A feature interaction response module is used to process the gun body temperature multi-scale time series feature vector and the furnace temperature multi-scale time series feature vector using a gamma interaction response module to obtain a furnace temperature-gun body temperature multi-scale time series interaction response feature vector as a furnace temperature-gun body temperature multi-scale time series interaction response feature; A control instruction generation module, used to obtain intelligent control instructions based on the multi-scale time series interactive response characteristics of the furnace temperature and the gun body temperature; The characteristic interaction response module is used to: process the multi-scale time series characteristic vector of the gun body temperature and the multi-scale time series characteristic vector of the furnace temperature by the following interaction response formula to obtain the furnace temperature-gun body temperature multi-scale time series interaction response characteristic vector; Wherein, the interactive response formula is: Among them, v 1i and v 2i are the eigenvalues ​​of the i-th position of the multi-scale time series feature vector of the gun body temperature and the multi-scale time series feature vector of the temperature in the furnace, respectively, i is the eigenvalue of the i-th position in the multi-scale time series interactive response eigenvector of the internal temperature-gun body temperature, and A, B, C and D are the corresponding adjustment parameters; Among them, the control instruction generation module is used to: pass the multi-scale time-series interactive response feature vector of the furnace temperature-gun body temperature through the classifier-based oxygen flow intelligent controller to obtain an intelligent control instruction, and the intelligent control instruction is used to indicate whether the oxygen flow value at the current time point should be increased, decreased or maintained.

2. The blast furnace oxygen lance fully intelligent online furnace opening system according to claim 1 is characterized in that: The time dimension arrangement module comprises: A time series arrangement unit, used for arranging the time series of the gun body temperature and the time series of the temperature inside the furnace into the gun body temperature time series input vector and the furnace temperature time series input vector respectively according to the time dimension; The difference calculation unit is used to calculate the difference between the characteristic values ​​of each two adjacent positions in the gun body temperature time series input vector and the furnace temperature time series input vector to obtain the gun body temperature time series fluctuation input vector and the furnace temperature time series fluctuation input vector.

3. The blast furnace oxygen lance fully intelligent online furnace opening system according to claim 2 is characterized in that: The time series association pattern feature extraction module is used to: pass the multi-scale time series input vector of the gun body temperature and the multi-scale time series input vector of the temperature inside the furnace through a time series association pattern feature extractor based on deep convolution to obtain a multi-scale time series feature vector of the gun body temperature and a multi-scale time series feature vector of the temperature inside the furnace.

4. The blast furnace oxygen lance fully intelligent online furnace opening system according to claim 3 is characterized in that: The temporal association pattern feature extractor based on deep convolution is a temporal association pattern feature extractor based on a one-dimensional convolutional layer.

5. The blast furnace oxygen lance fully intelligent online furnace opening system according to claim 4 is characterized in that: The control instruction generation module is used to: use the classifier-based oxygen flow intelligent controller to process the furnace temperature-gun body temperature multi-scale time series interactive response feature vector using the following classification formula to obtain a classification result; Wherein, the formula is: O = softmax{(W n ,B n ):…:(W1,B1)|X}, where W1 to W n is the weight matrix, B1 to B n is a bias vector, X is the multi-scale time series interactive response feature vector of the furnace temperature-gun body temperature, softmax represents the softmax function, and O represents the classification result.

6. The blast furnace oxygen lance fully intelligent online furnace opening system according to claim 5 is characterized in that: It also includes a training module for training the temporal correlation pattern feature extractor based on the one-dimensional convolution layer, the gamma interaction response module and the classifier-based oxygen flow intelligent controller.

7. The blast furnace oxygen lance fully intelligent online furnace opening system according to claim 6 is characterized in that: The training module comprises: A training data acquisition unit, used to obtain a time series of a training lance body temperature and a time series of a training furnace temperature of a blast furnace oxygen lance acquired by a temperature sensor, and a real value of the oxygen flow value at the current time point that should be increased, decreased or maintained; A training time dimension arrangement unit, used for arranging the time series of the training gun body temperature and the time series of the training furnace temperature into a training gun body temperature time series input vector and a training furnace temperature time series input vector respectively according to the time dimension; A training difference calculation unit, used for calculating the difference between the characteristic values ​​of each two adjacent positions in the training gun body temperature time series input vector and the training furnace temperature time series input vector to obtain a training gun body temperature time series fluctuation input vector and a training furnace temperature time series fluctuation input vector; A training gun body temperature cascade unit, used for cascading the training gun body temperature time series input vector and the training gun body temperature time series fluctuation input vector to obtain a training gun body temperature multi-scale time series input vector; A training furnace temperature cascade unit, used for cascading the training furnace temperature time series fluctuation input vector and the training furnace temperature time series input vector to obtain a training furnace temperature multi-scale time series input vector; A training data time series feature extraction unit, used for passing the training gun body temperature multi-scale time series input vector and the training furnace temperature multi-scale time series input vector through the time series association pattern feature extractor based on the one-dimensional convolution layer to obtain a training gun body temperature multi-scale time series feature vector and a training furnace temperature multi-scale time series feature vector; A training time series interactive response unit, used to process the training gun body temperature multi-scale time series feature vector and the training furnace temperature multi-scale time series feature vector using the gamma interactive response module to obtain a training furnace temperature-gun body temperature multi-scale time series interactive response feature vector; A feature optimization unit, used for optimizing the training furnace temperature-gun body temperature multi-scale time series interactive response feature vector to obtain an optimized training furnace temperature-gun body temperature multi-scale time series interactive response feature vector; A classification loss calculation unit, used for passing the optimized training furnace temperature-gun body temperature multi-scale time series interactive response feature vector through the classifier-based oxygen flow intelligent controller to obtain a classification loss function value; A model training unit is used to train the temporal association pattern feature extractor based on the one-dimensional convolution layer, the gamma interactive response module and the classifier-based oxygen flow intelligent controller based on the classification loss function value and through back propagation of gradient descent, wherein each time the training furnace temperature-gun body temperature multi-scale temporal interactive response feature vector is classified and iterated through the classifier-based oxygen flow intelligent controller, the training furnace temperature-gun body temperature multi-scale temporal interactive response feature vector is optimized.

Images