Furnace temperature change simulation model construction method, storage medium and furnace temperature prediction method

Abstract

The present invention provides a method for constructing a furnace temperature change simulation model, a storage medium and a furnace temperature prediction method, and relates to the technical field of furnace temperature control of a heating furnace. The method comprises: constructing a spatial graph structure of a heating furnace; determining a space-time graph structure of a heating furnace according to the spatial graph structure of the heating furnace; determining an adjacency matrix of the space-time graph structure according to nodes in the space-time graph structure of the heating furnace; processing the input characteristic time series and the output characteristic time series of the heating furnace by a sliding window technology to obtain input window data and output window data; and obtaining a furnace temperature change simulation model of the heating furnace according to the input window data and the output window data, combined with the adjacency matrix of the space-time graph structure. The present invention constructs a spatial graph structure of a heating furnace and then determines the space-time graph structure, takes the time factor into consideration, and realizes a more accurate prediction of the furnace temperature change trend by capturing the mutual influence between the various furnace sections of the heating furnace.

Description

Technical Field

[0001] The present invention relates to the technical field of furnace temperature control of a heating furnace, and in particular to a furnace temperature variation simulation model construction method, a storage medium and a furnace temperature prediction method. Background Art

[0002] With the advancement of controlled rolling and controlled cooling technology, enterprises have increasingly stringent requirements on the quality of steel billets, and the furnace temperature is an important factor affecting the quality of steel billets. Therefore, it is necessary to simulate the future temperature change trend of each furnace section in the heating furnace in order to predict the heating furnace temperature.

[0003] In the prior art, since the furnace temperatures between multiple furnace sections will affect each other and thus produce a coupling relationship, in the existing simulation of the furnace temperature change trend, only a single furnace section is analyzed and simulated, and only the temperature influencing factors within a single furnace section are considered, which makes it difficult to accurately predict the future change trend of the furnace temperature. Summary of the invention

[0004] The problem solved by the invention is how to improve the prediction accuracy of the furnace temperature variation trend of the furnace section.

[0005] In order to solve the above problems, the present invention provides a furnace temperature change simulation model construction method, a storage medium and a furnace temperature prediction method.

[0006] In a first aspect, the present invention provides a method for constructing a furnace temperature variation simulation model, comprising:

[0007] Construct the spatial graph structure of the heating furnace;

[0008] Determining a spatiotemporal graph structure of the heating furnace according to the spatial graph structure of the heating furnace;

[0009] Determining an adjacency matrix of the space-time graph structure according to the nodes in the space-time graph structure of the heating furnace;

[0010] The input characteristic time series and the output characteristic time series of the heating furnace are processed by a sliding window technology to obtain input window data and output window data; wherein the input characteristic time series represents a time series composed of temperature influence data of multiple time steps, the temperature influence data represents data having an influence on the temperature data of the heating furnace, and the output characteristic time series represents a time series composed of the temperature data of the heating furnace of multiple time steps;

[0011] According to the input window data and the output window data, combined with the adjacency matrix of the space-time graph structure, a furnace temperature change simulation model of the heating furnace is obtained.

[0012] Optionally, constructing the spatial graph structure of the heating furnace includes:

[0013] According to the working process of the heating furnace, a plurality of furnace sections are divided, and nodes corresponding to each of the furnace sections are set;

[0014] According to the workflow of the heating furnace, the nodes corresponding to the various furnace sections are connected through spatial connection lines to obtain the spatial graph structure of the heating furnace.

[0015] Optionally, determining the spatiotemporal graph structure of the heating furnace according to the spatial graph structure of the heating furnace includes:

[0016] Determine the size of the sliding window through time-lag analysis;

[0017] Determining the number of the spatial graph structures in the spatiotemporal graph structure according to the sliding window size;

[0018] The same nodes in each of the spatial graph structures are connected through a time connection line to obtain the space-time graph structure of the heating furnace.

[0019] Optionally, determining an adjacency matrix of the space-time graph structure according to the nodes in the space-time graph structure of the heating furnace includes:

[0020] Numbering all nodes in the space-time graph structure according to a preset order;

[0021] Determine a connection relationship between each of the nodes in the space-time graph structure, wherein the connection relationship includes one of being unconnected, being connected through the time connection line, and being connected through the space connection line;

[0022] The adjacency matrix of the space-time graph structure is determined according to the numbers of the nodes and the connection relationships between the nodes.

[0023] Optionally, the processing of the input characteristic time series and the output characteristic time series of the heating furnace by a sliding window technology to obtain input window data and output window data includes:

[0024] Acquire an input feature time series of the heating furnace, wherein the input feature time series includes a furnace section input feature set of multiple time steps arranged in chronological order, each furnace section input feature set includes a furnace section input feature corresponding to each furnace section at the same time step, and the furnace section input feature includes temperature influence data of the furnace section corresponding to each time step;

[0025] Sliding and intercepting the input feature time series according to a first preset time step through a sliding window to obtain a plurality of input window data;

[0026] Each of the input window data includes a set of furnace section input features at multiple time steps, and the number of the furnace section input feature sets corresponds to the size of the sliding window.

[0027] Optionally, the step of processing the input characteristic time series and the output characteristic time series of the heating furnace by a sliding window technology to obtain input window data and output window data further includes:

[0028] Acquire the output feature time series of the heating furnace, wherein the output feature time series includes a set of furnace section output features of multiple time steps arranged in chronological order, each of the furnace section output feature sets includes a furnace section output feature corresponding to each of the furnace sections at the same time step, and the furnace section output feature includes temperature influence data of the furnace section corresponding to each of the time steps;

[0029] Sliding and intercepting the output feature time series according to a second preset time step through a sliding window to obtain a plurality of output window data;

[0030] Each of the output window data includes a set of furnace section output features at multiple time steps, and the number of the furnace section output feature sets corresponds to the size of the sliding window.

[0031] Optionally, the obtaining of the furnace temperature variation simulation model of the heating furnace according to the input window data and the output window data in combination with the adjacency matrix of the space-time graph structure includes:

[0032] Constructing a spatiotemporal model of the heating furnace;

[0033] The space-time model is trained according to the input window data, the output window data and the adjacency matrix of the space-time graph structure to obtain a furnace temperature change simulation model of the heating furnace.

[0034] Optionally, the training of the spatiotemporal model according to the input window data, the output window data and the adjacency matrix of the spatiotemporal graph structure to obtain a furnace temperature variation simulation model of the heating furnace includes:

[0035] Dividing the input window data and the output window data into a training set and a test set according to a preset ratio, wherein the training set and the test set each include a plurality of data pairs, each of the data pairs includes one input window data and one output window data, and a time step corresponding to the input window data in each of the data pairs is earlier than a time step corresponding to the output window data;

[0036] Training the spatiotemporal model using the data pairs in the training set and the adjacency matrix;

[0037] Whether the training of the space-time model is completed is determined by using the data pairs in the test set and the adjacency matrix, and the trained space-time model is used as the furnace temperature change simulation model of the heating furnace.

[0038] In a second aspect, the present invention provides a computer-readable storage medium having a computer program stored thereon, and when the computer program is executed by a processor, the above-mentioned method for constructing a furnace temperature variation simulation model is implemented.

[0039] In a third aspect, the present invention provides a furnace temperature prediction method, comprising: performing prediction using a furnace temperature change simulation model constructed by the furnace temperature change simulation model construction method as described above to obtain a furnace temperature prediction result.

[0040] The furnace temperature variation simulation model construction method, storage medium and furnace temperature prediction method of the present invention, by constructing the spatial graph structure of the heating furnace, clarify the spatial relationship and mutual influence between different furnace sections in the furnace, and provide a structured analysis basis for furnace temperature simulation. By determining the spatiotemporal graph structure, the time factor is further taken into consideration, and the influence of node historical data on current data can be reflected by the time connection line, so as to capture the time dependence of the furnace temperature of the heating furnace. Then, by determining the adjacency matrix, the coupling relationship between nodes is quantified, and accurate topological structure information is provided for the model. The input and output features are processed by sliding window technology to form input window data and output window data containing time series information, which not only retains the time continuity of the data, but also enhances the learning ability of the model to the mutual influence between node features and furnace sections. Finally, in combination with the spatiotemporal graph structure and the adjacency matrix, the constructed furnace temperature variation simulation model can comprehensively consider the spatial coupling effect and time dependence, so as to achieve a more accurate prediction of the furnace temperature variation trend. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] Figure 1 A flow chart of a method for constructing a furnace temperature variation simulation model according to an embodiment of the present invention;

[0042] Figure 2 A schematic diagram of a spatial graph structure according to another embodiment of the present invention;

[0043] Figure 3 A schematic diagram of a space-time graph structure according to another embodiment of the present invention;

[0044] Figure 4 A schematic diagram of a structure of a space-time graph after encoding according to another embodiment of the present invention;

[0045] Figure 5 The simulation result of the future change trend of the preheating furnace temperature in another embodiment of the present invention;

[0046] Figure 6It is a simulation result of the future change trend of the furnace temperature in a further embodiment of the present invention;

[0047] Figure 7 The simulation result of the future change trend of the second-stage furnace temperature in another embodiment of the present invention;

[0048] Figure 8 The simulation result of the future change trend of the temperature in the soaking section of another embodiment of the present invention;

[0049] Fig. 9 The figure is a schematic diagram of the structure of an electronic device according to another embodiment of the present invention. DETAILED DESCRIPTION

[0050] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the specific embodiments of the present invention are described in detail below in conjunction with the accompanying drawings. Although certain embodiments of the present invention are shown in the accompanying drawings, it should be understood that the present invention can be implemented in various forms and should not be interpreted as being limited to the embodiments described herein. On the contrary, these embodiments are provided to provide a more thorough and complete understanding of the present invention. It should be understood that the drawings and embodiments of the present invention are only for exemplary purposes and are not intended to limit the scope of protection of the present invention.

[0051] It should be understood that the various steps described in the method embodiments of the present invention 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 invention is not limited in this respect.

[0052] The term "including" and its variations used in this document are open inclusions, that is, "including but not limited to"; the term "based on" means "based at least in part on"; the term "one embodiment" means "at least one embodiment"; the term "another embodiment" means "at least one other embodiment"; the term "some embodiments" means "at least some embodiments"; the term "optionally" means "optional embodiments". The relevant definitions of other terms will be given in the following description. It should be noted that the concepts of "first", "second", etc. mentioned in the present invention are only used to distinguish different devices, modules or units, and are not used to limit the order or interdependence of the functions performed by these devices, modules or units.

[0053] It should be noted that the modifications of "one" and "plurality" mentioned in the present invention 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".

[0054] The names of the messages or information exchanged between multiple devices in the embodiments of the present invention are only used for illustrative purposes, and are not used to limit the scope of these messages or information.

[0055] In the related technology, the heating furnace is large in size, high in heat capacity, and the combustion process is relatively slow. The chemical heat of gas combustion is ultimately manifested in the changes in furnace temperature and billet temperature, which have thermal inertia and time lag characteristics. The interconnected structure of the heating furnace causes energy exchange between adjacent furnace sections, and there is a spatial coupling effect. In the field of heating furnace temperature simulation, most of the existing methods only analyze and simulate a single furnace section, ignoring the coupling relationship between the temperatures of multiple furnace sections, making it difficult to accurately predict the future trend of furnace temperature changes.

[0056] In view of the problems existing in the above-mentioned related technologies, this embodiment provides a method for constructing a furnace temperature change simulation model, a storage medium and a furnace temperature prediction method.

[0057] Combination Figure 1 As shown, a method for constructing a furnace temperature change simulation model provided by an embodiment of the present invention includes:

[0058] S100: Construct a spatial graph structure of a heating furnace.

[0059] Specifically, the different furnace sections of the heating furnace are first simplified into spatial nodes, which can be used to represent the furnace temperature states of different areas in the heating furnace. Then, these nodes are connected to each other to form a spatial graph structure. The spatial coupling effect between furnace sections is reflected through the spatial graph structure of the heating furnace.

[0060] S200: Determine the spatiotemporal graph structure of the heating furnace according to the spatial graph structure of the heating furnace.

[0061] Specifically, the time factor is introduced when determining the space-time graph structure. The space-time graph structure is composed of spatial graph structures of multiple time periods, in which the same spatial nodes are connected to form a sequence with time order.

[0062] S300: Determine an adjacency matrix of the space-time graph structure according to the nodes in the space-time graph structure of the heating furnace.

[0063] Specifically, when determining the adjacency matrix of the spatiotemporal graph structure according to the nodes in the spatiotemporal graph structure of the heating furnace, the nodes in the spatiotemporal graph structure of the furnace section can be numbered, and a matrix whose size is the total number of nodes multiplied by the total number of nodes can be created as the adjacency matrix. Among them, the adjacency matrix is ​​a tool for describing the graph structure in graph theory. It provides the model with an accurate representation of the spatial and temporal relationships between nodes and is a key data structure for graph attention analysis and time series analysis.

[0064] S400: Processing the input feature time series and output feature time series of the heating furnace by means of sliding window technology to obtain input window data and output window data; wherein, the input feature time series represents a time series composed of temperature influence data of multiple time steps, the temperature influence data represents data that has an influence on the temperature data of the heating furnace, and the output feature time series represents a time series composed of the temperature data of the heating furnace of multiple time steps;.

[0065] Specifically, the key features of the time series data are extracted through the sliding window technology and organized into a format suitable for model input. The input window data and the output window data contain the input and output features of the furnace section at different time steps respectively. After rearrangement, these data provide structured input and expected output for model training. The output feature time series contains the furnace temperature changes at subsequent time points of the input feature time series, which are the targets that the model needs to predict. Among them, the time period of the input feature time series is before the time period of the output feature time series, ensuring that the model can predict future trends based on historical data.

[0066] S500: Obtaining a furnace temperature variation simulation model of the heating furnace according to the input window data and the output window data in combination with the adjacency matrix of the space-time graph structure.

[0067] Specifically, finally, all the previous preparations are combined to construct a simulation model of furnace temperature change. In a preferred embodiment of the present invention, a GAT-GRU model can be used to combine input window data, output window data and adjacency matrix to train a model that can predict the future trend of furnace temperature change. This model can comprehensively consider spatial coupling and time dependence to improve the accuracy and reliability of furnace temperature prediction.

[0068] In this embodiment, by constructing the spatial graph structure of the heating furnace, the spatial relationship and mutual influence between different furnace sections in the furnace are clarified, providing a structured analysis basis for furnace temperature simulation. By determining the spatiotemporal graph structure, the time factor is further taken into consideration, and the influence of node historical data on current data can be reflected through the time connection line, thereby capturing the time dependence of the furnace temperature of the heating furnace. Then, by determining the adjacency matrix, the coupling relationship between nodes is quantified, providing accurate topological structure information for the model. The input and output features are processed using sliding window technology to form input data and output window data containing time series information, which not only retains the time continuity of the data, but also enhances the model's learning ability for node features and mutual influence between furnace sections. Finally, combined with the spatiotemporal graph structure and the adjacency matrix, the constructed furnace temperature change simulation model can comprehensively consider the spatial coupling effect and time dependence, thereby achieving a more accurate prediction of the furnace temperature change trend.

[0069] Optionally, constructing the spatial graph structure of the heating furnace includes:

[0070] According to the working process of the heating furnace, a plurality of furnace sections are divided, and nodes corresponding to each of the furnace sections are set;

[0071] According to the workflow of the heating furnace, the nodes corresponding to the various furnace sections are connected through spatial connection lines to obtain the spatial graph structure of the heating furnace.

[0072] Specifically, in the process of constructing the spatial graph structure of the heating furnace, the heating furnace is divided into multiple key furnace sections according to the working process of the heating furnace. In a preferred embodiment of the present invention, the furnace sections of the heating furnace can be divided into a preheating section, a first heating section, a second heating section and a soaking section. These furnace sections are key areas in the heating process of the steel billet in the heating furnace, and each area has different contributions and characteristics to the heating effect of the steel billet. Corresponding to these furnace sections, nodes are set to represent the state and characteristics of each furnace section. Subsequently, these nodes are interconnected by spatial connecting lines to form a spatial graph structure. These connecting lines not only represent the physical connection between the furnace sections, but also reflect their energy exchange and mutual influence during the heating process. Combined with Figure 2 As shown, the constructed spatial graph structure includes spatial connection lines between the preheating section and the first heating section, the first heating section and the second heating section, and the second heating section and the soaking section. These connection lines have bidirectional arrows, representing the mutual influence between two nodes.

[0073] In this optional embodiment, the constructed spatial graph structure provides an intuitive framework for furnace temperature simulation, which enables the simulation and analysis of the spatial coupling effect between furnace sections. This structured representation method enables the model to more accurately capture the propagation and variation of furnace temperature between different furnace sections, and provides a solid foundation for the subsequent construction of the space-time graph structure and the training of the furnace temperature simulation model.

[0074] Optionally, determining the spatiotemporal graph structure of the heating furnace according to the spatial graph structure of the heating furnace includes:

[0075] Determine the size of the sliding window through time-lag analysis;

[0076] Determining the number of the spatial graph structures in the spatiotemporal graph structure according to the sliding window size;

[0077] The same nodes in each of the spatial graph structures are connected through a time connection line to obtain the space-time graph structure of the heating furnace.

[0078] Specifically, firstly, the time-lag analysis method is used to determine the size of the sliding window. In the preferred embodiment of the present invention, the size of the sliding window is set to 40. This size is obtained through the time-lag analysis method, which ensures that the model can capture the time characteristics of the furnace temperature change of the heating furnace. Then, according to the size of the sliding window, the number of spatial graph structures in the space-time graph structure is determined, that is, 40 furnace section spatial graph structures, each representing the furnace temperature state of a time step. Finally, the same nodes in these spatial graph structures are connected through time connecting lines to form a sequence with time order, thereby constructing the space-time graph structure of the heating furnace. Combined with Figure 3 The space-time graph structure shown in the figure has a one-way arrow in which the time connection line can only be transmitted in one direction, ensuring that the historical data of the node can affect the current data, while the current data will not affect the historical data in turn, which reflects the time lag characteristics of the temperature change of the heating furnace.

[0079] In this optional embodiment, by constructing a space-time graph structure, the model can not only capture the spatial distribution and coupling effects of the furnace temperature, but also take into account the impact on the time series, thereby more comprehensively simulating the changing trend of the furnace temperature and significantly improving the accuracy and reliability of the furnace temperature simulation.

[0080] Optionally, determining an adjacency matrix of the space-time graph structure according to the nodes in the space-time graph structure of the heating furnace includes:

[0081] Numbering all nodes in the space-time graph structure according to a preset order;

[0082] Determine a connection relationship between each of the nodes in the space-time graph structure, wherein the connection relationship includes one of being unconnected, being connected through the time connection line, and being connected through the space connection line;

[0083] The adjacency matrix of the space-time graph structure is determined according to the numbers of the nodes and the connection relationships between the nodes.

[0084] Specifically, all nodes in the spatiotemporal graph structure are numbered. In a preferred embodiment of the present invention, Figure 4The schematic diagram of the numbered spatiotemporal graph structure shown is that the numbering is performed in a preset order, which is from left to right and from top to bottom. The total number of nodes in the furnace section spatiotemporal graph structure is 160, and these nodes correspond to the four spatial nodes of each furnace section in 40 time steps. Next, for each node, other nodes connected to it by time connecting lines and / or spatial connecting lines are identified. These connecting lines define the relationships between nodes, including the interactions between the preheating section, the first adding section, the second adding section, and the soaking section nodes. Finally, an adjacency matrix is ​​constructed based on these connection relationships and the node numbers, where if node i points to node j through a connecting line, the corresponding element in the adjacency matrix is ​​set to 1, otherwise it is 0, and the elements on the main diagonal are all 0, indicating that there is no self-connection.

[0085] In this optional embodiment, the spatial and temporal relationships between the nodes of each furnace section of the heating furnace are accurately quantified and expressed through the adjacency matrix, providing a mathematical description of the interaction between the nodes, so that the model can take into account the energy exchange and influence between the furnace sections, thereby more accurately predicting the changing trend of the furnace temperature and improving the generalization ability and prediction efficiency of the model.

[0086] Optionally, the step of processing the input characteristic time series and the output characteristic time series of the heating furnace by using a sliding window technology to obtain input window data and output window data includes:

[0087] Acquire an input feature time series of the heating furnace, wherein the input feature time series includes a furnace section input feature set of multiple time steps arranged in chronological order, each furnace section input feature set includes a furnace section input feature corresponding to each furnace section at the same time step, and the furnace section input feature includes temperature influence data of the furnace section corresponding to each time step;

[0088] Sliding and intercepting the input feature time series according to a first preset time step by sliding a sliding window to obtain a plurality of input window data;

[0089] Each of the input window data includes the furnace section input feature sets of multiple time steps, and the number of the furnace section input feature sets corresponds to the size of the sliding window.

[0090] Specifically, first obtain the input feature time series of the heating furnace, which refers to the set of furnace section input features collected at multiple time points and arranged in time order. Each furnace section input feature set contains a series of related features of each furnace section at a specific time point, such as chemical heat, physical heat, heat of heating steel billets, etc. The features in the furnace section input feature set are represented by X1 to X20, covering the features of the preheating section, the first heating section, the second heating section and the soaking section. Each feature set represents the input data of a time step, and every 5 features correspond to a furnace section, covering the input of the above various heat. Among them, the preheating section input features include the chemical heat of fuel combustion in the preheating section (X1), the physical heat brought in by air in the preheating section (X2), the physical heat brought in by fuel in the preheating section (X3), the heat of heating steel billets in the preheating section (X4), and the heat taken away by flue gas in the preheating section (X5). The input features of the first heating stage include the chemical heat of fuel combustion in the first heating stage (X6), the physical heat brought in by the air in the first heating stage (X7), the physical heat brought in by the fuel in the first heating stage (X8), the heat of heating the steel billet in the first heating stage (X9), and the heat carried away by the flue gas in the first heating stage (X10). The input features of the second heating stage include the chemical heat of fuel combustion in the second heating stage (X11), the physical heat brought in by the air in the second heating stage (X12), the physical heat brought in by the fuel in the second heating stage (X13), the heat of heating the steel billet in the second heating stage (X14), and the heat carried away by the flue gas in the second heating stage (X15). The input features of the soaking stage include the chemical heat of fuel combustion in the soaking stage (X16), the physical heat brought in by the air in the soaking stage (X17), the physical heat brought in by the fuel in the soaking stage (X18), the heat of heating the steel billet in the soaking stage (X19), and the heat carried away by the flue gas in the soaking stage (X20). Then, the sliding window technology is used to slide from the first row of the input feature time series according to the preset time step. In another preferred embodiment of the present invention, the size of the input sliding window is defined as 40 time steps, that is, the window contains 40 rows of data, each row corresponding to a furnace section input feature set of a time step. When the sliding window moves one time step, a new input window is generated according to the furnace section input feature data in the window. A series of input windows are generated by continuous sliding, each of which contains a furnace section input feature set within a certain time range.

[0091] In this optional embodiment, the time series information of the heating furnace input characteristics is captured by sliding window technology, providing historical data for the model, including the interaction between furnace sections and the temporal evolution trend, so as to better understand and predict the changes in furnace temperature and improve the accuracy and reliability of the simulation.

[0092] Optionally, the step of processing the input characteristic time series and the output characteristic time series of the heating furnace by a sliding window technology to obtain input window data and output window data further includes:

[0093] Acquire the output feature time series of the heating furnace, wherein the output feature time series includes a set of furnace section output features of multiple time steps arranged in chronological order, each of the furnace section output feature sets includes a furnace section output feature corresponding to each of the furnace sections at the same time step, and the furnace section output feature includes temperature influence data of the furnace section corresponding to each of the time steps;

[0094] Sliding and intercepting the output feature time series according to a second preset time step through a sliding window to obtain a plurality of output window data;

[0095] Each of the output window data includes a set of furnace section output features at multiple time steps, and the number of the furnace section output feature sets corresponds to the size of the sliding window.

[0096] Specifically, the input features and output features of the heating furnace are processed by the sliding window technology, and the model's understanding of the working state of the heating furnace is further refined. In the processing of the input features, the features have been divided into furnace section input features corresponding to each furnace section, and the input window is generated by the sliding window technology. Each furnace section output feature set reflects the output features of each furnace section at the same time point, such as the furnace temperature change. In a preferred embodiment of the present invention, the features in the furnace section output feature set are represented by Y1 to Y4, which correspond to the furnace temperature changes of the preheating section, the first heating section, the second heating section and the soaking section, respectively. These furnace section output features reflect the temperature change of the furnace section and are the key targets of model prediction. Based on this, a specific sliding window size is used, for example, the input window size is 40, and the output window size is determined according to the number of future simulation time steps, for example, 10 time steps. The output feature time series is slidably intercepted according to the second preset time step to form multiple output window data. Each output window data set contains the furnace section output feature set at the time point corresponding to the sliding window size. For example, if the output window size is set to 10, each output window will cover 10 time steps of furnace temperature change data, which will be used in the prediction stage of the model. In a preferred embodiment of the present invention, when the input window covers data from time step 1 to time step 40, the output window will start from time step 41 and cover data to time step 50. After each second preset time step, a new output window will be generated based on the furnace segment output characteristics in the window.

[0097] In this optional embodiment, the furnace temperature change in the next period of time is predicted by the input features within a certain time range. In this way, the model can learn the complex relationship between the input features and the future furnace temperature change, thereby improving the accuracy of the prediction.

[0098] Optionally, the obtaining of the furnace temperature variation simulation model of the heating furnace according to the input window data and the output window data in combination with the adjacency matrix of the space-time graph structure includes:

[0099] Constructing a spatiotemporal model of the heating furnace;

[0100] The space-time model is trained according to the input window data, the output window data and the adjacency matrix of the space-time graph structure to obtain a furnace temperature change simulation model of the heating furnace.

[0101] Specifically, a spatiotemporal model is first constructed, and the input window data contains the input features of each furnace section of the heating furnace at a series of time steps, while the output window data contains the corresponding furnace temperature changes. These data provide the features and targets required for training the spatiotemporal model. In a preferred embodiment of the present invention, the input window data and the output window data can be divided into a training set and a test set, wherein the training set is used for model training, and the test set is used for model verification and evaluation. Then, the spatiotemporal model is trained in combination with the adjacency matrix of the spatiotemporal graph structure, which quantifies the spatial and temporal relationship between the furnace section nodes. During the training process, the graph attention layer learns the importance weights between nodes through the attention mechanism, and these weights reflect the mutual influence between different furnace sections. In this way, the model can capture the complex space-time dependencies inside the heating furnace. Among them, the Max-Min normalization method, the Adam optimizer, and the mean square error as the loss function can be used to train the spatiotemporal model.

[0102] In this optional embodiment, by combining the space-time model and the adjacency matrix, the model can not only learn the spatial coupling effect inside the heating furnace, but also take into account the influence on the time series, so as to more comprehensively simulate the changing trend of the furnace temperature and provide more accurate furnace temperature prediction.

[0103] Optionally, the training of the spatiotemporal model according to the input window data, the output window data and the adjacency matrix of the spatiotemporal graph structure to obtain a furnace temperature variation simulation model of the heating furnace includes:

[0104] Dividing the input window data and the output window data into a training set and a test set according to a preset ratio, wherein the training set and the test set each include a plurality of data pairs, each of the data pairs includes one input window data and one output window data, and a time step corresponding to the input window data in each of the data pairs is earlier than a time step corresponding to the output window data;

[0105] Training the spatiotemporal model using the data pairs in the training set and the adjacency matrix;

[0106] Whether the training of the space-time model is completed is determined by using the data pairs in the test set and the adjacency matrix, and the trained space-time model is used as the furnace temperature change simulation model of the heating furnace.

[0107] Specifically, the input window data and the output window data are divided into a training set and a test set according to a preset ratio, in order to ensure that the model can learn the inherent laws of the data through the training set, and evaluate the generalization ability of the model through the test set. In a preferred embodiment of the present invention, the preset ratio is set to 8:2 for division, that is, 80% of the data is used for training and 20% of the data is used for testing. Subsequently, the data pairs of the input window data, the output window data and the adjacency matrix in the training set are used as input to train the spatiotemporal model. The graph attention layer uses the attention mechanism to learn the relationship weights between nodes in different furnace sections, and the adjacency matrix provides the spatial and temporal connection information of these nodes. During the training process, the model adjusts the parameters by minimizing the loss function, thereby learning the mapping relationship between the input features and the furnace temperature changes. The spatiotemporal model is evaluated through the test set to determine whether the model training has converged and achieved the expected prediction accuracy. The GAT-GRU model is trained using the training set input window data, the output window data and the node adjacency matrix M, and the training principle is as follows:

[0108] ;

[0109] Where H is the number of preset simulation time steps, is the t-th output window data in the training set, is the tth input window data in the training set, , and are the furnace output feature sets in time steps t+1, t+2, and t+H respectively, , as well as are the input feature sets of the furnace section within time steps t-d+1, t-d+2, and t respectively, d is the number of time steps included in the input window, M is the adjacency matrix, and F(x) function is the spatiotemporal model being trained.

[0110] After the training is completed, the furnace temperature change simulation model of the heating furnace is obtained, and its principle is described as follows:

[0111] ;

[0112] in, The future H-step furnace temperature change trend obtained by simulation is: This is the furnace temperature change simulation model obtained after training. is the tth input window in the test set. Figures 5 to 8As shown, the temperature change of each furnace section is predicted by the furnace temperature change simulation model, and the simulation results of the preheating section, the first heating section, the second heating section and the soaking section are obtained, wherein each future time step corresponds to a predicted temperature change.

[0113] In this optional embodiment, by combining the spatiotemporal model and the adjacency matrix of the spatiotemporal graph structure, the model can comprehensively consider the spatial coupling effect and time series characteristics of the heating furnace, thereby improving the accuracy of furnace temperature prediction. Through the systematic training process, the spatiotemporal model can learn the complex space-time dependency relationship inside the heating furnace, so that the model can capture the dynamic characteristics of furnace temperature changes, including the interaction between different furnace sections and the evolution trend of furnace temperature over time.

[0114] An embodiment of the present invention provides a computer-readable storage medium on which a computer program is stored. When the computer program is executed by a processor, the above-mentioned method for constructing a furnace temperature change simulation model is implemented.

[0115] In other words, a non-volatile computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the processor performs the following operations:

[0116] Construct the spatial graph structure of the heating furnace;

[0117] Determining a spatiotemporal graph structure of the heating furnace according to the spatial graph structure of the heating furnace;

[0118] Determining an adjacency matrix of the space-time graph structure according to the nodes in the space-time graph structure of the heating furnace;

[0119] The input characteristic time series and the output characteristic time series of the heating furnace are processed by a sliding window technology to obtain input window data and output window data; wherein the input characteristic time series represents a time series composed of temperature influence data of multiple time steps, the temperature influence data represents data having an influence on the temperature data of the heating furnace, and the output characteristic time series represents a time series composed of the temperature data of the heating furnace of multiple time steps;

[0120] According to the input window data and the output window data, combined with the adjacency matrix of the space-time graph structure, a furnace temperature change simulation model of the heating furnace is obtained.

[0121] The computer-readable storage medium of the present invention clarifies the spatial relationship and mutual influence between different furnace sections in the furnace by constructing the spatial graph structure of the heating furnace, and provides a structured analysis basis for furnace temperature simulation. By determining the spatiotemporal graph structure, the time factor is further taken into consideration, and the influence of node historical data on current data can be reflected by the time connection line, thereby capturing the time dependence of the furnace temperature of the heating furnace. Then, by determining the adjacency matrix, the coupling relationship between nodes is quantified, and accurate topological structure information is provided for the model. The input and output features are processed using sliding window technology to form input and output windows containing time series information, which not only retains the temporal continuity of the data, but also enhances the model's learning ability for node features and mutual influence between furnace sections. Finally, combined with the spatiotemporal graph structure and the adjacency matrix, the constructed furnace temperature change simulation model can comprehensively consider the spatial coupling effect and time dependence, thereby achieving a more accurate prediction of the furnace temperature change trend.

[0122] A furnace temperature prediction method provided by an embodiment of the present invention includes: performing prediction using a furnace temperature change simulation model constructed by the furnace temperature change simulation model construction method as described above to obtain a furnace temperature prediction result.

[0123] Specifically, the furnace temperature change simulation model constructed above is used to predict the furnace temperature. The furnace temperature change simulation model comprehensively considers the spatial structure of the heating furnace, the time series characteristics and the mutual influence between nodes. First, the space-time graph structure is further determined by constructing the furnace section space graph structure, and then the adjacency matrix is ​​calculated based on this structure. Then, the input and output features are processed using the sliding window technology to form input and output windows. These window data reflect the changes in furnace temperature over time and the interactions between furnace sections. These data are combined with the adjacency matrix and input into the space-time model for training, and finally a simulation model that can predict furnace temperature changes is formed.

[0124] The furnace temperature prediction method of the present invention clarifies the spatial relationship and mutual influence between different furnace sections in the furnace by constructing the spatial graph structure of the heating furnace, and provides a structured analysis basis for furnace temperature simulation. By determining the spatiotemporal graph structure, the time factor is further taken into consideration, and the influence of the node historical data on the current data can be reflected by the time connection line, thereby capturing the time dependence of the furnace temperature of the heating furnace. Then, by determining the adjacency matrix, the coupling relationship between the nodes is quantified, and accurate topological structure information is provided for the model. The input and output features are processed using the sliding window technology to form an input and output window containing time series information, which not only retains the time continuity of the data, but also enhances the model's learning ability for the node features and the mutual influence between the furnace sections. Finally, combined with the spatiotemporal graph structure and the adjacency matrix, the constructed furnace temperature change simulation model can comprehensively consider the spatial coupling effect and time dependence, thereby achieving a more accurate prediction of the furnace temperature change trend.

[0125] Combination Fig. 9 As shown, an electronic device 900 provided by an embodiment of the present invention includes a memory 910 and a processor 920; the memory 910 is used to store a computer program; the processor 920 is used to implement the furnace temperature change simulation model construction method as described above when executing the computer program.

[0126] In other words, an electronic device 900 includes a memory 910 and a processor 920 coupled to the memory 910; the memory 910 is configured to store a computer program; and the processor 920 is configured to perform the following operations when executing the computer program:

[0127] Construct the spatial graph structure of the heating furnace;

[0128] Determining a spatiotemporal graph structure of the heating furnace according to the spatial graph structure of the heating furnace;

[0129] Determining an adjacency matrix of the space-time graph structure according to the nodes in the space-time graph structure of the heating furnace;

[0130] The input characteristic time series and the output characteristic time series of the heating furnace are processed by a sliding window technology to obtain input window data and output window data; wherein the input characteristic time series represents a time series composed of temperature influence data of multiple time steps, the temperature influence data represents data having an influence on the temperature data of the heating furnace, and the output characteristic time series represents a time series composed of the temperature data of the heating furnace of multiple time steps;

[0131] According to the input window data and the output window data, combined with the adjacency matrix of the space-time graph structure, a furnace temperature change simulation model of the heating furnace is obtained.

[0132] The electronic device of the present invention, by constructing the spatial graph structure of the heating furnace, clarifies the spatial relationship and mutual influence between different furnace sections in the furnace, and provides a structured analysis basis for furnace temperature simulation. By determining the spatiotemporal graph structure, the time factor is further taken into consideration, and the influence of the node historical data on the current data can be reflected by the time connection line, thereby capturing the time dependence of the furnace temperature of the heating furnace. Then, by determining the adjacency matrix, the coupling relationship between the nodes is quantified, and accurate topological structure information is provided for the model. The input and output features are processed using the sliding window technology to form an input and output window containing time series information, which not only retains the time continuity of the data, but also enhances the model's learning ability for the node features and the mutual influence between the furnace sections. Finally, combined with the spatiotemporal graph structure and the adjacency matrix, the constructed furnace temperature change simulation model can comprehensively consider the spatial coupling effect and time dependence, thereby achieving a more accurate prediction of the furnace temperature change trend.

[0133] A person of ordinary skill in the art can understand that all or part of the processes in the above-mentioned embodiment method can be completed by instructing the relevant hardware through a computer program, and the program can be stored in a non-volatile computer-readable storage medium. When the program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, storage, database or other media used in the embodiments provided by the present invention can include non-volatile and / or volatile memory. Non-volatile memory may include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory may include random access memory (RAM) or external cache memory. As an illustration and not limitation, RAM is available in many forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM).

[0134] It should be noted that, in this article, relational terms such as "first" and "second" are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device 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 device. In the absence of further restrictions, the elements defined by the sentence "comprise a ..." do not exclude the existence of other identical elements in the process, method, article or device including the elements.

[0135] The foregoing is merely a specific embodiment of the present invention, which enables those skilled in the art to understand or implement the present invention. Various modifications to these embodiments will be apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention will not be limited to the embodiments shown herein, but rather to the widest scope consistent with the principles and novel features claimed herein.

Claims

1. A method for constructing a furnace temperature variation simulation model, characterized in that: include: Constructing a spatial graph structure of a heating furnace, specifically comprising: dividing a plurality of furnace sections according to the workflow of the heating furnace, and setting nodes corresponding to each of the furnace sections; connecting the nodes corresponding to each of the furnace sections through spatial connection lines according to the workflow of the heating furnace, to obtain the spatial graph structure of the heating furnace; Determining a spatiotemporal graph structure of the heating furnace according to the spatial graph structure of the heating furnace; Determining an adjacency matrix of the space-time graph structure according to the nodes in the space-time graph structure of the heating furnace; The input characteristic time series and the output characteristic time series of the heating furnace are processed by a sliding window technology to obtain input window data and output window data; wherein the input characteristic time series represents a time series composed of temperature influence data of multiple time steps, the temperature influence data represents data having an influence on the temperature data of the heating furnace, and the output characteristic time series represents a time series composed of the temperature data of the heating furnace of multiple time steps; The method of processing the input feature time series and the output feature time series of the heating furnace by the sliding window technology to obtain input window data and output window data specifically includes: obtaining the input feature time series of the heating furnace, wherein the input feature time series includes a set of furnace section input features of multiple time steps arranged in chronological order, each of the furnace section input feature sets includes a furnace section input feature corresponding to each furnace section at the same time step, and the furnace section input feature includes the temperature influence data of the corresponding furnace section; slidingly intercepting the input feature time series according to a first preset time step by sliding a sliding window to obtain multiple input window data; obtaining the output feature time series of the heating furnace, wherein the output feature time series The sequence includes a set of furnace section output features of multiple time steps arranged in chronological order, each of the furnace section output feature sets includes a furnace section output feature corresponding to each furnace section at the same time step, and the furnace section output feature includes the temperature influence data of the corresponding furnace section; the output feature time sequence is intercepted by sliding a sliding window according to a second preset time step to obtain a plurality of output window data; wherein each of the output window data includes a set of furnace section output features of multiple time steps, and the number of the furnace section output feature sets corresponds to the size of the sliding window; each of the input window data includes a set of furnace section input features of multiple time steps, and the number of the furnace section input feature sets corresponds to the size of the sliding window; According to the input window data and the output window data, combined with the adjacency matrix of the space-time graph structure, a furnace temperature change simulation model of the heating furnace is obtained.

2. The method for constructing a furnace temperature variation simulation model according to claim 1, characterized in that: Determining the spatiotemporal graph structure of the heating furnace according to the spatial graph structure of the heating furnace includes: Determine the size of the sliding window through time-lag analysis; Determining the number of the spatial graph structures in the spatiotemporal graph structure according to the sliding window size; The same nodes in each of the spatial graph structures are connected through a time connection line to obtain the space-time graph structure of the heating furnace.

3. The method for constructing a furnace temperature variation simulation model according to claim 2, characterized in that: The step of determining the adjacency matrix of the space-time graph structure according to the nodes in the space-time graph structure of the heating furnace comprises: Numbering all nodes in the space-time graph structure according to a preset order; Determine a connection relationship between each of the nodes in the space-time graph structure, wherein the connection relationship includes one of being unconnected, being connected through the time connection line, and being connected through the space connection line; The adjacency matrix of the space-time graph structure is determined according to the numbers of the nodes and the connection relationships between the nodes.

4. The method for constructing a furnace temperature variation simulation model according to claim 1, characterized in that: The method of obtaining a furnace temperature variation simulation model of the heating furnace according to the input window data and the output window data in combination with the adjacency matrix of the space-time graph structure includes: Constructing a spatiotemporal model of the heating furnace; The space-time model is trained according to the input window data, the output window data and the adjacency matrix of the space-time graph structure to obtain a furnace temperature change simulation model of the heating furnace.

5. The method for constructing a furnace temperature variation simulation model according to claim 4, characterized in that: The step of training the spatiotemporal model according to the input window data, the output window data and the adjacency matrix of the spatiotemporal graph structure to obtain a furnace temperature variation simulation model of the heating furnace includes: Dividing the input window data and the output window data into a training set and a test set according to a preset ratio, wherein the training set and the test set each include a plurality of data pairs, each of the data pairs includes one input window data and one output window data, and a time step corresponding to the input window data in each of the data pairs is earlier than a time step corresponding to the output window data; Training the spatiotemporal model using the data pairs in the training set and the adjacency matrix; Whether the training of the space-time model is completed is determined by using the data pairs in the test set and the adjacency matrix, and the trained space-time model is used as the furnace temperature change simulation model of the heating furnace.

6. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the method for constructing a furnace temperature variation simulation model according to any one of claims 1 to 5 is implemented.

7. A furnace temperature prediction method, characterized in that: include: The furnace temperature prediction result is obtained by performing prediction using the furnace temperature variation simulation model constructed by the furnace temperature variation simulation model construction method according to any one of claims 1 to 5.

Images