Continuous casting process knowledge graph construction and process parameter prediction method, system and equipment

Abstract

The invention provides a continuous casting process knowledge graph construction and process parameter prediction method, system and device, and the method comprises the following steps: 1, obtaining a natural language description corresponding to a continuous casting process, and obtaining structured data and unstructured data; 2, corresponding triads are extracted from the obtained structured data and the obtained unstructured data respectively; 3, aligning the triple corresponding to the structured data and the triple corresponding to the unstructured data, and constructing to obtain a continuous casting process knowledge graph; according to the method, entities and relationships in the continuous casting field can be accurately identified, a relatively complete process knowledge graph is established, and process parameters required by a new steel grade are predicted through a parameter prediction method; therefore, the automation level of the production process is improved, and the production efficiency and the product quality are improved.

Description

Technical Field

[0001] The present invention belongs to the field of knowledge graph construction and process parameter prediction, and specifically relates to a method, system and equipment for continuous casting process knowledge graph construction and process parameter prediction. Background Art

[0002] The continuous casting production process is a collaborative operation of multiple equipment and multiple processes, which requires workers to accurately control various parameters. This requires workers to master a lot of production knowledge in order to deal with various possible problems. However, since there are so many equipment and processes related to continuous casting production, and the related production knowledge is also very complicated, it is often difficult for workers to master all the details and technical points. In addition, the knowledge construction of the continuous casting production process is extremely scarce, and there is a lack of a complete knowledge system in the field of continuous casting. When encountering the production needs of new steel grades, it often takes too long a pre-research stage. How to improve the production knowledge level of staff and how to use modern information technology to analyze and process production data have become important issues faced by steel companies in the continuous casting production process.

[0003] The difficulty in constructing a knowledge graph in the field of continuous casting lies in how to make full use of various types of production data to enrich the knowledge graph. At the same time, how to effectively extract entities and relationships in the field from existing texts is also one of the technical difficulties. The knowledge in the field of continuous casting lacks effective knowledge expression, and a large amount of knowledge is hidden in text materials. How to fully combine the existing extraction methods with the characteristics of the field is undoubtedly a technical difficulty. The technical difficulty in predicting continuous casting process parameters lies in the fact that parameter prediction in the existing production process is too dependent on expert knowledge, and some existing parameter prediction methods are not highly applicable in the field of continuous casting. At the same time, how to add knowledge graph assistance to the prediction of continuous casting process parameters and make full use of the structural information of the knowledge graph also poses technical challenges. Summary of the invention

[0004] The purpose of the present invention is to provide a method, system and equipment for constructing a continuous casting process knowledge graph and predicting process parameters, which solves the problems in the prior art of how to effectively extract entities and relationships in the field from existing texts and how to use knowledge graphs to complete process parameter prediction in the field of continuous casting.

[0005] In order to achieve the above object, the technical solution adopted by the present invention is:

[0006] The present invention provides a method for constructing a continuous casting process knowledge graph, comprising the following steps:

[0007] Step 1, obtaining a natural language description corresponding to the continuous casting process to obtain structured data and unstructured data;

[0008] Step 2, extracting corresponding triples from the obtained structured data and unstructured data respectively;

[0009] Step 3: Align the triples corresponding to the structured data and the triples corresponding to the unstructured data to construct a continuous casting process knowledge graph.

[0010] Preferably, in step 2, the corresponding triples are manually extracted from the structured data.

[0011] Preferably, in step 2, an entity relationship joint extraction algorithm based on span entity tags and self-attention mechanism is used to extract the corresponding triples from the unstructured data.

[0012] Preferably, in step 3, the triples corresponding to the structured data and the triples corresponding to the unstructured data are aligned to construct a continuous casting process knowledge graph, and the specific method is:

[0013] The multi-knowledge graph entity alignment model is used to align the triples corresponding to the structured data and the triples corresponding to the unstructured data to construct the continuous casting process knowledge graph.

[0014] A continuous casting process knowledge graph construction system, comprising:

[0015] A data acquisition unit, used to acquire a natural language description corresponding to the continuous casting process to obtain structured data and unstructured data;

[0016] A triple extraction unit, used for extracting corresponding triples from the obtained structured data and unstructured data respectively;

[0017] The graph construction unit is used to align the triples corresponding to the structured data and the triples corresponding to the unstructured data to construct a continuous casting process knowledge graph.

[0018] A computer device comprising:

[0019] a processor suitable for executing a computer program;

[0020] A computer-readable storage medium stores a computer program, and when the computer program is executed by the processor, the method described is performed.

[0021] A method for predicting continuous casting process parameters comprises the following steps:

[0022] Step 1, based on the continuous casting process knowledge graph described in claim 1, obtaining feature representations of key process nodes;

[0023] Step 2, clustering the acquired continuous casting production data to obtain clustering features corresponding to the continuous casting production data;

[0024] Step 3, fusing the obtained feature representation of the key process nodes and the clustering features to obtain fused features;

[0025] Step 4, using the obtained fused features to train the set scoring function to obtain a trained scoring function;

[0026] Step 5: Use the trained scoring function to predict the parameters of the continuous casting process.

[0027] Preferably, in step 1, a graph attention algorithm is used to obtain feature representations of key process nodes from the continuous casting process knowledge graph.

[0028] A continuous casting process parameter prediction system, comprising:

[0029] A feature acquisition unit is used to obtain feature representations of key process nodes based on the continuous casting process knowledge graph;

[0030] A clustering unit, used for clustering the acquired continuous casting production data to obtain clustering features corresponding to the continuous casting production data;

[0031] A fusion unit is used to fuse the obtained feature representation of the key process node and the clustering feature to obtain a fused feature;

[0032] A function training unit, used to train a set scoring function using the obtained fused features to obtain a trained scoring function;

[0033] The parameter prediction unit is used to predict the parameters of the continuous casting process using the trained scoring function.

[0034] A computer device comprising:

[0035] a processor suitable for executing a computer program;

[0036] A computer-readable storage medium stores a computer program, and when the computer program is executed by the processor, the method described is performed.

[0037] Compared with the prior art, the present invention has the following beneficial effects:

[0038] The present invention provides a method for constructing a continuous casting process knowledge graph and a method for predicting process parameters. The entity and relationship in the field are extracted from unstructured text by an entity relationship joint extraction algorithm based on span entity tags and self-attention mechanism, and the entity alignment between the artificially established knowledge graph and the triples obtained by the entity relationship extraction algorithm is completed by using a graph fusion algorithm based on a graph convolutional network to achieve the purpose of knowledge graph fusion. At the same time, in order to use the knowledge graph and historical production data to achieve the task of predicting the process parameters of new steel grades, the historical data is first clustered according to key processes to obtain the data features of each process classification as parameter prediction, and then the continuous casting process knowledge graph is divided into subgraphs according to key processes, and the knowledge graph divided into key process subgraphs is extracted by a graph attention network as the structural features during parameter prediction, and then the feature fusion module is used to fuse the two types of features as the features of the process parameter prediction method, so as to complete the process parameter prediction; the present invention can accurately identify entities and relationships in the continuous casting field, establish a relatively complete process knowledge graph, and predict the process parameters required for the new steel grade by the parameter prediction method; and thus the present invention improves the automation level of the production process, thereby improving production efficiency and product quality. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] Figure 1 A diagram showing the construction process of the knowledge graph in the field of continuous casting according to the present invention;

[0040] Figure 2 It is a schematic diagram of the construction process of the knowledge graph in the field of continuous casting according to the present invention;

[0041] Figure 3 It is a structural diagram of the entity relationship joint extraction algorithm based on span entity labeling and attention mechanism described in the present invention;

[0042] Figure 4 This is a structural diagram of the multi-knowledge graph entity alignment model described in the present invention;

[0043] Figure 5 A network structure diagram for prediction of continuous casting process parameters based on knowledge graph according to the present invention;

[0044] Figure 6 The figure is a flow chart of the continuous casting process clustering algorithm of the present invention. DETAILED DESCRIPTION

[0045] In the following description, specific details such as specific system structures, technologies, etc. are provided for the purpose of illustration rather than limitation, so as to provide a thorough understanding of the embodiments of the present application. However, it should be clear to those skilled in the art that the present application may also be implemented in other embodiments without these specific details. In other cases, detailed descriptions of well-known systems, devices, circuits, and methods are omitted to prevent unnecessary details from obstructing the description of the present application.

[0046] It should be understood that when used in the present specification and the appended claims, the term "comprising" indicates the presence of described features, wholes, steps, operations, elements and / or components, but does not exclude the presence or addition of one or more other features, wholes, steps, operations, elements, components and / or combinations thereof.

[0047] It should also be understood that the term “and / or” used in the specification and appended claims refers to any and all possible combinations of one or more of the associated listed items, and includes these combinations.

[0048] As used in the specification and appended claims of this application, the term "if" can be interpreted as "when" or "uponce" or "in response to determining" or "in response to detecting", depending on the context. Similarly, the phrase "if it is determined" or "if [described condition or event] is detected" can be interpreted as meaning "uponce it is determined" or "in response to determining" or "uponce [described condition or event] is detected" or "in response to detecting [described condition or event]", depending on the context.

[0049] In addition, in the description of the present application specification and the appended claims, the terms "first", "second", "third", etc. are only used to distinguish the descriptions and cannot be understood as indicating or implying relative importance.

[0050] References to "one embodiment" or "some embodiments" etc. described in the specification of this application mean that one or more embodiments of the present application include specific features, structures or characteristics described in conjunction with the embodiment. Therefore, the statements "in one embodiment", "in some embodiments", "in some other embodiments", "in some other embodiments", etc. that appear in different places in this specification do not necessarily refer to the same embodiment, but mean "one or more but not all embodiments", unless otherwise specifically emphasized in other ways. The terms "including", "comprising", "having" and their variations all mean "including but not limited to", unless otherwise specifically emphasized in other ways.

[0051] Example 1

[0052] See also Figures 1 to 6, this embodiment discloses a method for constructing a continuous casting process knowledge graph and predicting process parameters, which applies the knowledge graph to the continuous casting field, and uses an entity relationship joint extraction algorithm based on span entity tags and self-attention mechanism to extract domain entities and relationships from unstructured text. Through a graph fusion algorithm based on a graph convolutional network, entity alignment and fusion of artificial knowledge graphs and entity relationship extraction results are achieved. In order to complete the process parameter prediction of new steel grades, historical production data is used to extract data features according to key process clusters, and the key process subgraphs are extracted in combination with the graph attention network. Finally, the process parameter prediction is achieved through the feature fusion module, which specifically includes the following steps:

[0053] Step 1. According to the continuous casting production records and production data, obtain the relevant natural language description of the continuous casting process, organize and preprocess it, and obtain structured text and unstructured text;

[0054] Among them, the preprocessing of continuous casting production records and production data specifically includes the following aspects:

[0055] Classify all data into structured text and unstructured text, and pre-process the unstructured text in accordance with algorithm training.

[0056] Step 2. Manually obtain triples for structured texts, and use the entity relationship joint extraction algorithm based on span entity tags and self-attention mechanism to complete triple extraction for unstructured texts;

[0057] The entity relationship joint extraction algorithm based on span entity tags and self-attention mechanism is used to extract triples from unstructured text. The specific method is:

[0058] First, the unstructured text is input into the pre-trained BERT model to obtain the feature map of the input text;

[0059] Then, the obtained feature map is input into the two classifiers in the BERT model, and the two classifiers are used to predict the starting position and the ending position of the entity in the unstructured text respectively, and use them as the possible head entity and tail entity in the knowledge graph triple; then the CLS token of the BERT model is projected into the relationship detection space using a multi-label classification strategy to complete the encoding of entities and relationships; finally, the detected head entity set and relationship set are used to predict the starting position of the tail entity to ensure the accurate identification of the head entity and tail entity and their relationships in the knowledge graph triple.

[0060] The basis of the BERT model is the transformer model, which uses an encoder-decoder architecture. In the entire transformer architecture, it uses the attention mechanism and fully connected layers to process text. This attention method can be seen as finding hidden relationships within the input information sequence. In order to effectively obtain entities and their positions, a span-based annotation model is used, and four binary classifiers are used to predict the start and end positions of the entities respectively.

[0061] X=BERT(T)

[0062]

[0063] Where: X is the semantic feature map, BERT is the BERT model, T is the text sentence, are the probabilities of the i-th token in the input being the start and end positions of the head and tail entities, respectively, and W head_start , W head_end , W tail_start , W tail_end are the weight matrices of the corresponding classifiers, and σ is the sigmoid activation function.

[0064] In order to better complete the entity extraction task, set the loss function to train the classifier, and put the obtained head entity and tail entity into the set E respectively. h and E t middle.

[0065] Loss entity =Loss head +Loss tail

[0066]

[0067] Where: Loss entity is the loss function for entity recognition, Loss head 、Loss tail are the loss functions for head entity prediction and tail entity prediction, respectively, θ (s|X) is the binary classification loss function.

[0068] In the BERT model, the CLS tag is added to the beginning of each input sequence. Specifically, when a series of text is input into the BERT model for processing, the first input tag of the model is this special CLS tag. After the CLS tag has been transformed by BERT's multiple layers, it can capture the global information of the entire input sequence. Its hidden state encodes the information of the entire input sequence and can be used to predict downstream tasks. In order to identify relevant relationships in text, a multi-label classification strategy is adopted, and the CLS tag is applied to the relationship detection space of multi-label classification. CLS is its loss function as follows. The detected relationship is placed in the set R.

[0069] relation=σ(W relation X CLS )

[0070]

[0071] Where: relation is the relationship detection function, W relation is the relationship detection classifier weight matrix, X CLS CLS label representation for BERT model, Loss relation is the loss function for relationship detection.

[0072] Use a fully connected layer to encode the relationship as follows:

[0073]

[0074] Where: is the relationship span representation, FullyConnect is the fully connected layer function, is the relation embedding vector.

[0075] Finally, a binary classifier is used to identify the starting position of the tail entity given the current head entity and relationship, as shown in the following formula:

[0076]

[0077] Where: is the probability of the starting position of the tail entity, W select Select the feature matrix for the tail entity, A i Characterize the results for attention.

[0078] The loss function at this stage is:

[0079]

[0080] Where: Loss translate is the loss function for the tail entity translation stage.

[0081] The joint loss function is defined as follows:

[0082] Loss = aLoss entity +bLoss relation +cLoss translate

[0083] Where: Loss is the loss function of the entity relationship joint extraction algorithm, and a, b, and c are constants.

[0084] Step 3. Use the multi-knowledge graph entity alignment model to align the triples obtained in the previous step, thereby completing the construction of the continuous casting process knowledge graph;

[0085] The update rules for entity and relationship embeddings are as follows:

[0086]

[0087] Where: N r is the set of relations connected to entity i, is the set of tail entities connected to entity i by relationship r, is the set of head entities connected to entity i by relation r, are the l-th layer embeddings representing the head entity, relation entity and tail entity in the neural network, σ is the activation function, and W l is the weight matrix of the lth layer of the neural network; is the parameter of the next layer of the graph neural network.

[0088] Using the known aligned entities and relations, the objective function of the following formula is trained so that in the shared embedding space, the distance between aligned entities and relations should be minimized, and the distance between unaligned entities and relations should be maximized.

[0089]

[0090] Where: F is the objective function, (e x ,e y ) is a pair of aligned entities, (e′ x ,e′ y ) is a pair of entities that will not be aligned, d(…,…) is the distance between the two entities in space, η is the hyperplane that separates whether they are aligned, (r x ,r y ) is a pair of aligned relations, (r′ x ,r′ y ) is a pair of relationships that will not be aligned; α is a hyperparameter, E a is the set of known aligned entities.

[0091] Step 4. Based on the knowledge graph obtained in the previous step, the graph attention algorithm is used to obtain the feature representation of key process nodes;

[0092] The update formula of the graph attention network is:

[0093]

[0094] Where: e ij is the importance of the feature of node j to node i, α ij is the weight coefficient calculated by the attention mechanism, W is the shared weight matrix, is the feature vector of node i and node j.

[0095] Step 5. Cluster the production data using a clustering algorithm to obtain its clustering features;

[0096] The objective function and constraints of the continuous casting process clustering algorithm are as follows:

[0097]

[0098] Where: u ij is the sample point x i With cluster center v j The membership degree, m is the fuzzy index, m>1, d ij is the sample point x i With cluster center v j distance.

[0099] In order to minimize the objective function J(U,V), the Lagrange multiplier method is used on the objective function under the premise of satisfying the constraints to obtain the membership matrix U and the cluster center v j , as shown below:

[0100]

[0101] Where: u ij is the sample point x i With cluster center v j The membership degree, d ij is the sample point x i With cluster center v j The distance, v j is the cluster center of a certain class.

[0102] Step 6. Use the feature fusion algorithm to complete the fusion of the feature representation and clustering features of the key process nodes obtained in step 4, as the classification basis for process parameter prediction, and train the scoring function to complete the process parameter prediction.

[0103] The fusion formula is as follows:

[0104]

[0105] Where: Z is the fused feature representation, is the initial feature integration, M is the attention weight function, is matrix multiplication.

[0106] The scoring function for selecting the prediction results is as follows:

[0107] z(h,r,t)=‖h°rt‖

[0108] Where: z is the scoring function, h is the head entity, r is the relationship, t is the tail entity, and ° is the Hadamard product.

[0109] Example 2

[0110] This embodiment also discloses a continuous casting process knowledge graph construction and process parameter prediction system, including:

[0111] A continuous casting process knowledge graph construction system, characterized by comprising:

[0112] A data acquisition unit, used to acquire a natural language description corresponding to the continuous casting process to obtain structured data and unstructured data;

[0113] A triple extraction unit, used for extracting corresponding triples from the obtained structured data and unstructured data respectively;

[0114] The graph construction unit is used to align the triples corresponding to the structured data and the triples corresponding to the unstructured data to construct a continuous casting process knowledge graph.

[0115] A feature acquisition unit is used to obtain feature representations of key process nodes based on the continuous casting process knowledge graph;

[0116] A clustering unit, used for clustering the acquired continuous casting production data to obtain clustering features corresponding to the continuous casting production data;

[0117] A fusion unit is used to fuse the obtained feature representation of the key process node and the clustering feature to obtain a fused feature;

[0118] A function training unit, used to train a set scoring function using the obtained fused features to obtain a trained scoring function;

[0119] The parameter prediction unit is used to predict the parameters of the continuous casting process using the trained scoring function.

[0120] Example 3

[0121] This embodiment 3 provides a computer device, including: a memory for storing a computer program; and a processor for implementing the steps of a computer method when executing the computer program.

[0122] When the processor executes the computer program, the steps of the above-mentioned computer method are implemented, or when the processor executes the computer program, the functions of each module in the above-mentioned system are implemented.

[0123] Exemplarily, the computer program may be divided into one or more modules / units, which are stored in the memory and executed by the processor to implement the present invention. The one or more modules / units may be a series of computer program instruction segments capable of implementing preset functions, and the instruction segments are used to describe the execution process of the computer program in the computer device.

[0124] The computer device may be a computing device such as a desktop computer, a notebook, a PDA, and a cloud server. The computer device may include, but is not limited to, a processor and a memory. Those skilled in the art will appreciate that the above are examples of computer devices and do not constitute a limitation on computer devices. The computer device may include more components than the above, or a combination of certain components, or different components. For example, the computer device may also include input and output devices, network access devices, buses, etc.

[0125] The processor may be a central processing unit (CPU), other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor may be a microprocessor or any conventional processor, etc. The processor is the control center of the computer device, and uses various interfaces and lines to connect various parts of the entire computer device.

[0126] The memory may be used to store the computer programs and / or modules, and the processor implements various functions of the computer device by running or executing the computer programs and / or modules stored in the memory and calling the data stored in the memory.

[0127] The memory may mainly include a program storage area and a data storage area, wherein the program storage area may store an operating system, an application required for at least one function (such as a sound playback function, an image playback function, etc.), etc.; the data storage area may store data created according to the use of the mobile phone (such as audio data, a phone book, etc.), etc. In addition, the memory may include a high-speed random access memory, and may also include a non-volatile memory, such as a hard disk, a memory, a plug-in hard disk, a smart memory card (SmartMediaCard, SMC), a secure digital (SecureDigital, SD) card, a flash card (FlashCard), at least one disk storage device, a flash memory device, or other volatile solid-state storage devices.

[0128] The embodiments described above are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application has been described in detail with reference to the aforementioned embodiments, a person skilled in the art should understand that the technical solutions described in the aforementioned embodiments may still be modified, or some of the technical features may be replaced by equivalents. Such modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present application, and should all be included in the protection scope of the present application.

Claims

1. A method for constructing a continuous casting process knowledge graph, characterized in that: The following steps are involved: Step 1, obtaining a natural language description corresponding to the continuous casting process to obtain structured data and unstructured data; Step 2, extracting corresponding triples from the obtained structured data and unstructured data respectively; Step 3: Align the triples corresponding to the structured data and the triples corresponding to the unstructured data to construct a continuous casting process knowledge graph.

2. A method for constructing a continuous casting process knowledge graph according to claim 1, characterized in that: In step 2, the corresponding triples are manually extracted from the structured data.

3. A method for constructing a continuous casting process knowledge graph according to claim 1, characterized in that: In step 2, the entity relationship joint extraction algorithm based on span entity tags and self-attention mechanism is used to extract the corresponding triples from the unstructured data.

4. A method for constructing a continuous casting process knowledge graph according to claim 1, characterized in that: In step 3, the triples corresponding to the structured data and the triples corresponding to the unstructured data are aligned to construct a continuous casting process knowledge graph. The specific method is: The multi-knowledge graph entity alignment model is used to align the triples corresponding to the structured data and the triples corresponding to the unstructured data to construct the continuous casting process knowledge graph.

5. A continuous casting process knowledge graph construction system, characterized in that: include: A data acquisition unit, used to acquire a natural language description corresponding to the continuous casting process to obtain structured data and unstructured data; A triple extraction unit, used for extracting corresponding triples from the obtained structured data and unstructured data respectively; The graph construction unit is used to align the triples corresponding to the structured data and the triples corresponding to the unstructured data to construct a continuous casting process knowledge graph.

6. A computer device, characterized in that: include: a processor suitable for executing a computer program; A computer-readable storage medium, wherein a computer program is stored in the computer-readable storage medium, and when the computer program is executed by the processor, the method according to any one of claims 1 to 4 is performed.

7. A method for predicting continuous casting process parameters, characterized in that: The following steps are involved: Step 1, based on the continuous casting process knowledge graph described in claim 1, obtaining feature representations of key process nodes; Step 2, clustering the acquired continuous casting production data to obtain clustering features corresponding to the continuous casting production data; Step 3, fusing the obtained feature representation of the key process nodes and the clustering features to obtain fused features; Step 4, using the obtained fused features to train the set scoring function to obtain a trained scoring function; Step 5: Use the trained scoring function to predict the parameters of the continuous casting process.

8. A method for predicting continuous casting process parameters according to claim 7, characterized in that: In step 1, the graph attention algorithm is used to obtain the feature representation of key process nodes from the continuous casting process knowledge graph.

9. A continuous casting process parameter prediction system, characterized in that: include: A feature acquisition unit is used to obtain feature representations of key process nodes based on the continuous casting process knowledge graph; A clustering unit, used for clustering the acquired continuous casting production data to obtain clustering features corresponding to the continuous casting production data; A fusion unit is used to fuse the obtained feature representation of the key process node and the clustering feature to obtain a fused feature; A function training unit, used to train a set scoring function using the obtained fused features to obtain a trained scoring function; The parameter prediction unit is used to predict the parameters of the continuous casting process using the trained scoring function.

10. A computer device, characterized in that: include: a processor suitable for executing a computer program; A computer-readable storage medium, wherein a computer program is stored in the computer-readable storage medium, and when the computer program is executed by the processor, the method according to claim 7 or 8 is performed.

Images