Data processing method based on combination of data driving and continuous casting production process

Abstract

The invention discloses a data processing method based on combination of data driving and a continuous casting production process, and belongs to the technical field of industrial data management. The method comprises the following steps: extracting multi-source heterogeneous data in the continuous casting production process according to an IT class and an OT class, and mapping a physical model of a physical sensor in a digital space; the Internet of Things equipment accesses collected data to an industrial gateway, an industrial grid carries out analysis, format conversion and downsampling on the data and then reports the data to a data engine, and the data engine stores multi-source heterogeneous data into a business database of a corresponding object model in a message queue publishing / subscribing mode; performing real-time online data management including abnormal state detection, data completion and the like; the production process data and the production materials are aligned in time and space; and a metallurgical affair map and a digital casting blank expansion map are constructed. According to the invention, online data management is realized, the data reliability is improved, and a data management scheme is provided.

Description

Technical Field

[0001] The present invention relates to the technical field of industrial data management, and in particular to a data management method based on the combination of data-driven and continuous casting production process. Background Art

[0002] The domestic steel industry is developing in a coordinated manner towards digitalization, greening and intelligence. With the gradual upgrading of the industrial structure, my country's steel industry has developed rapidly. Various monitoring instruments at steel production sites have gradually developed towards multifunctionality, systematization and intelligence. The level of informatization and automation in the steel industry is also constantly improving. Enterprises are becoming more and more capable of acquiring data, and through the use of manufacturing execution systems, enterprise resource planning management systems, master data management and other systems to centrally manage and control production activities, the actual industrial production data obtained is becoming more and more comprehensive. The resulting production data is growing explosively in milliseconds. These massive data contain important information closely related to the production status and the quality of the final product, which has changed the quality control mode of steel products from the original "offline analysis" to "online control", from "short process" to "cross-process", and from "single mode" to "multi-mode".

[0003] Although the existing industrial data management platform has made great progress, there are still many problems in the data collection-analysis-processing-storage-application process:

[0004] In terms of data collection, since the quality of steel products is affected by the composition of molten steel and multiple production process parameters, these data are scattered in different equipment, sensors, and production systems. They are affected by the production environment, IoT ring network, edge gateway, scheduling process and other conditions. The collected data are difficult to efficiently parse and transmit in a structured manner. The data collection process consumes a lot of manpower and costs, resulting in certain difficulties in collecting multi-source heterogeneous data.

[0005] In terms of data analysis, facing different data sources in industrial production, since the connected devices, sensors, and production systems all come from different manufacturers, there is no unified data transmission protocol and interface, which makes analysis difficult. The data analysis module is highly dependent on the hardware part, and has the problem of strong coupling between software and hardware development. Finally, as the number of industry applications and equipment increases, new applications need to be repeatedly developed for different equipment protocols, which is difficult to scale.

[0006] In terms of data processing, due to the complexity of the on-site production environment and the diversity of business scenarios, customized data processing logic needs to be developed each time. At the same time, the data platform needs to pre-process the collected real-time data according to business needs. The data processing process must meet high performance, flexibility, low latency and other characteristics, resulting in staggered and repeated data processing processes, which consumes a lot of manpower and economic costs.

[0007] In terms of data management, different factories have different processes, and the business rules are complex and diverse. The production process data has multiple types of data and complex time-space correspondences. It is necessary to comprehensively match data from different sources in multiple information dimensions such as time-space. Traditional data management is mostly based on equipment and data storage in table form, which results in the relationship between data being hidden and makes it difficult to turn data into assets. Summary of the invention

[0008] In response to the problems existing in the existing industrial data management platform, the present invention provides a data management method based on the combination of data-driven and continuous casting production process, which adopts data collection, data access, data management, metallurgical principle map, and billet data expansion map data management solution to assetize and aggregate the metallurgical production process time series data from a discrete state, with material management as the core model, to achieve the purpose of online management of factory production holographic data and eliminate data islands.

[0009] The present invention provides a data management method based on the combination of data drive and continuous casting production process, comprising the following steps:

[0010] Step 1: Extract the multi-source heterogeneous data of the continuous casting production process according to IT (information technology) and OT (operation technology) categories, including: collecting the multi-source heterogeneous data of the continuous casting production process according to IT and OT categories, and collecting OT data in different areas according to the area to which the physical model belongs; mapping the physical "physical model" of the virtual sensor in the digital space based on the physical sensor, and using the "physical model" management method to directly collect time series data from the sensors of the continuous casting production process monitoring system.

[0011] Step 2: Multi-source heterogeneous data access and storage for the continuous casting production process, including: IoT devices access the collected time series data to the industrial gateway, the industrial gateway parses, converts the format and downsamples the time series data, and reports it to the data engine. The device access service of the data engine uses the message queue publish / subscribe method to store multi-source heterogeneous data in the business database of the corresponding object model.

[0012] Step 3: Real-time online data governance for the continuous casting production process, including: the data engine obtains the time series data of the continuous casting production process from the subscribed message queue, monitors the abnormal status of equipment signals, identifies the type of data anomalies, removes abnormal data, completes missing data, and performs standardized data processing operations on process rules.

[0013] Step 4: Spatiotemporal alignment of data for the continuous casting production process, including: for the stored business database, including OT data and IT data, based on metallurgical specifications, manufacturing process parameters, process mechanism knowledge and production data drive, OT / IT data spatiotemporal matching and process connection for data, obtaining production material tracking information for each order furnace, obtaining the effective working time of production materials passing through the sensor, segmenting the time series data collected by the sensor, using the order furnace and billet number to manage the OT data and IT data of the same order furnace, setting different codes for production materials in different production processes, and setting different codes for production materials in different production processes. The production material code is uniquely marked with furnace number + billet number + production process code to achieve the alignment of production process data and production materials in time and space.

[0014] Step 5: Constructing a metallurgical event graph for the continuous casting production process, including: the data engine takes production units, production equipment, production procedures, production processes and production materials as entities of the knowledge graph, and uses the metallurgical production process knowledge to construct the flow relationship between production units, production equipment, production procedures and production processes for production materials, constructing the event knowledge graph from top to bottom, and showing the evolution path of the production material production process derivative events.

[0015] Step 6: Construct a digital ingot expansion map for the continuous casting production process, including: a business database based on the object model, with production materials as the core nodes, and a top-down data structure based on the material flow. For each production process, corresponding production process data, production performance data, inspection and testing data, basic product information, and production metallurgical specification data are obtained, and data nodes are constructed. The relationship between each data node and the core node is established to obtain a digital ingot expansion map.

[0016] The data governance method based on data drive combined with continuous casting production process of the present invention has the following advantages: (1) The method of the present invention realizes online data governance, adapts to multi-state abnormal data collection, and can prevent the missing and abnormal phenomena of curve data in the collection process through real-time data completion, thereby improving the reliability of data; (2) The method of the present invention realizes data alignment based on process metallurgical knowledge, aggregates production process data, and aligns data in time and space based on the material production process model, thereby avoiding the problem of "simultaneous different sources" of data and ensuring the reliability and homology of data; (3) The method of the present invention realizes a data management mode based on the data expansion graph of the whole life cycle of materials, constructs a data asset shell management structure graph of the casting production process, and performs data aggregation and management. The method of the present invention expands the graph of production metallurgical specification data, thereby facilitating the study of taking measures to improve the production process, ensuring that data is managed based on the material model, and can better cover the relationship between data in the material production process, which is conducive to mining how to affect the working conditions of product quality. The method of the present invention has the advantages of transforming data into assets and improving data quality and value, which is of great significance to improving the convenience of data use by staff in the later stage. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 It is an overall flow chart of the data management method based on the combination of data-driven and continuous casting production process of the present invention;

[0018] Figure 2 This is a flow chart of pre-online management of time series data in an embodiment of the present invention;

[0019] Figure 3 It is a flow chart of the spatiotemporal alignment of data in the continuous casting production process according to an embodiment of the present invention;

[0020] Figure 4 It is a schematic diagram of a continuous casting production process diagram in an embodiment of the present invention;

[0021] Figure 5 It is a schematic diagram of the digital ingot expansion map for the continuous casting production process in an embodiment of the present invention. DETAILED DESCRIPTION

[0022] The technical solution of the present invention is described below in conjunction with the accompanying drawings and embodiments.

[0023] The present invention provides a data management method based on the combination of data-driven and continuous casting production process, such as Figure 1As shown, a data management scheme of data collection, data access, data management, metallurgical principle map, and ingot data expansion map is adopted to aggregate the metallurgical production process time series data from a discrete state into assets, with material management as the core mode, to achieve the purpose of online management of factory production holographic data, eliminate data islands, and has the advantages of transforming data into assets and improving data quality and value, which is of great significance to improving the convenience of data use by staff in the later stage. The implementation of the method of the present invention is explained in five steps below.

[0024] Step 1: Extraction of multi-source heterogeneous data for continuous casting production process. The data sources of the data engine are mainly divided into two parts, namely, the real-time high-frequency data of production process parameter variables collected from the sensors of the monitoring equipment of the production process, and the data obtained from the historical production system database. The method of the present invention constructs a data "physical model" from the data form or sensor. The "physical model" refers to the mapping of virtual sensor objects in the digital space, which has the same definition as the sensors in the production process. The physical model contains the basic attributes of the entity, such as the production unit, process, unit and other information to which the equipment belongs.

[0025] The data engine collects data by category for IT and OT based on the physical model, and collects OT data in partitions according to the area to which the physical model belongs. It can efficiently distinguish and classify multi-source heterogeneous data according to different attribute information, and store them in categories, providing important support for subsequent data governance and data management.

[0026] The types of data collection are mainly divided into IT data and OT data. IT data mainly comes from the primary system, and OT data is recorded data generated by the equipment, mainly from the secondary production system, MES system and other production subsystems. Based on the physical sensor, the virtual sensor's physical model is mapped in the digital space, and the physical model management method is used to directly collect time series data from the sensors of the continuous casting production process monitoring system. The physical model of the required historical production data form is constructed and mapped to the digital space. The historical production data form is stored according to the IT and OT categories, and the OT data is stored according to the area to which the physical model belongs.

[0027] The present invention embodiment studies the ingot production process data: the continuous casting process is the core link in the modern steel production process. As an intermediate process, it affects the further processing of steel products and the quality of the final product. It is a complex dynamic change process system in which the multi-system alloy in the high-temperature molten steel is cooled and solidified to form a steel ingot material. It has many advantages. It can not only greatly shorten the production process and reduce the intermediate process, but also improve the quality of the ingot and enhance the utilization of molten steel and energy. The main equipment of the continuous casting process includes: a ladle turntable, a tundish, a crystallizer, a straightening machine, a cutting machine, etc. In the production process, the ladle transports the high-temperature molten steel liquid after the refining process to the turntable. After the ladle turntable rotates to the pouring position, the molten steel is injected into the tundish. The tundish then distributes the molten steel to each crystallizer through the water distribution port, and rapidly solidifies and crystallizes in the crystallizer. Under the joint action of the straightening machine and the crystallization vibration device, the casting in the crystallizer is pulled out. After secondary cooling, it is cut into a slab of a certain length by the cutting machine after infrared measurement.

[0028] Step 2: Multi-source heterogeneous data access and storage for the continuous casting process. The access methods for object model data collection are divided into general protocol access and third-party protocol access. IoT devices can access the industrial gateway through protocols such as MQTT, Modbus, OPCUA, OPCDA, S7, JDBC, etc. The industrial gateway converts the format of the parsed data and then reports the data to the device access service of the data engine platform in the form of JSON. Since the data collection frequency of each parameter is different, in order to ensure that their timestamps are aligned and facilitate subsequent time series management, the industrial gateway then downsamples the time series data and selects the downsampling frequency according to the different distribution of parameter data points. The device access service of the data engine uses the publish-subscribe tool RabbitMQ for data transmission, and stores the collected time series data in the business database of the corresponding object model. Rabbit MQ is a message queue divided into producers and consumers. The message producer publishes the data to the Rabbit MQ message queue. Rabbit MQ has multiple Exchange and Queue queues. Different types of message data will be stored in different queues. Message consumers will subscribe to the queue where the message is stored to read the time series data. The data transmission and storage method is the consumer role. You can directly subscribe to the corresponding queue to obtain the data in Json format, parse it into structured data, and store the parsed structured data in the corresponding business.

[0029] Real-time data storage is achieved by reading the data in the subscribed message queue and writing it into the TDengine database in real time according to the object model. At the current data sampling frequency, the database is monitored in real time to obtain whether there is data written at the current time and whether the written data is valid. If the data is invalid, the data at the current time point is marked to provide a basis for subsequent data governance work.

[0030] Step 3: Real-time online data governance for the continuous casting production process. Data governance refers to the real-time online data processing operations on the time series data obtained by the data engine from the subscribed message queue, mainly including the abnormal status monitoring of equipment signals, data abnormality type identification, abnormal data elimination, missing data completion operations, and process rule standardization data processing for the time series data in the message queue. Figure 2 As shown, the data engine first monitors the abnormal state of the device signal in the data transmission channel, which means: monitoring the network status of the sensor device, hardware communication, etc., and timely monitoring and alarming the data status judgment caused by the device abnormality. If the device is abnormal, the data governance configuration item of this device will be assigned in the business database, and an alarm will be issued to ensure the accuracy of real-time data processing; secondly, the data upsampling time series filling means: real-time data upsampling filling for time series data, and filling the missing time positions in the data with null values ​​according to the resampling frequency; then, the data abnormality type judgment means: real-time category judgment for the abnormal state of the data, and analysis of data abnormality attribute classification, such as time delay, data missing, data failure, data noise, data exceeding the standard, data granularity abnormality, etc.; finally, according to the abnormal state of the data after judgment, standardized data processing operations are performed according to the continuous casting production process rules, that is, different metallurgical rules are used to unify the data format.

[0031] The advantages of this step include the following two aspects: data stability: by monitoring equipment anomalies, data anomalies, etc., and performing data governance online in real time, it can eliminate data omissions and frame drops, and improve the overall data quality by improving data stability; data accuracy: data filling based on the combination of process rules and data-driven can improve data quality for businesses with high requirements for data accuracy.

[0032] Step 4: Spatiotemporal alignment of data for the continuous casting production process. Since the continuous casting production measurement sensors are arranged along the longitudinal space of the production line, and the data collection process is carried out according to time, but the time when the production materials pass through the equipment sensors is not at the same time, the production process of the continuous casting process is a continuous state, so the production record data is misaligned in time and space. The data collected at the same time belongs to different products, resulting in a decrease in the value of the collected data and the collected data is "simultaneous but different sources". Data spatiotemporal alignment is based on the combination of metallurgical specifications, manufacturing process parameters, process mechanism knowledge and production data drive, and performs spatiotemporal matching of OT / IT data and process connection for data, and aligns the production process data with product materials in time and space.

[0033] like Figure 3As shown, firstly, the production material tracking information of the entire order furnace is obtained from the production system, such as the entry time and end time of the production material of a certain furnace entering the converter, refining furnace, and crystallizer, the start time and stop time of the ladle pouring, and the start time of the crystallizer dummy rod movement; secondly, the time series data and event information collected based on the physical model in the production process are obtained, such as the data of the production process recording sensors between the molten steel from the large ladle, the middle ladle, the crystallizer, the fan-shaped area, the secondary cooling area, the straightening area to the flame cutting machine in the production process, that is, the time series data collected based on the physical model in the data acquisition step, to provide data support for the information matching of the continuous casting billet belonging to a certain order and the molten steel of a certain furnace; then, When the flame cutting completion signal of the continuous casting billet is obtained, it means that the production process of the billet is over. At this time, the billet will have its own production information, namely the billet number, and will be associated with the current production furnace number information, and the billet will be divided into the furnace number. The furnace number information and the billet number information obtained after cutting are combined to mark the billet; finally, the continuous casting production process data and the billet time direction are transformed and aligned in time and space through the billet production equipment sensor position information, casting information, cutting information, molten steel weight, casting speed, crystallizer pulling speed, billet size and other information based on volume transformation calculation, and the effective working time of the continuous casting billet passing the sensor is obtained, and the data of the material passing the sensor is segmented. An order contains multiple furnaces, each furnace corresponds to a production process, and the furnace number + billet number is set as the unique number of the production process. The production materials have different codes in different production processes. According to the unique number of the production process + production material code, the generated material of a certain generation process can be located.

[0034] The time series data of the same order heat can be uniformly associated through the order heat code, and the data of each process can be managed by combining the order heat and production material code.

[0035] The advantages of this step include the following two aspects: data homology, by matching the production record data in time and space, ensuring the effective working time of the material passing through the sensor, and ensuring the "homology" of the collected data is of great significance for data-based modeling and analysis in the metallurgical industry.

[0036] Step 5: Construct a knowledge graph for the continuous casting production process. Use the knowledge of the metallurgical production process to construct a knowledge graph, and use the graphical representation of logical relationships such as sequential relationships, causal relationships, and hierarchical relationships between data to accurately obtain the evolution path of the material production process derivative events, helping metallurgical workers quickly and intuitively understand the evolutionary laws of the development of things.

[0037] The knowledge graph of facts and principles is a graphical representation method that describes the logical relationships such as the sequential relationship, causal relationship, and superior-subordinate relationship between events, and is used to discover the evolution law of events and predict subsequent events. The knowledge graph of facts and principles used in the method of the present invention is based on production units, production equipment, production processes, and production materials. The graph is constructed from top to bottom by combing the flow relationship between production materials in production units, production equipment, production processes, and production processes. A mature commercial knowledge graph database Neo4j is used as a knowledge graph development and storage tool. Neo4j is a high-performance NoSQL graph database that stores structured data on the network instead of in tables. It is an embedded, disk-based, Java persistence engine with complete transaction characteristics. Compared with traditional relational databases, Neo4j realizes the storage of data relationships and improves query efficiency; it breaks the data representation restrictions in the form of Key-Value and improves the expression ability of stored data. Neo4j also stores knowledge in the form of nodes, relationships, and attribute values. The constructed knowledge graph can provide more stable knowledge support for other topics and provide a solid data foundation for the subsequent development of knowledge reasoning tools.

[0038] The main steps of constructing the metallurgical knowledge graph for the continuous casting production process in the embodiment of the present invention are as follows: First, collect and organize the production unit, production equipment, production process, production time, production process and other information of the production materials to establish a material production process database; production unit such as 1# steel mill, 2# steel mill, 3# steel mill and other production workshop information; production equipment, such as 1# BOF converter, 2# BOF converter, 1# LF furnace, 2# LF furnace, 3# RH furnace, 3# continuous casting machine information; production process, such as BOF (converter process), LF (LF refining process), RH (RH refining process), CC (continuous casting process) and other related production information. Secondly, based on the production material information flow, the nodes of material attributes are associated with the production equipment and the production process by node tuples, and this information is organized into triple information, that is, it is composed as follows: (h, r, t), where r represents the relationship, h represents the head entity, and t represents the tail entity. Then, construct the production process of digital ingots in equipment and processes, associate the production equipment with the production process and production flow, associate the generated processes in sequence according to the production flow, and associate the production equipment in sequence according to the production flow. Each production process node represents a process of the production process. For example, a material goes through the ironmaking process of 1# blast furnace, enters the 2# converter for steelmaking process after leaving the 1# blast furnace, and is transported to the 2# LF refining furnace for steelmaking process after leaving the 2# converter, etc. Finally, associate the generated process with the production unit, complete the association of all tuples, and finally construct a Figure 4 The metallurgical principle diagram based on material flow is shown in Figure 1. Figure 4As shown, in the embodiment of the present invention, each production workshop, each production equipment, each process, each production flow, and production material are taken as entities of the knowledge graph, such as ironmaking, steelmaking, continuous casting, rolling, and quality inspection, which are process entities, and ordinary steel production line 1 is an abstract entity of the production flow. The relationship between entities is as follows: Figure 4 It includes equipment, next process, equipment owner, asset owner, etc. Figure 4 These are only some examples and are not an exhaustive list of all relationships and entities.

[0039] The advantage of this step is that it uses the top-down knowledge graph construction technology based on the idea of ​​evolution of things, which has the advantage of being able to accurately obtain the evolution path of events derived from the material production process, helping metallurgical workers to quickly and intuitively understand the development path of things. It is of great significance for the subsequent construction of digital ingot knowledge graphs, the development of reasoning tools based on knowledge graphs, the discovery of evolution laws of events and the prediction of quality events.

[0040] Step 6: Construct an extended graph of digital ingots for the continuous casting production process. For the business database stored in the "object model", use data opening tools to build a data structure from top to bottom, take production materials as the core node, and expand the graph of production process data, production performance data, inspection and testing data, product basic information, and production metallurgical specification data involved in the material flow in the production process.

[0041] The digital ingot body contains holographic data of the physical ingot production process. The data assets have multiple data branch nodes. The data is divided into structured data and unstructured data according to the structure, and divided into production process data, production performance data, inspection and testing data, product basic information, and production metallurgical specification data according to the data content. The data nodes are connected in series according to the production process. Combined with expert opinions, the digital ingot body graph structure is constructed from top to bottom and bottom to top based on the material production data flow. It is necessary to construct the digital ingot by combining production units, production equipment, production processes, production process data, production performance data, inspection and testing data, product basic information, and production metallurgical specification data. Each node represents the process, data, and classification relationship of the production process. The main structure of the graph includes nodes, directed edges, and values. Among them, nodes represent different events themselves; directed edges represent the logical relationship between events, including causal relationships, sequential relationships, and hierarchical relationships; values ​​represent the probability of measuring the possibility of uncertain evolutionary relationships of events.

[0042] The steps for constructing the digital ingot extension map for the continuous casting production process are as follows: First, collect and organize the production process data, production performance data, inspection and testing data, basic product information, and production metallurgical specification data, and use these data in each production process as a data node. Secondly, adopt a top-down data structure, take the material as the core node, associate the production process data, production performance data, inspection and testing data, basic product information, production metallurgical specification data and other data nodes with the core node, and associate the production process data with the equipment variable node to expand the map. Then, the map is expanded in real time. As the material production process proceeds and real-time data is continuously generated, the data information is tupled and the map is completed. Production process data refers to the production process time series data collected from sensors. Production performance data refers to information about input and output quantity, energy consumption, etc. collected during the production process, which is used to reflect the specific operation status and material status of the production process. Inspection and testing data refers to the data obtained by quality inspection and testing of production materials during the production process, which is used to monitor product quality and ensure the stability of the production process and the qualified rate of products. Product basic information refers to the attributes of the product itself, such as material, specification parameters, implementation standards, product quality grade, etc. Production metallurgical specification data refers to the set of operation instructions for each process defined in the production process according to product standards and process requirements. The instructions include quality requirements and process requirements and are used to guide production operations.

[0043] like Figure 5 As shown in the figure, it is an example of the constructed digital ingot expansion map, with production materials as the core nodes. The time series data in the figure generally refers to the OT data in the business database, and the process data refers to the relational data recorded in the production process, that is, the IT data in the business database. The time series data nodes must be associated with the equipment variable nodes involved. The equipment refers to the production equipment involved in the process. The several blank circles associated with the equipment variable nodes represent different sensor nodes, such as sensors for measuring temperature, pressure, and vibration frequency. The sensors collect data, images, text and other information. Figure 5 The production material node in Figure 4 Each process is associated with a digital casting billet expansion map, such as Figure 4 There is a corresponding ironmaking process in Figure 5 Spectral data shown.

[0044] Advantages of constructing a digital billet knowledge data expansion map: The holographic expansion map data management mode, with materials as the core node, expands the production process data, production performance data, inspection and testing data, product basic information, and production metallurgical specification data, which can better cover the relationship between data in the material production process, and is conducive to mining how the working mode affects product quality, so as to take measures to improve the production process; the data management mode with materials as the core, manages data based on materials, and all data related to materials are centrally managed, which is different from the traditional data management mode based on equipment and files. For quality prediction and prevention issues, this management mode can quickly find the problem of continuous casting billet quality defects, and assist process personnel to take preventive measures to reduce the occurrence of product defects and quality problems; for research reference, the mined fixed working mode working subgraph can provide useful reference information for subsequent metallurgical product quality research and process path optimization, helping researchers to better understand the production process.

[0045] Based on the holographic expansion graph data management mode constructed by the method of the present invention, the method of the present invention provides data retrieval and management functions for multi-source heterogeneous data in the continuous casting production process. The staff can perform data retrieval according to the material ID or other conditions, such as Figure 4 In the process shown, after obtaining the production material code ID associated with the continuous casting unit, such as the batch number of the continuous casting billet material number, the material can be used as the root node to obtain Figure 5 The holographic production process data, production performance data, inspection data, basic product information, and production metallurgical specification data about this material within the current process scope. At the same time, data queries can be performed on forward material traceability (molten steel) or backward material traceability (hot-rolled steel coils). In addition, data query personnel can obtain local data of the holographic extended graph data according to their own data needs, combination and retrieval depth. The present invention classifies and manages discrete multi-source heterogeneous data through the above steps, and stores them in the knowledge graph network. The constructed extended graph structure is the data management method. When it is necessary to add, delete, or modify the data, it is achieved by adding, deleting, and modifying the knowledge graph nodes.

[0046] Except for the technical features described in the specification, all of them are known technologies to those skilled in the art. The present invention omits the description of known components and known technologies to avoid redundancy and unnecessary limitation of the present invention. The implementation methods described in the above embodiments do not represent all implementation methods consistent with the present application. On the basis of the technical solution of the present invention, various modifications or variations that can be made by those skilled in the art without creative labor are still within the protection scope of the present invention.

Claims

1. A data management method based on the combination of data-driven and continuous casting production process, characterized in that: The steps include: Step 1: The data engine extracts multi-source heterogeneous data of the continuous casting production process according to IT and OT categories, including: collecting the data of the continuous casting production process according to IT and OT categories, and collecting OT data in different areas according to the areas to which the physical model belongs; mapping the physical model of the virtual sensor in the digital space based on the physical sensor, and using the physical model management method to directly collect time series data from the sensors of the continuous casting production process monitoring system; IT stands for information technology, and OT stands for operational technology; Step 2: The IoT device connects the collected time series data to the industrial gateway. The industrial grid parses, converts the format and downsamples the time series data, and then reports it to the data engine. The device access service of the data engine uses the message queue publish / subscribe method to store multi-source heterogeneous data in the business database of the corresponding object model. Step 3: Conduct real-time online data management of the continuous casting production process, including: the data engine obtains the time series data of the continuous casting production process from the subscribed message queue, monitors the abnormal status of equipment signals, removes abnormal data, completes data, identifies data anomaly types, and processes process rule standardized data; Step 4: The data engine aligns the OT data and IT data stored in the business database with the production materials in time and space, including: the data engine obtains the production material tracking information for each order furnace, obtains the effective working time of the production material passing through the sensor, segments the time series data collected by the sensor, and uses the furnace number and billet number to manage the OT data and IT data of the same order furnace; production materials are set with different codes in different production processes, and the production material code is uniquely marked with the furnace number + billet number + production process code; Step 5: The data engine constructs a knowledge graph of metallurgical facts and principles for the continuous casting production process, including: taking production units, production equipment, production procedures, production processes and production materials as entities of the knowledge graph, using metallurgical production process knowledge to construct the flow relationship between production units, production equipment, production procedures and production processes for production materials, constructing a knowledge graph of facts and principles from top to bottom, and displaying the evolution path of the production material production process derivative events; Step 6: The data engine constructs a digital ingot expansion map for the continuous casting production process, including: a business database based on the object model, with production materials as the core nodes, and a top-down data structure based on material flow. For each production process, corresponding production process data, production performance data, inspection and testing data, basic product information, and production metallurgical specification data are obtained, and data nodes are constructed, and the relationship between each data node and the core node is established to obtain a digital ingot expansion map.

2. The method according to claim 1, characterized in that In the step 1, the data source of the data engine also includes the historical production system database, and the historical production data form constructs the object model and maps it into the digital space. The historical production data form is stored according to IT and OT categories, and the OT data is partitioned and stored according to the area to which the object model belongs.

3. The method according to claim 1, characterized in that: In the step 2, the industrial gateway converts the parsed data into JSON format for reporting, and selects the downsampling frequency according to the different distribution of parameter data points to ensure the alignment of parameter timestamps.

4. The method according to claim 1, characterized in that: In step 3, the data engine monitors the abnormal state of the device signal in the data transmission channel. If the device is abnormal, the data management configuration item of the device will be assigned in the business database and an alarm will be issued; secondly, the time series data is upsampled and filled; Then, the data anomalies are classified in real time, including time delay, data missing, data failure, data noise, data exceeding the standard and data granularity abnormality; finally, based on the identified data abnormality status, standardized data processing is carried out in accordance with the continuous casting production process rules.

5. The method according to claim 1, characterized in that In the step 4, the data engine first obtains the production material tracking information of the entire order furnace, including the entry time and end time of the production materials of the furnace into the converter, refining furnace, and crystallizer, the start time and stop time of the ladle pouring, and the start time of the crystallizer dummy rod movement; secondly, the time series data and event information collected based on the physical model in the production process are obtained, including the data collected by the production process monitoring system sensors between the molten steel from the ladle, the tundish, the crystallizer, the fan-shaped area, the secondary cooling area, the straightening area to the flame cutting machine; Then, when the flame cutting completion signal of the continuous casting billet is obtained, it indicates that the casting billet production process is finished. At this time, the casting billet number is obtained, and the current production furnace number is combined with the casting billet number as the code for this casting billet process; finally, the volume is transformed through the working parameters of the casting billet production equipment and the production material size information, and the effective working time of the production material passing through the sensor in the continuous casting production process is obtained. The time series data collected by the sensor is segmented according to the effective working time, and the time series data of the same order furnace are associated using the same casting billet process code.

6. The method according to claim 1, characterized in that In the step 5, constructing a metallurgical knowledge graph of the continuous casting production process includes: first, obtaining the production unit, production equipment, production process, production time and production process of the production materials, and establishing a material production process database; second, based on the metallurgical production process knowledge, the production materials are associated with the production equipment and the production process by node tuple association, the production equipment is associated with the production process and the production process, the generation process is sequentially associated according to the generation process, the production equipment is associated according to the production process sequence, the generation process is associated with the production unit, and a metallurgical knowledge graph based on material flow is constructed.

7. The method according to claim 1, characterized in that In the step 6, constructing a digital ingot expansion map includes: for each production process, obtaining the production process data, production performance data, inspection and testing data, product basic information and production metallurgical specification data in the corresponding process from the business database, associating the data node with the core node, associating the production process data node with the equipment variable node involved, and associating the equipment variable node with the sensor of the monitoring system; as the material production process proceeds, a digital ingot expansion map of each process is generated according to the acquired real-time data.

8. The method according to claim 1, 6 or 7, characterized in that: The method described sets up data retrieval function and data management function based on the constructed metallurgical knowledge graph and digital ingot extension graph; when data retrieval is performed according to the production material code, the production process data, production performance data, inspection and testing data, product basic information and production metallurgical specification data in the digital ingot extension graph with the production material code as the root node are obtained; when the data needs to be added, deleted or modified, it is achieved by adding, deleting or modifying the graph nodes.

Images