Knowledge graph-based ethylene cracking furnace tube coking reasoning method

Abstract

The invention provides an ethylene cracking furnace tube coking reasoning method based on a knowledge graph, and the method comprises the steps: building an ethylene cracking knowledge ontology model for the knowledge of ethylene cracking related fields, and obtaining ethylene industrial field data; the method comprises the following steps: constructing a historical ethylene cracking furnace tube coking knowledge graph and a current ethylene cracking furnace tube coking knowledge graph, reasoning coking attributes according to the historical ethylene cracking furnace tube coking knowledge graph, updating the current ethylene cracking furnace tube coking knowledge graph according to the reasoned coking attributes, and when the current operation period is ended, determining the current ethylene cracking furnace tube coking knowledge graph according to the updated current ethylene cracking furnace tube coking knowledge graph. And storing the current ethylene cracking furnace tube coking knowledge graph as a historical ethylene cracking furnace tube coking knowledge graph, and entering a new operation period. Compared with the prior art, the method has the advantages that by continuously adding the historical ethylene cracking furnace tube coking knowledge graph and enriching the reasoning scene, the reasoning capability can be effectively improved, and the coking attribute of the ethylene cracking furnace can be more accurately reasoned.

Description

Technical Field

[0001] The present invention relates to the fields of fault diagnosis and computer technology, and more specifically, to a knowledge graph-based reasoning method for coking of furnace tubes in an ethylene cracking furnace. Background Art

[0002] Ethylene is the most basic chemical raw material in the field of petrochemicals and the basis for the production of various organic chemical products. The production scale and production technology of ethylene also measure the development level of a country's petrochemical technology. In the process of using a cracking furnace to produce ethylene, a side reaction will occur to form coke particles that adhere to the inner wall of the furnace tube in the radiation section of the cracking furnace. These coke particles continue to accumulate during the continuous production process, causing the local temperature of the furnace tube in the radiation section of the cracking furnace to be too high, increasing energy consumption in the production process, reducing ethylene production and other adverse effects. In severe cases, the furnace tube may even be blocked, leading to production accidents. Therefore, how to judge the coking condition of the furnace tube is crucial to the safe production of ethylene cracking equipment.

[0003] The existing technology usually relies on on-site workers to measure and judge the degree of coking based on the measured indicators and their own experience. However, different people have different understandings of the severity of the coking degree and which production process factors cause coking, which leads to inaccurate judgment results of coking. Summary of the invention

[0004] The present invention aims to overcome at least one defect of the above-mentioned prior art, and provides a coking reasoning method for ethylene cracking furnace tubes based on knowledge graph, and provides a method that can accurately and effectively perform coking reasoning on the coking properties of ethylene cracking furnace tubes.

[0005] The technical solution adopted by the present invention is:

[0006] The present invention provides a knowledge graph-based reasoning method for coking of ethylene cracking furnace tubes, the reasoning method comprising:

[0007] Construct the knowledge ontology model of ethylene cracking based on the knowledge related to ethylene cracking;

[0008] Acquire ethylene industry field data of several operation cycles, the several operation cycles include current operation cycle and historical operation cycle; the operation cycle represents the time of continuous feeding operation between two coke cleaning treatments of ethylene cracking furnace, the current operation cycle represents the operation time when the previous coke cleaning is completed at the current moment and the next coke cleaning time has not yet arrived, and the historical operation cycle represents the complete operation time between two coke cleanings;

[0009] A knowledge graph of coking of furnace tubes of a historical ethylene cracking furnace is constructed according to the ethylene cracking knowledge ontology model and the acquired ethylene industry field data of the historical operation cycle; the knowledge graph of coking of furnace tubes of a historical ethylene cracking furnace contains at least one knowledge relationship of coking of furnace tubes of the historical operation cycle;

[0010] Constructing a knowledge graph of coking of furnace tubes of a current ethylene cracking furnace according to the ethylene cracking knowledge ontology model, and updating the knowledge graph of coking of furnace tubes of a current ethylene cracking furnace according to the acquired ethylene industry field data of the current operation cycle;

[0011] According to the ethylene cracking knowledge ontology model, combined with the historical ethylene cracking furnace tube coking knowledge graph, the coking attributes in the current ethylene cracking furnace tube coking knowledge graph are inferred, and the current ethylene cracking furnace tube coking knowledge graph is updated according to the inference result;

[0012] If the current operation cycle corresponding to the current ethylene cracking furnace tube coking knowledge graph ends, the current ethylene cracking furnace tube coking knowledge graph is used as the historical ethylene cracking furnace tube coking knowledge graph, and the ethylene industry field data of a new round of operation cycle is obtained, and the current ethylene cracking furnace tube coking knowledge graph corresponding to the operation cycle is constructed.

[0013] Based on the knowledge graph technology, by extracting knowledge from the relevant fields of ethylene cracking, an ethylene cracking knowledge ontology model is constructed, and ethylene industry field data is obtained to construct the historical ethylene cracking furnace tube coking knowledge graph and the current ethylene cracking furnace tube coking knowledge graph. The historical ethylene cracking furnace tube coking knowledge graph is constructed based on the historical ethylene industry field data, and has complete data reasoning capabilities, and can accurately infer the coking properties of the current ethylene cracking furnace based on the ethylene industry field data in the current operation cycle;

[0014] At the same time, the coking properties obtained by reasoning are updated, and the current ethylene cracking furnace tube coking knowledge graph constructed according to the ethylene industry field data of the current operation cycle is continuously updated so that the current ethylene cracking furnace tube coking knowledge graph has the ability of data reasoning. When the current operation cycle ends, the current ethylene cracking furnace tube coking knowledge graph is saved as the historical ethylene cracking furnace tube coking knowledge graph, and enters a new operation cycle. At the same time, a new current ethylene cracking furnace tube coking knowledge graph is constructed. By continuously adding historical ethylene cracking furnace tube coking knowledge graphs, the reasoning scenarios are enriched, the reasoning ability can be effectively improved, and the coking properties of the ethylene cracking furnace can be more accurately inferred.

[0015] Furthermore, the knowledge graph of coking of furnace tubes of historical ethylene cracking furnaces is constructed based on the ethylene cracking knowledge ontology model and the acquired ethylene industry field data of the historical operation cycle, specifically including:

[0016] Acquiring the ethylene industry field data at each time point in each of the historical operation cycles;

[0017] The knowledge description of the relationship between entities in the ethylene cracking coking knowledge ontology model and the ethylene industry field data are combined to determine the time-varying relationship between entities and between entities and attributes. The entities and entity attributes are used as vertices and the relationships are used as edges to form several historical ethylene cracking furnace tube coking knowledge graphs with time-varying characteristics within a period.

[0018] Furthermore, the knowledge graph of coking of furnace tubes of ethylene cracking furnace is constructed according to the ethylene cracking knowledge ontology model, and the knowledge graph of coking of furnace tubes of ethylene cracking furnace is updated according to the acquired ethylene industrial field data of the current operation cycle, specifically including:

[0019] Determine whether the current ethylene cracking ethylene cracking furnace tube coking knowledge graph exists;

[0020] If it does not exist, determine the relationship between entities and entities, and between entities and attributes according to the basic relationship between entities in the ethylene cracking coking knowledge ontology model, take entities and entity attributes as vertices, and take relationships as edges to construct the knowledge graph of ethylene cracking furnace tube coking in the current period;

[0021] If it exists, determine whether the ethylene industrial field data at the current time point in the current operation cycle has changed: if so, update the current ethylene cracking furnace tube coking knowledge graph according to the current time point and the ethylene industrial field data at the current time point; if not, only update the current ethylene cracking furnace tube coking knowledge graph according to the current time point.

[0022] During the operation of the ethylene cracking furnace, industrial field data is constantly changing, and the coking properties of the ethylene cracking furnace will also change accordingly. It is necessary to periodically record various parameters related to the operation of the cracking furnace, and update the data of the current knowledge graph to ensure that the current ethylene cracking furnace tube coking knowledge graph can use the latest operating parameters for coking reasoning, thereby ensuring the timeliness of the graph data and subsequent coking reasoning.

[0023] Furthermore, the coking attributes in the current ethylene cracking furnace tube coking knowledge graph are inferred based on the ethylene cracking knowledge ontology model and combined with the historical ethylene cracking furnace tube coking knowledge graph, specifically including:

[0024] Constructing a coking diagnosis factor set according to the updated knowledge graph of coking in the current ethylene cracking furnace tube;

[0025] According to each coking diagnosis factor in the coking diagnosis factor set, a parameter item corresponding to each coking diagnosis factor is obtained from the current ethylene cracking furnace tube coking knowledge graph, and a parameter query condition set is constructed according to the parameter item; the parameter query condition set contains at least one parameter item;

[0026] According to the parameter query condition set, query the historical coking attribute through the historical ethylene cracking furnace tube coking knowledge graph, if the query result is empty, adjust the parameter query condition set, and query the historical coking attribute from the historical ethylene cracking furnace tube coking knowledge graph again;

[0027] According to the current ethylene cracking furnace tube coking knowledge graph, the coking attribute before the data is updated is obtained as a comparison coking attribute;

[0028] Comparing the historical coking attributes with the compared coking attributes, reasoning the comparison result according to the knowledge of the ethylene cracking knowledge ontology model, and inferring and predicting the coking attributes;

[0029] The coking attribute of the current ethylene cracking furnace tube coking knowledge graph is updated to the predicted coking attribute.

[0030] After the current ethylene cracking furnace tube coking knowledge graph is continuously updated by the ethylene industrial field data, the current ethylene cracking furnace tube coking knowledge graph can "remember" the parameters of the ethylene cracking furnace at each moment of the industrial field, and thus can conveniently obtain the coking diagnosis factors related to the coking of the ethylene cracking furnace according to the parameters of the industrial field at each moment, and thus can effectively infer the coking conditions of the ethylene cracking furnace tubes according to the coking diagnosis factors.

[0031] At the same time, by obtaining coking-related factors from the coking diagnosis factor set and using them as query conditions, the coking attributes corresponding to the current ethylene industrial field data are obtained in the historical ethylene cracking furnace tube coking knowledge graph. While realizing the inference of the coking attributes of the ethylene cracking furnace, the reasoning and prediction accuracy of the historical ethylene cracking furnace tube coking knowledge graph can be effectively improved through continuous comparison and combination with the ethylene industrial field data.

[0032] Furthermore, the coking diagnosis factor set is extracted according to the ethylene cracking furnace tube coking knowledge graph after data update, specifically including:

[0033] Taking coking as the final tail entity of the triple path, searching for a non-cyclic triple path in the current ethylene cracking furnace tube coking knowledge graph; the triple path is formed by correspondingly connecting the tail entities and the head entities of several triples in sequence, wherein the tail entity of the previous triple is the same as the head entity of the next triple;

[0034] The coking diagnosis factor set is constructed according to the acquired head entities of all the triplet paths.

[0035] By taking coking as the final tail entity and querying the non-cyclic triple path, all entities associated with coking during the operation of the ethylene cracking furnace can be found. These entities are used as coking diagnostic factors, and actual ethylene industry field data can be effectively obtained based on the coking diagnostic factors, and coking can be inferred based on the ethylene industry field data.

[0036] Further, the query condition set is adjusted to query the historical coking attributes from the historical ethylene cracking furnace tube coking knowledge graph again, specifically including:

[0037] Adjusting at least one of the parameter items in the parameter query condition set to range data by a preset threshold;

[0038] According to the adjusted range data, the historical coking attribute is again queried from the historical ethylene cracking furnace tube coking knowledge graph;

[0039] If the query result is an empty set, the predicted coking attribute is set to be the same as the coking attribute most recently updated during this operation cycle.

[0040] Further, the query condition set is adjusted to query the historical coking attributes from the historical ethylene cracking furnace tube coking knowledge graph again, specifically including:

[0041] Obtaining several subsets of the parameter query condition set;

[0042] The obtained subset of the parameter query condition set is queried again through the historical ethylene cracking furnace tube coking knowledge graph;

[0043] If the query result is an empty set, the predicted coking attribute is set to be the same as the coking attribute most recently updated during this operation cycle.

[0044] When the coking attribute search result is empty, the coking attribute is queried again by appropriately adjusting the query scope or relaxing the query conditions, trying to ensure that the corresponding coking attribute can be queried from the historical ethylene cracking furnace tube coking knowledge graph, avoiding manual detection of coking attributes at the industrial site, and ensuring query efficiency.

[0045] Furthermore, the historical coking attributes are compared with the compared coking attributes, and the comparison result is inferred based on the knowledge of the ethylene cracking knowledge ontology model to predict the coking attributes, which specifically includes:

[0046] Determining whether the numerical changes of the historical coking attributes and the compared coking attributes are consistent with the coking knowledge description in the ethylene cracking knowledge ontology model;

[0047] If the descriptions are consistent, the historical coking attribute is used as the predicted coking attribute;

[0048] If they are inconsistent, the coking properties are obtained from the industrial site as the predicted coking properties.

[0049] When the corresponding coking attribute cannot be found in the historical ethylene cracking furnace tube coking knowledge graph, the industrial coking attribute of the current ethylene industrial field data is taken as the inference result, and then the current ethylene cracking furnace tube coking knowledge graph is updated according to the inference result of the coking attribute of the industrial field, so that the current ethylene cracking furnace tube coking knowledge graph can "remember" the corresponding coking attribute of the ethylene industrial field data. When the current ethylene cracking furnace tube coking knowledge graph is saved as the historical ethylene cracking furnace tube coking knowledge graph, it can enrich the query reasoning scope of the historical ethylene cracking furnace tube coking knowledge graph and more effectively realize the query of coking attributes.

[0050] Furthermore, the core ontology in the ethylene cracking knowledge ontology model at least includes cracking raw materials, cracking furnace, furnace tubes, process parameters, products and yields, and coking.

[0051] Compared with the prior art, the present invention has the following beneficial effects:

[0052] The present invention continuously obtains ethylene industrial field data of each operation cycle of an ethylene cracking furnace by utilizing knowledge graph technology, wherein a historical ethylene cracking furnace tube coking knowledge graph is constructed according to the ethylene industrial field data of the historical operation cycle, and a current ethylene cracking furnace tube coking knowledge graph is constructed according to the ethylene industrial field data of the current operation cycle. The coking properties of the ethylene cracking furnace are inferred according to the current ethylene industrial field data by utilizing the historical ethylene cracking furnace tube coking knowledge graph, and the current ethylene cracking furnace tube coking knowledge graph is updated according to the inference result. After completing the update of the ethylene industrial field data and the coking properties, the current ethylene cracking furnace tube coking knowledge graph is transformed into the historical ethylene cracking furnace tube coking knowledge graph. Through continuous transformation, the data content in the historical ethylene cracking furnace tube coking knowledge graph can be continuously enriched, and the reasoning ability of the coking properties can be improved, thereby effectively realizing accurate coking reasoning of the ethylene cracking furnace. BRIEF DESCRIPTION OF THE DRAWINGS

[0053] Figure 1 It is a flowchart of the steps of the reasoning method of the present invention.

[0054] Figure 2 It is a structural schematic diagram of the ethylene cracking knowledge ontology model of the present invention.

[0055] Figure 3 It is a structural schematic diagram of the current ethylene cracking furnace tube coking knowledge map of the present invention.

[0056] Figure 4 It is a flow chart of the reasoning steps of the knowledge graph of coking of historical ethylene cracking furnace tubes of the present invention. DETAILED DESCRIPTION

[0057] The drawings of the present invention are only for illustrative purposes and should not be construed as limiting the present invention. In order to better illustrate the following embodiments, some parts of the drawings may be omitted, enlarged or reduced, and do not represent the size of the actual product; it is understandable to those skilled in the art that some well-known structures and their descriptions in the drawings may be omitted.

[0058] Ethylene is the most basic chemical raw material in the field of petrochemicals and the basis for the production of various organic chemical products. The production scale and production technology of ethylene also measure the development level of a country's petrochemical technology. In the process of using a cracking furnace to produce ethylene, a side reaction will occur to form coke particles that adhere to the inner wall of the furnace tube in the radiation section of the cracking furnace. These coke particles continue to accumulate during the continuous production process, causing the local temperature of the furnace tube in the radiation section of the cracking furnace to be too high, increasing energy consumption in the production process, reducing ethylene production and other adverse effects. In severe cases, the furnace tube may even be blocked, leading to production accidents. Therefore, how to judge the coking condition of the furnace tube is crucial to the safe production of ethylene cracking equipment.

[0059] There are three main methods in the prior art to judge and infer the coking of ethylene cracking furnaces: the first is to rely on the measurement of workers on site, and judge the degree of coking based on the measured indicators and their own experience, but different people have different understandings of the severity of the coking degree and which production process factors cause coking. The second is to examine the factors affecting coking based on the coking mechanism, and derive the relationship between these factors and coking, establish a coking model to calculate the coking rate, and then calculate the thickness of the coke layer, but the disadvantage is that these coking models only examine some coking factors, and the established models are only applicable to cracking certain raw materials and equipment; the third method is to use the currently popular machine learning or deep learning technology, with the real-time production data such as cracking temperature, absolute pressure ratio, and tube wall temperature in the production process as support, and learn these data through machine learning models, and finally obtain the prediction of the coking condition of the radiation section of the cracking furnace tube. Although this method can more accurately predict the coking condition of the cracking furnace tube and is also applicable to different cracking equipment, it lacks interpretability and can only answer the question of "whether there is" but not the question of "why there is".

[0060] In order to solve the problems existing in the above-mentioned prior art, this example provides an ethylene cracking furnace tube coking reasoning method based on knowledge graph.

[0061] Specifically, Figure 1 As shown, the reasoning method may include:

[0062] S1: Construct the knowledge ontology model of ethylene cracking based on the knowledge related to ethylene cracking;

[0063] Specifically, in this embodiment, the knowledge in the field of ethylene cracking can be sorted out through the seven-step method of the existing knowledge graph to construct the ethylene cracking knowledge ontology model, which will not be further elaborated here.

[0064] Among them, the constructed ethylene cracking knowledge ontology model is a directed graph structure, with the ontology as the vertex of the directed graph, and the ontologies are connected by directed edges, and the directed edges represent the relationship between the ontologies; through the ethylene cracking knowledge ontology model, the relationship between the ethylene cracking related ontologies can be intuitively observed, and then the coking properties of the ethylene cracking furnace and the corresponding factors can be intuitively "explained".

[0065] Specifically, in order to more accurately utilize the ethylene cracking knowledge ontology model to realize the reasoning of the coking properties of the ethylene cracking furnace, in this embodiment, the ethylene cracking knowledge ontology model includes at least the following core ontology: cracking raw materials, cracking furnace, furnace tube, process parameters, products and yields, coking; the ethylene cracking knowledge ontology model constructed in this embodiment can be as follows Figure 2 As shown;

[0066] The meanings of the core entities are as follows:

[0067] Cracking raw materials: Cracking raw materials refer to the raw materials for the preparation of chemical products such as ethylene. These raw materials will be put into the cracking furnace for production after certain pre-treatment.

[0068] Cracking furnace: Cracking furnace is a key equipment for producing chemical products such as ethylene.

[0069] Furnace tube: The furnace tube is an important component of the cracking furnace. It is the reaction site of the cracking reaction and also the place where coking is produced.

[0070] Process parameters: Process parameters include variable parameters and measurable parameters that control the cracking reaction process.

[0071] Products and Yields: The main products of the cleavage reaction and the yields of these products.

[0072] Coking: A phenomenon in which coke is deposited on the inner wall of the furnace tube during the cracking reaction, which affects the normal operation of the cracking furnace tube.

[0073] It should be noted that the configuration of the core ontology can be added or modified according to specific production needs, and its main purpose is to better include coking-related factors and thus accurately realize the reasoning of coking properties.

[0074] In this embodiment, the various ontologies in the ethylene cracking knowledge ontology model can be described by a triple in the form of <head entity, relationship, tail entity>, such as <cracking raw material, raw material, cracking furnace>, <cracking furnace, product, product and yield>, <cracking furnace, reactor, furnace tube>, <furnace tube, parameters, process parameters>, <furnace tube, coking reaction, coking>, <coking, attribute, coking rate>, <coking, attribute, coking layer thickness>, etc.

[0075] S2: Acquire ethylene industry field data of a plurality of operation cycles, wherein the plurality of operation cycles include a current operation cycle and a historical operation cycle;

[0076] In this embodiment, the operation cycle indicates the time of continuous feeding operation between two coke cleaning processes of the ethylene cracking furnace, the current operation cycle indicates the operation time when the previous coke cleaning has been completed and the next coke cleaning time has not yet arrived, and the historical operation cycle indicates the complete operation time between two coke cleaning processes;

[0077] In the step S1, the ethylene cracking knowledge ontology model is constructed. In order to construct a specific knowledge graph, it is necessary to further obtain specific entities, entity attributes and entity relationships, so it is necessary to collect ethylene industry field data to realize the construction of the knowledge graph.

[0078] In this embodiment, ethylene industry field data of the current operation cycle and the historical operation cycle are collected. It can be understood that the current operation cycle can be converted into the historical operation cycle, and when the current operation cycle ends, the current operation cycle is converted into the historical operation cycle.

[0079] Therefore, in a specific implementation of the present embodiment, the ethylene industry field data of several operation cycles can be obtained by collecting the ethylene industry field data of the current operation cycle. When the current operation cycle completes the coke clearing operation and ends the current operation cycle, the ethylene industry field data of the current operation cycle is used as the ethylene industry field data of the historical operation cycle, and the ethylene industry field data of a new round of operation cycle is continuously collected, and the above-mentioned collection process is continuously repeated to obtain the ethylene industry field data of several operation cycles.

[0080] By continuously collecting the ethylene industry field data, the database corresponding to the constructed knowledge graph can be continuously enriched, providing a sufficient data basis for the reasoning of coking properties.

[0081] S3: constructing a knowledge graph of coking of furnace tubes of historical ethylene cracking furnaces according to the ethylene cracking knowledge ontology model and the acquired ethylene industry field data of the historical operation cycle;

[0082] In this embodiment, the historical ethylene cracking furnace tube coking knowledge graph contains at least one tube coking knowledge relationship of the historical operation cycle; wherein, the construction of the historical ethylene cracking furnace tube coking knowledge graph can be based on the ethylene cracking knowledge ontology model, and the ethylene industrial field data of the historical operation cycle is extracted to obtain the time-varying relationship between entities and between entities and attributes, and according to the extracted entities, entity attributes and time-varying relationships, the entities and entity attributes are used as vertices, and the relationships are used as edges to form a historical ethylene cracking furnace tube coking knowledge graph with time-varying characteristics within the cycle; corresponding to the ethylene cracking knowledge ontology model, the historical ethylene cracking furnace tube coking knowledge graph can be described by a triple of <head entity, relationship, tail entity>.

[0083] Exemplarily, the ethylene industry field data of the historical operation cycle may include the following:

[0084] The cracking furnace equipment uses a USC-U type cracking furnace, and its radiation section has a total of 48 groups of furnace tubes, each furnace tube is 25m long and 50mm in diameter. The cracking raw material is light diesel produced in a certain place. After inspection, it is found that the relative density of the raw material at 15.6℃ is 0.826g / cm 3 , the aromatics correlation index BMCI is 26.7. On the 10th day of the corresponding historical operation cycle, the monitoring equipment measured the 48 furnace tubes of the cracking furnace in turn, among which the inlet and outlet temperatures of furnace tube 1 were measured to be 908°C and 971°C respectively, the outlet pressure was 189kPa, the coke layer thickness was 0.07mm, and the coking rate was 0.00026kg / h·m 2 , coking grade is 0; the inlet and outlet temperatures of furnace tube 24 are 1090℃ and 1077℃ respectively, the outlet pressure is 165kPa, the coke layer thickness is 2.0mm, and the coking rate is 0.00765kg / h·m 2 , coking grade is 2; the inlet and outlet temperatures of furnace tube 37 are 1003℃ and 933℃ respectively, the outlet pressure is 180kPa, the coke layer thickness is 0.8mm, and the coking rate is 0.00302kg / h·m 2 , the coking grade is 1, and the yields of trienes (ethylene, propylene, butadiene) are measured to be 24.76%, 15.31% and 5.40% respectively.

[0085] In this embodiment, the historical ethylene cracking furnace tube coking knowledge graph constructed can be as follows: Figure 3 shown.

[0086] S4: constructing a knowledge graph of coking of furnace tubes of a current ethylene cracking furnace according to the ethylene cracking knowledge ontology model, and updating the knowledge graph of coking of furnace tubes of a current ethylene cracking furnace according to the acquired ethylene industrial field data of the current operation cycle;

[0087] When the ethylene industry field data is first acquired in the current operation cycle, the current ethylene cracking furnace tube coking knowledge graph of the current operation cycle has not yet been established, and the objective conditions for updating the current ethylene cracking furnace tube coking knowledge graph of the current operation cycle are not met, and the update step cannot be performed. Therefore, it is necessary to first construct the current ethylene cracking furnace tube coking knowledge graph of the current operation cycle. Therefore, in this embodiment, the construction and update of the current ethylene cracking furnace tube coking knowledge graph may include:

[0088] Determine whether the current ethylene cracking furnace tube coking knowledge graph exists;

[0089] If it does not exist, it means that the current ethylene cracking furnace tube coking knowledge graph has not been established, then the relationship between entities and entities, entities and attributes is determined according to the basic relationship between entities in the ethylene cracking coking knowledge ontology model, and the entities and entity attributes are used as vertices, and the relationship is used as an edge to construct the ethylene cracking furnace tube coking knowledge graph of the current cycle; in this embodiment, as described above, since the current operation cycle can be transformed into the historical operation cycle, the construction of the current ethylene cracking furnace tube coking knowledge graph can refer to the construction of the historical ethylene cracking furnace tube coking knowledge graph in step S3, which will not be further elaborated here. Among them, each entity attribute in the ethylene cracking furnace tube coking knowledge graph of the current operation cycle constructed for the first time is filled with "0".

[0090] If it exists, as described above, the current operation cycle represents the operation event of continuous feeding of the ethylene cracking furnace, so the ethylene industrial field data of the current operation cycle is constantly changing, and the corresponding current ethylene cracking furnace tube coking knowledge map also needs to be continuously updated.

[0091] It can be understood that the current ethylene cracking furnace tube coking knowledge graph is the same as the historical ethylene cracking furnace tube coking knowledge graph, and there is a temporal relationship between the entities and between the entities and attributes. During the updating process, the corresponding entities and attributes need to be associated and updated with the time point of the current operation cycle.

[0092] Specifically, when the industrial parameters in the ethylene cracking furnace tube change, that is, the ethylene industrial field data changes, it will affect the coking properties of the ethylene cracking furnace tube; the coking properties represent the coking state parameters of the ethylene cracking furnace tube, which are usually difficult to observe, and the ethylene industrial field data as shown above mainly include the industrial parameters of the ethylene cracking furnace tube, and these industrial parameters can be obtained by measurement.

[0093] Among them, in this embodiment, the acquisition of the ethylene industrial field data can be carried out according to a certain acquisition cycle, and the specific acquisition cycle can be set according to the actual production situation, and can be set to one day, one hour, or 20 minutes, etc. After the industrial parameters are acquired in each acquisition cycle, the industrial parameters are compared with the corresponding entity attributes in the current ethylene cracking furnace tube coking knowledge graph. If the two are different, the acquired ethylene industrial field data is updated to the entity attributes corresponding to the current ethylene cracking furnace tube coking knowledge graph. At the same time, the current time point is recorded in the current ethylene cracking furnace tube coking knowledge graph, and the updated entity attributes are associated with the recorded current time point. After that, the industrial parameters of the ethylene cracking furnace can be acquired through the time point to understand the working status of the ethylene cracking furnace; if the two are the same, only the current time point is updated, that is, the current time point is recorded in the current ethylene cracking furnace tube coking knowledge graph and associated with the corresponding entity attributes.

[0094] S5: according to the ethylene cracking knowledge ontology model, combined with the historical ethylene cracking furnace tube coking knowledge graph, reasoning about the coking attributes in the current ethylene cracking furnace tube coking knowledge graph, and updating the current ethylene cracking furnace tube coking knowledge graph according to the reasoning result;

[0095] Specifically, Figure 4 As shown, in this step, the coking attribute is inferred based on the ethylene industry field data in the current operation cycle through the historical ethylene cracking furnace tube coking knowledge graph, which may specifically include:

[0096] A1: constructing a coking diagnosis factor set according to the updated knowledge graph of coking in the current ethylene cracking furnace tube;

[0097] Specifically, the extracting of the coking diagnosis factor set according to the current ethylene cracking furnace tube coking knowledge graph after data update specifically includes:

[0098] Taking coking as the final tail entity of the triple path, searching for a non-cyclic triple path in the current ethylene cracking furnace tube coking knowledge graph; the triple path is formed by correspondingly connecting the tail entities and the head entities of several triples in sequence, wherein the tail entity of the previous triple is the same as the head entity of the next triple;

[0099] Exemplarily, take the following triples as an example <cracking raw material, raw material, cracking furnace>, <cracking furnace, product, product and yield>, <cracking furnace, reactor, furnace tube>, <furnace tube, parameter, process parameter>, <furnace tube, coking reaction, coking>, <coking, attribute, coking rate>, <coking, attribute, coke layer thickness>;

[0100] Then, by taking coking as the tail entity, two triplet paths can be obtained: <furnace tube, coking reaction, coking> and <cracking raw material, raw material-reactor-coking reaction, coking>;

[0101] Then, the coking diagnosis factor set is constructed according to the head entities of all the acquired triple paths.

[0102] Specifically, obtaining the coking diagnosis factor set may include:

[0103] Taking the above example as an example, the head entity "furnace tube" of the ternary path <furnace tube, coking reaction, coking> obtains the specific "furnace tube 16", "furnace tube 16" represents the furnace tube with a preset number of 16, and further, the three process parameters of "furnace tube 16" can be obtained: inlet temperature, outlet temperature, outlet pressure, recorded as parameter set p1 = (inlet temperature, outlet temperature, outlet pressure), and the equipment parameters of "furnace tube 16" are also obtained: length, tube diameter, recorded as parameter set p2 = (length, tube diameter); the head entity "cracking raw material" of the ternary path <cracking raw material, raw material-reactor-coking reaction, coking> can obtain parameters related to the raw material of furnace tube 16: cracking raw material, BMCI of cracking raw material, relative density, recorded as parameter set p3 = (cracking raw material composition, BMCI, relative density). The parameter sets obtained in the above process are sorted into a set, that is, p = (p1, p2, p3), and the obtained set p is the coking diagnosis factor set.

[0104] It should be noted that, in this embodiment, the above example uses the triple of the core ontology as an example in order to facilitate the understanding of the acquisition of the triple path. In the actual application process, it is necessary to obtain the triple path with the specific entities of the current ethylene cracking furnace tube coking knowledge graph, that is, to be able to directly obtain various specific parameters in the above example.

[0105] In a specific implementation of this embodiment, the triple path can be obtained through the ethylene cracking knowledge ontology model. Referring to the above example, after obtaining the triple path based on the ontology, the ontology therein is further replaced with concepts through the current ethylene cracking furnace tube coking knowledge graph to obtain the triple path about the entity.

[0106] A2: According to each coking diagnostic factor in the coking diagnostic factor set, a number of parameter items corresponding to each coking diagnostic factor are obtained from the current ethylene cracking furnace tube coking knowledge graph, and a parameter query condition set is constructed according to the parameter items, wherein the parameter query condition set contains at least one parameter item; Taking the coking diagnostic factor set p obtained in the above example as an example of the ethylene cracking furnace tube coking knowledge graph, it can be seen that the coking diagnostic factor set p contains 3 coking diagnostic factors, and each of the coking diagnostic factors contains a certain number of parameter items, such as the coking diagnostic factor "furnace tube 16" contains three parameter items: inlet temperature, outlet temperature and outlet pressure. In order to realize specific reasoning, it is necessary to obtain these parameter items and the corresponding parameter item values ​​from the current corresponding ethylene industrial field data, such as inlet temperature 895°C, outlet temperature 973°C, outlet pressure 166kPa.

[0107] At the same time, in this embodiment, the construction of the parameter query condition set can select at least one from the parameter items in the coking diagnosis factor set for construction. The parameter items are all obtained through the current ethylene industry field data, so the corresponding coking properties can be queried to a certain extent. The more parameter items included in the parameter query condition set, the more accurate the query result. The specific settings can be set according to the actual scenario.

[0108] A3: According to the parameter query condition set, query the historical coking attribute through the historical ethylene cracking furnace tube coking knowledge graph. If the query result is empty, adjust the parameter query condition set and query the historical coking attribute from the historical ethylene cracking furnace tube coking knowledge graph again;

[0109] As described above, theoretically, the change from the compared coking properties to the predicted coking properties is consistent with the relevant knowledge description in the ethylene cracking knowledge ontology model. However, there may be a situation where the query scope is reduced due to unreasonable setting of the parameter query condition set, and no corresponding results are found. Therefore, when the judgment result is inconsistent, the parameter query condition set can be appropriately adjusted.

[0110] In a preferred implementation of this embodiment, the adjustment of the parameter query condition set can adjust at least one of the parameter items in the parameter query condition set to range data by a preset threshold. Exemplarily, if the parameter query condition set is: (inlet temperature 895°C, outlet temperature 973°C, outlet pressure 166kPa), the inlet temperature is adjusted to range data according to the preset threshold ±5°C, then the adjusted parameter query condition set is (inlet temperature 895±5°C, outlet temperature 973°C, outlet pressure 166kPa).

[0111] Then, the historical coking attribute is queried again from the historical ethylene cracking furnace tube coking knowledge graph based on the adjusted range data; at this time, if the query result is an empty set, the predicted coking attribute is set to be the same as the coking attribute most recently updated in this operation cycle.

[0112] In another preferred implementation of this embodiment, the adjustment of the parameter query condition set can be achieved by obtaining several subsets of the parameter query condition set; the obtained subsets of the parameter query condition set are queried and compared again through the historical ethylene cracking furnace tube coking knowledge graph. Exemplarily, if the parameter query condition set is: (inlet temperature 895°C, outlet temperature 973°C, outlet pressure 166kPa), then the subsets of the parameter condition set may include (inlet temperature 895°C, outlet temperature 973°C), (outlet temperature 973°C, outlet pressure 166kPa), etc. (not all are listed), and one of the subsets is queried again through the historical ethylene cracking furnace tube coking knowledge graph.

[0113] Then, the obtained subset of the parameter query condition set is re-queried through the historical ethylene cracking furnace tube coking knowledge graph; at this time, if the query result is an empty set, the predicted coking attribute is set to be the same as the coking attribute most recently updated in this operation cycle.

[0114] It is understandable that when the amount of data in the ethylene cracking furnace tube coking knowledge graph is sufficiently large, through the above two adjustment methods of the parameter query condition set, the predicted coking attribute that meets the query parameter condition set can be found in the historical ethylene knowledge graph within a limited number of steps. When the current ethylene cracking furnace tube coking knowledge graph is constructed for the first time, the historical ethylene cracking furnace tube coking knowledge graph does not exist at this time, and the result returned for the predicted coking attribute is empty, and the predicted coking attribute is set to be the same as the coking attribute last updated in this operation cycle.

[0115] A4: According to the current ethylene cracking furnace tube coking knowledge graph, the coking attribute before the data is updated is obtained as a comparison coking attribute;

[0116] A5: comparing the historical coking attributes with the compared coking attributes, and inferring the comparison result according to the knowledge of the ethylene cracking knowledge ontology model to infer and predict the coking attributes;

[0117] Specifically, in this step, the reasoning process may specifically include:

[0118] Determining whether the numerical changes of the historical coking attributes and the compared coking attributes are consistent with the coking knowledge description in the ethylene cracking knowledge ontology model;

[0119] If the descriptions are consistent, the historical coking attribute is used as the predicted coking attribute;

[0120] As described above, in this embodiment, the historical ethylene cracking furnace tube coking knowledge graph contains the ethylene industrial field data of the entire complete operation cycle, and the coking attributes obtained by querying the parameter query condition set related to the current ethylene industrial field data can reflect the coking state at the historical time point under the parameters reflected by the corresponding ethylene industrial field data. At this time point, the current ethylene cracking furnace tube coking knowledge graph has not completed the update of the coking attributes, so the historical coking attributes obtained by querying the historical ethylene cracking furnace tube coking knowledge graph are equivalent to the prediction of the current coking attributes relative to the current ethylene cracking furnace tube coking knowledge graph; also because the coking attributes of the current ethylene cracking furnace tube coking knowledge graph have not been updated, the coking attributes before the data of the current ethylene cracking furnace tube coking knowledge graph are obtained as the comparison coking attributes.

[0121] Exemplarily, the coking knowledge ontology model describes that "the coking thickness of the furnace tube becomes thicker as the number of operating days increases". After obtaining the historical coking attributes and the comparison coking attributes, the historical coking attributes are compared with the comparison coking attributes, and the attribute changes of the historical coking attributes and the comparison coking attributes are obtained as the comparison results, which are specifically manifested in whether the numerical changes of the coking attributes are consistent with the description in the coking knowledge ontology model. If they are consistent with the description in the coking knowledge ontology model, the historical coking attributes are used as the predicted coking attributes.

[0122] Because the compared coking attribute reflects the coking status before the data is updated, and the historical coking attribute reflects the coking status under the current ethylene industry field data, under normal circumstances, the numerical change from the compared coking attribute to the historical coking attribute is consistent with the relevant knowledge description in the ethylene cracking knowledge ontology model.

[0123] Therefore, to determine whether the comparison result is consistent with the coking knowledge in the ethylene cracking knowledge ontology model, specifically, the theoretical numerical changes between the ethylene industrial field data corresponding to the historical coking attributes and the compared coking attributes in the ethylene cracking knowledge ontology model can be obtained; and whether the comparison result is consistent with the theoretical numerical changes can be determined.

[0124] The theoretical value change may include the increase or decrease of coking thickness, the increase or decrease rate of coking thickness, etc.

[0125] If they are inconsistent, the coking properties are obtained from the industrial site as the predicted coking properties.

[0126] In actual situations, there may also be a situation where the comparison result is completely inconsistent with the knowledge description in the ethylene cracking knowledge ontology model. The occurrence of this situation usually indicates that the data of the constructed historical ethylene cracking furnace tube coking knowledge graph is incomplete, resulting in inaccurate reasoning results. This situation mainly occurs in the early stage of the establishment of the historical ethylene cracking furnace tube coking knowledge graph. At this time, the historical ethylene cracking furnace tube coking knowledge graph does not exist (i.e., the current ethylene cracking furnace tube coking knowledge graph is constructed for the first time), or the number of constructed historical ethylene cracking furnace tube coking knowledge graphs is small. At this time, the historical ethylene cracking furnace tube coking knowledge graph is queried, and the returned result is the coking attribute of the most recent update, or the returned result is the result obtained by the query after adjustment, resulting in a certain deviation from the actual coking attribute. Therefore, it is necessary to update and correct the constructed current ethylene cracking furnace tube coking knowledge graph through the actual on-site industrial coking attributes.

[0127] If the comparison result is inconsistent with the coking knowledge in the ethylene cracking knowledge ontology model, the industrial coking attributes under the corresponding ethylene industrial field data are obtained, and the industrial coking attributes are used as the reasoning result, that is, the predicted coking attributes.

[0128] A6: Update the coking attribute of the current ethylene cracking furnace tube coking knowledge graph to the predicted coking attribute; complete the update of the coking attribute at the current time point of the current operation cycle.

[0129] Updating the current ethylene cracking furnace tube coking knowledge graph can facilitate the subsequent conversion of the current ethylene cracking furnace tube coking knowledge graph into the historical ethylene cracking furnace tube coking knowledge graph.

[0130] S6: If the current operation cycle corresponding to the current ethylene cracking furnace tube coking knowledge graph ends, the current ethylene cracking furnace tube coking knowledge graph is used as the historical ethylene cracking furnace tube coking knowledge graph, and the ethylene industry field data of a new round of operation cycle is obtained, and the current ethylene cracking furnace tube coking knowledge graph corresponding to the operation cycle is constructed.

[0131] The current ethylene cracking furnace tube coking knowledge graph, which is the same as the current operation cycle and the historical operation cycle, can be converted into the historical ethylene cracking furnace tube coking knowledge graph; in fact, the step S6 is the basis of the above steps. In this embodiment, by continuously predicting the coking properties, the predicted coking properties are updated to the current ethylene cracking furnace tube coking knowledge graph, and the current ethylene cracking furnace tube coking knowledge graph is continuously converted into the historical ethylene cracking furnace tube coking knowledge graph, so as to realize the continuous enrichment of the knowledge graph database, and then realize the inference prediction of the coking properties under different ethylene industrial field data.

[0132] In the present embodiment, the coking properties are manually detected only when the current ethylene cracking furnace tube coking knowledge graph is first established and when reasoning cannot be performed through query; and as the number of historical ethylene cracking furnace tube coking knowledge graphs continues to increase, the knowledge graph database is continuously enriched, and eventually the corresponding coking properties can be queried in one of the historical ethylene cracking furnace tube coking knowledge graphs. From an overall perspective, the present embodiment can automatically realize the reasoning of the coking properties of the ethylene cracking furnace, reduce manual detection, and effectively improve the efficiency of the reasoning of the coking properties of the ethylene cracking furnace.

[0133] In a specific implementation of the present embodiment, when the ethylene cracking furnace tube coking knowledge graph is first established, there is no corresponding historical ethylene cracking furnace tube coking knowledge graph, so the predicted coking attributes cannot be obtained through the historical ethylene cracking furnace tube coking knowledge graph. Therefore, in the initial construction process of the current ethylene cracking furnace tube coking knowledge graph, the current ethylene cracking furnace tube coking knowledge graph is updated entirely by manually detecting the industrial coking attributes of the industrial site; when the first operating cycle ends and enters the second operating cycle, the second operating cycle is used as the current operating cycle, and the first operating cycle is used as the historical operating cycle. Correspondingly, the first constructed current ethylene cracking furnace tube coking knowledge graph is transformed into A new current ethylene cracking furnace tube coking knowledge graph is reconstructed for the historical ethylene cracking furnace tube coking knowledge graph; by acquiring the ethylene industrial field data of the current operation cycle, the historical ethylene cracking furnace tube coking knowledge graph is used for query to obtain the corresponding predicted coking attributes, and at the same time, the corresponding compared coking attributes are obtained through the current ethylene cracking furnace tube coking knowledge graph, and the compared results are compared with the knowledge description in the ethylene cracking knowledge ontology model. If the final comparison results are inconsistent after multiple comparisons, the current ethylene cracking furnace tube coking knowledge graph is updated by manually detecting the industrial coking attributes of the industrial site; if there is a consistency, the current ethylene cracking furnace tube coking knowledge graph is updated with the predicted coking attributes. Repeat the above operation until the end of the second round of operation cycle, convert the current ethylene cracking furnace tube coking knowledge graph into the historical ethylene cracking furnace tube coking knowledge graph, and enter the third round of operation cycle. By analogy, continuously construct the current ethylene cracking furnace tube coking knowledge graph and add the historical ethylene cracking furnace tube coking knowledge graph.

[0134] Obviously, the above embodiments of the present invention are merely examples for clearly illustrating the technical solution of the present invention, and are not intended to limit the specific implementation methods of the present invention. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the claims of the present invention shall be included in the protection scope of the claims of the present invention.

Claims

1. A knowledge graph-based reasoning method for ethylene cracking furnace tube coking, characterized in that: The reasoning method includes: Construct the knowledge ontology model of ethylene cracking based on the knowledge related to ethylene cracking; Acquire ethylene industry field data of several operation cycles, the several operation cycles include current operation cycle and historical operation cycle; the operation cycle represents the time of continuous feeding operation between two coke cleaning treatments of ethylene cracking furnace, the current operation cycle represents the operation time when the previous coke cleaning is completed at the current moment and the next coke cleaning time has not yet arrived, and the historical operation cycle represents the complete operation time between two coke cleanings; According to the ethylene cracking knowledge ontology model and the acquired ethylene industry field data of the historical operation cycle, a knowledge graph of coking of furnace tubes of historical ethylene cracking furnaces is constructed; the knowledge graph of coking of furnace tubes of historical ethylene cracking furnaces contains at least one knowledge relationship of coking of furnace tubes of the historical operation cycle; Constructing a knowledge graph of coking of furnace tubes of a current ethylene cracking furnace according to the ethylene cracking knowledge ontology model, and updating the knowledge graph of coking of furnace tubes of a current ethylene cracking furnace according to the acquired ethylene industry field data of the current operation cycle; According to the ethylene cracking knowledge ontology model, combined with the historical ethylene cracking furnace tube coking knowledge graph, the coking attributes in the current ethylene cracking furnace tube coking knowledge graph are inferred, and the current ethylene cracking furnace tube coking knowledge graph is updated according to the inference result; If the current operation cycle corresponding to the current ethylene cracking furnace tube coking knowledge graph ends, the current ethylene cracking furnace tube coking knowledge graph is used as the historical ethylene cracking furnace tube coking knowledge graph, and the ethylene industry field data of a new round of operation cycle is obtained, and the current ethylene cracking furnace tube coking knowledge graph corresponding to the operation cycle is constructed.

2. The method for reasoning about coking in ethylene cracking furnace tubes based on knowledge graph according to claim 1, characterized in that: The method of constructing a knowledge graph of coking of furnace tubes of ethylene cracking furnaces based on the ethylene cracking knowledge ontology model and the acquired ethylene industry field data of the historical operation cycle specifically includes: Acquiring the ethylene industry field data at each time point in each of the historical operation cycles; The knowledge description of the relationship between entities in the ethylene cracking coking knowledge ontology model and the ethylene industry field data are combined to determine the time-varying relationship between entities and between entities and attributes. The entities and entity attributes are used as vertices and the relationships are used as edges to form a historical ethylene cracking furnace tube coking knowledge graph with time-varying characteristics within a period.

3. The method for reasoning about coking of ethylene cracking furnace tubes based on knowledge graph according to claim 1, characterized in that: The method of constructing a knowledge graph of coking of furnace tubes of a current ethylene cracking furnace according to the ethylene cracking knowledge ontology model, and updating the knowledge graph of coking of furnace tubes of a current ethylene cracking furnace according to the acquired ethylene industrial field data of the current operation cycle, specifically includes: Determine whether the current ethylene cracking ethylene cracking furnace tube coking knowledge graph exists; If it does not exist, determine the relationship between entities and entities, and between entities and attributes according to the basic relationship between entities in the ethylene cracking coking knowledge ontology model, take entities and entity attributes as vertices, and take relationships as edges to construct the knowledge graph of ethylene cracking furnace tube coking in the current period; If it exists, determine whether the ethylene industrial field data at the current time point in the current operation cycle has changed: if so, update the current ethylene cracking furnace tube coking knowledge graph according to the current time point and the ethylene industrial field data at the current time point; if not, only update the current ethylene cracking furnace tube coking knowledge graph according to the current time point.

4. The method for reasoning about coking of ethylene cracking furnace tubes based on knowledge graph according to claim 3 is characterized in that: The method of inferring the coking attributes in the current ethylene cracking furnace tube coking knowledge graph based on the ethylene cracking knowledge ontology model and combining the historical ethylene cracking furnace tube coking knowledge graph specifically includes: Constructing a coking diagnosis factor set according to the updated knowledge graph of coking in the current ethylene cracking furnace tube; According to each coking diagnosis factor in the coking diagnosis factor set, a parameter item corresponding to each coking diagnosis factor is obtained from the current ethylene cracking furnace tube coking knowledge graph, and a parameter query condition set is constructed according to the parameter item; the parameter query condition set contains at least one parameter item; According to the parameter query condition set, query the historical coking attribute through the historical ethylene cracking furnace tube coking knowledge graph, if the query result is empty, adjust the parameter query condition set, and query the historical coking attribute from the historical ethylene cracking furnace tube coking knowledge graph again; According to the current ethylene cracking furnace tube coking knowledge graph, the coking attribute before the data is updated is obtained as a comparison coking attribute; Comparing the historical coking attributes with the compared coking attributes, reasoning the comparison result according to the knowledge of the ethylene cracking knowledge ontology model, and inferring and predicting the coking attributes; The coking attribute of the current ethylene cracking furnace tube coking knowledge graph is updated to the predicted coking attribute.

5. The method for reasoning about coking of ethylene cracking furnace tubes based on knowledge graph according to claim 4 is characterized in that: The coking diagnosis factor set is constructed according to the updated knowledge graph of coking of the current ethylene cracking furnace tube, specifically including: Taking coking as the final tail entity of the triple path, searching for a non-cyclic triple path in the current ethylene cracking furnace tube coking knowledge graph; the triple path is formed by correspondingly connecting the tail entities and the head entities of several triples in sequence, wherein the tail entity of the previous triple is the same as the head entity of the next triple; The coking diagnosis factor set is constructed according to the acquired head entities of all the triplet paths.

6. The method for reasoning about coking of ethylene cracking furnace tubes based on knowledge graph according to claim 4, characterized in that: The adjusting query condition set, and again querying the historical coking attributes from the historical ethylene cracking furnace tube coking knowledge graph, specifically includes: Adjusting at least one of the parameter items in the parameter query condition set to range data by a preset threshold; According to the adjusted range data, the historical coking attribute is again queried from the historical ethylene cracking furnace tube coking knowledge graph; If the query result is an empty set, the predicted coking attribute is set to be the same as the coking attribute most recently updated during this operation cycle.

7. The method for reasoning about coking in ethylene cracking furnace tubes based on knowledge graph according to claim 4, characterized in that: The adjusting query condition set, and again querying the historical coking attributes from the historical ethylene cracking furnace tube coking knowledge graph, specifically includes: Obtaining several subsets of the parameter query condition set; The obtained subset of the parameter query condition set is queried again through the historical ethylene cracking furnace tube coking knowledge graph; If the query results are all empty sets, the predicted coking attribute is set to be the same as the coking attribute most recently updated during this operation cycle.

8. The method for reasoning about coking in ethylene cracking furnace tubes based on knowledge graph according to claim 4, characterized in that: The comparing the historical coking attributes with the compared coking attributes, and reasoning the comparison result according to the knowledge of the ethylene cracking knowledge ontology model to predict the coking attributes specifically include: Determining whether the numerical changes of the historical coking attributes and the compared coking attributes are consistent with the coking knowledge description in the ethylene cracking knowledge ontology model; If the descriptions are consistent, the historical coking attribute is used as the predicted coking attribute; If they are inconsistent, the coking properties are obtained from the industrial site as the predicted coking properties.

9. The method for reasoning about coking in ethylene cracking furnace tubes based on knowledge graph according to any one of claims 1 to 8, characterized in that: The core ontology in the ethylene cracking knowledge ontology model at least includes cracking raw materials, cracking furnace, furnace tubes, process parameters, products and yields, and coking.

Images