Public engineering and energy management method, device and system for chemical production

Abstract

The embodiment of the invention provides a public engineering and energy management method, device and system for chemical production. In response to the existence of the data source exception, performing data source exception root cause query in the operation knowledge graph so as to determine a first exception KPI of the root cause data source exception from an exception KPI set obtained in the public engineering and energy management operation process, and for a second exception KPI in the exception KPI set, performing data source exception root cause query on the first exception KPI of the root cause data source exception; and performing a KPI entity root cause query based on an upstream and downstream relationship in the operation knowledge graph to determine a first KPI entity as a root cause of each second abnormal KPI, the second abnormal KPIs including remaining KPIs of which the first abnormal KPI is removed from the abnormal KPI set. According to the method, the operation knowledge graph used for reflecting the semantic association between the KPI and the service scene, between the KPI and the KPI and between the KPI and the underlying data is constructed, and the KPI anomaly root cause analysis is performed based on the operation knowledge, so that the root of the KPI anomaly problem can be quickly positioned, and the accuracy of the KPI anomaly root cause analysis is improved.

Description

Technical Field

[0001] The embodiments in this specification generally relate to the field of chemical production, and in particular to methods, apparatus and systems for utilities and energy management in chemical production. Background Technology

[0002] Utility and energy management are essential in chemical production processes. In chemical production, utilities are the "infrastructure and auxiliary systems" that support the stable operation of core production units. While they do not directly participate in product synthesis, they determine the continuity, safety, and economy of chemical production. Energy management is the comprehensive management of all types of energy consumed in the chemical production process, involving "overall planning, optimized allocation, and monitoring and accounting," to achieve energy conservation, cost reduction, compliant emission reduction, and efficient and stable supply.

[0003] In the operation of utilities and energy management, a large amount of operational data needs to be collected through operational data acquisition components (e.g., terminal sensing devices, process control components, metering and accounting devices) deployed at various nodes throughout the entire process of the utility system (e.g., energy supply, media transportation, consumption terminals, environmental emissions). The collected operational data may include key parameters in chemical production, such as flow rate, pressure, and temperature. This operational data can be transformed into various Key Performance Indicators (KPIs) to help users understand the current operational status of utilities and energy management and to support decision-making.

[0004] When a KPI anomaly alarm occurs, it is necessary to determine the potential risks behind the alarm and take appropriate measures to resolve the issue. The existing KPI anomaly alarm handling mechanism relies on collaboration between digital and operational technical personnel to build static Business Intelligence (BI) reports that display the KPIs that triggered the alarm, followed by manual analysis of the current alert status and potential causes.

[0005] Existing KPI anomaly alert handling mechanisms require manual analysis, making them highly dependent on user expertise and unable to automatically identify anomalies, trends, or perform root cause analysis, thus reducing the efficiency of decision support. Furthermore, the displayed content needs to be predefined, such as pre-set charts and data views, lacking flexibility and unable to be dynamically adjusted according to specific problems or application scenarios. Moreover, the displayed content lacks contextual adaptability, failing to automatically adjust according to the current business background, roles, or tasks, resulting in irrelevant or inaccurate content. Additionally, the fragmented information in the displayed content requires users to switch between multiple dashboards to piece together a complete business scenario, increasing the cognitive burden and operational complexity for users. Summary of the Invention

[0006] In view of the above, embodiments of this specification provide methods, apparatus, and systems for utilities and energy management in chemical production. Using this method, by constructing an operational knowledge graph reflecting the semantic relationships between KPIs and business scenarios, between KPIs themselves, and between KPIs and underlying data, and by performing root cause analysis of KPI anomalies based on operational knowledge, the root causes of KPI anomalies can be quickly located, improving the accuracy of KPI anomaly root cause analysis.

[0007] According to one aspect of the embodiments of this specification, a method for utilities and energy management in chemical production is provided, comprising: after obtaining an abnormal key performance indicator set in the operation process of utilities and energy management, determining whether there is a data source abnormality; in response to the existence of a data source abnormality, performing a root cause query of the data source abnormality in an operational knowledge graph to determine a first abnormal key performance indicator whose root cause is the data source abnormality from the abnormal key performance indicator set; and for a second abnormal key performance indicator in the abnormal key performance indicator set, performing a key performance indicator entity root cause query based on upstream and downstream relationships in the operational knowledge graph to determine a first key performance indicator entity as the root cause of each second abnormal key performance indicator, wherein the second abnormal key performance indicators include the remaining abnormal key performance indicators after removing the first abnormal key performance indicator from the abnormal key performance indicator set, wherein the operational knowledge graph includes entity nodes and entity node relationships, the entity nodes include key performance indicator entity nodes, data source entity nodes, and business scenario entity nodes, and the entity node relationships are used to reflect the logical relationships between entity nodes, the logical relationships include the affiliation relationship of key performance indicator entity nodes to business scenario entity nodes, the data source relationship of key performance indicator entity nodes to data source entity nodes, and the upstream and downstream relationships between key performance indicator entity nodes.

[0008] Optionally, in one example of the above aspects, performing a root cause query for data source anomalies in the operational knowledge graph to determine a first abnormal key performance indicator (KPI) from the set of abnormal KPIs whose root cause is data source anomalies may include: querying the operational knowledge graph to see if a third abnormal KPI exists in the set of abnormal KPIs; if a third abnormal KPI exists, querying the operational knowledge graph to see a fourth abnormal KPI that has an upstream relationship with the third abnormal KPI; and determining the third abnormal KPI and the fourth abnormal KPI as the first abnormal KPI whose root cause is data source anomalies.

[0009] Optionally, in one example of the above aspects, the entity node relationship also includes the causal relationship between key performance indicator (KPI) entity nodes. The method may further include: performing a causal relationship query in the operational knowledge graph to identify second KPI entities that have a causal relationship with the first KPI entities that are the root cause of each second abnormal KPI; performing semantic analysis based on a key performance indicator knowledge base for the first KPI entities that have normal or no corresponding second KPI entities; and generating root cause analysis results for each second abnormal KPI based on the semantic analysis results of the abnormal second KPI entities and / or the first KPI entities that have normal or no corresponding second KPI entities.

[0010] Optionally, in one example of the above aspects, the entity node further includes entity attributes. The method may further include: generating an operational anomaly diagnostic report based on the root cause analysis results of the first abnormal key performance indicator and the second abnormal key performance indicator, the operational anomaly diagnostic report including the anomaly attribution analysis results of the key performance indicator anomalies.

[0011] Optionally, in one example of the above aspects, the operational anomaly diagnostic report may also include measures to address anomalies in key performance indicators.

[0012] Optionally, in one example of the above aspects, generating an operational anomaly diagnostic report based on the root cause analysis results of the first abnormal key performance indicator and the second abnormal key performance indicator may include: providing the root cause analysis results and root cause analysis process of the first abnormal key performance indicator and the second abnormal key performance indicator to a large language model to generate an operational anomaly diagnostic report with a structured data structure.

[0013] Optionally, in one example of the above aspects, the operational knowledge graph can be constructed based on at least one of the following knowledge graph construction methods: constructing the operational knowledge graph based on entity nodes, entity node relationships, and entity node attributes input by the user in a specified format; constructing the operational knowledge graph by performing semantic parsing based on a large language model on the operational business scenario description, data source description, and key performance indicator definition output by the user in natural language to obtain entity nodes, entity node relationships, and entity node attributes; constructing the operational knowledge graph by performing semantic extraction based on a large language model on the unstructured text data output by the user to obtain entity nodes, entity node relationships, and entity node attributes, wherein the unstructured text data includes operational documents, operating procedures, and key performance indicator definition manuals; and constructing the operational knowledge graph based on the historical operational data of the public works and energy management system.

[0014] Optionally, in one example of the above aspects, the method may further include: obtaining a key performance indicator (KPI) query request initiated by a user, the KPI query request including at least one of historical trends of KPIs, influencing factors, and upstream and downstream KPIs; querying the query results of the KPI query request in the operational knowledge graph; and generating a KPI query report based on the query results of the KPI query request to provide to the user.

[0015] According to another aspect of the embodiments of this specification, an apparatus for utilities and energy management in chemical production is provided, comprising: a data source anomaly determination unit configured to determine whether a data source anomaly exists after acquiring an abnormal key performance indicator set during the operation of the utilities and energy management; a data source anomaly root cause determination unit configured to, in response to the existence of a data source anomaly, perform a data source anomaly root cause query in an operational knowledge graph to determine a first abnormal key performance indicator from the abnormal key performance indicator set whose root cause is a data source anomaly; and a key performance indicator entity root cause determination unit configured to, for a second abnormal key performance indicator in the abnormal key performance indicator set, perform a key performance indicator entity root cause determination based on upstream and downstream relationships in the operational knowledge graph. A query is performed to identify the first key performance indicator entity that is the root cause of each second abnormal key performance indicator. The second abnormal key performance indicators include the remaining abnormal key performance indicators after removing the first abnormal key performance indicator from the set of abnormal key performance indicators. The operational knowledge graph includes entity nodes and entity node relationships. The entity nodes include key performance indicator entity nodes, data source entity nodes, and business scenario entity nodes. The entity node relationships are used to reflect the logical relationships between entity nodes. The logical relationships include the affiliation relationship of key performance indicator entity nodes with respect to business scenario entity nodes, the data source relationship of key performance indicator entity nodes with respect to data source entity nodes, and the upstream and downstream relationships between key performance indicator entity nodes.

[0016] According to another aspect of the embodiments of this specification, a system for utilities and energy management in chemical production is provided, comprising: an operations knowledge graph construction device configured to construct an operations knowledge graph of the operations of utilities and energy management in chemical production; an operations monitoring device configured to perform data source anomaly detection and key performance indicator anomaly detection on the operations of the utilities and energy management; and the aforementioned apparatus for utilities and energy management in chemical production.

[0017] According to another aspect of the embodiments of this specification, an apparatus for utilities and energy management in chemical production is provided, comprising: at least one processor; a memory coupled to the at least one processor; and a computer program stored in the memory, wherein the at least one processor executes the computer program to implement the method for utilities and energy management in chemical production as described above.

[0018] According to another aspect of the embodiments of this specification, a computer-readable storage medium is provided that stores executable instructions, which, when executed, cause a processor to perform the method for utilities and energy management for chemical production as described above. Attached Figure Description

[0019] A further understanding of the nature and advantages of this specification can be achieved by referring to the following figures. In the figures, similar components or features may have the same reference numerals.

[0020] Figure 1 An example block diagram of a chemical production system according to an embodiment of this specification is shown.

[0021] Figure 2 An example schematic diagram of an operational knowledge graph for utilities and energy management according to an embodiment of this specification is shown.

[0022] Figure 3 An example flowchart of the operational knowledge graph construction process according to an embodiment of this specification is shown.

[0023] Figure 4 Another example flowchart of the operational knowledge graph construction process according to an embodiment of this specification is shown.

[0024] Figure 5 Another example flowchart of the operational knowledge graph construction process according to an embodiment of this specification is shown.

[0025] Figure 6 An example flowchart of a KPI anomaly analysis method according to an embodiment of this specification is shown.

[0026] Figure 7 An example flowchart of a data source root cause determination process according to an embodiment of this specification is shown.

[0027] Figure 8 An example flowchart of a KPI root cause determination process according to an embodiment of this specification is shown.

[0028] Figure 9 A schematic diagram illustrating an example of a KPI anomaly root cause analysis process according to an embodiment of this specification is shown.

[0029] Figure 10 An example block diagram of a KPI anomaly analysis apparatus according to an embodiment of this specification is shown.

[0030] Figure 11 An example schematic diagram of a KPI anomaly analysis device implemented using a computer system according to an embodiment of this specification is shown. Detailed Implementation

[0031] The subject matter described herein will now be discussed with reference to exemplary embodiments. It should be understood that these embodiments are discussed merely to enable those skilled in the art to better understand and implement the subject matter described herein, and are not intended to limit the scope, applicability, or examples set forth in the claims. The function and arrangement of the elements discussed may be changed without departing from the scope of this specification. Various processes or components may be omitted, substituted, or added as needed in the various examples. For example, the described methods may be performed in a different order than described, and steps may be added, omitted, or combined. Furthermore, features described in some examples may be combined in other examples.

[0032] As used herein, the term "comprising" and its variations are open terms meaning "including but not limited to". The term "based on" means "at least partially based on". The terms "one embodiment" and "an embodiment" mean "at least one embodiment". The term "another embodiment" means "at least one other embodiment". The terms "first", "second", etc., may refer to different or the same objects. Other definitions, whether explicit or implicit, may be included below. Unless explicitly indicated by the context, the definition of a term shall remain consistent throughout the specification.

[0033] Figure 1 An example block diagram of a chemical production system 100 according to an embodiment of this specification is shown.

[0034] like Figure 1 As shown, the chemical production system 100 includes a chemical production unit 110, an operation monitoring unit 120, an operation knowledge graph construction unit 130, an operation knowledge graph storage unit 140, and a KPI anomaly analysis unit 150. The chemical production unit 110, operation monitoring unit 120, operation knowledge graph construction unit 130, operation knowledge graph storage unit 140, and KPI anomaly analysis unit 150 can be communicatively connected via a network 160. In some embodiments, some or all components of the chemical production unit 110, operation monitoring unit 120, operation knowledge graph construction unit 130, operation knowledge graph storage unit 140, and KPI anomaly analysis unit 150 can communicate directly without needing to communicate through the network 160.

[0035] In some embodiments, network 160 can be any one or more of wired or wireless networks. Examples of network 160 may include, but are not limited to, cable networks, fiber optic networks, telecommunications networks, corporate intranets, the Internet, local area networks (LANs), wide area networks (WANs), wireless local area networks (WLANs), metropolitan area networks (MANs), public switched telephone networks (PSTNs), Bluetooth networks, ZigBee networks, near field communication (NFC), device internal buses, device internal lines, and any combination thereof.

[0036] Chemical production unit 110 is used to perform chemical production using raw materials that meet the requirements of production orders after receiving them. Utilities and energy management are required during chemical production in chemical production unit 110. Utilities are the "infrastructure and auxiliary systems" that support the stable operation of the core production units in chemical production. They do not directly participate in product synthesis but determine the continuity, safety, and economy of chemical production. Utilities may include, for example, energy supply systems, industrial gas systems, water supply and drainage systems, environmental protection systems, and other auxiliary systems. Energy management is the comprehensive management of all types of energy consumed in the chemical production process, including "overall planning, optimized allocation, and monitoring and accounting," to achieve energy conservation, cost reduction, compliance and emission reduction, and efficient and stable supply. Energy management may include, for example, energy metering and data acquisition, energy structure optimization, operational optimization and technological transformation, energy cost control and assessment, and compliance and emission reduction management.

[0037] In the operation of public utilities and energy management, a large amount of operational data needs to be collected through operational data acquisition components (e.g., terminal sensing devices, process control components, metering and accounting devices) deployed at various nodes throughout the entire process of public utility systems (e.g., energy supply, media transportation, consumption terminals, environmental emissions). The collected operational data may include key parameters in chemical production, such as flow rate, pressure, and temperature. This operational data can be transformed into various KPIs to help users understand the current operational status of public utilities and energy management and to support decision-making.

[0038] The operation monitoring device 120 is used to detect data source anomalies and KPI anomalies in the operation process of utilities and energy management. Data sources refer to the sources that generate operational data during the operation process of utilities and energy management. Examples of data sources include, but are not limited to, operational data generating devices that generate various operational data during the operation process of utilities and energy management, such as flow meters, pressure gauges, and thermometers. KPI anomaly detection can include single KPI anomalies and multiple KPI joint anomalies. For example, multiple KPI joint anomalies can be detected in the steam cost KPI monitoring, specifically targeting the amount of self-produced fuel, the amount of purchased steam fuel, and boiler load. If the amount of self-produced fuel is sufficient to bring the boiler to its minimum load, and more purchased natural gas is introduced, then a multiple KPI joint anomaly targeting the amount of self-produced fuel, the amount of purchased steam fuel, and boiler load is considered. The detected abnormal KPIs can be uniformly aggregated into an abnormal KPI pool for subsequent root cause analysis and intelligent diagnosis.

[0039] The operational knowledge graph construction device 130 is used to construct an operational knowledge graph of the operational processes of public works and energy management, and stores the constructed operational knowledge graph in the operational knowledge graph storage device 140. The KPI anomaly analysis device 150 uses the constructed operational knowledge graph to perform root cause analysis on anomaly KPIs in the anomaly KPI pool, thereby determining the causes of anomalies and providing corresponding countermeasures to ensure production safety in chemical production. In some examples, the operational knowledge graph constructed by the operational knowledge graph construction device 130 can be directly stored in the KPI anomaly analysis device 150, eliminating the need for the operational knowledge graph storage device 140.

[0040] An operational knowledge graph can be built based on user-provided operational system knowledge or historical operational data from utility and energy management processes. An operational knowledge graph can include entity nodes and the relationships between them. Entity nodes can include, for example, KPI entity nodes, business scenario entity nodes, and data source entity nodes. KPI entity nodes represent KPIs in the utility and energy management domain, such as energy consumption, unit energy consumption, and energy conversion rate. Business scenario entity nodes represent the business context in which the KPI exists, reflecting the semantic relationship between the KPI and actual operational activities. Data source entities represent the data sources from which the underlying data is transformed into KPIs, such as Process Information Management Systems (PIMS) and SAP systems.

[0041] Entity node relationships include logical relationships between entity nodes. In an operational knowledge graph, entity node relationships can include, but are not limited to, relationships between KPI entity nodes and business scenario entity nodes, relationships between KPI entity nodes and data source entity nodes, and upstream and downstream relationships between KPI entity nodes. The entity relationship between a KPI entity node and a business scenario entity node indicates that the KPI serves the business scenario (belong_to). The entity relationship between a KPI entity node and a data source entity node indicates that the KPI is calculated based on the underlying data generated by the data source (derived_from), and this entity relationship can support data tracing and computational chain construction. Upstream and downstream relationships between KPI entity nodes represent the upstream and downstream relationships between KPIs, such as aggregation / decomposition relationships (aggregate_to / decomposed_from). If KPI A and KPI B are aggregated into KPI C, then KPI A and KPI B are downstream of KPI C, and KPI C is upstream of KPI A and KPI B. If KPI A is decomposed into KPI B and KPI C, then KPI A is upstream of KPI B and KPI C, and KPI B and KPI C are downstream of KPI A. In some examples, entity node relationships can also include causal relationships between KPI entity nodes (e.g., causal direction, confidence level, etc.). Confidence level indicates the strength of the causal relationship; the stronger the confidence level, the more significant the causal relationship. A causal relationship between KPI entity nodes indicates that one KPI may cause another KPI to be abnormal (impacted_by).

[0042] Figure 2 An example schematic diagram of an operational knowledge graph for utilities and energy management according to an embodiment of this specification is shown. Figure 2 In the example shown, the operational knowledge graph 200 includes business scenario entity nodes, data source entity nodes, and KPI entity nodes KPI 1 to KPI 3. There is a service entity relationship (belong_to) between KPI 1 and the business scenario entity nodes, a derived entity relationship (derived_from) between KPI 1 and the data source entity nodes, an aggregation / decomposition relationship (aggregates_to / decomposed_from) between KPI 1 and KPI 2, and a causal relationship between KPI 1 and KPI 3, meaning that KPI 3 causes KPI 1 to become abnormal.

[0043] In some embodiments, entity nodes may further include entity attributes. KPI entity attributes may include, but are not limited to, KPI name, KPI definition, KPI calculation formula, KPI unit, utility and energy type, KPI-assigned device, and KPI update frequency. Business scenario entity nodes may include, but are not limited to, business scenario name and business scenario type. Data source entity nodes may include, but are not limited to, data source name, data source type, data source-assigned device, data acquisition frequency, data unit, and data quality.

[0044] In some examples, users possess relatively rich operational knowledge. In such cases, users can input entity nodes, entity relationships (and entity node attributes, if they exist), in a specified format through the front-end interface, and then the operational knowledge graph can be directly generated based on the user-input entity nodes, entity relationships, and entity node attributes.

[0045] In some examples, an operational knowledge graph can be constructed by semantically parsing the operational business scenario descriptions, data source descriptions, and KPI definitions that users input in natural language through the front-end interface.

[0046] Figure 3 An example flowchart of an operational knowledge graph construction process 300 according to an embodiment of this specification is shown.

[0047] like Figure 3 As shown in section 310, the operational business scenario description, data source description, and KPI definition input by the user in natural language form through the front-end interface are obtained. In section 320, the input operational business scenario description, data source description, and KPI definition are semantically parsed using a large language model, thereby resolving entity nodes, entity node relationships (and entity node attributes, if present). For example, through semantic parsing, the relationship between KPI definitions and business affiliation, potential upstream and downstream relationships, and causal relationships are automatically identified. In section 330, an operational knowledge graph is constructed based on the semantic parsing results (the parsed entity nodes, entity node relationships, and entity node attributes).

[0048] In some examples, an operational knowledge graph can be constructed by semantically extracting operational guidelines documents in unstructured text format that users input through a front-end interface.

[0049] Figure 4 Another example flowchart of an operational knowledge graph construction process 400 according to an embodiment of this specification is shown.

[0050] like Figure 4As shown in section 410, the operation specification document with unstructured text format input by the user through the front-end interface is obtained. The input operation specification document may include, for example, operation documents, operation specifications, and KPI definition manuals.

[0051] At 420, semantic extraction is performed on the user-input operation specification document via a large language model to extract entity nodes, entity node relationships (and entity node attributes, if they exist) from the operation specification document.

[0052] In section 430, an operational knowledge graph is constructed based on the semantic extraction results (extracted entity nodes, entity node relationships, and entity node attributes). For example, the extracted entity nodes, entity node relationships (and entity node attributes, if they exist) are structured, and the operational knowledge graph is constructed based on the structured entity nodes, entity node relationships (and entity node attributes, if they exist).

[0053] In some examples, an operational knowledge graph can be built based on historical operational data of utilities and energy management.

[0054] Figure 5 Another example flowchart of an operational knowledge graph construction process 500 according to an embodiment of this specification is shown.

[0055] like Figure 5 As shown in Figure 510, historical operational data for utilities and energy management are obtained.

[0056] In step 520, entity nodes, entity node relationships (and entity node attributes, if they exist), are obtained from historical operational data. For example, correlation modeling between KPIs can be performed based on historical operational data, and statistical analysis and machine learning methods can be used to identify causal relationships between KPIs.

[0057] In 530, an operational knowledge graph is constructed based on the acquired entity nodes, entity node relationships (and entity node attributes, if they exist).

[0058] After constructing the operational knowledge graph as described above, KPI anomaly analysis can be performed based on the operational knowledge graph.

[0059] Figure 6 An example flowchart of a KPI anomaly analysis method 600 according to an embodiment of this specification is shown.

[0060] like Figure 6As shown in section 610, after obtaining the set of abnormal KPIs in the operation of public works and energy management, it is determined whether there are any data source anomalies. The set of abnormal KPIs can be obtained in batches, for example, from an abnormal KPI pool. The abnormal KPI pool summarizes the abnormal KPIs monitored by the operation monitoring device. Data source anomalies can be determined through monitoring by the operation monitoring device.

[0061] In step 620, in response to the existence of a data source anomaly, a root cause query for the data source anomaly is performed in the operational knowledge graph to identify the first anomalous KPI from the set of anomalous KPIs whose root cause is the data source anomaly.

[0062] Figure 7 An example flowchart of a data source root cause determination process 700 according to an embodiment of this specification is shown.

[0063] like Figure 7 As shown in Figure 710, the system queries the operational knowledge graph to determine if a third abnormal KPI exists in the abnormal KPI set, with its computational basis originating from data generated by an abnormal data source. If no third abnormal KPI exists with its computational basis originating from data generated by an abnormal data source, then it is considered that there are no abnormal KPIs whose root cause is an abnormal data source.

[0064] In 720, when there is a third abnormal KPI whose computational basis originates from data generated by an abnormal data source, query the operational knowledge graph for a fourth abnormal KPI that has an upstream relationship with the third abnormal KPI.

[0065] In 730, the third and fourth abnormal KPIs were identified as the first abnormal KPI whose root cause was the abnormality of the data source.

[0066] Back Figure 6 In step 630, for the second abnormal KPI in the abnormal KPI set, a root cause query based on upstream and downstream relationships is performed in the operational knowledge graph to determine the first KPI entity that is the root cause of each second abnormal KPI. The second abnormal KPIs include the remaining abnormal KPIs after removing the first abnormal KPI from the abnormal KPI set.

[0067] For the second abnormal KPI, upstream and downstream relationships are established within the operational knowledge graph. If the KPI entity corresponding to the second abnormal KPI has no downstream KPI entities, or if it has downstream relationships with other KPI entities but all of its downstream KPI entities are normal, then the second abnormal KPI entity is identified as the root cause of the second abnormal KPI. If the KPI entity corresponding to the second abnormal KPI has downstream relationships with other KPI entities and one of its downstream KPI entities is abnormal, then its downstream abnormal KPI entity is identified as a possible root cause. Subsequently, the same downstream abnormal KPI entity identification process is repeated for each downstream abnormal KPI entity until all downstream entities are normal KPI entities. Finally, all identified downstream abnormal KPI entities are determined as the root cause of the second abnormal KPI.

[0068] Optionally, the entity node relationships in the operational knowledge graph also include causal relationships between KPI entity nodes. Furthermore, causal relationship queries can be performed in the operational knowledge graph to determine the root cause analysis results of abnormal KPIs.

[0069] Figure 8 An example flowchart of a KPI root cause determination process 800 according to an embodiment of this specification is shown.

[0070] like Figure 8 As shown in 810, a causal relationship query is performed in the operational knowledge graph to identify the second KPI entity that has a causal relationship with the first KPI entity that is the root cause of each second abnormal KPI.

[0071] In section 820, semantic analysis is performed on the first KPI entity, which has a corresponding second KPI entity (either normal or nonexistent), based on the KPI knowledge base. For example, the first KPI entity, which has a corresponding second KPI entity (either normal or nonexistent), can be provided to a large language model for semantic analysis based on the KPI knowledge base.

[0072] In step 830, based on the semantic analysis results of the abnormal second KPI entity and / or the corresponding second KPI entity, the normal or non-corresponding first KPI entity is generated to generate the root cause analysis results of each abnormal second KPI.

[0073] Optionally, in some examples, after determining the root cause analysis results of the data source for the first abnormal KPI and the root cause analysis results of the second abnormal KPI, an operational anomaly diagnostic report can be generated based on the root cause analysis results of the first and second abnormal KPIs. The generated operational anomaly diagnostic report includes the anomaly attribution analysis results of the KPI anomalies. For example, the generated operational anomaly diagnostic report includes which abnormal KPIs were caused by which data source anomalies, and which abnormal KPIs were caused by which KPI entity anomalies. In some examples, entity nodes also include entity attributes. Accordingly, the operational anomaly diagnostic report may also include the entity attributes of the abnormal KPI entities corresponding to the abnormal KPIs.

[0074] In some examples, the generated operational anomaly diagnostic report may also include measures to address KPI anomalies. For instance, if the KPI anomaly is caused by a data source anomaly, the report would restore the anomaly data source to normal. If the KPI anomaly is caused by another anomaly KPI entity, that other anomaly KPI entity would be identified as an improvement target.

[0075] In some examples, when generating an operational anomaly diagnostic report based on the root cause analysis results of the first and second anomalous KPIs, the root cause analysis results of the first and second anomalous KPIs can be provided to the large language model to generate a structured operational anomaly diagnostic report. That is, the operational anomaly diagnostic report has a fixed format, and the output results of the large language model are filled into the corresponding fixed positions in the operational diagnostic report. In some examples, the root cause analysis results of the first and second anomalous KPIs and the root cause analysis process can also be provided to the large language model to generate an operational anomaly diagnostic report with a structured data structure.

[0076] Figure 9 A schematic diagram illustrating an example of a KPI anomaly root cause analysis process according to an embodiment of this specification is shown. Figure 9 The example uses the cost analysis of public works projects to illustrate the root cause analysis process for KPI anomalies.

[0077] exist Figure 9The example demonstrates generating a utility and energy management operations knowledge graph based on the user's operational knowledge system. In this exemplary operations knowledge graph, the business scenario entity node is utility cost analysis, and the KPI entity nodes include total utility cost, variable cost of self-produced utilities, variable cost of purchased utilities, variable steam cost, fuel cost for steam production, boiler feedwater cost for steam production, total fuel consumption for steam production, purchased fuel cost for steam production, the ratio of boiler feedwater, steam condensate, and demineralized water makeup water, fuel consumption for steam production in department A, unit price of purchased natural gas, and the difference between planned and actual natural gas quantities. The operations knowledge graph can also include data source entity nodes, such as meter P1, SAP, PIMS (not shown), etc. The upstream and downstream relationships between KPI entity nodes include: the KPI entity node "Variable Cost of Self-Produced Utilities" is a downstream indicator of the KPI entity node "Total Cost of Utilities". The causal relationships between KPI entity nodes include: the ratio of boiler feedwater, steam condensate, and demineralized water makeup water in the KPI entity node "Boiler feedwater cost for steam generation" is a possible cause of the KPI entity node "Difference between planned and actual natural gas volume" is a possible cause of the KPI entity node "Abnormal unit price of purchased natural gas," etc. When constructing the operational knowledge graph, users primarily rely on the front-end interface to input entity nodes, entity node relationships, and entity node attributes in a specified format. In addition, users can provide relevant operational documents and operating procedures, and use LLM (Local Management Model) for semantic extraction to automatically identify KPI definitions and business affiliations to supplement the operational knowledge graph.

[0078] When conducting public works and energy management operations, multi-dimensional anomaly detection is implemented. For example, real-time anomaly monitoring is performed on data source and steam cost-related KPIs. Anomaly detection includes data source anomalies, single KPI anomalies, and combined anomalies of multiple KPIs. For instance, "total fuel consumption for steam production" or "fuel consumption for steam production in Department A" is often difficult to determine directly through a single KPI anomaly. In such cases, it is necessary to perform joint analysis of these KPIs with related KPIs such as "production plan" and "unit load." When a significant deviation occurs between the trend of fuel consumption for steam production and the production plan or unit operating load, it is determined to be a combined anomaly of multiple KPIs.

[0079] The detected abnormal KPIs will be aggregated into an abnormal KPI pool for subsequent root cause analysis and intelligent diagnosis of KPI anomalies.

[0080] Figure 9The example illustrates an example of an abnormal KPI entity when the utility costs KPI is abnormally high. Through anomaly detection methods such as underlying data reading anomalies, KPI trend anomaly detection, KPI threshold detection, and multi-KPI joint anomaly detection, abnormal KPIs such as "Abnormally high self-produced utility costs KPI," "Abnormally high steam variable costs," "Abnormally high fuel costs for steam production," "Abnormally high total fuel consumption for steam production," "Abnormally high externally purchased fuel costs for steam production," "Abnormally high unit price of externally purchased natural gas," "Fuel consumption for steam production by Department A," and "Abnormal meter reading P1" are aggregated into an abnormal KPI pool.

[0081] For the seven abnormal KPIs mentioned above, a data source anomaly detection was first performed, which identified the abnormal KPI "Abnormal Meter P1 Reading" as having an abnormal data source. After identifying the KPI "Abnormal Meter P1 Reading" with an abnormal data source, a graph query was performed in the operational knowledge graph. It was found that the KPI "Increased Fuel Consumption of Steam Production in Department A," which was directly calculated based on the abnormal meter P1 reading, was the root cause of the abnormality of the KPI "Increased Fuel Consumption of Steam Production in Department A" and its upstream abnormal KPI entities "Variable Cost of Self-Produced Utilities," "Variable Cost of Steam," "Fuel Cost Used for Steam Production," and "Total Fuel Consumption Used for Steam Production." These abnormal KPIs were then removed from the abnormal KPI pool.

[0082] Next, for the remaining abnormal KPIs in the abnormal KPI pool, "Abnormal increase in purchased fuel cost for steam production" and "Abnormal increase in purchased natural gas unit price," upstream and downstream relationships were queried. The KPI entity "purchased natural gas unit price" has a downstream relationship with the KPI entities "purchased fuel cost for steam production," "fuel cost for steam production," and "steam variable cost," and this KPI entity has no other downstream KPIs still in the abnormal KPI pool. Therefore, it was determined that the KPI entity "increased purchase natural gas unit price" is the root cause of the KPI entity "increased steam variable cost." Then, for the KPI entity identified as the root cause, a graph query was performed in the operational knowledge graph to find its corresponding causal relationship. The entity identified as the root cause was the KPI entity "purchased natural gas unit price." Furthermore, in the defined cost analysis scenario, the KPI entity "Purchased Natural Gas Unit Price" and the KPI entity "Difference Between Planned and Actual Natural Gas Quantity" have a causal relationship. Therefore, the improvement plan for the abnormal KPI "Purchased Fuel Cost for Steam Production" is determined to be "Reducing the Difference Between Planned and Actual Natural Gas Quantity". The improvement plan for the abnormal KPI "Purchased Fuel Cost for Steam Production" is also indicated to be "Reducing the Difference Between Planned and Actual Natural Gas Quantity".

[0083] After identifying the root causes of each abnormal KPI, the root cause analysis process is summarized using a large language model, and a structured diagnostic report is automatically generated to help users quickly understand the causes of the problem and formulate response strategies.

[0084] If users wish to gain a deeper understanding of certain KPIs, they can interact with the large language model through natural language to query information such as the historical trends, influencing factors, and upstream and downstream KPIs for specific KPIs. The large language model can automatically generate visual reports based on the query content to improve the efficiency of data insight. For example, users can view the real-time status and historical trends of the KPI entity steam variable cost.

[0085] In some examples, KPI query requests initiated by users (e.g., via a user interface) can be retrieved. These requests may include at least one of the following: historical trends of the KPI, influencing factors, and upstream / downstream KPIs. Upon receiving a KPI query request, the query results are retrieved from the operational knowledge graph. Then, a KPI query report is generated based on these results and provided to the user.

[0086] By using the above methods, an operational knowledge graph can be constructed to reflect the semantic relationships between KPIs and business scenarios, between KPIs, and between KPIs and underlying data. Based on the operational knowledge, KPI anomaly root cause analysis can be performed to quickly locate the root cause of KPI anomalies and improve the accuracy of KPI anomaly root cause analysis.

[0087] Figure 10 An example block diagram of an anomaly KPI analysis device 1000 according to an embodiment of this specification is shown. Figure 10 As shown, the abnormal KPI analysis device 1000 includes a data source abnormality determination unit 1010, a data source abnormality root cause determination unit 1020, and a KPI entity root cause determination unit 1030.

[0088] The data source anomaly determination unit 1010 is configured to determine whether a data source anomaly exists after acquiring a set of abnormal KPIs from the utility and energy management operation process. The operation of the data source anomaly determination unit 1010 can be referenced above. Figure 6 The operation described in 610.

[0089] The data source anomaly root cause determination unit 1020 is configured to perform a data source anomaly root cause query in the operational knowledge graph in response to the existence of a data source anomaly, in order to determine the first anomaly KPI whose root cause is a data source anomaly from the set of anomaly KPIs. The operation of the data source anomaly root cause determination unit 1020 can be referenced above. Figure 6 The operation described in 620.

[0090] The KPI entity root cause determination unit 1030 is configured to perform a KPI entity root cause query based on upstream and downstream relationships in the operational knowledge graph for the second abnormal KPI in the abnormal KPI set, in order to determine the first KPI entity that is the root cause of each second abnormal KPI. The second abnormal KPIs include the remaining abnormal KPIs in the abnormal KPI set after removing the first abnormal KPI. The operation of the KPI entity abnormal root cause determination unit 1030 can be referred to the above. Figure 6 The operation described in 630.

[0091] In some examples, the abnormal KPI analysis device 1300 may further include a root cause result analysis unit (not shown). The result analysis unit is configured to perform a causal relationship query in an operational knowledge graph to identify second KPI entities that have a causal relationship with the first KPI entities that are the root causes of each second abnormal KPI; perform semantic analysis based on a KPI knowledge base for the first KPI entities that are normal or do not have corresponding second KPI entities for the corresponding second KPIs; and generate root cause analysis results for each second abnormal KPI based on the semantic analysis results of the abnormal second KPI entities and / or the first KPI entities that are normal or do not have corresponding second KPI entities. Furthermore, the root cause result analysis unit also identifies the abnormal data source corresponding to the first abnormal KPI as the root cause analysis result of the first abnormal KPI.

[0092] In some examples, the abnormal KPI analysis device 1300 may also include a diagnostic report generation unit (not shown). The diagnostic report generation unit is configured to generate an operational anomaly diagnostic report based on the root cause analysis results of a first abnormal KPI and a second abnormal KPI. The generated operational anomaly diagnostic report includes the anomaly attribution analysis results of the KPI anomaly. In some examples, the generated operational anomaly diagnostic report may also include countermeasures for the KPI anomaly.

[0093] In some examples, the diagnostic report generation unit can provide the root cause analysis results of the first and second anomalous KPIs to the large language model to generate an operational anomaly diagnostic report. In other examples, the diagnostic report generation unit can also provide the root cause analysis results of the first and second anomalous KPIs, as well as the root cause analysis process, to the large language model to generate an operational anomaly diagnostic report with a structured data structure.

[0094] As per the above reference Figures 1 to 10 This specification describes a KPI anomaly analysis method, a KPI anomaly analysis apparatus, and a KPI anomaly analysis system according to embodiments thereof. The aforementioned KPI anomaly analysis apparatus can be implemented in hardware, software, or a combination of both.

[0095] Figure 11A schematic diagram of an example of a KPI anomaly analysis device 1100 implemented using a computer system according to an embodiment of this specification is shown. Figure 11 As shown, the KPI anomaly analysis device 1100 may include at least one processor 1110, a memory (e.g., non-volatile memory) 1120, a RAM 1130, and a communication interface 1140, and the at least one processor 1110, memory 1120, RAM 1130, and communication interface 1140 are connected together via a bus 1160. The at least one processor 1110 executes at least one computer-readable instruction (i.e., the elements implemented in software described above) stored or encoded in the memory.

[0096] In one embodiment, computer-executable instructions are stored in memory, which, when executed, cause at least one processor 1110 to: after acquiring a set of abnormal KPIs in the operation of utility and energy management, determine whether there is a data source abnormality; in response to the existence of a data source abnormality, perform a root cause query of the data source abnormality in an operational knowledge graph to identify a first abnormal KPI from the set of abnormal KPIs whose root cause is a data source abnormality; and for a second abnormal KPI in the set of abnormal KPIs, perform a root cause query of the KPI entity based on upstream and downstream relationships in the operational knowledge graph to identify a first KPI entity as the root cause of each second abnormal KPI, the second abnormal KPIs including the remaining abnormal KPIs in the set of abnormal KPIs after removing the first abnormal KPI, wherein the operational knowledge graph includes entity nodes and entity node relationships, the entity nodes including KPI entity nodes, data source entity nodes and business scenario entity nodes, and entity node relationships used to reflect the logical relationships between entity nodes, the logical relationships including the affiliation relationship of KPI entity nodes to business scenario entity nodes, the data source relationship of KPI entity nodes to data source entity nodes, and the upstream and downstream relationships between KPI entity nodes.

[0097] It should be understood that the computer-executable instructions stored in memory, when executed, cause at least one processor 1110 to perform the above-described combinations in the various embodiments of this specification. Figures 1-10 The description includes various operations and functions.

[0098] According to one embodiment, a program product, such as a machine-readable medium (e.g., a non-transitory machine-readable medium), is provided. The machine-readable medium may have instructions (i.e., the elements implemented in software as described above), which, when executed by a machine, cause the machine to perform the above-described combinations of the various embodiments of this specification. Figures 1-10The various operations and functions described. Specifically, a system or apparatus equipped with a readable storage medium storing software program code that implements the functions of any of the embodiments described above, and enabling the computer or processor of the system or apparatus to read and execute the instructions stored in the readable storage medium.

[0099] In this case, the program code itself, which can be read from a readable medium, can perform the functions of any of the above embodiments. Therefore, the machine-readable code and the readable storage medium storing the machine-readable code constitute a part of the present invention.

[0100] Examples of readable storage media include floppy disks, hard disks, magneto-optical disks, optical disks (such as CD-ROM, CD-R, CD-RW, DVD-ROM, DVD-RAM, DVD-RW, DVD-RW), magnetic tapes, non-volatile memory cards, and ROMs. Alternatively, program code can be downloaded from a server computer or the cloud via a communication network.

[0101] According to one embodiment, a computer program product is provided, the computer program product including a computer program, which, when executed by a processor, causes the processor to perform the above-described combinations of the various embodiments of this specification. Figures 1-10 The description includes various operations and functions.

[0102] Those skilled in the art will understand that the various embodiments disclosed above can be modified and varied without departing from the spirit of the invention. Therefore, the scope of protection of this invention should be defined by the appended claims.

[0103] It should be noted that not all steps and units in the above process and system structure diagrams are mandatory; some steps or units can be omitted according to actual needs. The execution order of each step is not fixed and can be determined as needed. The device structure described in the above embodiments can be a physical structure or a logical structure; that is, some units may be implemented by the same physical entity, or some units may be implemented by multiple physical entities, or they may be jointly implemented by certain components in multiple independent devices.

[0104] In the above embodiments, the hardware units or modules can be implemented mechanically or electrically. For example, a hardware unit, module, or processor may include permanent dedicated circuitry or logic (such as a dedicated processor, FPGA, or ASIC) to perform the corresponding operation. The hardware unit or processor may also include programmable logic or circuitry (such as a general-purpose processor or other programmable processor), which can be temporarily configured by software to perform the corresponding operation. The specific implementation method (mechanical, dedicated permanent circuitry, or temporarily configured circuitry) can be determined based on cost and time considerations.

[0105] The specific embodiments described above with reference to the accompanying drawings are exemplary embodiments, but do not represent all embodiments that can be implemented or fall within the scope of the claims. The term "exemplary" as used throughout this specification means "serving as an example, instance, or illustration" and does not imply that it is "preferred" or "advantageous" compared to other embodiments. Specific details are included to provide an understanding of the described techniques. However, these techniques can be practiced without these specific details. In some instances, well-known structures and apparatuses are shown in block diagram form to avoid obscuring the concepts of the described embodiments.

[0106] The foregoing description of this disclosure is provided to enable any person skilled in the art to implement or use this disclosure. Various modifications to this disclosure will be apparent to those skilled in the art, and the general principles defined herein can be applied to other variations without departing from the scope of this disclosure. Therefore, this disclosure is not limited to the examples and designs described herein, but is consistent with the widest scope of the principles and novel features disclosed herein.

Claims

1. A method for utility and energy management in chemical production, characterized in that, The method includes: After obtaining the set of abnormal key performance indicators in the operation of public works and energy management, determine whether there are any abnormal data sources; In response to the existence of a data source anomaly, a root cause query for the data source anomaly is performed in the operational knowledge graph to determine the first abnormal key performance indicator (KPI) from the set of abnormal KPIs whose root cause is the data source anomaly; and For the second abnormal key performance indicator in the set of abnormal key performance indicators, a root cause query based on upstream and downstream relationships is performed on the operational knowledge graph to determine the first key performance indicator entity that serves as the root cause of each second abnormal key performance indicator. The second abnormal key performance indicators include the remaining abnormal key performance indicators in the set of abnormal key performance indicators after removing the first abnormal key performance indicator. The operational knowledge graph includes entity nodes and entity node relationships. The entity nodes include key performance indicator entity nodes, data source entity nodes, and business scenario entity nodes. The entity node relationships are used to reflect the logical relationships between entity nodes. The logical relationships include the affiliation of key performance indicator entity nodes with respect to business scenario entity nodes, the data source relationship of key performance indicator entity nodes with respect to data source entity nodes, and the upstream and downstream relationships between key performance indicator entity nodes.

2. The method as described in claim 1, characterized in that, The step of performing a root cause query on the data source anomaly in the operational knowledge graph to determine the first abnormal key performance indicator from the set of abnormal key performance indicators as the root cause of data source anomaly includes: In the operational knowledge graph, query whether there is a third abnormal key performance indicator in the abnormal key performance indicator set whose calculation basis is derived from data generated by abnormal data sources. When a third anomalous key performance indicator (KPI) exists, its computational basis originates from data generated by an anomalous data source; in the operational knowledge graph, a fourth anomalous KPI with an upstream relationship to the third anomalous KPI is queried; and The third and fourth abnormal key performance indicators are identified as the first abnormal key performance indicators whose root cause is the abnormality of the data source.

3. The method as described in claim 1 or 2, characterized in that, The entity node relationship also includes the causal relationship between key performance indicator entity nodes, and the method further includes: A causal relationship query is performed in the operational knowledge graph to identify second key performance indicator entities that have a causal relationship with the first key performance indicator entities that are the root causes of each second abnormal key performance indicator. For first key performance indicator entities that have a corresponding second key performance indicator entity (either normal or nonexistent), semantic analysis is performed based on the key performance indicator knowledge base; and Based on the semantic analysis results of the abnormal second key performance indicator entities and / or the corresponding second KPI entities, and the normal or non-corresponding first key performance indicator entities, root cause analysis results are generated for each abnormal second key performance indicator.

4. The method as described in claim 3, characterized in that, The entity node also includes entity attributes, and the method further includes: Based on the root cause analysis results of the first and second abnormal key performance indicators, an operational anomaly diagnostic report is generated, which includes the anomaly attribution analysis results of the key performance indicator anomalies.

5. The method as described in claim 4, characterized in that, The operational anomaly diagnostic report also includes measures to address anomalies in key performance indicators.

6. The method as described in claim 4, characterized in that, The generation of the operational anomaly diagnostic report based on the root cause analysis results of the first and second abnormal key performance indicators includes: The root cause analysis results and process of the first and second abnormal key performance indicators are provided to the large language model to generate an operational anomaly diagnostic report with a structured data structure.

7. The method as described in claim 1 or 2, characterized in that, The operational knowledge graph is constructed based on at least one of the following knowledge graph construction methods: The operational knowledge graph is constructed based on the entity nodes, entity node relationships, and entity node attributes input by the user in a specified format. The operational knowledge graph is constructed by performing semantic parsing based on a large language model on the user's description of operational business scenarios, data source descriptions, and key performance indicator definitions in natural language input to obtain entity nodes, entity node relationships, and entity node attributes. The operational knowledge graph is constructed by performing semantic extraction based on a large language model on the unstructured text format of the operational specification document input by the user to obtain entity nodes, entity node relationships and entity node attributes. The operational specification document includes operational documents, operating procedures and key performance indicator definition manuals. and The operational knowledge graph is constructed based on the historical operational data of the aforementioned public works and energy management system.

8. The method as described in claim 1, characterized in that, The method further includes: Obtain a key performance indicator (KPI) query request initiated by a user, wherein the KPI query request includes at least one of the following: historical trend of the KPI, influencing factors, and upstream and downstream KPIs. Query the results of the key performance indicator query request in the operational knowledge graph; and A key performance indicator (KPI) query report is generated based on the query results of the KPI query request and provided to the user.

9. A utility and energy management device for chemical production, characterized in that, The device includes: The data source anomaly determination unit is configured to determine whether there is a data source anomaly after obtaining the set of abnormal key performance indicators in the process of utility and energy management operations; The data source anomaly root cause determination unit is configured to, in response to the existence of a data source anomaly, perform a data source anomaly root cause query in the operational knowledge graph to determine a first abnormal key performance indicator (KPI) from the set of abnormal KPIs whose root cause is the data source anomaly; and The Key Performance Indicator (KPI) entity root cause determination unit is configured to perform a KPI entity root cause query based on upstream and downstream relationships in the operational knowledge graph for the second abnormal KPI in the abnormal KPI set, in order to determine the first KPI entity as the root cause of each second abnormal KPI. The second abnormal KPI includes the remaining abnormal KPIs in the abnormal KPI set after removing the first abnormal KPI. The operational knowledge graph includes entity nodes and entity node relationships. The entity nodes include key performance indicator entity nodes, data source entity nodes, and business scenario entity nodes. The entity node relationships are used to reflect the logical relationships between entity nodes. The logical relationships include the affiliation of key performance indicator entity nodes with respect to business scenario entity nodes, the data source relationship of key performance indicator entity nodes with respect to data source entity nodes, and the upstream and downstream relationships between key performance indicator entity nodes.

10. A utility and energy management system for chemical production, characterized in that, The system includes: An operational knowledge graph construction device is configured to construct an operational knowledge graph of the utilities and energy management operations in chemical production. The operation monitoring device is configured to detect data source anomalies and key performance indicator anomalies in the operation of the aforementioned utility works and energy management processes; and The utility and energy management apparatus for chemical production as described in claim 9.

11. A utility and energy management device for chemical production, characterized in that, The device includes: At least one processor; A memory coupled to the at least one processor; and A computer program stored in the memory, which is executed by the at least one processor to implement the method for utility and energy management for chemical production as described in any one of claims 1 to 8.

12. A computer-readable storage medium, characterized in that, It stores executable instructions that, when executed, cause the processor to perform the method for utilities and energy management for chemical production as described in any one of claims 1 to 8.

Images