Heating network operation and maintenance method and system based on hybrid knowledge graph

Abstract

This invention provides a method and system for the operation and maintenance of a heating network based on a hybrid knowledge graph. The method includes: collecting multi-source operation and maintenance information, including equipment parameters, real-time sensor data, historical fault records, and expert operation and maintenance rules; preprocessing the multi-source operation and maintenance information; performing entity recognition, relation extraction, and attribute definition on the preprocessed multi-source operation and maintenance information, and combining it with the expert operation and maintenance rules to construct triples and store them in a graph database, forming a hybrid knowledge graph; receiving operation and maintenance reasoning requests; and performing rule-based reasoning on the operation and maintenance reasoning requests based on the hybrid knowledge graph: if the rule reasoning matches successfully, the rule reasoning result is output; if the rule reasoning fails to match, graph reasoning is performed, and the corresponding graph reasoning result is output when the reasoning result is obtained. This invention significantly improves the intelligence level of heating network operation and maintenance and enhances the accuracy and efficiency of fault diagnosis.

Description

Technical Field

[0001] This invention relates to the field of smart heating technology, specifically to a heating network operation and maintenance method and system based on a hybrid knowledge graph. Background Technology

[0002] Heating network operation and maintenance (O&M) is a core component of ensuring the stable operation of heating systems. Its purpose is to ensure the efficient and reliable operation of the heating system through monitoring equipment status, diagnosing faults, and making maintenance decisions. Traditional heating network O&M relies on a vast amount of O&M knowledge, which is scattered across various media, including equipment parameter manuals, real-time monitoring sensor data, historical fault records, and expert experience. The dispersed and unstructured nature of this O&M knowledge leads to difficulties in management and use, especially in complex fault diagnosis and handling processes, where fragmented information often increases the risk of misjudgments and omissions.

[0003] Currently, many heating network operation and maintenance systems rely on traditional knowledge management methods and single reasoning mechanisms. While traditional knowledge graph technology plays a role in achieving structured integration of knowledge, its reasoning process primarily depends on rule-based reasoning, which typically cannot simultaneously meet the needs of rapid response to simple faults and in-depth analysis of complex ones. Furthermore, these existing reasoning methods largely rely on human experience to verify the correlations between devices and levels, resulting in time-consuming and inaccurate reasoning processes. This is especially problematic when facing complex faults that cross devices and levels, often failing to efficiently identify and resolve issues.

[0004] Therefore, under current technological conditions, the operation and maintenance of heating networks still faces problems such as low efficiency of knowledge integration, long fault diagnosis time, and poor reasoning accuracy. There is an urgent need for an innovative method to optimize operation and maintenance decision support and improve the efficiency and accuracy of fault diagnosis. Summary of the Invention

[0005] This invention provides a method and system for thermal network operation and maintenance based on a hybrid knowledge graph. By combining multiple operation and maintenance data sources and employing a hybrid reasoning mechanism, it overcomes the shortcomings of existing technologies in thermal network operation and maintenance. Existing technologies mainly rely on a single reasoning method, making it difficult to simultaneously meet the needs of rapid response to simple faults and in-depth tracing of complex faults. Traditional knowledge graphs often focus on static knowledge storage, with the reasoning process relying on manual experience matching, and the fragmented information leads to low efficiency and a high risk of errors in fault diagnosis.

[0006] This invention employs a hybrid knowledge graph-based operation and maintenance method that deeply integrates rule-based reasoning and graph reasoning. It can select the appropriate reasoning method based on the varying complexity of fault phenomena, providing accurate fault diagnosis and maintenance decisions. In simple fault cases, rule-based reasoning enables rapid matching, reducing manual intervention. In complex fault scenarios, the graph reasoning module performs multi-hop path searching to uncover deep causal relationships, improving the depth and accuracy of fault tracing. Furthermore, the knowledge graph of this invention supports an incremental update mechanism, enabling real-time collection and updating of new operation and maintenance data, ensuring the knowledge base remains up-to-date and guaranteeing long-term knowledge accumulation and optimization. Through these innovations, this invention effectively solves the problems of insufficient knowledge integration, low reasoning efficiency, and inaccurate fault diagnosis in existing technologies.

[0007] In a first aspect, the present invention provides a hot network operation and maintenance method based on a hybrid knowledge graph, characterized in that the method includes: Collect multi-source operation and maintenance information, including equipment parameters, real-time sensor data, historical fault records, and expert operation and maintenance rules; The multi-source operation and maintenance information is preprocessed; Entity identification, relation extraction, and attribute definition are performed on the preprocessed multi-source operation and maintenance information. Combined with the expert operation and maintenance rules, triples are constructed and stored in the graph database to form a hybrid knowledge graph. Receive operation and maintenance inference requests; Based on the hybrid knowledge graph, rule-based reasoning is performed on the operation and maintenance reasoning request: if the rule reasoning is successfully matched, the rule reasoning result is output; if the rule reasoning is not successfully matched, graph reasoning is performed, and the corresponding graph reasoning result is output when the reasoning result is obtained.

[0008] Secondly, the present invention also provides a heat network operation and maintenance system based on a hybrid knowledge graph, characterized in that the system comprises: The data acquisition module is used to collect multi-source operation and maintenance information, including equipment parameters, real-time sensor data, historical fault records, and expert operation and maintenance rules. The preprocessing module is used to preprocess the multi-source operation and maintenance information; The construction module is used to perform entity recognition, relation extraction and attribute definition on the preprocessed multi-source operation and maintenance information, and combine it with the expert operation and maintenance rules to construct triples and store them in the graph database to form a hybrid knowledge graph. The receiving module is used to receive operation and maintenance inference requests; The reasoning module is used to perform rule-based reasoning on the operation and maintenance reasoning request based on the hybrid knowledge graph: if the rule reasoning is successfully matched, the rule reasoning result is output; if the rule reasoning is not successfully matched, graph reasoning is performed, and the corresponding graph reasoning result is output when the reasoning result is obtained.

[0009] The present invention provides a heating network operation and maintenance method and system based on hybrid knowledge graphs: First, unlike traditional fragmented knowledge management methods, this invention integrates multi-source operation and maintenance information (such as equipment parameters, sensor data, historical fault records, and expert operation and maintenance rules) through a structured knowledge graph. It can clearly define entities, attributes, and relationships, and store them in a graph database in the form of triples, which greatly improves data integration and query efficiency, and avoids the inefficiency caused by information fragmentation.

[0010] Secondly, this invention employs a combination of rule-based reasoning and graph reasoning, enabling the selection of an appropriate reasoning mechanism based on the complexity of the fault. For simple faults, rule-based reasoning can respond quickly and provide fault causes and maintenance solutions; for complex faults, the graph reasoning module can deeply analyze the relationships between entities, trace the relationship paths between devices, quickly uncover potential fault causes, and provide accurate source information, greatly improving the accuracy and efficiency of fault diagnosis. Especially in complex fault scenarios involving multiple devices and levels, it can achieve efficient fault tracing, shorten fault response time, and improve maintenance accuracy.

[0011] Third, this invention supports an incremental update mechanism for hybrid knowledge graphs, which can acquire and process new operation and maintenance data in real time, ensuring that the knowledge base of the operation and maintenance system is always up-to-date and avoiding reasoning errors caused by outdated data, thereby improving the timeliness and accuracy of operation and maintenance decisions.

[0012] Fourth, this invention can generate a visual file of the reasoning process, helping operations and maintenance personnel to better understand the reasoning process and its results, and improve the transparency and credibility of decision-making. This not only helps improve the work efficiency of operations and maintenance personnel, but also enhances the transparency of operations and maintenance decisions and reduces the risks caused by misjudgment or misunderstanding.

[0013] Fifth, through the knowledge graph and hybrid reasoning engine of this invention, the heating network operation and maintenance system can provide intelligent fault diagnosis and maintenance decision support when facing a variety of complex situations, reducing manual intervention and improving the scientificity and reliability of operation and maintenance decisions.

[0014] Sixth, when constructing the hybrid knowledge graph, after performing entity identification, relation extraction, and attribute definition on the preprocessed multi-source operation and maintenance information, the system further accelerates the fault reasoning process and data query by establishing entity attribute indexes and relation path indexes. This significantly improves the intelligence level of the heating network operation and maintenance system, ensuring rapid response and accurate decision-making when facing complex faults. Attached Figure Description

[0015] To more clearly illustrate the technical solutions in the embodiments of the present invention, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0016] Figure 1 This is a flowchart of the heat network operation and maintenance method based on hybrid knowledge graph provided in the embodiments of the present invention; Figure 2a and 2b The present invention provides a schematic diagram of heat network operation and maintenance based on a hybrid knowledge graph. Figure 3 This is a block diagram of a heat network operation and maintenance system based on a hybrid knowledge graph, provided in an embodiment of the present invention. Detailed Implementation

[0017] The technical solution of the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. Invention Overview As mentioned above, this invention provides a method and system for the operation and maintenance of a heat network based on a hybrid knowledge graph. By constructing a hybrid knowledge graph and combining rule-based reasoning and graph reasoning mechanisms, the intelligent level of heat network operation and maintenance is greatly improved, and the accuracy and efficiency of fault diagnosis are enhanced.

[0019] Exemplary methods Figure 1 This is a flowchart of a heat network operation and maintenance method based on a hybrid knowledge graph provided in an embodiment of the present invention. This embodiment includes the following steps: S101: Collect multi-source operation and maintenance information, including equipment parameters, real-time sensor data, historical fault records, and expert operation and maintenance rules.

[0020] Among them, equipment parameters are used to describe the static attributes and design specifications of the heating network operating equipment, including equipment model, rated power, rated flow, pressure rating and installation location, etc., to support equipment-level operation analysis.

[0021] Real-time sensor data is used to reflect the dynamic status information of the heating network system during operation, including temperature, pressure, flow rate, energy consumption, current and voltage, etc. It is acquired through a real-time acquisition interface to ensure the timeliness and continuity of the data.

[0022] Historical fault records are used to accumulate fault events that have occurred during the operation of the heating network and their corresponding handling results, including fault occurrence time, fault type, scope of impact, root cause analysis, maintenance measures and effect evaluation, which are used to support the confidence calculation of the fault mode mining and reasoning process.

[0023] Expert operation and maintenance rules are summarized by operation and maintenance personnel based on practical experience. They are usually presented in the form of conditions and results, covering the judgment conditions of abnormal equipment status, possible causes of failure and recommended maintenance solutions, providing prior knowledge for subsequent rule reasoning.

[0024] This invention comprehensively collects multi-source operation and maintenance information, not only gathering static parameters of heating network equipment but also integrating real-time sensor data, historical fault records, and expert operation and maintenance rules to achieve the fusion of multi-source information. Compared with existing technologies that rely solely on sensor data or human experience, this invention, through the integration of static information, dynamic operational data, and empirical rules, enables more comprehensive and accurate modeling and analysis of the heating network system, thus providing a solid data foundation for subsequent knowledge graph construction and operation and maintenance reasoning.

[0025] S102: Preprocess the multi-source operation and maintenance information.

[0026] Specifically, the following preprocessing operations can be performed on the collected equipment parameters, real-time sensor data, historical fault records, and expert operation and maintenance rules: Data cleaning involves processing outliers and missing values ​​in real-time sensor data, such as by using interpolation to fill in missing data or by using statistical methods to identify and remove outlier data points; merging redundant and duplicate entries in historical fault records to avoid knowledge redundancy; and performing semantic consistency checks on expert operation and maintenance rules to remove conflicting or logically incomplete rules.

[0027] Data standardization involves unifying the unit conversion and formatting of equipment parameters and operating indicators collected from different data sources, such as standardizing temperature to degrees Celsius and pressure to megapascals; and performing natural language normalization processing on operation and maintenance records and rule texts, including word segmentation, part-of-speech tagging, and synonym normalization, to facilitate subsequent entity recognition and relation extraction.

[0028] Data alignment and fusion map device parameters and real-time sensor data according to device identifiers to achieve alignment of static and dynamic data; bind historical fault records with corresponding devices and time periods to achieve time-series management; and establish mapping relationships between expert operation and maintenance rules and device types and operating conditions to ensure their availability in the knowledge graph.

[0029] Data quality assessment evaluates the completeness, consistency, timeliness, and accuracy of preprocessed data and generates quality labels; low-confidence data is marked for subsequent manual review or rule correction.

[0030] In addition to conventional data cleaning, standardization, and alignment, this invention introduces data quality labeling and confidence management mechanisms. Compared to traditional methods that rely solely on manual cleaning or simple formatting, this invention can assess the completeness, consistency, timeliness, and accuracy of various operational data, and mark or further process low-confidence data, thereby effectively ensuring the reliability and consistency of knowledge sources and avoiding problems caused by low-quality data, thus providing a solid foundation for building high-quality hybrid knowledge graphs.

[0031] S103: Perform entity identification, relation extraction, and attribute definition on the preprocessed multi-source operation and maintenance information, and combine it with the expert operation and maintenance rules to construct triples and store them in the graph database to form a hybrid knowledge graph.

[0032] The hybrid knowledge graph includes the pipeline network topology (physical connection relationships), equipment failure modes, and operation and maintenance rules.

[0033] By simultaneously describing the physical connections between devices, possible fault types and their modes, and corresponding operation and maintenance rules, a complete knowledge system for heating network operation and maintenance can be formed. This comprehensive knowledge coverage enables the simultaneous use of topological information to determine fault propagation paths during the reasoning process, and the rapid location of problems and provision of solutions by combining fault modes and operation and maintenance rules. Compared to traditional knowledge graphs that only store a single type of information, the hybrid knowledge graph of this invention enables collaborative work between rule-based reasoning and graph reasoning, thereby improving the accuracy of fault diagnosis and the operability of operation and maintenance decisions. In addition, the hybrid knowledge graph facilitates the dynamic updating and maintenance of new devices, fault modes, or operation and maintenance operations, ensuring the integrity and timeliness of the knowledge base and providing reliable knowledge support for intelligent operation and maintenance of heating networks.

[0034] Specifically, the system first automatically identifies core entities related to the operation of the heating network from the pre-processed multi-source operation and maintenance information. These entities include, but are not limited to, heating network equipment (such as heat exchangers, circulating pumps, pipelines, valves, and sensors), fault types (such as leaks, blockages, over-temperature, over-pressure, and sensor failures), and / or operation and maintenance operations (such as repair, replacement, cleaning, and adjusting valve openings). Attributes are defined for each entity, such as the equipment's rated parameters, operating status, and installation location; the severity and probability of the fault; and the execution conditions and scope of application for the operation and maintenance operations.

[0035] Secondly, the relationships between entities are identified. By analyzing multi-source data and expert operation and maintenance rules, the semantic relationships between different entities are automatically identified, including but not limited to the connection relationship between equipment (such as pump A connecting to water supply pipe B), the association relationship between faults and equipment (such as the failure mode of a circulating pump being bearing wear) and / or the correspondence between faults and operation and maintenance operations (such as the maintenance operation for a pipeline leak being replacing the pipe section).

[0036] Finally, entities, attributes, and relationships are organized into triples (subject-predicate-objects) that conform to knowledge graph standards. For example, circulation pump A is connected to water supply pipe B; pipe C experiences a leak; the leak corresponds to the pipe replacement operation. These triples are stored in a graph database to form a semantic, reasoning-enabled, structured hybrid knowledge graph.

[0037] In constructing a hybrid knowledge graph, rule-based knowledge summarized by operations and maintenance experts based on experience is transformed into a structured representation within the knowledge graph that can be stored, retrieved, and reasoned about. Specifically, these expert operations and maintenance rules typically exist in a condition-result form. For example, if the equipment temperature exceeds a threshold, the valve opening needs to be adjusted. Such rules can be transformed into triples or logical constraints using formal modeling methods. The condition part is bound to entity attributes or relationships in the knowledge graph, while the conclusion part is associated with the corresponding operations and maintenance operation or fault cause.

[0038] For example, if the water supply pressure is less than 0.2 MPa and the pump frequency is normal, it can be determined as a pipeline leak, which can be transformed into a logical rule in the knowledge graph to support subsequent reasoning and invocation.

[0039] In summary, this invention transforms expert operation and maintenance rules from traditional textual experience documents into reasonable logical relationships, and directly embeds them into a hybrid knowledge graph in a structured manner. Unlike existing knowledge graphs that can only serve as relationship display tools, the knowledge graph of this invention can not only store operation and maintenance rules, but also actively invoke and participate in reasoning calculations during the reasoning process, thereby significantly improving the intelligence level of the knowledge graph and achieving a functional upgrade from knowledge display to knowledge-driven operation and maintenance.

[0040] It can also establish entity attribute indexes and relation path indexes to accelerate the retrieval and reasoning process in graph databases, and support fast calculation of multi-hop relation paths to ensure real-time reasoning performance under complex graph scales.

[0041] The resulting hybrid knowledge graph can not only intuitively reflect the topology and operating status of heating network equipment, but also combine historical experience and expert knowledge to describe typical failure modes and maintenance strategies, providing a high-quality knowledge foundation for subsequent rule-based reasoning and graph reasoning.

[0042] S104: Receive operation and maintenance inference request.

[0043] Operational reasoning requests typically include the fault symptoms to be diagnosed, relevant equipment information, and necessary operating parameters. These requests can originate from human operational staff, automated alarms from monitoring systems, or other management systems.

[0044] During the receiving process, basic checks can be performed on the operation and maintenance inference requests, such as parameter integrity, format validity, and data validity, to ensure the accuracy and stability of the inference process.

[0045] Upon receiving an operation and maintenance reasoning request, it will parse the request and map the fault phenomenon to the corresponding entity in the hybrid knowledge graph, so as to carry out rule reasoning and graph reasoning in the future.

[0046] S105: Based on the hybrid knowledge graph, perform rule-based reasoning on the operation and maintenance reasoning request: if the rule-based reasoning matches successfully, output the rule-based reasoning result; if the rule-based reasoning does not match successfully, perform graph reasoning and output the corresponding graph reasoning result when the reasoning result is obtained.

[0047] Upon receiving an operation and maintenance reasoning request, reasoning is performed using a hybrid knowledge graph. The reasoning process includes two parts: rule-based reasoning and graph reasoning.

[0048] This invention employs a hybrid reasoning mode that prioritizes rules and supplements them with graph reasoning. For simple faults, the rule-based reasoning module can quickly match triggering conditions and output corresponding fault causes and maintenance solutions, achieving efficient response. For complex faults or faults not covered by rules, the graph reasoning module is activated, using multi-hop relationship path analysis combined with historical fault data for in-depth tracing. Compared to existing single reasoning methods that rely solely on rules or solely on graph algorithms, this invention balances real-time performance and complexity, significantly improving the accuracy and response efficiency of fault diagnosis.

[0049] This invention utilizes multi-hop relationship path analysis combined with historical fault data for in-depth fault tracing. It can uncover potential deep-seated fault causes by tracing multi-level associations among equipment, pipelines, and fault modes, achieving more comprehensive fault localization. Simultaneously, it quantitatively evaluates candidate fault causes using historical fault occurrence frequency and association confidence levels, improving the reliability and scientific rigor of the inference results. This method not only distinguishes between common and complex rare faults, enabling targeted diagnosis and optimized operation and maintenance strategies, but also significantly enhances the intelligence level of heating network operation and maintenance systems, providing reliable and interpretable decision support for complex heating network systems.

[0050] Specifically, key feature parameters of the fault phenomena in the maintenance inference request are extracted, such as equipment operating parameters like temperature, pressure, and flow rate. These key feature parameters are then compared and logically combined with trigger conditions in structured rules extracted from a hybrid knowledge graph. The structured rules are organized in IF-THEN format, with each rule containing a trigger condition, a fault cause, and a corresponding maintenance plan. The IF-THEN format directly reads "If a certain equipment parameter or state meets the condition (IF), then the corresponding fault cause and maintenance operation (THEN)" for automatic matching, eliminating the need for manual judgment. For example, if the temperature > 90℃ (trigger condition), the corresponding fault cause can be identified, and valve opening adjustment (maintenance plan) can be suggested, achieving rapid response and automated decision-making. Furthermore, the rules are clear, facilitating subsequent additions or modifications.

[0051] If the rule-based reasoning matches successfully, the corresponding fault cause and maintenance plan will be output directly to achieve a rapid response.

[0052] If the rule-based reasoning fails to match, the system starts by searching the hybrid knowledge graph for candidate fault cause entities associated with the fault phenomenon entity. If a candidate fault cause entity is found, the system further determines the relationship path between the fault phenomenon entity and the candidate fault cause entity, and calculates the correlation degree of each candidate fault cause entity by combining the historical fault occurrence frequency and relationship path confidence. Based on the calculated correlation degree, the most likely fault cause is selected, and the fault cause and corresponding maintenance plan are output. At the same time, a visualization file of the reasoning process can be generated to display the entity relationship path, correlation degree weight, and historical case references, improving the interpretability of the reasoning results. If no candidate fault cause entity or relationship path is found, the system prompts that the fault phenomenon has not been recorded in the hybrid knowledge graph, and prompts the operation and maintenance personnel to carry out subsequent processing or knowledge base updates.

[0053] As an optional embodiment, the correlation degree can be calculated using a weighted scoring method, which uses the historical occurrence frequency and relationship path confidence of the candidate fault cause entity as the main factors, and performs a weighted sum according to preset weights to obtain the comprehensive score of each candidate fault cause, as shown in the following formula: Score(i) =α* Freq(i) +β* Conf(i) Where Freq(i) represents the historical occurrence frequency of candidate fault cause i, with a value range of [0,1]; Conf(i) represents the confidence level of the relationship path between the candidate fault cause and the fault phenomenon, with a value range of [0,1]; α and β are weight parameters, and α+β= 1, which can be configured according to the actual operation and maintenance scenario. Based on the Score(i) of each candidate fault cause, the one with the highest score is selected as the most likely fault cause, and the corresponding maintenance plan is output.

[0054] As another optional embodiment, the correlation degree calculation method can be based on a probabilistic inference model. Using Bayes' theorem, the historical occurrence frequency of candidate fault cause entities is used as the prior probability, and the relationship path confidence is used as the likelihood probability to calculate the conditional probability of each candidate fault cause under a given fault phenomenon. P(Cause(i) | Symptom)∝P(Symptom | Cause(i)) * P(Cause(i)) Where P(Cause(i)) represents the prior probability of candidate fault cause i, which can be obtained from historical data statistics; P(Symptom | Cause(i)) represents the probability of the target fault phenomenon occurring when the candidate fault cause exists, which can be estimated from the relationship path confidence. Finally, the candidate fault cause with the largest P(Cause(i) | Symptom) is selected, and the corresponding maintenance plan is output.

[0055] Furthermore, in practical applications, the calculation of the correlation degree can also incorporate time factors, environmental factors, or expert experience factors to obtain more accurate reasoning results through multi-factor comprehensive scoring, thereby improving the reliability and robustness of reasoning.

[0056] The present invention may also include real-time acquisition of new operation and maintenance data to dynamically update the hybrid knowledge graph, thereby ensuring that the knowledge system remains synchronized with the actual operating environment.

[0057] Specifically, the knowledge graph incremental update mechanism includes the following steps: collecting new operation and maintenance data through real-time interfaces (such as sensor interfaces, monitoring system interfaces, and manual input interfaces); automatically identifying the entities, relationships, attributes, and operation and maintenance rules of the new operation and maintenance data using natural language processing and pattern recognition algorithms; verifying the accuracy and reliability of the extraction results, which may include manual review, rule-based consistency verification, statistical confidence verification, or a combination of manual and automatic verification methods to avoid knowledge graph pollution caused by the introduction of erroneous information; updating the verified data to the hybrid knowledge graph in the form of triples to supplement or correct existing knowledge, realize the continuous accumulation and iterative optimization of operation and maintenance experience, and improve the timeliness and accuracy of fault diagnosis, risk warning, and decision support.

[0058] Compared with existing technologies that rely on manual input and periodic batch updates, this invention achieves rapid accumulation and timely updates of operation and maintenance knowledge through a collaborative mechanism that combines real-time interface with automatic identification and manual verification. This avoids the shortcomings of traditional methods, such as delayed updates and incomplete knowledge, and thus significantly improves the level of intelligence in heating network operation and maintenance.

[0059] As an optional embodiment, such as Figure 2a and 2b As shown, the system automatically calls the data acquisition interface to collect multi-source operation and maintenance information from sources such as the heating network monitoring system, equipment management database, and document storage system. If no data is collected, an error message is displayed prompting the user to check the data source.

[0060] The collected information undergoes preprocessing, including removing invalid data, supplementing missing information, and standardizing data formats. After processing, entity recognition and relation extraction technologies are used to extract entities, relations, and attributes from the knowledge graph, construct triplet data, and store it in a graph database to form a knowledge graph for heat network operation and maintenance.

[0061] Upon receiving an operation and maintenance inference request, the hybrid inference engine is invoked. The hybrid inference engine consists of a rule-based inference module and a graph inference module. The rule-based inference module, based on operation and maintenance rules extracted from the hybrid knowledge graph, compares the input fault phenomena with the rule conditions using parameters and performs logical judgments to achieve rapid inference for simple faults. The graph inference module utilizes the relationships between entities in the hybrid knowledge graph, starting with the entity corresponding to the fault phenomenon, and through multi-hop path search, combined with historical fault occurrence frequency and relationship confidence, to uncover deep causal relationships, enabling the tracing and inference of complex faults. Simultaneously, it generates a visual inference path containing entity relationship paths, correlation weights, and historical case references.

[0062] First, the rule reasoning module is started. This module includes a rule base and a rule matching engine. The rules stored in the rule base are structured rules extracted from the entity relationships of the hybrid knowledge graph and are organized in IF-THEN format. Each rule includes a triggering condition, a conclusion, and a corresponding maintenance plan. The rule matching engine breaks down the fault phenomenon in the request into key feature parameters and compares them parameter by parameter with the triggering conditions in the rule base and performs logical combination judgment. If the match is successful, the cause of the fault and the maintenance plan are directly output. If the match is unsuccessful, the graph reasoning module is started.

[0063] The graph reasoning module, based on a knowledge graph, starts with the fault phenomenon in the request, searches for related devices, historical faults, and other entities, analyzes the relationship paths between entities, calculates the correlation degree of each possible fault cause, filters out the most likely fault cause and corresponding maintenance decisions, and generates a visualization file of the reasoning path. After the reasoning is complete, the reasoning result is output. If the search fails, a message is displayed indicating that the relevant fault has not been entered into the knowledge base.

[0064] The incremental update program is launched periodically, acquiring newly added operational data through a real-time interface, automatically identifying new entities, relationships, and rules, and generating data to be updated. After manual review and confirmation, this data is updated into the knowledge graph, while historical versions are backed up to ensure knowledge traceability. If the update fails, the system prompts the user to check the update data or restart the update program.

[0065] In summary, this invention integrates multi-source operation and maintenance information such as equipment parameters and sensor data to construct a hybrid knowledge graph covering pipeline topology, fault modes, and operation and maintenance rules. Its core lies in the deep integration of structured storage of the knowledge graph with a hybrid inference engine, distinguishing it from traditional knowledge graphs used only for static knowledge display. A hybrid inference engine combining rule-based reasoning and graph reasoning is designed to enable fault tracing and maintenance decision generation. Simultaneously, based on the incremental update mechanism of the hybrid knowledge graph, real-time data collection and multi-dimensional verification support iterative optimization of operation and maintenance experience, improving the structured level of knowledge management and the interpretability of reasoning, thus providing efficient knowledge service support for intelligent operation and maintenance of heating networks.

[0066] In summary, this invention deeply integrates multi-source operation and maintenance information, expert knowledge, and knowledge graphs, and designs a hybrid engine and incremental update mechanism that combines rule-based reasoning and graph reasoning. Unlike the static graphs and single reasoning methods of existing technologies, it can achieve rapid response, in-depth tracing, and continuous knowledge evolution, significantly improving the intelligence level of heating network operation and maintenance.

[0067] Exemplary System Accordingly, embodiments of the present invention also provide a heat network operation and maintenance system based on a hybrid knowledge graph. Figure 3 This is a block diagram of a heat network operation and maintenance system based on a hybrid knowledge graph provided in an embodiment of the present invention, such as... Figure 3As shown, the system 100 provided in this embodiment includes: The acquisition module 101 is used to collect multi-source operation and maintenance information, which includes equipment parameters, real-time sensor data, historical fault records and expert operation and maintenance rules. Preprocessing module 102 is used to preprocess the multi-source operation and maintenance information; The construction module 103 is used to perform entity recognition, relation extraction and attribute definition on the preprocessed multi-source operation and maintenance information, and combine it with the expert operation and maintenance rules to construct triples and store them in the graph database to form a hybrid knowledge graph. Receiver module 104 is used to receive operation and maintenance inference requests; The reasoning module 105 is used to perform rule reasoning on the operation and maintenance reasoning request based on the hybrid knowledge graph: if the rule reasoning is successfully matched, the rule reasoning result is output; if the rule reasoning is not successfully matched, graph reasoning is performed, and the corresponding graph reasoning result is output when the reasoning result is obtained.

[0068] It should be noted that although the operation of the heat network operation and maintenance method based on hybrid knowledge graph of the present invention is described in a specific order in the accompanying drawings, this does not require or imply that these operations must be performed in that specific order, or that all the operations shown must be performed to achieve the desired result. Additionally or alternatively, certain steps may be omitted, multiple steps may be combined into one step, and / or one step may be broken down into multiple steps.

[0069] Furthermore, although several devices, units, or modules of the heat network operation and maintenance system based on hybrid knowledge graphs have been mentioned in the detailed description above, this division is merely exemplary and not mandatory. In fact, according to embodiments of the present invention, the features and functions of two or more modules described above can be embodied in one module. Conversely, the features and functions of one module described above can be further divided and embodied by multiple modules.

[0070] While the spirit and principles of the invention have been described with reference to several specific embodiments, it should be understood that the invention is not limited to the disclosed specific embodiments, and the division of aspects does not imply that features in these aspects cannot be combined for benefit; such division is merely for ease of description. The invention is intended to cover various modifications and equivalent arrangements included within the spirit and scope of the appended claims.

Claims

1. A method for operating and maintaining a hot network based on a hybrid knowledge graph, characterized in that, The method includes: Collect multi-source operation and maintenance information, including equipment parameters, real-time sensor data, historical fault records, and expert operation and maintenance rules; The multi-source operation and maintenance information is preprocessed; Entity identification, relation extraction, and attribute definition are performed on the preprocessed multi-source operation and maintenance information. Combined with the expert operation and maintenance rules, triples are constructed and stored in the graph database to form a hybrid knowledge graph. Receive operation and maintenance inference requests; Based on the hybrid knowledge graph, rule-based reasoning is performed on the operation and maintenance reasoning request: if the rule reasoning is successfully matched, the rule reasoning result is output; if the rule reasoning is not successfully matched, graph reasoning is performed, and the corresponding graph reasoning result is output when the reasoning result is obtained.

2. The heat network operation and maintenance method based on hybrid knowledge graph as described in claim 1, characterized in that, The method further includes: New operational and maintenance data is acquired in real time to update the hybrid knowledge graph.

3. The heat network operation and maintenance method based on hybrid knowledge graph as described in claim 2, characterized in that, The specific steps for acquiring new operation and maintenance data in real time to update the hybrid knowledge graph are as follows: New operation and maintenance data is collected through real-time interfaces; The system automatically identifies the entities, relationships, attributes, and operation and maintenance rules of the newly added operation and maintenance data, and updates them to the hybrid knowledge graph.

4. The heat network operation and maintenance method based on hybrid knowledge graph according to any one of claims 1-3, characterized in that, Based on the hybrid knowledge graph, rule-based reasoning is performed on the operation and maintenance reasoning request: if the rule reasoning match is successful, the specific steps for outputting the rule reasoning result are as follows: Extract the key feature parameters of the fault phenomenon from the operation and maintenance inference request; The key feature parameters are compared and logically combined with the triggering conditions in the structured rules extracted from the hybrid knowledge graph. If the rule reasoning is successful, the corresponding fault cause and maintenance plan are output.

5. The heat network operation and maintenance method based on hybrid knowledge graph according to claim 4, characterized in that, The structured rules are organized in IF-THEN format, and each structured rule includes the triggering condition, the cause of the fault, and the corresponding maintenance plan.

6. The heat network operation and maintenance method based on hybrid knowledge graph according to any one of claims 1-3, characterized in that, If the rule-based reasoning fails to match, graph reasoning is performed, and the corresponding graph reasoning result is output when the reasoning result is obtained. The specific steps are as follows: Starting with the entity corresponding to the fault phenomenon, search for candidate fault cause entities associated with the fault phenomenon entity in the hybrid knowledge graph; If candidate fault cause entities are found, the relationship path between the fault phenomenon entity and the candidate fault cause entity is determined, and the correlation degree of each candidate fault cause entity is calculated by combining the historical fault occurrence frequency and the confidence of the relationship path. Filter the most likely causes of failure based on correlation, and output the causes of failure and corresponding maintenance solutions.

7. The heat network operation and maintenance method based on hybrid knowledge graph as described in claim 6, characterized in that, The steps of filtering the most likely causes of failure based on correlation and outputting the causes of failure and corresponding maintenance solutions also include: Generate visualization files to demonstrate the reasoning process.

8. The heat network operation and maintenance method based on hybrid knowledge graph according to any one of claims 1-3, characterized in that, The hybrid knowledge graph includes pipeline topology, equipment failure modes, and operation and maintenance rules.

9. The heat network operation and maintenance method based on hybrid knowledge graph according to any one of claims 1-3, characterized in that, The steps of performing entity identification, relation extraction, and attribute definition on the preprocessed multi-source operation and maintenance information, and constructing triples stored in a graph database in conjunction with the expert operation and maintenance rules to form a hybrid knowledge graph also include: Establish entity attribute indexes and relationship path indexes.

10. A heat network operation and maintenance system based on a hybrid knowledge graph, characterized in that, The system includes: The data acquisition module is used to collect multi-source operation and maintenance information, including equipment parameters, real-time sensor data, historical fault records, and expert operation and maintenance rules. A preprocessing module is used to preprocess the multi-source operation and maintenance information; The construction module is used to perform entity recognition, relation extraction and attribute definition on the preprocessed multi-source operation and maintenance information, and combine it with the expert operation and maintenance rules to construct triples and store them in the graph database to form a hybrid knowledge graph. The receiving module is used to receive operation and maintenance inference requests; The reasoning module is used to perform rule-based reasoning on the operation and maintenance reasoning request based on the hybrid knowledge graph: if the rule reasoning is successfully matched, the rule reasoning result is output; if the rule reasoning is not successfully matched, graph reasoning is performed, and the corresponding graph reasoning result is output when the reasoning result is obtained.

Images