Accident early warning method based on digital twinborn evolution and prediction data
By embedding knowledge graphs into digital twins and performing joint reasoning, the problems of insufficient knowledge utilization and inadequate adaptability in existing technologies are solved, achieving highly accurate and interpretable equipment status prediction and improving the reliability of accident early warning.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-11-11
- Publication Date
- 2026-03-10
AI Technical Summary
Existing digital twin systems suffer from insufficient knowledge utilization, poor interpretability of prediction results, and poor model adaptability in state prediction. There is an urgent need for a new type of digital twin system that can deeply integrate domain knowledge and has self-learning and self-optimization capabilities.
By constructing a domain knowledge graph for the device and embedding it into a digital twin, joint reasoning is performed using the prediction model and the knowledge graph to generate device status prediction results. The knowledge graph and prediction model are then updated through an optimization mechanism triggered by prediction deviations, thus achieving closed-loop optimization.
It significantly improves the accuracy and interpretability of state prediction results. The system has self-evaluation and self-learning capabilities, can maintain prediction accuracy over a long period of time, and reduces maintenance risks.
Smart Images
Figure CN121638448A_ABST
Abstract
Description
Technical Field
[0001] This application relates to the field of industrial intelligence and predictive maintenance technology, and more specifically, to an accident early warning method based on digital twin evolution and predictive data. Background Technology
[0002] With the deepening development of intelligent complex equipment, digital twin technology has become crucial for realizing the interactive integration of physical entities and virtual models. Existing digital twin systems primarily focus on the reproduction of geometric and physical rules and the mapping of real-time data. In terms of state prediction, they typically employ single prediction methods based on physical models or data-driven approaches. This results in problems such as insufficient knowledge utilization, poor interpretability of prediction results, and poor model adaptability. There is an urgent need for a new type of digital twin system that can deeply integrate domain knowledge and possess self-learning and self-optimization capabilities to achieve more accurate, reliable, and interpretable equipment state prediction and health management. Summary of the Invention
[0003] In view of at least one defect or improvement need of the prior art, this application provides an accident early warning method based on digital twin evolution and prediction data, which can solve at least one of the problems existing in the above background art.
[0004] To achieve the above objectives, according to the first aspect of this application, an accident early warning method based on digital twin evolution and prediction data is provided, the method comprising: Collect operational data from the equipment, and build and update its digital twin; Construct a domain knowledge graph for the device and embed the knowledge graph into the digital twin; Based on the real-time and / or historical data of the digital twin, a prediction model is used to predict the future state of the device, and relevant knowledge in the knowledge graph is used for joint reasoning to generate the prediction result of the future state of the device. The state prediction result is compared with the subsequently acquired actual monitoring data to generate a prediction deviation. When the prediction deviation exceeds a preset threshold, an optimization mechanism is triggered to update the knowledge graph and the prediction model using the prediction deviation and the subsequently acquired actual monitoring data. The updated knowledge graph and the updated prediction model are fed back to the digital twin to complete the closed-loop optimization.
[0005] Furthermore, in the above-mentioned accident early warning method based on digital twin evolution and prediction data, the constructed domain knowledge graph of the device includes nodes that include at least one of the following: device components, failure modes, symptoms, maintenance measures, or expert rule entities, and edges that represent the relationships between the entities.
[0006] Furthermore, the aforementioned accident early warning method based on digital twin evolution and prediction data, wherein the joint reasoning using relevant knowledge in the knowledge graph specifically includes dynamically querying and using relevant knowledge in the knowledge graph for joint reasoning through graph neural network technology or knowledge graph embedding technology, and outputting prediction results and explanations including data-driven and knowledge-driven ones.
[0007] Furthermore, in the aforementioned accident early warning method based on digital twin evolution and prediction data, the updating of the knowledge graph specifically includes: Based on the prediction deviation data, new symptom-fault association knowledge is generated; Update and incrementally learn the knowledge graph.
[0008] Furthermore, in the aforementioned accident early warning method based on digital twin evolution and prediction data, updating the prediction model specifically includes: The parameters of the prediction model are dynamically adjusted using incremental learning or online learning algorithms; When the performance degradation of the prediction model reaches a preset level, the model structure optimization or reconstruction process is initiated.
[0009] Furthermore, the aforementioned accident early warning method based on digital twin evolution and prediction data also includes: The state prediction results, the associated paths in the knowledge graph, and the system's self-updated state are visualized through a human-computer interaction interface.
[0010] According to a second aspect of this application, an accident early warning device based on digital twin evolution and prediction data is also provided, comprising: The data acquisition and processing module is used to collect the equipment's operating data and build and update its digital twin; A digital twin module is used to construct a domain knowledge graph of the device and embed the knowledge graph into the digital twin; The fusion prediction module is used to predict the future state of the device based on the real-time and / or historical data of the digital twin, and to perform joint reasoning using relevant knowledge in the knowledge graph to generate the prediction result of the future state of the device. The self-updating optimization module is used to compare the state prediction result with the subsequently acquired actual monitoring data to generate a prediction deviation. When the prediction deviation exceeds a preset threshold, an optimization mechanism is triggered to update the knowledge graph and the prediction model using the prediction deviation and the subsequently acquired actual monitoring data. The updated knowledge graph and the updated prediction model are fed back to the digital twin to complete the closed-loop optimization.
[0011] According to a third aspect of this application, an accident early warning device based on digital twin evolution and prediction data is also provided, which includes at least one processing unit and at least one storage unit, wherein the storage unit stores a computer program that, when executed by the processing unit, causes the processing unit to perform the steps of any of the methods described above.
[0012] According to a fourth aspect of this application, a storage medium is also provided, which stores a computer program executable by an accident early warning device based on digital twin evolution and prediction data, wherein when the computer program is run on the accident early warning device based on digital twin evolution and prediction data, the accident early warning device based on digital twin evolution and prediction data performs the steps of any of the methods described above.
[0013] In summary, compared with the prior art, the above-described technical solutions conceived in this application can achieve the following beneficial effects: The accident early warning method based on digital twin evolution and prediction data provided in this application effectively integrates the advantages of data-driven and knowledge-driven approaches by embedding knowledge graphs into digital twins and performing joint reasoning. This significantly improves the accuracy and interpretability of state prediction results, making predictions not only based on data trends but also consistent with domain logic. By comparing prediction results with actual data and generating prediction deviations, a self-evaluation mechanism is established, enabling the system to have continuous self-adaptation and self-learning capabilities. This allows the system to maintain prediction accuracy over a long period, effectively cope with changes in equipment operating conditions and performance degradation, thereby greatly improving the reliability of accident early warning and reducing maintenance risks. Attached Figure Description
[0014] To more clearly illustrate the technical solutions in the embodiments of this application, the accompanying drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.
[0015] Figure 1 This is a flowchart illustrating an accident early warning method based on digital twin evolution and prediction data, provided as an embodiment of this application. Detailed Implementation
[0016] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application. Furthermore, the technical features involved in the various embodiments described below can be combined with each other as long as they do not conflict with each other.
[0017] The terms "first," "second," "third," etc., in the specification, claims, and accompanying drawings of this application are used to distinguish different objects, not to describe a specific order. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion. For example, a process, method, system, product, or apparatus that includes a series of steps or units is not limited to the listed steps or units, but may optionally include steps or units not listed, or may optionally include other steps or units inherent to these processes, methods, products, or apparatuses.
[0018] Figure 1 This application provides a flowchart illustrating an accident early warning method based on digital twin evolution and prediction data, as shown in the following embodiment. Figure 1 As shown in the embodiments of this application, the accident early warning method based on digital twin evolution and prediction data includes the following steps: Collect operational data from the equipment, and build and update its digital twin; Construct a domain knowledge graph for the device and embed the knowledge graph into the digital twin; Based on the real-time and / or historical data of the digital twin, a prediction model is used to predict the future state of the device, and relevant knowledge in the knowledge graph is used for joint reasoning to generate the prediction result of the future state of the device. The state prediction result is compared with the subsequently acquired actual monitoring data to generate a prediction deviation. When the prediction deviation exceeds a preset threshold, an optimization mechanism is triggered to update the knowledge graph and the prediction model using the prediction deviation and the subsequently acquired actual monitoring data. The updated knowledge graph and the updated prediction model are fed back to the digital twin to complete the closed-loop optimization.
[0019] Specifically, various parameter values reflecting the state of the physical device are acquired from its corresponding sensors and other measurement units. A model that maps the main characteristics of the physical device is created in virtual space. Using continuously collected operational data, the state of this virtual model is driven to change, ensuring consistency with the real-time or historical state of the physical device. This step ensures that the digital twin is not a static model, but a dynamic image synchronized with the physical entity.
[0020] The knowledge graph about the device's structure, function, behavior, and failure modes is organized into a structured, machine-readable network in the form of nodes and relationships. This knowledge graph is then deeply integrated with the aforementioned digital twin, making it a queryable and accessible component within the digital twin. This empowers the digital twin to utilize prior knowledge for cognition and reasoning.
[0021] The data relied upon for prediction comes from a constructed and updated dataset that reflects the actual state of the equipment. An algorithm or model is used to analyze and calculate this data to infer the state of the equipment at a future point in time. This is a deductive process based on data patterns. Simultaneously, an embedded knowledge graph is invoked, and logical inferences about the development of the equipment's state are made based on the logical rules and relationships contained within it. This is a deductive process based on knowledge logic. The combined effect of these two processes, "prediction" and "reasoning," generates the predicted state of the equipment. The final prediction result is produced through the synergistic and joint action of data-driven and knowledge-driven approaches.
[0022] The system waits for the predicted future moment to become the present, acquires the actual state data of the device at that moment, and compares it with the generated prediction results. The difference between the predicted and actual values is calculated through this comparison. The system has a predefined judgment standard (threshold); when the deviation becomes too large, it indicates that the current system's predictive ability or knowledge reserve is insufficient, requiring the initiation of an automatic optimization program. The optimization mechanism utilizes newly generated data (actual monitoring data) and performance feedback (prediction deviation) to modify and adjust the two core components of the system: the knowledge graph reflecting cognition and the prediction model responsible for computation, in order to improve their future performance.
[0023] The optimized new version of the knowledge graph and predictive model is reintegrated or loaded into the digital twin system, replacing the old version. The system then restarts from the beginning using the optimized new components, enabling it to continuously improve its performance by leveraging the latest data and feedback, achieving continuous and spontaneous evolution.
[0024] The accident early warning method based on digital twin evolution and prediction data provided in this application effectively integrates the advantages of data-driven and knowledge-driven approaches by embedding knowledge graphs into digital twins and performing joint reasoning. This significantly improves the accuracy and interpretability of state prediction results, making predictions not only based on data trends but also consistent with domain logic. By comparing prediction results with actual data and generating prediction deviations, a self-evaluation mechanism is established, enabling the system to have continuous self-adaptation and self-learning capabilities. This allows the system to maintain prediction accuracy over a long period, effectively cope with changes in equipment operating conditions and performance degradation, thereby greatly improving the reliability of accident early warning and reducing maintenance risks.
[0025] Optionally, the accident early warning method based on digital twin evolution and prediction data provided in this application embodiment constructs a domain knowledge graph of the device, whose nodes include at least one of device components, fault modes, symptoms, maintenance measures or expert rule entities, and edges are used to represent the relationships between the entities.
[0026] Specifically, nodes include at least one of the following entities: equipment components, failure modes, symptoms, maintenance measures, or expert rules, thus clarifying the specific types of nodes that constitute the knowledge graph of this domain. These five types of entities together form a complete knowledge system oriented towards equipment health management: equipment components describe the physical or functional composition of the equipment itself; failure modes describe the possible forms of functional failure of equipment components; symptoms describe the observable internal or external phenomena exhibited by the failure modes; maintenance measures describe the corrective or repair actions taken for specific failure modes; and expert rules describe the complex logical judgment conditions between the above entities summarized by domain experts.
[0027] Edges are used to represent the relationships between entities, clarifying the specific semantic relationships carried by the edges connecting different nodes (entities) in the knowledge graph. Dispersed entity nodes are organized into a network with rich semantic associations, making path traversal and logical reasoning based on the graph possible.
[0028] Optionally, the accident early warning method based on digital twin evolution and prediction data provided in this application embodiment, wherein the joint reasoning using relevant knowledge in the knowledge graph specifically includes dynamically querying and using relevant knowledge in the knowledge graph for joint reasoning through graph neural network technology or knowledge graph embedding technology, and outputting prediction results and explanations including data-driven and knowledge-driven methods.
[0029] Specifically, the joint reasoning is not an abstract conceptual combination, but rather achieved through two computational techniques: graph neural network technology and knowledge graph embedding technology. This transforms structured, symbolic knowledge graph information into numerical vector representations, enabling deep integration with data-driven prediction models, which are also numerical, within the same mathematical space. This process is dynamic; the system actively queries and utilizes relevant subsets of knowledge from the graph based on the prediction target. The output not only contains a simple predicted value but also an explanation derived from the logical reasoning path based on the knowledge graph. This achieves an organic combination of data-driven and knowledge-driven approaches at the technical level, and endows the prediction results with interpretability.
[0030] Optionally, the accident early warning method based on digital twin evolution and prediction data provided in this application embodiment, wherein updating the knowledge graph specifically includes: Based on the prediction deviation data, new symptom-fault association knowledge is generated; Update and incrementally learn the knowledge graph.
[0031] Specifically, the update is not a simple data entry, but a feedback-based automated knowledge discovery and learning process. The system can automatically analyze the differences between predicted results and actual data, identify new causal relationships not covered by the existing knowledge system, and formally define these rules as new elements in the knowledge graph. The update method is non-destructive and gradual. The system does not rebuild the entire graph, but integrates the newly discovered relationships into the existing knowledge graph by adding nodes or edges, or adjusting the weights of existing edges. This mechanism enables the knowledge graph to continuously learn from experience and improve itself, which is the core of the system's long-term adaptability.
[0032] Optionally, the accident early warning method based on digital twin evolution and prediction data provided in this application embodiment, wherein updating the prediction model specifically includes: The parameters of the prediction model are dynamically adjusted using incremental learning or online learning algorithms; When the performance degradation of the prediction model reaches a preset level, the model structure optimization or reconstruction process is initiated.
[0033] Specifically, while maintaining the basic mathematical structure of the model, the weight parameters within the model are fine-tuned using new monitoring data. This is a lightweight, high-frequency optimization designed to enable the model to quickly adapt to the slow, dynamic evolution of equipment conditions. A more thorough approach is taken when parameter adjustments cannot reverse the continuous degradation of model performance: modifying or completely replacing the model's basic framework to address the fundamental mismatch between the model's capabilities and complex operating conditions. These two stages work together to ensure that the predictive model possesses continuous and robust adaptive capabilities. Optionally, the accident early warning method based on digital twin evolution and prediction data provided in this application embodiment further includes: The state prediction results, the associated paths in the knowledge graph, and the system's self-updated state are visualized through a human-computer interaction interface.
[0034] Specifically, visualizing the state prediction results transforms the numerical or probabilities output by the algorithm into intuitive charts, trend lines, and warning levels, enabling users to quickly understand the future state of the device. Visualizing the association paths in the knowledge graph presents the internal reasoning logic chains of the system in an interactive graph format, providing interpretable evidence to support the prediction conclusions and enhancing the credibility of the results. Visualizing the system's self-updating status makes the system's own health and learning progress, such as model performance indicators and knowledge evolution logs, transparent, facilitating user monitoring of system reliability and manual intervention when necessary.
[0035] This application also provides an accident early warning device based on digital twin evolution and prediction data, including: The data acquisition and processing module is used to collect the equipment's operating data and build and update its digital twin; A digital twin module is used to construct a domain knowledge graph of the device and embed the knowledge graph into the digital twin; The fusion prediction module is used to predict the future state of the device based on the real-time and / or historical data of the digital twin, and to perform joint reasoning using relevant knowledge in the knowledge graph to generate the prediction result of the future state of the device. The self-updating optimization module is used to compare the state prediction result with the subsequently acquired actual monitoring data to generate a prediction deviation. When the prediction deviation exceeds a preset threshold, an optimization mechanism is triggered to update the knowledge graph and the prediction model using the prediction deviation and the subsequently acquired actual monitoring data. The updated knowledge graph and the updated prediction model are fed back to the digital twin to complete the closed-loop optimization.
[0036] This application also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the above-described method. The computer-readable storage medium may include, but is not limited to, any type of disk, including floppy disks, optical disks, DVDs, CD-ROMs, microdrives, as well as magneto-optical disks, ROMs, RAMs, EPROMs, EEPROMs, DRAMs, VRAMs, flash memory devices, magnetic cards or optical cards, nanosystems (including molecular memory ICs), or any type of medium or device suitable for storing instructions and / or data.
[0037] It should be noted that, for the sake of simplicity, the foregoing method embodiments are all described as a series of actions. However, those skilled in the art should understand that this application is not limited to the described order of actions, as some steps may be performed in other orders or simultaneously according to this application. Furthermore, those skilled in the art should also understand that the embodiments described in the specification are preferred embodiments, and the actions and modules involved are not necessarily essential to this application.
[0038] In the above embodiments, the descriptions of each embodiment have different focuses. For parts not described in detail in a certain embodiment, please refer to the relevant descriptions in other embodiments.
[0039] In the several embodiments provided in this application, it should be understood that the disclosed apparatus can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some service interface; the indirect coupling or communication connection between devices or units may be electrical or other forms.
[0040] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.
[0041] Furthermore, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.
[0042] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage device (CMD). Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a memory and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned memory includes various media capable of storing program code, such as USB flash drives, read-only memory (ROM), random access memory (RAM), portable hard drives, magnetic disks, or optical disks.
[0043] Those skilled in the art will understand that all or part of the steps in the various methods of the above embodiments can be implemented by a program instructing related hardware. The program can be stored in a computer-readable storage medium, which may include: a flash drive, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk, etc.
[0044] The foregoing description is merely an exemplary embodiment of this disclosure and should not be construed as limiting the scope of this disclosure. Any equivalent changes and modifications made in accordance with the teachings of this disclosure shall still fall within the scope of this disclosure. Those skilled in the art will readily conceive of embodiments of this disclosure upon considering the specification and practicing the disclosure herein. This application is intended to cover any variations, uses, or adaptations of this disclosure that follow the general principles of this disclosure and include common knowledge or customary techniques in the art not described herein. The specification and embodiments are to be considered exemplary only, and the scope and spirit of this disclosure are defined by the claims.
[0045] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.
[0046] Those skilled in the art will readily understand that the above description is merely a preferred embodiment of this application and is not intended to limit this application. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this application should be included within the scope of protection of this application.
Claims
1. An accident early warning method based on digital twin evolution and prediction data, characterized in that, The method comprises the following steps: collecting operation data of the equipment, constructing and updating a digital twin of the equipment; constructing a domain knowledge graph of the equipment and embedding the knowledge graph into the digital twin; based on real-time and / or historical data of the digital twin, using a prediction model to predict the future state of the equipment, using related knowledge in the knowledge graph for joint reasoning, and generating a prediction result of the future state of the equipment; comparing the prediction result with actual monitoring data obtained subsequently, generating a prediction deviation, and when the prediction deviation exceeds a preset threshold, triggering an optimization mechanism, and using the prediction deviation and the actual monitoring data obtained subsequently to update the knowledge graph and the prediction model; feeding the updated knowledge graph and the updated prediction model back to the digital twin to complete closed-loop optimization.
2. The method of claim 1, wherein, The method of constructing the domain knowledge graph of the equipment, wherein nodes of the knowledge graph comprise at least one of equipment components, failure modes, symptoms, maintenance measures or expert rule entities, and edges are used to represent relationships between entities.
3. The accident early warning method based on digital twin evolution and prediction data according to claim 1 or 2, characterized in that, The method of using related knowledge in the knowledge graph for joint reasoning comprises dynamically querying and using related knowledge in the knowledge graph for joint reasoning through a graph neural network technology or a knowledge graph embedding technology, and outputting prediction results and explanations including data-driven and knowledge-driven.
4. The method of claim 1, wherein, The method of updating the knowledge graph comprises: based on the prediction deviation data, generating new symptom-failure association knowledge; updating and incrementally learning the knowledge graph.
5. The method of claim 1, wherein, The method of updating the prediction model comprises: using an incremental learning or online learning algorithm to dynamically adjust parameters of the prediction model; when the performance of the prediction model degrades to a preset degree, starting a model structure optimization or reconstruction process.
6. The method of claim 1, wherein, The method further comprises: visually displaying the prediction result, an associated path in the knowledge graph, and a state of system self-updating through a human-computer interaction interface.
7. An accident early warning device based on digital twin evolution and prediction data, characterized in that, The method comprises: a data collection and processing module for collecting operation data of the equipment, constructing and updating a digital twin of the equipment; a digital twin module for constructing a domain knowledge graph of the equipment and embedding the knowledge graph into the digital twin; a fusion prediction module for predicting the future state of the equipment based on real-time and / or historical data of the digital twin, using a prediction model, using related knowledge in the knowledge graph for joint reasoning, and generating a prediction result of the future state of the equipment; a self-updating optimization module for comparing the prediction result with actual monitoring data obtained subsequently, generating a prediction deviation, and when the prediction deviation exceeds a preset threshold, triggering an optimization mechanism, and using the prediction deviation and the actual monitoring data obtained subsequently to update the knowledge graph and the prediction model; feeding the updated knowledge graph and the updated prediction model back to the digital twin to complete closed-loop optimization.
8. An accident early warning device based on digital twin evolution and prediction data, characterized by, The method comprises at least one processing unit and at least one storage unit, wherein the storage unit stores a computer program, and when the computer program is executed by the processing unit, the processing unit executes the steps of the method of any one of claims 1-6.
9. A storage medium, characterized by which stores the computer program executable by the accident early warning device based on digital twin evolution and prediction data, when the computer program runs on the accident early warning device based on digital twin evolution and prediction data, makes the accident early warning device based on digital twin evolution and prediction data execute the steps of the method of any one of claims 1-6.