A method and system for constructing a wheel axle full life cycle maintenance model

Abstract

The application relates to the technical field of data management, and relates to a construction method and system of a wheel shaft full-life cycle maintenance model. The method comprises the following steps: acquiring original data and a data access request; processing the data access request and the original data to obtain temporary authorization instructions; acquiring encrypted data packets based on the temporary authorization instructions, and extracting original heterogeneous data in the encrypted data packets, wherein the original heterogeneous data comprises various types of data in different data forms, formats and standards generated in each stage of the wheel shaft full-life cycle; performing semantic standardization processing based on the original heterogeneous data to obtain a knowledge graph; constructing a prediction model based on the knowledge graph; and performing optimization feedback processing on the prediction model to obtain the wheel shaft full-life cycle maintenance model. By constructing a trusted and interconnected data space, the application not only improves the maintenance efficiency and safety of a single wheel shaft, but also promotes the collaborative innovation and continuous optimization of each link of design, manufacturing, service and maintenance.

Description

Technical Field

[0001] This invention relates to the field of data management technology, and more specifically, to a method and system for constructing a wheel and axle full life cycle maintenance model. Background Technology

[0002] As core load-bearing and running components of rail transit vehicles, the health of wheel and axle directly determines the safety of train operation. Currently, wheel and axle maintenance is constrained by multiple bottlenecks in data management and application. The industry generally faces four core challenges: First, the phenomenon of data silos is serious. The entire life cycle data of wheel and axle is scattered across different main systems such as design, manufacturing, service, and maintenance. The data chain is broken, making it impossible to form a complete digital archive and supporting full-process traceability and collaborative management. Second, the contradiction between data sharing and security is prominent. Enterprises lack a reliable sharing mechanism due to concerns about data sovereignty, trade secrets, and responsibilities. Traditional interaction methods are prone to abuse and are not traceable. Third, multi-source heterogeneous data. The integration is difficult. The entire life cycle of wheel and axle generates data in different forms, formats, and standards, such as design drawings, process parameters, sensor timing data, flaw detection images, and maintenance records. Semantic standards are not unified, and there is a lack of effective standardization and semantic alignment methods. Fourth, data application has not formed a closed-loop optimization. Data is limited to use in a single link and cannot reach all participants in the entire wheel and axle life cycle. Back-end maintenance and service data are difficult to effectively feed back to the front end to optimize design, manufacturing processes, and maintenance strategies. Overall, it is still in an experience-driven rather than data-driven development stage. There is an urgent need to build a wheel and axle life cycle maintenance system that covers reliable data flow, deep integration, and predictive maintenance. Summary of the Invention

[0003] The purpose of this invention is to provide a method and system for constructing a wheel and axle full life cycle maintenance model to improve the above-mentioned problems.

[0004] To achieve the above objectives, the embodiments of this application provide the following technical solutions:

[0005] On the one hand, embodiments of this application provide a method for constructing a wheel and axle full life cycle maintenance model, the method comprising:

[0006] Acquire raw data and data access requests, wherein the raw data includes flaw detection image files obtained during wheel and axle maintenance;

[0007] The data access request and the original data are processed to obtain a temporary authorization instruction;

[0008] Based on the temporary authorization instruction, an encrypted data packet is obtained, and the original heterogeneous data in the encrypted data packet is extracted. The original heterogeneous data includes various types of data with different data forms, formats and standards generated at each stage of the wheel and axle's entire life cycle.

[0009] Based on the original heterogeneous data, semantic standardization processing is performed to obtain a knowledge graph, which includes all relevant data for the entire life cycle of the wheel and axle.

[0010] Based on the knowledge graph, a prediction model is constructed;

[0011] The prediction model is optimized and feedback is applied to obtain the wheel and axle full life cycle maintenance model.

[0012] Secondly, this application provides a system for constructing a wheel and axle full life cycle maintenance model, the system comprising:

[0013] The acquisition module is used to acquire raw data and data access requests, wherein the raw data includes flaw detection image files obtained during wheel and axle maintenance;

[0014] The first processing module is used to process the data access request and the original data to obtain a temporary authorization instruction;

[0015] The second processing module is used to obtain the encrypted data packet based on the temporary authorization instruction and extract the original heterogeneous data in the encrypted data packet. The original heterogeneous data includes various types of data with different data forms, formats and standards generated at each stage of the wheel axle's entire life cycle.

[0016] The third processing module is used to perform semantic standardization processing on the original heterogeneous data to obtain a knowledge graph, which includes all relevant data of the entire life cycle of the wheel and axle.

[0017] The fourth processing module is used to construct a prediction model based on the knowledge graph;

[0018] The fifth processing module is used to perform optimization feedback processing on the prediction model to obtain the wheel and axle full life cycle maintenance model.

[0019] The beneficial effects of this invention are:

[0020] By introducing a mechanism that generates temporary authorization tokens based on raw data and access requests, combined with smart contract verification and encrypted delivery, a multi-source data fusion framework is constructed. This framework standardizes and semantically aligns heterogeneous data throughout the entire lifecycle of axles, forming a knowledge graph. Based on the knowledge graph, a predictive model is built and optimized in a closed loop, resulting in an integrated maintenance model covering the entire lifecycle of axles. This solves the problems of lack of trust in cross-entity data sharing, semantic barriers between multi-source heterogeneous data, and application fragmentation. It improves data integrity and consistency, significantly enhances maintenance efficiency and operational security, promotes feedback from backend maintenance facts to frontend design optimization and model iteration, and facilitates the digitalization and intelligentization of axle lifecycle data management.

[0021] Other features and advantages of the invention will be set forth in the following description, and will be apparent in part from the description, or may be learned by practicing embodiments of the invention. The objects and other advantages of the invention may be realized and obtained by means of the structures particularly pointed out in the written description, claims, and drawings. Attached Figure Description

[0022] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the embodiments will be briefly introduced below. It should be understood that the following drawings only show some embodiments of the present invention and should not be regarded as a limitation on the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0023] Figure 1 This is a flowchart illustrating the construction method of the wheel and axle full life cycle maintenance model described in this embodiment of the invention.

[0024] Figure 2 This describes the workflow of the wheel and axle full life cycle maintenance model as described in this embodiment of the invention. Detailed Implementation

[0025] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. The components of the embodiments of the present invention described and shown in the accompanying drawings can generally be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of the present invention provided in the accompanying drawings is not intended to limit the scope of the claimed invention, but merely to illustrate selected embodiments of the invention. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without inventive effort are within the scope of protection of the present invention.

[0026] It should be noted that similar reference numerals and letters in the following figures indicate similar items; therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures. Furthermore, in the description of this invention, terms such as "first," "second," etc., are used only to distinguish descriptions and should not be construed as indicating or implying relative importance.

[0027] Example 1

[0028] This embodiment provides a method for constructing a full life-cycle maintenance model for wheel and axle. It can be understood that a scenario can be set up in this embodiment, for example: in the advanced maintenance workshop of a railway group's vehicle depot, thousands of wheel and axle of EMU trains that have returned from the front line of operation need to be maintained every year. In the traditional maintenance mode, the design data of wheel and axle is held by the main engine manufacturer, the manufacturing and assembly data are scattered among various parts manufacturers, and the load, vibration and flaw detection data during service are kept within the railway group. Due to the serious data silos and lack of a reliable sharing mechanism, maintenance decisions mainly rely on static procedures and human experience, making it difficult to conduct differentiated and accurate life assessment and maintenance process optimization based on the actual condition of a single wheel and axle.

[0029] like Figure 1 As shown, this invention provides a method for constructing a wheel and axle full life cycle maintenance model, the method comprising steps S1-S6:

[0030] Step S1: Obtain raw data and data access request, wherein the raw data includes flaw detection image files obtained during wheel and axle maintenance;

[0031] In this step, data access requests include requests initiated by the user, which carry their own digital identity certificate and a statement of intent to use.

[0032] Step S2: Process the data access request and the original data to obtain a temporary authorization instruction;

[0033] Step S2 also includes steps S21, S22, and S23, which are specifically as follows:

[0034] Step S21: Construct a smart contract based on the original data;

[0035] In this step, based on a wheel axle flaw detection image file in the data provider's local database, a multimodal file (image, text, video, etc.) is selected. The system calls a hash algorithm to calculate the file content, generating a fixed-length data fingerprint as the unique identifier of the data asset. The provider selects a template from a predefined smart contract template library and configures the policy on the graphical interface. For example, the authorized user is "OEM A", the usage period is "December 31, 2024", the purpose of use is "wheel axle crack mode analysis", the authorized scope is "importing image metadata and defect annotation information", and "original image download" is prohibited. Based on the configuration policy, data fingerprint, timestamp, etc., an executable code deployed on the blockchain is automatically generated. This executable code is a smart contract instance. This smart contract is unique. By using consortium blockchain technology to build a decentralized trust foundation, combined with smart contracts, the programmable automatic execution of data sharing rules can be achieved, enabling fine-grained control over data access permissions.

[0036] Step S22: Perform verification processing based on the smart contract and the data access request to obtain the authorization result;

[0037] In this step, the user system initiates a request, which is routed to the corresponding smart contract on the blockchain. The smart contract begins execution, verifying whether the requester's identity matches the "authorized user," whether the current time is within the "usage period," and whether the usage intent matches the "usage purpose."

[0038] Step S23: Based on the authorization result, determine and obtain a temporary authorization instruction.

[0039] In this step, if the verification passes, the smart contract generates a temporary authorization token; if it fails, an "access denied" record is generated.

[0040] Step S3: Obtain the encrypted data packet based on the temporary authorization instruction, and extract the original heterogeneous data in the encrypted data packet. The original heterogeneous data includes various types of data with different data forms, formats and standards generated at each stage of the wheel and axle's entire life cycle.

[0041] In this step, the data user initiates a data request to the data provider's local data agent using a temporary authorization token. The data provider's agent verifies the validity of the temporary authorization token and, according to the authorization scope of the smart contract, extracts only the metadata corresponding to the image and the defect locations pre-annotated by the AI ​​model from the local database. This data is then encrypted using the data user's public key and transmitted to the data user through a secure channel. Simultaneously, all information related to this delivery is encapsulated into an audit log transaction and submitted to the blockchain for permanent storage. This information includes the data delivery time, user, amount of data delivered, and data hash value. This achieves a shift from human agreement to code enforcement, giving the data provider full lifecycle control over its data assets. All operations are traceable and tamper-proof, fundamentally solving the problem of data users being hesitant or unwilling to share due to concerns about data sovereignty and privacy, and ensuring a high degree of trust in the data sharing process.

[0042] Step S4: Perform semantic standardization processing on the original heterogeneous data to obtain a knowledge graph, which includes all relevant data for the entire life cycle of the wheel and axle;

[0043] Step S4 also includes steps S41, S42, and S43, which are specifically as follows:

[0044] Step S41: Obtain the preset knowledge model;

[0045] In this step, the preset knowledge model includes a pre-constructed "wheel and axle full life cycle domain ontology", which defines the knowledge model of concepts such as "wheel and axle", "hardness", "crack", "heat treatment" and their relationships.

[0046] Step S42: Encapsulate the original heterogeneous data to obtain a standardized data packet, wherein the standardized data packet includes a metadata file associated with the original heterogeneous data;

[0047] In this step, the raw heterogeneous data includes heterogeneous data from different source systems, such as structured data from the MES system, time-series data from vibration sensors, and unstructured data from flaw detection equipment. Each data source corresponds to an adapter. The adapter polls or listens for data updates. For example, for crack_image_01.jpg, the adapter does not modify its original file, but calculates its hash value, file size, creation time, etc., and generates an associated metadata file. The adapter encapsulates all data, including the original file, time-series CSV, and original file data, according to a pre-defined, lightweight standard exchange format to form a standardized data packet. Through the adapter and the standard encapsulation format, standard units with different interfaces and formats provide a unified input form for subsequent semantic processing.

[0048] Step S43: Process the standardized data package and the preset knowledge model to obtain a knowledge graph.

[0049] Step S43 further includes steps S431, S432, S433, and S434, which are specifically as follows:

[0050] Step S431: Read and parse the standardized data packet to obtain parsed data, which includes fields and content corresponding to structured data, time-series data, and unstructured data;

[0051] Step S432: Process the parsed data and the preset knowledge model to obtain path triples;

[0052] In this step, the semantic mapping engine loads standardized data packages and reads their contents. For example, when it reads "'hardness':280", the engine queries the domain ontology and finds the standard attribute name of the concept "hardness" as hasHardnessValue, with the unit HB. The value "280" of the source field hardness is converted into a triple form "wheel_W-001, hasHardnessValue,280^^xsd:integer" and associated with the unit HB. The crack_image_01.jpg and defect annotation information are also converted into triples. The preset knowledge model adopts the form of an attribute graph, with wheel and axle entities, components, events, observation parameters, etc. as nodes. All extracted triples are inserted into a graph database in real time or in batches, forming a continuously growing "wheel and axle knowledge graph". Through the semantic mapping engine, based on the pre-established domain ontology, the labels and fields in the source data are automatically mapped to the standard concepts and attributes in the unified data management, solving the problems of data name disputes and synonyms.

[0053] Step S433: Write the path triples into the graph database to construct the basic framework of the knowledge graph;

[0054] In this step, all generated triples are submitted to the graph database in batches. The graph database creates or updates nodes and relation edges based on the subject-verb-object structure of the triples, forming a dynamic knowledge graph with the wheel-axis entity as the central node. At the same time, full-text indexes and spatial indexes are built to support fast retrieval.

[0055] Step S434: Process the knowledge graph according to the knowledge graph basic framework to obtain the knowledge graph.

[0056] In this step, for unstructured data such as flaw detection images, after completing the pre-processing triplet transformation, the system additionally routes the data to an AI feature extraction service. It then calls a pre-trained computer vision model to infer the image and automatically outputs structured detection results such as crack location coordinates, crack type, confidence level, and estimated length. These results are converted into RDF triples and stored in association with the original image path. By constructing a "wheel and axle digital twin" knowledge graph, not only is a deep association between structured and unstructured data achieved, but traditionally difficult-to-use images, documents, and other data are transformed into analyzable structured data. This significantly improves the integrity, consistency, and usability of the data, providing a high-quality data foundation for building a complete wheel and axle digital twin model.

[0057] Step S5: Construct a prediction model based on the knowledge graph;

[0058] In this step, heterogeneous data from multiple sources, such as design parameters, manufacturing processes, service load spectra, vibration time series data, structured results of flaw detection images, and maintenance records, are transformed into unified RDF triples through adapter encapsulation and semantic mapping. These triples are then stored in a graph database to form a knowledge graph centered on the wheel axle, thereby constructing a complete, interconnected, and standardized initial prediction model and solving the problems of data heterogeneity and fragmentation.

[0059] Step S6: Perform optimization feedback processing on the prediction model to obtain the wheel and axle full life cycle maintenance model.

[0060] Step S6 further includes steps S61, S62, S63, and S64, which are specifically as follows:

[0061] Step S61: Obtain the axle factory serial number and fault event;

[0062] In this step, failure events include records from the maintenance depot, such as an axle that, after 450,000 kilometers of service, was found to have a 15mm transverse crack 5mm below the tread during flaw detection.

[0063] Step S62: Based on the knowledge graph, process the axle manufacturing serial number to obtain a digital identity identifier and an initial node;

[0064] In this step, a new digital profile is created for each physical axle based on its factory serial number. A node of type "axle" is created in the knowledge graph, with attributes digitalID:987654321, physicalCode:W-001. When subsequent axles go through any stage of design, manufacturing, service, or maintenance, all relevant data must be associated with this DigitalID:987654321 when entering the system, obtaining a digital identity identifier for each physical axle and the initial node bound to it in the knowledge graph. This creates a unique ID for each physical axle. All data scattered across different times and systems are precisely linked together through this ID, forming an unbreakable digital thread. This ID vertically links all information such as the design bill of materials, manufacturing batches, vehicle loading history, real-time sensor data, and all maintenance records, forming a complete digital twin.

[0065] Step S63: Process the fault event, the digital identity identifier and the initial node to obtain a design feedback data packet, which includes the original fault data and the comparison data;

[0066] In this step, when the background monitoring service detects a new node of type "fault event" in the knowledge graph, once it finds the record "When the axle has been in service for 450,000 kilometers, flaw detection revealed a 15mm transverse crack 5mm below the tread," the system immediately uses the axle's digital identifier as the center, aggregates its actual cumulative load spectrum for the past 12 months from the graph, and the original fatigue assumption load spectrum of the axle number. It then calls a load spectrum comparison algorithm, such as the KL divergence algorithm, to calculate the distribution difference between the two. The "fault event + actual load spectrum + design load spectrum + difference" is encapsulated into a structured report and transmitted through the data space channel. The feedback interface automatically sends data to the design department. By comparing newly added fault events with the data in the associated nodes and encapsulating them into a design feedback report data package, the actual failure evidence detected by the backend is automatically and reliably fed back in a format and path that can be directly used by the frontend design system. A large number of actual load spectra, typical failure modes and occurrence mileage of axles under the same conditions are fed back to the simulation model of the design department. This is used to correct load assumptions, optimize fatigue life design and reliability growth, realize the assessment of axle health status and prediction of remaining life, and transform design iteration from experience-driven to data-driven, forming a quantitative closed loop and improving the economic benefits of the entire life cycle.

[0067] Step S64: Optimize the prediction model based on the design feedback data package to obtain the wheel and axle full life cycle maintenance model;

[0068] Step S64 further includes steps S641, S642, S643, and S644, specifically as follows:

[0069] Step S641: Obtain a segment of vibration timing data prior to the occurrence of the fault event;

[0070] In this step, the timing data of the same wheel axis mainline of the newly added "fault event" node in the above steps is obtained for a period of time in the previous period.

[0071] Step S642: Process the predicted model and the vibration time series data to obtain fault samples;

[0072] In this step, the model training service extracts vibration data from the 30 days prior to the fault based on the vibration time series data and marks it as a "positive sample," i.e., a fault sample.

[0073] Step S643: Extract vibration data from healthy wheel axles to obtain healthy samples;

[0074] In this step, vibration data from the same axle or other healthy axles are randomly extracted and labeled as "negative samples," i.e., healthy samples.

[0075] Step S644: Based on the fault samples, the healthy samples and the prediction model, perform incremental training to obtain the wheel and axle full life cycle maintenance model.

[0076] In this step, time-domain features such as root mean square, peak value, and kurtosis, as well as frequency-domain features such as specific frequency band energy, are calculated from the original vibration data in the positive and negative samples. After generating feature vectors, they are input into the deployed prediction model for incremental training. The model parameters are fine-tuned according to the new data distribution. The updated performance is evaluated on the validation set. For example, if the accuracy improves by 2%, or if the performance improves to a threshold, the new model version is automatically deployed to the online prediction service, resulting in a more accurate prediction model with updated parameters. The fault warnings predicted based on monitoring data are compared with the facts that occur during the actual maintenance process. The newly generated fault samples are continuously used to periodically iterate and optimize the predictive maintenance algorithm model, making the prediction model smarter with use. The fault facts discovered during the maintenance process are continuously used to correct and improve the model, making the next prediction more accurate and improving the model's accuracy.

[0077] See Figure 2 The workflow of the wheel and axle full life cycle maintenance model described in this application includes: First, participating parties join the consortium blockchain, deploy local data agents, and register their available data assets to the global directory. Then, the data provider selects data assets, configures sharing strategies, selects or edits smart contract templates, and publishes authorization offers. Data users discover the required data assets through the directory, submit usage applications, trigger smart contract execution, and after contract verification, the data is delivered to the user through a secure channel. At the same time, the original data or the results after privacy processing can be used for collaborative analysis. Finally, the metadata, contract status, and access logs of the entire process are synchronously recorded on the blockchain for regulatory and audit queries. Through the above technical solution, this invention constructs a complete system from trusted data connection to the data application layer, providing a solution for the digital transformation of wheel and axle full life cycle management.

[0078] This invention addresses the core issues of trust deficiency, semantic interaction bottlenecks, and data fragmentation in the entire lifecycle data management of wheel and axle from three levels: trust mechanism establishment, data fusion processing, and model closed-loop optimization. By constructing a trustworthy, interoperable, and intelligent knowledge model, it not only improves the maintenance efficiency and safety of individual wheel and axle, but also promotes collaborative updates and continuous optimization throughout the entire lifecycle of wheel and axle.

[0079] Example 2

[0080] This invention provides a system for constructing a wheel and axle full life cycle maintenance model. The system includes an acquisition module, a first processing module, a second processing module, a fourth processing module, and a fifth processing module, specifically comprising:

[0081] The acquisition module is used to acquire raw data and data access requests, wherein the raw data includes flaw detection image files obtained during wheel and axle maintenance;

[0082] The first processing module is used to process the data access request and the original data to obtain a temporary authorization instruction;

[0083] The second processing module is used to obtain the encrypted data packet based on the temporary authorization instruction and extract the original heterogeneous data in the encrypted data packet. The original heterogeneous data includes various types of data with different data forms, formats and standards generated at each stage of the wheel axle's entire life cycle.

[0084] The third processing module is used to perform semantic standardization processing on the original heterogeneous data to obtain a knowledge graph, which includes all relevant data of the entire life cycle of the wheel and axle.

[0085] The fourth processing module is used to construct a prediction model based on the knowledge graph;

[0086] The fifth processing module is used to perform optimization feedback processing on the prediction model to obtain the wheel and axle full life cycle maintenance model.

[0087] In one specific embodiment of this disclosure, the first processing module includes a first processing unit, a second processing unit, and a judgment unit, specifically:

[0088] The first processing unit is used to construct a smart contract based on the original data;

[0089] The second processing unit is used to perform verification processing based on the smart contract and the data access request to obtain an authorization result;

[0090] The judgment unit is used to make a judgment based on the authorization result and obtain a temporary authorization instruction.

[0091] In one specific embodiment of this disclosure, the third processing module includes a first acquisition unit, a third processing unit, and a fourth processing unit, specifically:

[0092] The first acquisition unit is used to acquire a preset knowledge model;

[0093] The third processing unit is used to encapsulate the original heterogeneous data to obtain a standardized data packet, wherein the standardized data packet includes a metadata file associated with the original heterogeneous data.

[0094] The fourth processing unit is used to process the standardized data packet and the preset knowledge model to obtain a knowledge graph.

[0095] In one specific embodiment of this disclosure, the fifth processing module includes a second acquisition unit, a fifth processing unit, a sixth processing unit, and a seventh processing unit, specifically:

[0096] The second acquisition unit is used to acquire the wheel axle factory serial number and fault events;

[0097] The fifth processing unit is used to process the knowledge graph according to the axle manufacturing serial number to obtain a digital identity identifier and an initial node.

[0098] The sixth processing unit is used to process the fault event, the digital identity identifier and the initial node to obtain a design feedback data packet, which includes the original fault data and the comparison data.

[0099] The seventh processing unit is used to perform optimization feedback processing on the prediction model based on the design feedback data packet to obtain the wheel and axle full life cycle maintenance model.

[0100] In one specific embodiment of this disclosure, the seventh processing unit includes a third acquisition unit, an eighth processing unit, a ninth processing unit, and a tenth processing unit, specifically:

[0101] The third acquisition unit is used to acquire a period of vibration timing data before the fault event occurs.

[0102] The eighth processing unit is used to process the prediction model and the vibration time series data to obtain fault samples;

[0103] The ninth processing unit is used to extract vibration data from healthy wheel axles to obtain healthy samples;

[0104] The tenth processing unit is used to perform incremental training processing based on the fault samples, the healthy samples and the prediction model to obtain the wheel and axle full life cycle maintenance model.

[0105] In one specific embodiment of this disclosure, the fourth processing unit includes an eleventh processing unit, a twelfth processing unit, a thirteenth processing unit, and a fourteenth processing unit, specifically as follows:

[0106] The eleventh processing unit reads and parses the standardized data packet to obtain parsed data, which includes fields and contents corresponding to structured data, time-series data, and unstructured data.

[0107] The twelfth processing unit processes the parsed data and the preset knowledge model to obtain path triples.

[0108] The thirteenth processing unit writes the path triples into the graph database to construct the basic framework of the knowledge graph.

[0109] The fourteenth processing unit processes the knowledge graph according to the knowledge graph framework to obtain the knowledge graph.

[0110] The above description is merely a preferred embodiment of the present invention and is not intended to limit the invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

[0111] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A method for constructing a full life-cycle maintenance model for wheel and axle, characterized in that, include: Acquire raw data and data access requests, wherein the raw data includes flaw detection image files obtained during wheel and axle maintenance; The data access request and the original data are processed to obtain a temporary authorization instruction; Based on the temporary authorization instruction, an encrypted data packet is obtained, and the original heterogeneous data in the encrypted data packet is extracted. The original heterogeneous data includes various types of data with different data forms, formats and standards generated at each stage of the wheel and axle's entire life cycle. Based on the original heterogeneous data, semantic standardization processing is performed to obtain a knowledge graph, which includes all relevant data for the entire life cycle of the wheel and axle. Based on the knowledge graph, a prediction model is constructed; The prediction model is optimized and feedback is applied to obtain the wheel and axle full life cycle maintenance model.

2. The method for constructing the wheel and axle full life cycle maintenance model according to claim 1, wherein the data access request and the original data are processed, characterized in that, include: Construct a smart contract based on the original data; The authorization result is obtained by verifying the smart contract and the data access request. Based on the authorization result, a temporary authorization instruction is obtained.

3. The method for constructing the wheel and axle full life cycle maintenance model according to claim 1, characterized in that, semantic standardization processing is performed based on the original heterogeneous data, include: Acquire a pre-defined knowledge model; The original heterogeneous data is encapsulated to obtain a standardized data packet, which includes a metadata file associated with the original heterogeneous data. The knowledge graph is obtained by processing the standardized data package and the preset knowledge model.

4. The method for constructing the wheel and axle full life cycle maintenance model according to claim 1, wherein a prediction model is constructed based on the knowledge graph, and the prediction model is optimized and fed back based on fault events, characterized in that, include: Obtain the axle's factory serial number and fault events; Based on the knowledge graph, the axle manufacturing serial number is processed to obtain a digital identity identifier and an initial node; The fault event, the digital identity identifier, and the initial node are processed to obtain a design feedback data packet, which includes the original fault data and the comparison data; Based on the design feedback data package, the prediction model is optimized and fed back to obtain the wheel and axle full life cycle maintenance model.

5. The method for constructing the wheel and axle full life cycle maintenance model according to claim 4, wherein the prediction model is optimized and fed back based on the design feedback data package, characterized in that, include: Acquire vibration time-series data prior to the occurrence of the fault event; The fault samples are obtained by processing the prediction model and the vibration time series data. Vibration data of healthy wheel axles are extracted to obtain healthy samples; Incremental training is performed on the fault samples, the healthy samples, and the prediction model to obtain a wheel and axle full life cycle maintenance model.

6. A system for constructing a full life-cycle maintenance model for wheel and axle, characterized in that, include: The acquisition module is used to acquire raw data and data access requests, wherein the raw data includes flaw detection image files obtained during wheel and axle maintenance; The first processing module is used to process the data access request and the original data to obtain a temporary authorization instruction; The second processing module is used to obtain the encrypted data packet based on the temporary authorization instruction and extract the original heterogeneous data in the encrypted data packet. The original heterogeneous data includes various types of data with different data forms, formats and standards generated at each stage of the wheel axle's entire life cycle. The third processing module is used to perform semantic standardization processing on the original heterogeneous data to obtain a knowledge graph, which includes all relevant data of the entire life cycle of the wheel and axle. The fourth processing module is used to construct a prediction model based on the knowledge graph; The fifth processing module is used to perform optimization feedback processing on the prediction model to obtain the wheel and axle full life cycle maintenance model.

7. The system for constructing a wheel and axle full life cycle maintenance model according to claim 6, wherein the first processing module is characterized in that, include: The first processing unit is used to construct a smart contract based on the original data; The second processing unit is used to perform verification processing based on the smart contract and the data access request to obtain an authorization result; The judgment unit is used to make a judgment based on the authorization result and obtain a temporary authorization instruction.

8. The system for constructing a wheel and axle full life cycle maintenance model according to claim 6, wherein the third processing module is characterized in that, include: The first acquisition unit is used to acquire a preset knowledge model; The third processing unit is used to encapsulate the original heterogeneous data to obtain a standardized data packet, wherein the standardized data packet includes a metadata file associated with the original heterogeneous data. The fourth processing unit is used to process the standardized data packet and the preset knowledge model to obtain a knowledge graph.

9. The system for constructing a wheel and axle full life cycle maintenance model according to claim 6, wherein the fifth processing module is characterized in that, include: The second acquisition unit is used to acquire the wheel axle factory serial number and fault events; The fifth processing unit is used to process the knowledge graph according to the axle manufacturing serial number to obtain a digital identity identifier and an initial node. The sixth processing unit is used to process the fault event, the digital identity identifier and the initial node to obtain a design feedback data packet, which includes the original fault data and the comparison data. The seventh processing unit is used to perform optimization feedback processing on the prediction model based on the design feedback data packet to obtain the wheel and axle full life cycle maintenance model.

10. The system for constructing a wheel and axle full life cycle maintenance model according to claim 9, wherein the seventh processing unit is characterized in that, include: The third acquisition unit is used to acquire a period of vibration timing data before the fault event occurs. The eighth processing unit is used to process the prediction model and the vibration time series data to obtain fault samples; The ninth processing unit is used to extract vibration data from healthy wheel axles to obtain healthy samples; The tenth processing unit is used to perform incremental training processing based on the fault samples, the healthy samples and the prediction model to obtain the wheel and axle full life cycle maintenance model.

Images