An aviation equipment manufacturing digital main line engine system supporting multi-source heterogeneous data access

Abstract

This invention relates to a digital mainline engine system for aerospace equipment manufacturing that supports multi-source heterogeneous data access, belonging to the field of aerospace manufacturing data management technology. The system includes a multi-source heterogeneous data access adaptation module, a data fusion processing module, and a full-process data link construction module. The multi-source heterogeneous data access adaptation module utilizes data interface protocols and format conversion components; the data standardization and fusion processing module leverages a data element standard library in the aerospace manufacturing field and completes correlation and fusion through an aerospace heterogeneous data collaborative computing framework to generate a unified heterogeneous data model. This framework includes a data sharding layer, a parallel node layer, and a result aggregation layer; the full-process data link construction module constructs a data link covering the entire product lifecycle based on a time-series correlation algorithm and a product unique identifier mapping mechanism. The system effectively solves the problem of integrating and managing multi-source heterogeneous data in aerospace equipment manufacturing, improving data processing efficiency and full-process data traceability capabilities.

Description

Technical Field

[0001] This invention belongs to the field of aviation manufacturing data management technology, specifically relating to an aviation equipment manufacturing digital mainline engine system that supports multi-source heterogeneous data access. Background Technology

[0002] The entire lifecycle of aerospace equipment manufacturing encompasses multiple stages, including design, process, and production. This process generates heterogeneous data from various sources, such as 3D models, process parameters, inspection reports, and operational data streams. These data originate from different platforms, including CAD and MES. Significant differences in data formats and semantics, coupled with a lack of unified standards, make data association and sharing difficult. Traditional computing architectures struggle to handle the demands of massive data processing, resulting in high latency in data fusion and analysis. Furthermore, the fragmented storage of data across different stages leads to broken data links throughout the entire process, resulting in insufficient data traceability and decision support capabilities. Existing solutions are either limited to point-to-point transmission of specific data or lack customized design for the aerospace field, failing to meet the requirements for standardized fusion and end-to-end connectivity of heterogeneous multi-source data. Therefore, a corresponding digital mainline engine system is urgently needed to overcome this bottleneck. Summary of the Invention

[0003] To address the aforementioned problems in the existing technology, this invention provides a digital mainline engine system for aviation equipment manufacturing that supports multi-source heterogeneous data access;

[0004] The objective of this invention can be achieved through the following technical solutions:

[0005] A digital mainline engine system for aerospace equipment manufacturing that supports multi-source heterogeneous data access is characterized by comprising: a multi-source heterogeneous data access adaptation module, a data fusion processing module, and a full-process data link construction module.

[0006] The multi-source heterogeneous data access adaptation module is configured with data interface protocols and format conversion components to receive 3D model data, structured process parameters, unstructured test reports, and real-time operating condition data streams.

[0007] The data standardization and fusion processing module incorporates a data element standard library for the aerospace manufacturing field. It performs standardization transformation on the heterogeneous data through semantic mapping and rule engine. It adopts an aerospace heterogeneous data collaborative computing framework to complete the association and fusion of heterogeneous data. The aerospace heterogeneous data collaborative computing framework includes a data sharding layer, a parallel node layer, and a result aggregation layer. The data sharding layer shards heterogeneous data according to data type and time dimension. The parallel node layer performs feature extraction and association rule matching on the sharded data at the same time. The result aggregation layer integrates the processing results of each node to generate a unified heterogeneous data model, which includes data identifier, association weight, and timestamp information.

[0008] The end-to-end data link construction module builds a full lifecycle data link based on time-series association algorithms and product unique identifier mapping mechanisms.

[0009] Specifically, the data sharding layer of the aviation heterogeneous data collaborative computing framework includes a data feature identification component and a dynamic sharding strategy component. The data feature identification unit extracts the type labels, timestamps and data volume features of heterogeneous data. The dynamic sharding strategy component performs sharding operations based on the feature analysis results, according to the data type homogeneity and time continuity, and the sharding granularity is adaptively adjusted.

[0010] Specifically, the parallel processing node layer of the aviation heterogeneous data collaborative computing framework adopts a master-slave architecture. The master node is responsible for sharding task scheduling and node status monitoring, while the slave nodes deploy feature extractors and association rule engines. The feature extractors extract geometric topological features from 3D model data and process association features from process parameters. The association rule engine has a built-in association pattern mining algorithm.

[0011] Specifically, the result aggregation layer of the aviation heterogeneous data collaborative computing framework includes a data consistency verification component and an association weight calculation component. The data consistency verification component identifies abnormal data based on comparing the processing results of multiple nodes, and the association weight calculation component calculates the association strength between different types of data based on the cosine similarity algorithm to generate weight values ​​in the range of 0-1.

[0012] Specifically, the data sharding layer is configured with a sharding balancer to monitor the differences in heterogeneous data volume among the shards in real time, automatically trigger data redistribution, and ensure load balancing among the parallel nodes.

[0013] Specifically, the slave nodes of the parallel processing node layer are equipped with a fault-tolerant mechanism, so that when a node fails, the master node migrates its tasks to a backup node.

[0014] Specifically, the data standardization and fusion processing module's data element standard library for the aerospace manufacturing field is built based on the SAEAS5580 standard, including standardized data element definitions for material grades, heat treatment processes, and non-destructive testing methods.

[0015] Specifically, the time-series correlation algorithm is based on a time-series link learning model. By learning the timestamp sequence characteristics of heterogeneous data, it establishes the time dependency relationship between heterogeneous data in different processes and predicts the time deviation range of data flow.

[0016] Specifically, the product unique identifier mapping mechanism is based on the unique identifier of the AS9132 standard, and uses blockchain technology to store the mapping relationship between the identifier and heterogeneous data to ensure that the mapping record cannot be tampered with.

[0017] Specifically, the rule engine has a built-in business rule library for the aerospace manufacturing field, including process compliance verification rules and data correlation verification rules, and performs add, delete, modify and query operations on rules according to actual business needs.

[0018] The beneficial effects of this invention are as follows:

[0019] This invention provides a digital mainline engine system for aerospace equipment manufacturing that supports multi-source heterogeneous data access. It efficiently integrates various types of data through a multi-source heterogeneous data access adaptation module, solving the problems of inconsistent data formats and interoperability across different platforms. Leveraging a data element standard library, semantic mapping, rule engine, and a collaborative computing framework for heterogeneous aerospace data in the aerospace manufacturing field, it improves data standardization and fusion quality, generating a unified heterogeneous data model. A full-process data link construction module builds a fully traceable data link, utilizing blockchain technology to ensure the immutability of mapping records and enhance data traceability capabilities. The system's stability and adaptability are enhanced by a data sharding layer equalizer, a parallel node layer fault tolerance mechanism, and the flexibility of the rule engine. Simultaneously, the system's effective data processing and full-process management provide comprehensive and accurate data support for decisions such as process optimization and production scheduling, improving the level of intelligence in aerospace equipment manufacturing. Attached Figure Description

[0020] To facilitate understanding by those skilled in the art, the present invention will be further described below with reference to the accompanying drawings.

[0021] Figure 1 This is a flowchart illustrating a digital mainline engine system for aircraft equipment manufacturing that supports multi-source heterogeneous data access, according to the present invention. Detailed Implementation

[0022] To further illustrate the technical means and effects of the present invention in achieving its intended purpose, the following detailed description of the specific implementation methods, structures, features, and effects of the present invention, in conjunction with the accompanying drawings and preferred embodiments, is provided.

[0023] Please see Figure 1 A digital mainline engine system for aerospace equipment manufacturing that supports access to multi-source heterogeneous data includes: a multi-source heterogeneous data access adaptation module, a data fusion processing module, and a full-process data link construction module;

[0024] The multi-source heterogeneous data access adaptation module is configured with a data interface protocol and format conversion component to receive 3D model data, structured process parameters, unstructured test reports and real-time operating condition data streams.

[0025] The data standardization and fusion processing module incorporates a data element standard library for the aerospace manufacturing field. It performs standardization transformation on the accessed heterogeneous data through semantic mapping and a rule engine. An aerospace heterogeneous data collaborative computing framework is used to complete the association and fusion of the heterogeneous data. This framework includes a data sharding layer, a parallel node layer, and a result aggregation layer. The data sharding layer shards the heterogeneous data according to data type and time dimension. The parallel node layer simultaneously performs feature extraction and association rule matching on the sharded data. The result aggregation layer integrates the processing results of each node to generate a unified heterogeneous data model, including data identifiers, association weights, and timestamp information.

[0026] The end-to-end data link construction module constructs a full lifecycle data link based on a time-series association algorithm and a product unique identifier mapping mechanism.

[0027] Specifically, the multi-source heterogeneous data access adaptation module refers to a module used to connect different data sources and convert data formats. It is implemented based on data interface protocols and format conversion components. By adapting to the communication protocols and format differences of different data sources, it solves the problem of multi-source heterogeneous data access.

[0028] Specifically, the data fusion processing module refers to the module that associates and integrates standardized data. It is based on the data element standard library and semantic mapping rule engine in the aviation manufacturing field. By eliminating semantic differences and unifying data element definitions, it ensures the standardized fusion of heterogeneous data.

[0029] Specifically, the aviation heterogeneous data collaborative computing framework refers to a distributed architecture that enables efficient processing of massive heterogeneous data. It is implemented using a layered structure of data sharding layer, parallel node layer, and result aggregation layer. Data processing efficiency is improved through sharding and parallel computing, and a unified data model is generated through the aggregation layer.

[0030] Specifically, the data sharding layer divides heterogeneous data into data blocks that can be processed in parallel, shards them according to data type and time dimension, and optimizes the load balancing of parallel nodes and the utilization of computing resources by dynamically adjusting the sharding granularity.

[0031] The data sharding layer of the aerospace heterogeneous data collaborative computing framework includes a data feature recognition component and a dynamic sharding strategy component. The data feature recognition component employs machine learning algorithms, such as decision trees or support vector machines, to extract features from the input heterogeneous data. It identifies data type labels, such as 3D models, process parameters, and inspection reports, extracts timestamp information, and statistically analyzes data volumetric features, such as the number of data entries and file size. The dynamic sharding strategy component then formulates an adaptive sharding strategy based on the feature analysis results. Data of the same type are grouped into the same shard; data that are consecutive or close in time are also grouped into the same shard.

[0032] Specifically, the parallel node layer executes a cluster of computing nodes for feature extraction and association rule matching, employing a master-slave architecture to deploy the feature extractor and association rule engine. The master node is configured with a high-performance server, responsible for task scheduling and node status monitoring. Slave nodes are deployed on multiple computers, each with the feature extractor and association rule engine installed. The feature extractor extracts geometric topological features from the 3D model data, including boundary representations, surface information, and feature point coordinates of the parts; it also extracts process-related features from the process parameters, including the sequence of processing steps and trends in process parameter changes. The association rule engine incorporates an association pattern mining algorithm to uncover implicit relationships between data. The master node allocates computing tasks to slave nodes through a task queue management system and monitors the slave node status in real time through a heartbeat mechanism. After receiving a task, the slave node first performs feature extraction, then runs the association rule engine for pattern matching, and finally returns the processing result to the master node.

[0033] Specifically, the result aggregation layer integrates the processing results of each node. The result aggregation layer includes a data consistency verification component and a correlation weight calculation component. The data consistency verification component identifies abnormal data by comparing the processing results of multiple nodes. For each data item, it collects the processing results of all parallel nodes and calculates the mean and standard deviation of the results. Results deviating from the mean by more than three times the standard deviation are marked as potentially abnormal data. The correlation weight calculation component calculates the correlation strength between different types of data based on the cosine similarity algorithm, generating weight values ​​in the 0-1 interval. For 3D model data and process parameter data, their respective feature vectors are extracted. The dot product of the two vectors is calculated and divided by the product of their magnitudes to obtain the cosine similarity value. This value is mapped to the 0-1 interval as the correlation weight between the two types of data.

[0034] In this embodiment, taking the processing of multi-source heterogeneous data in the manufacturing process of high-pressure turbine blades for aero-engines as an example, the complete operation flow of the aerospace heterogeneous data collaborative computing framework is shown as follows:

[0035] The data sharding layer first receives four types of core data:

[0036] 3D model data: Formatted in STEP AP242, including a 3D solid model of the blade (containing 1287 geometric feature surfaces and 362 dimensional constraints) and material properties (GH4169 high-temperature alloy, density 8.2 g / cm³). 3 The file contains assembly baseline information, with a file size of 2.4GB and a timestamp accurate to 2024-08-01 09:00:15.327.

[0037] Structured process parameters: stored in the database, in the form of a process card consisting of 87 records, covering roughing, heat treatment, finishing, etc., timestamp 2024-08-01 10:30:08.751.

[0038] Unstructured inspection report: consists of a PDF document (including 3 pages of text description) + high-resolution image (4096×3072 pixels, close-up of scratches in the leaf tip area), recorded inspection time 2024-08-01 14:00:22, defect level is rated as "minor" (compliant with HB5226-2020 standard).

[0039] Real-time operating condition data stream: collected from a five-axis machining center, including spindle temperature, X / Y / Z axis vibration values, feed motor current, etc., with a time series covering 11:00:00 to 12:00:00 on August 1, 2024.

[0040] Data feature recognition component startup in-depth analysis:

[0041] Geometric feature extraction is performed on the 3D model, and the "Design Class - Turbine Blade - GH4169" type label is automatically assigned. Key topological structures are preserved through lightweight processing.

[0042] Semantic annotation is performed on process parameters to generate composite tags of "process type - machining - heat treatment" and to establish a mapping relationship between process number (such as P03) and parameter value;

[0043] The inspection report is processed using OCR text recognition and image feature extraction to generate a "Quality Category - Appearance Inspection - Scratch" label, and the defect coordinates are located (relative to the blade tip reference point X: 12.3mm, Y: -8.7mm).

[0044] The operating data was segmented into time series and categorized by the label "Production-5-axis Machining-Real-time Monitoring". Three abnormal fluctuation periods were identified (11:15-11:20 temperature surge).

[0045] The dynamic sharding strategy component enables multi-dimensional sharding:

[0046] Based on type isomorphism: "Design category + process category" is merged into segment S1 (due to the existence of direct design-process relationship), accounting for 38% of the data volume;

[0047] Based on temporal continuity: "Production" data is divided into S2 (11:00-11:30) and S3 (11:30-12:00) in 30-minute windows, with each window containing 18,000 records;

[0048] Based on the specific characteristics of the data: "Quality" is divided into S4 separately, preserving the original image and text association relationship;

[0049] The granularity of the fragments is dynamically adjusted: S1 uses 2000 records / fragment because it contains a 3D model, S2 / S3 uses 1000 records / fragment, and S4 uses 500 records / fragment because of its small data volume.

[0050] The parallel node layer adopts a "1 master, 4 slaves, 1 standby" architecture. The master node cluster manages the distribution of S1 to slave node 1, S2 to slave node 2, S3 to slave node 3, and S4 to slave node 4, and synchronously sends shard metadata (including data volume, feature tags, and processing priority).

[0051] Process S1 from node 1:

[0052] The feature extractor starts the CAD model parsing engine to extract the geometric topological features of the tenon: 10 tenon teeth, tooth angle 30°±0.5°, tooth pitch 4.2mm, and generates a lightweight model in STL format;

[0053] The process feature extraction module establishes a process dependency map and identifies a strong dependency relationship of "rough machining (P01) → flaw detection (P02) → finish machining (P03)" (confidence level 0.97).

[0054] The association rule engine runs an aerospace manufacturing association pattern mining algorithm, with a minimum support of 0.8 and a minimum confidence of 0.85. Through 12 rounds of iterative calculation, the rule "tenon angle 30° ∧ material GH4169 → finishing speed 5000 r / min" (support 0.89, confidence 0.92) is discovered and a visual association network graph is generated.

[0055] Process S2 from node 2:

[0056] The time-series feature extraction module uses a sliding window to calculate feature values: average temperature 52℃, vibration peak 3.2mm / s. 2 Current standard deviation 1.2A;

[0057] The anomaly detection submodule identifies the temperature anomaly between 11:18 and 11:20 (exceeding twice the standard deviation of the mean) using the 3σ criterion and marks it as an event requiring attention.

[0058] Association rule mining revealed: "Temperature > 60℃ ∧ Current > 10A → Vibration value > 3mm / s" 2 (Confidence level 0.88), and record the number of times the rule was triggered (5 times).

[0059] Process S3 from node 3:

[0060] Repeating the processing logic of S2, a similar temperature-vibration correlation pattern was found between 11:45 and 11:50, but the intensity was slightly weaker (confidence level 0.82).

[0061] The generated time-period comparison report indicates that the processing stability in the latter 30 minutes is better than that in the first 30 minutes (the fluctuation coefficient is reduced by 18%).

[0062] Process S4 from node 4:

[0063] The image feature extractor uses an edge detection algorithm to extract the scratch contour: length 3.2mm, maximum width 0.15mm, and direction at 30° to the blade axis;

[0064] The text semantic analysis module maps the defect description "minor scratches on the leaf tip" to the standard defect library (code Q012).

[0065] Establish a link between "scratch length - inspection time - inspector ID" to generate a quality feature vector.

[0066] During operation:

[0067] The load balancer monitors node load every 10 seconds and found that the GPU utilization of node 2 reached 92% (the average utilization of other nodes was 65%). It automatically migrated 15% of the data in S2 (from 11:25 to 11:30) to the backup node. The migration took 2.3 seconds. After balancing, the load difference between the nodes was less than 10%.

[0068] Simulate a sudden network outage on node 3 (fault injection test). The master node detects the loss of heartbeat within 380ms and immediately activates the standby node to take over S3 processing without losing any timing data.

[0069] As a result, the aggregation layer initiates a three-level processing flow:

[0070] Data consistency verification:

[0071] Calculate the hash value of the output results of the 4 slave nodes (e.g., hash of S1 result: a7b3...f2d9);

[0072] By comparing the shard hash digests pre-stored in the master node, it was found that the hash values ​​of the original result from slave node 3 and the recalculated result from the backup node were consistent (error <), confirming that the data had not been tampered with.

[0073] The system identifies a coordinate deviation (0.2mm) between the image and text description in S4 and automatically triggers a realignment process.

[0074] Association weight calculation:

[0075] An improved cosine similarity algorithm is adopted, and a weighting factor from the aerospace manufacturing field is introduced (a 20% weighting factor for design-process correlation):

[0076] 3D model and process parameters: Feature vector similarity 0.82×1.2=0.85 (meets the strong correlation standard);

[0077] Process parameters and production data: Calculated based on process time matching degree, the result is 0.72;

[0078] Production data and quality reports: 0.68 was obtained by correlating abnormal time periods with defect locations;

[0079] Design data and quality report: Indirect correlation weight 0.85×0.72×0.68=0.41.

[0080] Unified model generation:

[0081] Constructing a graph database storage structure:

[0082] Nodes: Design Model (ID:M20240801), Process Card (ID:P007), Production Period (ID:T11), Quality Report (ID:Q342);

[0083] Edges: include associated weights (e.g., edge weight 0.85 for M20240801-P007) and time differences (e.g., 30-minute interval for P007-T11).

[0084] Attributes: Each edge is appended with confidence score and computed timestamp metadata.

[0085] The final unified heterogeneous data model contains 28 nodes and 43 related edges, with a data compression ratio of 1:3.7, and can be directly imported into a digital platform for visualization.

[0086] The standardized data output by this process successfully supported the subsequent construction of the entire process data link: by linking design, process, production and quality data through the unique blade identifier (UID:B-20240801-003), the entire link traceability of "design parameters → process execution → production status → quality results" was realized, providing a quantitative basis for process optimization.

[0087] Specifically, the end-to-end data link construction module builds a full lifecycle data link based on a time-series correlation algorithm and a product unique identifier mapping mechanism. The time-series correlation algorithm establishes time-dimensional relationships, and the unique identifier mapping mechanism forms a cross-stage data traceability chain, ultimately constructing a data link covering the entire process of design, process, and production.

[0088] In this embodiment, the time-series correlation algorithm is based on the aviation time-series link learning model to perform in-depth mining of timestamp sequences in a unified heterogeneous data model:

[0089] Timestamp sequence extraction: Extract key timestamps from the data of each node to form a sequence set: Design completion (2024-08-01 09:00:15) → Process card release (2024-08-01 10:30:08) → Production and processing (2024-08-01 11:00:00-12:00:00) → Quality inspection (2024-08-01 14:00:22) → Warehousing (2024-08-02 08:15:00) → Installation and testing (2024-08-10 15:30:00) → Delivery and maintenance (2024-08-15 09:00:00).

[0090] Time dependency modeling: The sequence is trained using an LSTM neural network to calculate the probability distribution of time intervals for each stage. For example, the average interval from "process release to production start-up" is 26 minutes, with a standard deviation of 3.2 minutes and a 95% confidence interval of 20-32 minutes, which matches the actual interval (29 minutes), verifying the accuracy of the time series correlation.

[0091] Cross-process temporal correlation: Identify the temporal correlation between "abnormal production period (11:18-11:20) → quality inspection defect (blade tip scratch)", calculate the time difference as 2 hours and 42 minutes, and confirm the matching of the blade area processed during this period with the defect location in combination with the processing flow. Set the temporal correlation weight of the two to 0.76.

[0092] A unique identifier based on the AS9132 standard (UID: B-20240801-003) is used to achieve end-to-end data binding through blockchain technology.

[0093] Identifier Embedding and Mapping: UIDs are embedded in the 3D model metadata. Process cards form unique process identifiers using "process number + UID" (P007-B-20240801-003). Production data is recorded using "equipment ID + timestamp + UID" (MC01-202408011100-B-20240801-003), and inspection reports generate "inspection report number + UID" (Q342-B-20240801-003). Blockchain nodes (including one consensus node each from the design, production, and quality inspection departments) store the mapping relationships of these identifiers. Each mapping record contains a preceding hash, current data digest, and node signature to ensure immutability.

[0094] Cross-system identifier association: By associating UIDs with material codes in the Enterprise Resource Planning (ERP) system, work order numbers in the Manufacturing Execution System (MES) system, and asset numbers in the Operations and Maintenance (O&M) system, a cross-system identifier chain of "design-production-O&M" is formed. When querying a UID, the associated data in each system can be automatically traced.

[0095] Dynamic mapping update: Vibration monitoring data generated by the blades during the operation and maintenance phase is written into the chain through new blockchain transactions, extending the link to the operation and maintenance phase. The update of the mapping record needs to be confirmed by consensus of more than 3 nodes to ensure data integrity.

[0096] Link visualization: Generates a tree-shaped data link diagram with UID as the core. The main link is "Design → Process → Production → Inspection → Operation and Maintenance". The branch links contain sub-data of each link (such as equipment parameters and operation records in the production link). The links are marked with association weights (such as design-process 0.85, production-inspection 0.68) and time intervals, and support interactive expansion / collapse.

[0097] When blade performance degradation is detected during the operation and maintenance phase, it can be traced back by UID:

[0098] a. Call the design data to confirm the original geometric parameters;

[0099] b. Check the process card to verify whether the processing is in accordance with the standards;

[0100] c. Retrieve abnormal temperature data from the production period and combine it with scratch records from quality inspection to analyze the potential correlations that led to the defects;

[0101] d. Propose targeted maintenance solutions based on end-to-end data to shorten troubleshooting time.

[0102] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention in any way. Although the present invention has been disclosed above with reference to preferred embodiments, it is not intended to limit the present invention. Any person skilled in the art can make some modifications or alterations to the above-disclosed technical content to create equivalent embodiments without departing from the scope of the present invention. Any simple modifications, equivalent changes and alterations made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the scope of the present invention.

Claims

1. A digital mainline engine system for aerospace equipment manufacturing that supports multi-source heterogeneous data access, characterized in that, include: Multi-source heterogeneous data access and adaptation module, data standardization and fusion processing module, and end-to-end data link construction module; The multi-source heterogeneous data access adaptation module is configured with a data interface protocol and format conversion component to receive 3D model data, structured process parameters, unstructured test reports and real-time operating condition data streams. The data standardization and fusion processing module incorporates a data element standard library for the aerospace manufacturing field. It performs standardization transformation on the accessed heterogeneous data through semantic mapping and a rule engine. This data element standard library, built on the SAEAS5580 standard, includes standardized data element definitions for material grades and heat treatment processes. An aerospace heterogeneous data collaborative computing framework is used to complete the association and fusion of the heterogeneous data. This framework includes a data sharding layer, a parallel node layer, and a result aggregation layer. The data sharding layer shards the heterogeneous data according to data type and time dimension. The parallel node layer simultaneously performs feature extraction and association rule matching on the sharded data. The parallel node layer of the aerospace heterogeneous data collaborative computing framework adopts a master-slave architecture. The master node is responsible for sharding task scheduling and node status monitoring, while the slave nodes deploy feature extractors and association rule engines. The feature extractor extracts geometric topological features from 3D model data and process association features from process parameters. The association rule engine incorporates an association pattern mining algorithm. The result aggregation layer integrates the processing results of each node to generate a unified heterogeneous data model and constructs a graph database storage structure, which includes data identifiers, association weights and timestamp information. For 3D model data and process parameter data, the feature vectors of each are extracted and the cosine similarity value is calculated as the association weight between the two types of data. The end-to-end data link construction module constructs a full lifecycle data link based on a time-series association algorithm and a product unique identifier mapping mechanism.

2. The system according to claim 1, characterized in that, The data sharding layer of the aviation heterogeneous data collaborative computing framework includes a data feature identification component and a dynamic sharding strategy component. The data feature identification unit extracts the type label, timestamp, and data volume features of the heterogeneous data. The dynamic sharding strategy component performs sharding operations based on the feature analysis results, according to the data type homogeneity and time continuity, and the sharding granularity is adaptively adjusted.

3. The system according to claim 1, characterized in that, The result aggregation layer of the aviation heterogeneous data collaborative computing framework includes a data consistency verification component and an association weight calculation component. The data consistency verification component identifies abnormal data based on the comparison of multi-node processing results, and the association weight calculation component calculates the association strength between different types of data based on the cosine similarity algorithm and generates weight values ​​in the range of 0-1.

4. The system according to claim 1, characterized in that, The data sharding layer is configured with a sharding balancer to monitor the differences in the amount of heterogeneous data in each shard in real time, automatically trigger data redistribution, and ensure load balancing of each parallel node.

5. The system according to claim 1, characterized in that, The slave nodes of the parallel node layer are equipped with a fault-tolerant mechanism. When a node fails, the master node migrates its tasks to a backup node.

6. The system according to claim 1, characterized in that, The data standardization and fusion processing module's aerospace manufacturing data element standard library is built based on the SAE AS5580 standard, including standardized data element definitions for material grades, heat treatment processes, and non-destructive testing methods. The non-destructive testing methods are defined according to aerospace industry standards.

7. The system according to claim 1, characterized in that, The time-series correlation algorithm is based on a time-series link learning model. By learning the timestamp sequence characteristics of the heterogeneous data, it establishes the time dependency relationship between heterogeneous data in different processes and predicts the time deviation range of data flow.

8. The system according to claim 1, characterized in that, The product unique identifier mapping mechanism is based on the unique identifier of the AS9132 standard and uses blockchain technology to store the mapping relationship between the identifier and the heterogeneous data.

9. The system according to claim 1, characterized in that, The semantic mapping and rule engine includes a built-in business rule library for the aerospace manufacturing field, including process compliance verification rules and data correlation verification rules, which can perform add, delete, modify and query operations on rules according to actual business needs.

Images