System capable of efficiently processing massive aviation bus data
By constructing a distributed processing architecture and adopting parallel acquisition using multiple bus protocols and Spark in-memory computing, the bottleneck problem of the existing aviation bus data processing system has been solved, and efficient and real-time aviation bus data processing has been achieved.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-12-11
- Publication Date
- 2026-03-10
AI Technical Summary
Existing avionics bus data processing systems suffer from bottlenecks in bandwidth, real-time performance, flexibility, and resource utilization, making it difficult to meet the demands of next-generation aviation systems for efficient data processing.
A distributed processing architecture is constructed by employing a multi-bus protocol parallel acquisition module, a heterogeneous data parsing engine, a business computing and statistical analysis module, a hierarchical storage management module, and a resource scheduling and load management module. Through many-core parallel acquisition and preprocessing, combined with Spark in-memory computing and adaptive resource scheduling, hardware and software collaborative processing is achieved.
It significantly improved data throughput and real-time performance, achieved low latency and high throughput, optimized resource utilization and system elasticity, and met the high-efficiency processing requirements of massive aviation bus data.
Smart Images
Figure CN121636115A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the fields of avionics technology and big data processing, and in particular to a system capable of efficiently processing massive amounts of avionics bus data. Background Technology
[0002] Existing avionics bus data processing technologies primarily rely on a centralized processing architecture based on dedicated interface cards (such as FC and 1553B boards) and general-purpose processors. The core of this architecture is the parsing, transformation, and basic analysis of raw data on the bus based on a predefined Interface Control Document (ICD) database. These systems typically employ modular software designs, possessing real-time monitoring and post-analysis capabilities, and complete tasks through multi-process collaboration. When processing specific protocols (such as ARINC659 and CAN), table-driven architectures and multiple redundancy backups are used to ensure deterministic data transmission and high reliability, thereby meeting the basic processing requirements of traditional avionics systems for mission-critical data.
[0003] The main drawback of existing technologies lies in their bottlenecks in processing power in terms of bandwidth, real-time performance, flexibility, and resource utilization. These shortcomings stem from the fact that the traditional bus protocols they rely on (such as 1553B) have limited bandwidth and high protocol overhead. Furthermore, centralized, standalone system architectures cannot fully utilize the parallel computing capabilities of modern hardware (such as multi-core and distributed systems), resulting in poor scalability and low processing efficiency when faced with massive amounts of heterogeneous bus data. Consequently, they are unable to meet the growing demands of next-generation aviation systems for efficient data processing. Summary of the Invention
[0004] The purpose of this invention is to provide a system capable of efficiently processing massive amounts of aviation bus data, solving the problems of insufficient real-time performance, limited protocol compatibility, and low system resource utilization in existing technologies for aviation multi-bus data processing.
[0005] To solve the above-mentioned technical problems, the present invention adopts the following technical solution: A system capable of efficiently processing massive amounts of aviation bus data includes a multi-bus protocol parallel acquisition module, a heterogeneous data parsing engine, a business computing and statistical analysis module, a hierarchical storage management module, and a resource scheduling and load management module. The multi-bus protocol parallel acquisition module includes a protocol interface array, a many-core processing core, and a dynamic buffer management unit. The heterogeneous data parsing engine is connected to the multi-bus protocol parallel acquisition module and includes a configurable protocol parser, a data routing and framing unit, and a streaming data processing pipeline. The business computing and statistical analysis module comprises a distributed real-time computing unit, a multi-layer statistical analysis unit, and a semantic analysis unit. The hierarchical storage management module adopts a hot, warm, and cold three-tier storage architecture, with the hot data memory storage area, warm data solid-state storage area, and cold data disk storage area used to store real-time processed data, recent historical data, and long-term archived data, respectively. The resource scheduling and load management module includes a load prediction unit, an adaptive resource allocation unit, and a model migration and knowledge sharing unit.
[0006] This solution abandons the traditional model of relying on general-purpose processors for serial, centralized processing, and constructs a new distributed processing paradigm of "software and hardware collaboration and data-driven". This solution uses many-core cores to perform parallel acquisition and preprocessing of multiple bus protocols, replacing the traditional software polling. In terms of logical operations, it introduces configuration-based dynamic parsing and a two-level data routing mechanism, replacing the rigid hard-coded parsing logic. In terms of algorithms, the core adopts Spark in-memory computing for distributed parallel processing, combined with an adaptive resource scheduling algorithm based on load prediction, which completely changes the static and passive resource allocation method.
[0007] Employing parallel processing, stream computing, and resource scheduling, it boasts significant technical advantages: hardware parallelization directly leverages parallel processing theory, offloading protocol processing tasks to many-core devices, greatly reducing CPU overhead and transmission latency, fundamentally improving data throughput and real-time performance; the Spark-based stream processing architecture ensures low latency and high throughput in data processing, and combined with configurable routing logic, it achieves a transformation from "full slow storage" to "precise fast acquisition"; while the intelligent resource scheduling model applies control theory and machine learning, enabling the system to dynamically perceive load and predict future demand, thereby achieving optimal resource allocation, ultimately achieving a comprehensive and extreme improvement in processing efficiency, system elasticity, and resource utilization.
[0008] The tiered storage management module adopts a three-tiered storage architecture of hot, warm, and cold storage. This tiered storage strategy optimizes the utilization efficiency of storage resources while ensuring high data reliability and availability.
[0009] As a further preferred embodiment of the present invention, the protocol interface array is configured according to the bus type of the target avionics system, and simultaneously supports the synchronous acquisition of ARINC825, ARINC717, TTP, 1394 bus and TCN standard data. The many-core processing core is a CUDA device, which realizes parallel processing of FC, 1553B, ARINC825, ARINC717, TTP and 1394 bus protocols through programming. The dynamic buffer management unit adopts a hierarchical ring buffer structure.
[0010] The CUDA device internally implements a multi-channel DMA controller to ensure efficient data transfer to host memory. The many-core processing core uses a parallel processing architecture to simultaneously receive and preprocess data from multiple protocols, freeing the CPU from tedious and resource-intensive avionics bus data transmission and reception. The hierarchical ring buffer structure can cache data of different priorities in a hierarchical manner to avoid data loss.
[0011] As a further preferred embodiment of the present invention, the hierarchical ring buffer structure is specifically manifested in that high-priority data is allocated a larger buffer, and low-priority data is allocated a smaller buffer.
[0012] High-priority data, such as flight control data.
[0013] As a further preferred embodiment of the present invention, the configurable protocol parser dynamically loads protocol parsing rules based on a configuration file, the data routing and framing unit adopts a two-level filtering mechanism, and the streaming data processing pipeline is constructed based on SparkStructured Streaming structured stream processing technology.
[0014] The configurable protocol parser dynamically loads protocol parsing rules based on configuration files, flexibly handling different formats of bus data according to application requirements; the data routing and framing unit adopts a two-level filtering mechanism, performing round-robin routing of data frames based on a preset filter configuration table, extracting key byte data from massive data streams; the streaming data processing pipeline is built based on SparkStructured Streaming structured stream processing technology, which enables real-time pipelined data processing, ensuring data continuity.
[0015] Heterogeneous data parsing engine configuration: Based on the specific application scenario, prepare protocol configuration files, define the data format, parsing rules and field mapping relationships of each bus protocol, and configure two-level filters: the first filter configuration table is used to filter specific node data frames of specific types, such as only collecting key flight parameters; the second filter configuration table is used to extract key node data from the filtered data frames. The streaming data processing pipeline is built based on Spark Structured Streaming, and appropriate window sizes and watermark thresholds are set to balance processing latency and data integrity.
[0016] As a further preferred embodiment of the present invention, the first filter configuration table in the two-level filtering mechanism is used to filter specific node data frames of a specific type; the second filter configuration table is used to extract keyword node data from the filtered data frames.
[0017] As a further preferred embodiment of the present invention, the distributed real-time computing unit loads data into memory based on Spark RDD technology for parallel processing and analysis of CPK, the multi-layer statistical analysis unit is configured with anomaly detection algorithms and trend prediction models, and the semantic analysis unit loads aviation domain knowledge graphs and defines SPARQL query templates and inference rules.
[0018] The distributed real-time computing unit, based on Spark RDD technology, loads data into memory for parallel CPK processing and analysis, which can significantly improve the speed and efficiency of data processing. The multi-layer statistical analysis unit is configured with anomaly detection algorithms such as the Isolation Forest algorithm and trend prediction models such as the ARIMA model. The multi-layer statistical analysis unit is used to perform trend analysis, anomaly detection, and performance prediction on key feature datasets. The semantic analysis unit combines aviation domain knowledge graphs and uses entities and relationships to achieve data integration, query optimization, and semantic reasoning, providing deeper data analysis capabilities.
[0019] As a further preferred embodiment of the present invention, the load prediction unit uses a time series analysis model or a recurrent neural network to predict resource demand, the adaptive resource allocation unit sets a resource allocation strategy, and the model transfer and knowledge sharing unit configures a transfer learning framework to share models and knowledge among different analysis tasks through transfer learning, domain adaptation, or knowledge distillation techniques.
[0020] The time series analysis model can be a Prophet model, and the recurrent neural network can be an LSTM. It can predict future resource requirements based on historical load data. The adaptive resource allocation unit sets resource allocation strategies, such as allocating more computing cores to CPU-intensive tasks and more cache space to I / O-intensive tasks. The model transfer and knowledge sharing unit configures a transfer learning framework to enable the sharing of feature extractors and model parameters between different analysis tasks. The model transfer and knowledge sharing unit can transfer the trained model and knowledge in the source domain to the target domain through transfer learning, domain adaptation, and knowledge distillation techniques, and perform model training and optimization for different big data analysis tasks.
[0021] A system capable of efficiently processing massive amounts of aviation bus data, as described in any one of the above, includes the following steps for processing aviation bus data: A) A multi-bus protocol parallel acquisition module simultaneously receives data from various aviation buses. The CUDA processing core performs preliminary parsing and format standardization on the raw data, and then stores the processed data in a dynamic buffer; B) A heterogeneous data parsing engine reads data from the dynamic buffer, first performing complete parsing according to the protocol configuration file, then extracting key data through a two-level routing mechanism, and finally cleaning and transforming the data through a streaming processing pipeline; C) A business calculation and statistical analysis module receives the parsed data, performs real-time calculation and statistical analysis to generate real-time monitoring indicators, and performs deep semantic analysis using knowledge graphs to discover hidden patterns and relationships in the data; D) The processing results are stored in a hierarchical storage system according to a preset strategy: real-time results and frequently accessed data are stored in hot storage, recent data in warm storage, and historical data in cold storage; E) A resource scheduling and load management module monitors the system's operating status throughout the process, dynamically adjusting resource allocation according to load changes to ensure the system always operates in an optimal state.
[0022] Compared with the prior art, the present invention can achieve at least one of the following beneficial effects: 1. Abandoning the traditional model of relying on general-purpose processors for serial and centralized processing, this solution constructs a new distributed processing paradigm of "software and hardware collaboration and data-driven" processing. This solution uses many-core cores to perform parallel acquisition and preprocessing of multiple bus protocols, replacing the traditional software polling. In terms of logical operations, it introduces configuration-based dynamic parsing and a two-level data routing mechanism, replacing the rigid hard-coded parsing logic. In terms of algorithms, the core adopts Spark in-memory computing for distributed parallel processing, combined with an adaptive resource scheduling algorithm based on load prediction, which completely changes the static and passive resource allocation method.
[0023] 2. Employing parallel processing, stream computing, and resource scheduling, it boasts significant technical advantages: Hardware parallelization directly leverages parallel processing theory, offloading protocol processing tasks to many-core devices, greatly reducing CPU overhead and transmission latency, fundamentally improving data throughput and real-time performance; the Spark-based stream processing architecture ensures low latency and high throughput in data processing, and combined with configurable routing logic, it achieves a transformation from "full slow storage" to "precise fast acquisition"; while the intelligent resource scheduling model applies control theory and machine learning, enabling the system to dynamically perceive load and predict future demand, thereby achieving optimal resource allocation, ultimately achieving a comprehensive and extreme improvement in processing efficiency, system elasticity, and resource utilization. Attached Figure Description
[0024] Figure 1 This is a diagram showing the overall architecture of the system of the present invention.
[0025] Figure 2 This is a schematic diagram of the data processing flow of the present invention.
[0026] Figure 3 This is a timing diagram of resource scheduling and load in this invention. Detailed Implementation
[0027] 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 embodiments of the present invention, and not all embodiments. 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.
[0028] Therefore, the following detailed description of the embodiments of the 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 invention without inventive effort are within the scope of protection of the invention.
[0029] It should be noted that, unless otherwise specified, the embodiments and features described in this invention can be combined with each other.
[0030] It should be noted that similar labels 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.
[0031] In the description of this invention, it should be noted that the terms "center," "upper," "lower," "left," "right," "vertical," "horizontal," "inner," and "outer," etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings, or the orientation or positional relationship commonly used when the product of this invention is in use, or the orientation or positional relationship commonly understood by those skilled in the art. They are only used for the convenience of describing this invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation, and therefore should not be construed as a limitation of this invention. In addition, the terms "first," "second," etc., are only used to distinguish descriptions and should not be construed as indicating or implying relative importance.
[0032] In the description of this invention, it should also be noted that, unless otherwise explicitly specified and limited, the terms "set," "install," "connect," and "link" should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral connection; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; and they can refer to the internal connection of two components. Those skilled in the art can understand the specific meaning of the above terms in this invention based on the specific circumstances. Specific Implementation Example 1:
[0034] Figure 1 , Figure 2 , Figure 3 This paper presents a system capable of efficiently processing massive amounts of aviation bus data, including a multi-bus protocol parallel acquisition module, a heterogeneous data parsing engine, a business computing and statistical analysis module, a hierarchical storage management module, and a resource scheduling and load management module. The multi-bus protocol parallel acquisition module includes a protocol interface array, a many-core processing core, and a dynamic buffer management unit. The heterogeneous data parsing engine is connected to the multi-bus protocol parallel acquisition module and includes a configurable protocol parser, a data routing and framing unit, and a streaming data processing pipeline. The business computing and statistical analysis module consists of a distributed real-time computing unit, a multi-layer statistical analysis unit, and a semantic analysis unit. The hierarchical storage management module adopts a hot, warm, and cold three-layer storage architecture, where the hot data memory storage area, the warm data solid-state storage area, and the cold data disk storage area are used to store real-time processed data, recent historical data, and long-term archived data, respectively. The resource scheduling and load management module includes a load prediction unit, an adaptive resource allocation unit, and a model migration and knowledge sharing unit.
[0035] This solution abandons the traditional model of relying on general-purpose processors for serial, centralized processing, and constructs a new distributed processing paradigm of "software and hardware collaboration and data-driven". This solution uses many-core cores to perform parallel acquisition and preprocessing of multiple bus protocols, replacing the traditional software polling. In terms of logical operations, it introduces configuration-based dynamic parsing and a two-level data routing mechanism, replacing the rigid hard-coded parsing logic. In terms of algorithms, the core adopts Spark in-memory computing for distributed parallel processing, combined with an adaptive resource scheduling algorithm based on load prediction, which completely changes the static and passive resource allocation method.
[0036] Employing parallel processing, stream computing, and resource scheduling, it boasts significant technical advantages: hardware parallelization directly leverages parallel processing theory, offloading protocol processing tasks to many-core devices, greatly reducing CPU overhead and transmission latency, fundamentally improving data throughput and real-time performance; the Spark-based stream processing architecture ensures low latency and high throughput in data processing, and combined with configurable routing logic, it achieves a transformation from "full slow storage" to "precise fast acquisition"; while the intelligent resource scheduling model applies control theory and machine learning, enabling the system to dynamically perceive load and predict future demand, thereby achieving optimal resource allocation, ultimately achieving a comprehensive and extreme improvement in processing efficiency, system elasticity, and resource utilization.
[0037] The tiered storage management module adopts a three-tiered storage architecture of hot, warm, and cold storage. This tiered storage strategy optimizes the utilization efficiency of storage resources while ensuring high data reliability and availability. Specific Implementation Example 2:
[0039] This solution further describes the protocol interface array based on Specific Embodiment 1. The protocol interface array is configured according to the bus type of the target avionics system and supports synchronous acquisition of ARINC825, ARINC717, TTP, 1394 bus and TCN standard data. The many-core processing core is a CUDA device, which realizes parallel processing of FC, 1553B, ARINC825, ARINC717, TTP and 1394 bus protocols through programming. The dynamic buffer management unit adopts a hierarchical ring buffer structure.
[0040] The CUDA device internally implements a multi-channel DMA controller to ensure efficient data transfer to host memory. The many-core processing core uses a parallel processing architecture to simultaneously receive and preprocess data from multiple protocols, freeing the CPU from tedious and resource-intensive avionics bus data transmission and reception. The hierarchical ring buffer structure can cache data of different priorities in a hierarchical manner to avoid data loss. Specific Implementation Example 3:
[0042] This solution further illustrates the hierarchical ring buffer structure based on specific embodiment 2. Specifically, the hierarchical ring buffer structure is characterized by allocating a larger buffer for high-priority data and a smaller buffer for low-priority data.
[0043] High-priority data, such as flight control data. Specific Implementation Example 4:
[0045] This solution further describes the configurable protocol parser based on specific embodiment 1. The configurable protocol parser dynamically loads protocol parsing rules based on configuration files. The data routing and framing unit adopts a two-level filtering mechanism. The streaming data processing pipeline is built based on Spark Structured Streaming technology.
[0046] The configurable protocol parser dynamically loads protocol parsing rules based on configuration files, flexibly handling different formats of bus data according to application requirements; the data routing and framing unit adopts a two-level filtering mechanism, performing round-robin routing of data frames based on a preset filter configuration table, extracting key byte data from massive data streams; the streaming data processing pipeline is built based on SparkStructured Streaming structured stream processing technology, which enables real-time pipelined data processing, ensuring data continuity.
[0047] Heterogeneous data parsing engine configuration: Based on the specific application scenario, prepare protocol configuration files, define the data format, parsing rules and field mapping relationships of each bus protocol, and configure two-level filters: the first filter configuration table is used to filter specific node data frames of specific types, such as only collecting key flight parameters; the second filter configuration table is used to extract key node data from the filtered data frames. The streaming data processing pipeline is built based on Spark Structured Streaming, and appropriate window sizes and watermark thresholds are set to balance processing latency and data integrity. Specific Implementation Example 5:
[0049] This solution further explains the two-level filtering mechanism based on specific embodiment 4. The first filter configuration table in the two-level filtering mechanism is used to filter specific node data frames of specific types; the second filter configuration table is used to extract keyword node data from the filtered data frames. Specific Implementation Example 6:
[0051] This solution further describes the distributed real-time computing unit based on Specific Embodiment 1. The distributed real-time computing unit loads data into memory based on Spark RDD technology for parallel processing and analysis of CPK. The multi-layer statistical analysis unit is configured with anomaly detection algorithms and trend prediction models. The semantic analysis unit loads aviation domain knowledge graphs and defines SPARQL query templates and inference rules.
[0052] The distributed real-time computing unit, based on Spark RDD technology, loads data into memory for parallel CPK processing and analysis, which can significantly improve the speed and efficiency of data processing. The multi-layer statistical analysis unit is configured with anomaly detection algorithms such as the Isolation Forest algorithm and trend prediction models such as the ARIMA model. The multi-layer statistical analysis unit is used to perform trend analysis, anomaly detection, and performance prediction on key feature datasets. The semantic analysis unit combines aviation domain knowledge graphs and uses entities and relationships to achieve data integration, query optimization, and semantic reasoning, providing deeper data analysis capabilities. Specific Implementation Example 7:
[0054] This solution further describes the load prediction unit based on Specific Embodiment 1. The load prediction unit uses a time series analysis model or a recurrent neural network to predict resource demand. The adaptive resource allocation unit sets a resource allocation strategy. The model transfer and knowledge sharing unit configures a transfer learning framework and shares models and knowledge among different analysis tasks through transfer learning, domain adaptation, or knowledge distillation techniques.
[0055] The time series analysis model can be a Prophet model, and the recurrent neural network can be an LSTM. It can predict future resource requirements based on historical load data. The adaptive resource allocation unit sets resource allocation strategies, such as allocating more computing cores to CPU-intensive tasks and more cache space to I / O-intensive tasks. The model transfer and knowledge sharing unit configures a transfer learning framework to enable the sharing of feature extractors and model parameters between different analysis tasks. The model transfer and knowledge sharing unit can transfer the trained model and knowledge in the source domain to the target domain through transfer learning, domain adaptation, and knowledge distillation techniques, and perform model training and optimization for different big data analysis tasks. Specific Implementation Example 8:
[0057] A system capable of efficiently processing massive amounts of aviation bus data, as described in any one of the above, includes the following steps for processing aviation bus data: A) A multi-bus protocol parallel acquisition module simultaneously receives data from various aviation buses. The CUDA processing core performs preliminary parsing and format standardization on the raw data, and then stores the processed data in a dynamic buffer; B) A heterogeneous data parsing engine reads data from the dynamic buffer, first performing complete parsing according to the protocol configuration file, then extracting key data through a two-level routing mechanism, and finally cleaning and transforming the data through a streaming processing pipeline; C) A business calculation and statistical analysis module receives the parsed data, performs real-time calculation and statistical analysis to generate real-time monitoring indicators, and performs deep semantic analysis using knowledge graphs to discover hidden patterns and relationships in the data; D) The processing results are stored in a hierarchical storage system according to a preset strategy: real-time results and frequently accessed data are stored in hot storage, recent data in warm storage, and historical data in cold storage; E) A resource scheduling and load management module monitors the system's operating status throughout the process, dynamically adjusting resource allocation according to load changes to ensure the system always operates in an optimal state.
[0058] Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing embodiments or make equivalent substitutions for some of the technical features. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.
Claims
1. A system capable of efficiently processing massive amounts of avionics bus data, characterized by: The application comprises a multi-bus protocol parallel acquisition module, a heterogeneous data analysis engine, a business computing and statistical analysis module, a hierarchical storage management module, and a resource scheduling and load management module. The heterogeneous data analysis engine is connected to the multi-bus protocol parallel acquisition module, and comprises a configurable protocol analyzer, a data routing and framing unit, and a streaming data processing pipeline.
2. The system capable of efficiently processing massive amounts of avionics bus data of claim 1, wherein: The protocol interface array is configured according to the bus type of the target avionics system, and supports the synchronous acquisition of ARINC825, ARINC717, TTP, 1394 bus and TCN standard data. 3.The system capable of efficiently processing massive aviation bus data according to claim 2, characterized in that: The hierarchical ring buffer structure is characterized in that high-priority data is allocated to a larger buffer area, and low-priority data is allocated to a smaller buffer area.
4. The system capable of efficiently processing massive amounts of avionics bus data of claim 1, wherein: The configurable protocol analyzer dynamically loads protocol analysis rules based on a configuration file.
5. The system capable of efficiently processing massive amounts of avionics bus data of claim 4, wherein: The two-level filtering mechanism comprises a first filter configuration table for screening specific node data frames of a specific type, and a second filter configuration table for extracting key byte data from the screened data frames.
6. The system capable of efficiently processing massive amounts of avionics bus data of claim 1, wherein: The distributed real-time computing unit loads data into memory for parallel processing and analysis of CPK based on Spark RDD technology.
7. The system capable of efficiently processing massive amounts of avionics bus data of claim 1, wherein: The load prediction unit uses a time series analysis model or a recurrent neural network to predict resource requirements.
8. A process for processing the avionics bus data by the system capable of efficiently processing mass avionics bus data according to any one of claims 1-7, characterized in that: The application comprises the following steps: A, the multi-bus protocol parallel acquisition module simultaneously receives data from various types of aviation buses, the CUDA processing core performs preliminary analysis and format standardization on the raw data, and then stores the processed data in the dynamic buffer area. B, isomeric data analysis engine reads data from dynamic buffer, firstly complete analysis according to protocol configuration file, then extract key data through two level road mechanism, finally through the flow processing pipeline data cleaning and conversion; C, business computing and statistical analysis module receives the parsed data, on the one hand, real-time calculation and statistical analysis, generate real-time monitoring index; On the other hand, combined with knowledge graph for deep semantic analysis, find the implicit rules and association in data; D, the processing result is stored in the hierarchical storage system according to the preset strategy, the real-time result and the frequently accessed data are stored in the hot storage, the recent data is stored in the warm storage, and the historical data is stored in the cold storage; E, resource scheduling and load management module monitors system running state throughout, dynamically adjusts resource allocation according to load change, ensures that the system always runs in the optimal state.