A sysml heterogeneous model integration method

Abstract

The present application relates to the field of complex equipment development in the aviation industry, in particular to a SysML heterogeneous model integration method, aiming to solve the problems of data island and collaboration barriers caused by model heterogeneity. Based on a unified data management platform, the method realizes the normalization of multi-source SysML models, the unified integration and scheduling of heterogeneous models, the high-precision joint simulation of virtual and real systems, and the multi-dimensional intelligent analysis of simulation results through five core steps of intercommunication and synchronization, including SysML model analysis, functional unit conversion, multi-source heterogeneous model integration, virtual-real fusion test simulation, and comprehensive analysis and iterative optimization. The platform is equipped with four special function modules, which are compatible with multiple modeling tools and form a closed-loop design system through technologies such as full-factor mapping, middleware integration, gradient descent method parameter calibration, and three-dimensional visualization analysis. The method has been verified effective in the development of aircraft engines and airborne equipment, significantly improving model reuse rate and cross-disciplinary collaboration efficiency, shortening design optimization cycle, and providing replicable technical support for the implementation of MBSE in the aviation industry.

Description

Technical Field

[0001] This invention relates to the field of complex equipment development in the aviation industry, specifically a SysML heterogeneous model integration method. Background Technology

[0002] In the field of complex equipment development in the aviation industry, with the development of industrial digitalization and intelligence, model-based systems engineering (MBSE) has become a core industry practice. It replaces traditional document-based design with the establishment of digital system models, achieving model-driven development throughout the entire lifecycle of complex equipment development and significantly improving the standardization and scientific rigor of the design. Currently, major domestic aviation design institutes have purchased and used various system analysis and modeling tools, such as SysML modeling tools MagicDraw and Rhapsody, control system modeling tools Matlab / Simulink, 3D design tools CAD, and simulation analysis tools CAE. Different tools are suitable for different design stages and professional fields, becoming important technical support for the development of complex equipment.

[0003] However, the current practice of using multiple tools in parallel has led to widespread heterogeneity in system models between design units and among different design departments within a design unit. This issue is a legacy problem from the promotion of the MBSE methodology and a key area for resolution in the current information technology construction of aerospace design units. Model heterogeneity directly results in a large number of model fragments and data silos during system analysis and design. Historical design model assets cannot be effectively reused in new design projects, requiring designers to repeatedly model, which not only wastes a lot of human, material, and time costs but also seriously reduces the system development efficiency of complex aerospace equipment.

[0004] The current collaborative barriers created by the parallel use of multiple tools in the design process are mainly reflected in three core aspects: First, the fragmented format leads to parsing and semantic loss. SysML model files generated by different modeling tools use different storage formats and data structures, resulting in inherent parsing differences. When directly reusing models, it is easy to lose the attributes, constraints, and semantic information of key elements such as blocks, ports, and relationships in the model, leading to a significant reduction in the accuracy of model reuse. Second, the functional limitations of existing collaborative platforms. Currently, most collaborative platforms in the industry focus on the integration of a single toolchain. For example, the TWC platform for SysML modeling tools can only achieve model management and collaboration for tools of the same type, lacking support for parallel design scenarios with heterogeneous tools, and failing to meet the needs of cross-disciplinary and cross-tool model integration. Third, the lack of unified model standards. Not only are there different versions of SysML standards, but there are also significant compatibility issues between SysML models and UML models. The definitions of model elements and the expression of relationships under different standards are different, further exacerbating the collaborative difficulties caused by heterogeneous models.

[0005] To address these issues, some design firms have attempted manual model format conversion and data processing. However, manual operations are inefficient, error-prone, and lack real-time data synchronization. Other firms have introduced simple model conversion tools, but these tools only convert single formats, lacking data verification and cleaning, resulting in compromised data integrity and accuracy. Furthermore, they haven't established an integration framework for heterogeneous models, hindering cross-domain model co-simulation. In addition, current technologies for co-simulating virtual and physical systems only address simple data exchange, suffering from incompatible interfaces, low time synchronization accuracy, and poor matching between simulation models and physical systems. This leads to significant discrepancies between simulation results and actual system behavior, failing to provide reliable data for design optimization. Simulation result analysis often relies on traditional 0D / 1D viewing methods, lacking multi-dimensional visualization and automated verification capabilities. This makes it difficult for designers to quickly locate model problems, resulting in low efficiency in system design optimization.

[0006] In summary, existing technologies lack a complete and practical SysML heterogeneous model integration method, which cannot fundamentally solve the problems of data silos and collaboration barriers caused by model heterogeneity. It also fails to meet the needs of the aerospace industry for high-precision simulation and efficient design optimization in the development of complex equipment. Therefore, developing a SysML heterogeneous model integration method that can achieve multi-source SysML model normalization, unified integration of heterogeneous models, high-precision joint simulation of virtual and real systems, and comprehensive analysis of simulation results has become an urgent need in the field of MBSE in the aerospace industry. Summary of the Invention

[0007] The purpose of this invention is to overcome the shortcomings of existing technologies and provide a SysML heterogeneous model integration method. This method addresses the problems existing in existing technologies, such as difficulties in processing multi-source heterogeneous SysML models, low integration efficiency, inaccurate simulation, insufficient joint simulation of virtual and real systems, and low efficiency in system design optimization. It also solves pain points such as model fragmentation, data silos, semantic loss, and barriers to cross-team design collaboration. This method enables efficient reuse of SysML models, unified integration and scheduling of cross-domain heterogeneous models, high-precision joint simulation of virtual and real systems, and multi-dimensional comprehensive analysis of simulation results, thereby improving the efficiency and quality of complex equipment development in the aerospace industry.

[0008] The technical solution adopted by this invention to solve its technical problem is: a SysML heterogeneous model integration method, comprising the following steps: S1, SysML model parsing, sequentially completes specification clarification and interface preparation, data reading and preliminary processing, mapping transformation and data verification, and model storage, converting SysML models generated by different modeling tools into models that conform to a common architecture and standards and building a model library, realizing version management and secure backup; S2. Functional Unit Conversion: Identify and classify the functional modules of the general SysML model, clarify the input, output and internal logic of each functional unit, formulate conversion rules based on the target system requirements and architecture, and complete the functional unit adaptation. S3, Multi-source heterogeneous model integration, achieves unified registration, scheduling and collaborative operation of cross-domain multi-format models through model standardization processing, middleware-based integration, model conversion and adaptation, integration verification and performance tuning; S4. Virtual-real fusion test simulation: Complete the interface design of virtual and real systems, build data interaction and synchronization mechanisms, construct and optimize virtual models and calibrate model parameters, and realize bidirectional real-time data interaction and high-precision joint simulation between physical systems and virtual simulation models; S5. Comprehensive analysis and iterative optimization: Through real-time monitoring of indicators, model optimization and adjustment, and closed-loop iterative optimization, the simulation results are analyzed from multiple dimensions and the SysML model is continuously optimized until the performance indicators meet the design requirements; data is shared among all steps and results are synchronized in real time, forming a complete closed loop of model integration and design optimization.

[0009] Specifically, in step S1, the specification and interface preparation are defined as a dedicated API or general file parsing interface that automatically adapts to the modeling tool; when reading data, a data buffer and error capture mechanism are set up; in the preliminary processing stage, the original data is format-checked and abnormal data is marked; the mapping transformation covers the full-element mapping of elements such as blocks, ports, and relationships, as well as their attributes and constraints; the data verification performs logical consistency and data integrity checks and cleans and corrects problematic data.

[0010] Specifically, in step S2, the functional unit conversion rules include merging, splitting and reorganizing functional modules. The adaptation method is to adjust the input and output logic of the functional units according to the architectural requirements of the target system so that the converted functional units match the technical indicators of the target system.

[0011] Specifically, in step S3, the middleware has model management, data conversion, and communication coordination functions, realizing model registration, query, scheduling, and format conversion and semantic parsing of different model data; model conversion and adaptation achieve format compatibility of models from different modeling tools by developing dedicated format conversion tools, and designing a functional adaptation layer to complete signal conversion and interaction between models.

[0012] Specifically, in step S3, integration verification and performance tuning include unit testing, integration testing, and comparative testing with the actual system. The accuracy and reliability of the model are verified through comprehensive testing, and the model performance is tuned to address the test issues.

[0013] Specifically, in step S4, the virtual-physical system interface includes a physical interface for the physical system and a virtual interface for the virtual simulation model. The physical interface includes a signal conditioning and analog-to-digital conversion module for sensors and a digital-to-analog conversion and drive circuit for actuators. A one-to-one parameter mapping relationship is established between the virtual interface and the physical interface. The virtual-physical data interaction adopts one or more of the following communication protocols: real-time Ethernet, OPC UA, bus, etc. The data transmission is protected by AES128 encryption technology.

[0014] Specifically, in step S4, the virtual model construction and optimization adopts multibody dynamics modeling for mechanical systems and establishes an accurate circuit model for electronic systems. The model parameter calibration is achieved by comparing the output of the virtual model with the test data of the actual system and adjusting the model parameters using the gradient descent method.

[0015] Specifically, in step S5, real-time monitoring of indicators acquires performance indicators such as reliability, efficiency, and response time through sensors and data acquisition modules, sets indicator thresholds, and implements automatic alarms when limits are exceeded; data analysis uses visualization methods such as curves and charts, and at the same time completes the automated verification of performance indicators through a built-in verification rule base.

[0016] Specifically, in step S5, the comprehensive analysis adopts a three-dimensional lightweight model display method, which dynamically associates the three-dimensional geometric model of the system with the SysML functional logic model and simulation data to realize the three-dimensional visualization analysis of the system structure and the rapid location of faulty components.

[0017] Specifically, this method is based on a unified data management platform, which is equipped with a unified parsing and visualization editing module for multi-source SysML models, a multi-disciplinary simulation model scheduling engine module, a general real-time data interaction and closed-loop interface module, and a three-dimensional correlation visualization analysis module. The modules work together to achieve real-time data sharing and synchronization.

[0018] The beneficial effects of this invention are: It achieves normalization and efficient reuse of multi-source SysML models: The developed parsing and conversion component is compatible with multiple modeling tools and can directly read, parse and convert SysML models of different formats, realizing unified processing of model data and effectively solving the problems of model fragmentation and data silos; at the same time, it provides a graphical visual editing interface, supports viewing, modifying and synchronously updating all elements of the model, and completes model management and design modifications in a unified environment, improving the efficiency and accuracy of model reuse.

[0019] It breaks through the barriers to integration and collaboration of heterogeneous models across disciplines: the middleware-based heterogeneous model integration framework enables unified registration, scheduling, and collaborative operation of cross-domain and multi-format simulation models. The built-in multidisciplinary simulation model scheduling engine can automatically manage the calculation order and data exchange of different models, completing the cross-disciplinary joint simulation process. At the same time, through unified modeling specifications and functional adaptation layers, it solves the problems of model standard differences and functional incompatibility, and significantly improves the efficiency of cross-team design collaboration.

[0020] A high-precision co-simulation of virtual and real systems was achieved: the designed general real-time data interaction and closed-loop interface system supports multiple standard communication protocols, enabling bidirectional, real-time, and low-latency data interaction between the physical system and the virtual simulation model. At the same time, a high-precision time synchronization mechanism ensures the timeliness of data interaction and the causal correctness of the simulation. Hardware-in-the-loop testing effectively connects virtual verification and physical testing, significantly improving the matching degree between simulation results and actual system behavior, and solving the problems of inaccurate simulation and insufficient virtual-real co-simulation.

[0021] It improves the efficiency and intelligence of system design optimization: The constructed three-dimensional correlation visualization analysis system breaks through the limitations of traditional 0-dimensional / 1-dimensional structure viewing, realizes three-dimensional and visual analysis of system structure, and can dynamically correlate the three-dimensional geometric model with SysML functional logic model and simulation data to achieve rapid location of faulty components; at the same time, through multi-dimensional data visualization and automated model verification functions, it reduces the negligence of manual review, and realizes continuous improvement of system design through closed-loop iterative optimization, which greatly improves the efficiency of design optimization.

[0022] The technical solution of this invention covers the entire process of SysML model from analysis, transformation, integration to simulation, analysis and optimization, forming a closed-loop model-driven design system. It effectively solves the historical problems in the promotion of MBSE methodology, provides feasible and replicable technical support for the implementation of MBSE in the development of complex equipment in the aviation industry, and promotes the digital and intelligent upgrading of complex equipment development in the aviation industry. Attached Figure Description

[0023] The present invention will be further described below with reference to the accompanying drawings and embodiments.

[0024] Figure 1 The flowchart of the SysML heterogeneous model integration method of the present invention illustrates the four core steps of the method: SysML model analysis, multi-source heterogeneous model integration, virtual-real fusion experimental simulation, and comprehensive analysis, as well as the data interoperability relationship, reflecting a closed-loop model integration and design optimization system.

[0025] Figure 2This is a flowchart illustrating the specific process of parsing the SysML model in this invention. The flowchart shows the four main steps of the parsing process: reading the SysML model file, parsing model elements (blocks, ports, etc.), mapping to the platform's general specifications, and data verification. It also shows the error recording process for data verification failures and the process of generating the parsed model and visual editing display after successful verification.

[0026] Figure 3 This is an architecture diagram of the multi-source heterogeneous model integration of the present invention. The diagram shows the core components of the integration architecture: multi-source heterogeneous model, model scheduling engine, performance model access unit, model association unit, data fusion unit, and the allocation relationship between each unit and the SysML model.

[0027] Figure 4 This is an architecture diagram of the virtual-real fusion test simulation of the present invention. The diagram shows the core components of the simulation architecture: real-time synchronous controller, adapter, virtual interface, physical interface, virtual simulation model, and physical system, reflecting the interface correspondence and data interaction path of the virtual and real systems.

[0028] Figure 5 This is an architecture diagram of the comprehensive analysis of the present invention. The diagram shows the core components of the comprehensive analysis: integrated simulation environment, simulation engine, performance evaluation unit, visualization environment, analysis report, and the allocation relationship between each unit and the comprehensive analysis module. Detailed Implementation

[0029] To make the technical means, creative features, objectives and effects of this invention easier to understand, the invention will be further described below in conjunction with specific embodiments.

[0030] like Figures 1-5 As shown, the present invention discloses a SysML heterogeneous model integration method. This method is based on a unified data management platform and forms a closed-loop model integration and design optimization system through five core steps. It also includes four dedicated functional modules as technical support, comprising the following steps: SysML Model Parsing: First, the specific format, specifications, data storage structure, and element definition methods of SysML models generated by different modeling tools are clearly defined, automatically adapting to tool-specific APIs or general file parsing interfaces. Then, the original model file is read through the interface, with data buffering and error capture mechanisms set up to ensure reading stability. The original data undergoes format checks, and abnormal data is marked. Next, based on the general SysML model architecture and standards, full-element mapping rules for elements such as blocks, ports, and relationships, as well as their attributes and constraints, are formulated. The original model data is converted into a general model through a conversion program, performing logical consistency and data integrity checks during the conversion process, and cleaning and correcting problematic data. Finally, the verified general SysML model is stored in a standard format, a model library is built, and version management and secure backup are implemented.

[0031] Functional Unit Conversion: The functional modules in the parsed general SysML model are meticulously identified and classified, and the input, output and internal logic of each functional unit are clarified. Based on the requirements, architecture and technical indicators of the target system, functional unit conversion rules are formulated, including merging, splitting and reorganizing functional modules, and adjusting the input and output logic of functional units to make the converted functional units compatible with the target system.

[0032] Multi-source heterogeneous model integration: First, a unified modeling standard is established in the initial stage of model creation, unifying the element definitions, representation methods, coordinate systems, etc. of models from different domains. Then, middleware is introduced as the core coordination layer, utilizing its model management, data conversion, and communication coordination functions to realize model registration, query, scheduling, and data format conversion and semantic parsing. Next, a dedicated format conversion tool is developed to achieve format compatibility between models from different modeling tools, and a functional adaptation layer is designed to complete signal conversion and interaction between models. Finally, integration verification is completed through unit testing, integration testing, and comparative testing with the actual system, and model performance is optimized for test issues.

[0033] Virtual-Real Fusion Simulation Experiment: First, the virtual-real system interface was designed. Physical interfaces were developed for the physical system, including signal conditioning and analog-to-digital conversion modules for sensors, and digital-to-analog conversion and drive circuits for actuators. Virtual interfaces were defined to correspond one-to-one with the physical interfaces, establishing parameter mapping relationships. Next, a data interaction and synchronization mechanism was built, using real-time Ethernet and OPC UA protocols to achieve bidirectional data transmission. AES128 encryption technology was employed to ensure transmission security, and high-precision time synchronization was achieved through a dynamic time synchronization algorithm combined with the PTP precision clock protocol. Then, a high-precision virtual model was constructed based on the characteristics of the actual physical system. Multibody dynamics modeling was used for the mechanical system, and a precise circuit model was established for the electronic system. The models were then optimized. Finally, using actual system test data, the virtual model parameters were adjusted using the gradient descent method to complete parameter calibration, ensuring that the virtual model's behavior remained consistent with the actual system.

[0034] Comprehensive analysis and iterative optimization: During simulation, performance indicators such as reliability, efficiency, and response time are acquired in real time through sensors and data acquisition modules. Thresholds for these indicators are set, and automatic alarms are triggered when limits are exceeded. Visual methods such as curves and charts are used to display and analyze the indicator data from multiple dimensions. Simultaneously, a built-in configurable verification rule library is used to automatically verify simulation input, output, and process data, identifying potential problems. When performance indicators fail to meet design requirements, relevant data such as indicator values, simulation time, and model parameters are recorded. The root cause of the problem is analyzed, and the parameters or structure of the SysML model are adjusted. Finally, a closed-loop iterative process is established. After optimization, simulation and monitoring are repeated until the performance indicators meet design requirements. The iteration process and results are recorded to provide a reference for subsequent designs.

[0035] Meanwhile, this invention has developed four dedicated functional modules on a unified data management platform: a unified parsing and visualization editing module for multi-source SysML models, a multi-disciplinary simulation model scheduling engine, a general real-time data interaction and closed-loop interface system, and a three-dimensional correlation visualization analysis system. These modules work together to achieve real-time data sharing and synchronization.

[0036] Example 1: Application of SysML heterogeneous model integration for aero-engines: This embodiment applies the SysML heterogeneous model integration method of the present invention to the development process of aero-engines. As a core and complex piece of equipment in the aviation industry, the development of aero-engines involves multiple professional fields such as mechanics, electrical engineering, control, and hydraulics. It employs various modeling and simulation tools such as MagicDraw, Rhapsody, Matlab / Simulink, and CAE, and faces serious problems of SysML model heterogeneity and integration. The specific implementation steps of this embodiment are as follows: SysML Model Analysis: First, the native file formats of aero-engine SysML models generated by tools such as MagicDraw and Rhapsody are defined. MagicDraw models use the mdzip file format, while Rhapsody models use the rpy file format. These are adapted to MagicDraw's Open API and Rhapsody's dedicated parsing interface, respectively. A general XML file parsing interface is also configured as a backup. Then, the original SysML model file of the aero-engine is read through the aforementioned interfaces. A 1024KB data buffer pool and an exception handling block are set up to ensure the stability of data reading. XML format checks and element integrity checks are performed on the read raw data, identifying 23 instances of abnormal data, including missing port attributes and logical relationship errors. Next, based on the SysML... The v2.0 general standard establishes full-element mapping rules for blocks (such as combustion chambers, turbines, and compressors), ports (such as fuel input ports and exhaust output ports), relationships (such as mechanical connections and electrical connections), and their attributes and constraints in aero-engine models. A Python conversion program was developed to convert the original model data into a general SysML model. During the conversion process, the built-in verification rule library performs logical consistency and integrity checks on the data, cleansing and correcting 23 marked abnormal data, including adding 15 missing attributes and correcting 8 logical relationships. Finally, the verified general aero-engine SysML model is stored in the model library of the unified data management platform in standard XMI format, creating version V1.0 for the model, and performing off-site cloud backup to achieve version management and secure backup of the model.

[0037] Functional Unit Conversion: The functional modules of the analyzed SysML model of the general aviation engine were identified and classified, resulting in eight major functional modules, including fuel supply, ignition control, turbine rotation, exhaust cooling, and hydraulic regulation. The input and output parameters of each functional unit were defined. For example, the input of the fuel supply unit is fuel pressure and flow rate, and the output is atomized fuel. Based on the development indicators of the aero-engine (thrust-to-weight ratio ≥10, fuel consumption ≤0.6kg / (daN·h)) and the overall architecture requirements, functional unit conversion rules were formulated. The ignition control unit and the fuel supply unit were merged into a combustion control unit, and the hydraulic regulation unit was split into two sub-units: oil pressure regulation and mechanical regulation. At the same time, the input logic of the turbine rotation unit was adjusted to include the speed feedback signal as an input parameter, so that the converted functional units are adapted to the overall development requirements of the aero-engine.

[0038] Multi-source heterogeneous model integration: First, a unified modeling standard for heterogeneous aero-engine models is established. For mechanical models, the geometric representation adopts the STEP format, and material property definitions adopt the ISO standard. For control models, circuit component modeling adopts the IEC standard, and signal transmission representation adopts the CAN bus standard. Then, a middleware based on a microservice architecture is introduced as the core coordination layer. This middleware includes three main functional modules: model management, data conversion, and communication coordination. Control system models built with Matlab / Simulink, combustion chamber flow field simulation models built with CAE, and 3D geometric models built with CAD are all registered to the middleware. The model management module enables model querying and scheduling, the data conversion module converts CAE simulation data from VTK format to the platform-compatible CSV format, and the communication coordination module manages TCP / IP communication connections between models. Next, a dedicated format conversion tool is developed to... The 3D geometric model of the aero-engine created by CAD was converted into an STL format recognizable by the simulation platform. The control system model from Matlab / Simulink was converted into an FMU format that could run directly in the integrated environment. A functional adaptation layer was designed to convert the signals of the hydraulic adjustment unit, converting analog signals into digital signals to achieve signal interaction with the control system model. Finally, the integrated model was verified. First, unit testing was conducted to verify the simulation accuracy of individual functional units such as combustion control and turbine rotation, with a 100% pass rate. Then, integration testing was conducted to check the collaborative work between the models and correct the signal delay problem between the hydraulic adjustment unit and the turbine rotation unit. Finally, a comparative test was conducted with the physical test bench of the aero-engine to verify the rationality of the simulation results. The deviation between the simulation results and the physical test results was controlled within 5%. The model parameters were fine-tuned to address the simulation deviations, and performance optimization was completed.

[0039] Virtual-Real Fusion Simulation: First, the virtual-real system interface for the aero-engine was designed. For the sensors (pressure, speed, and temperature sensors) of the physical test bench, signal conditioning circuits and a 16-bit analog-to-digital converter (ADC) module were designed to convert the analog signals collected by the sensors into digital signals and transmit them to the virtual simulation model. For the actuators (fuel valves and igniters) of the physical test bench, a 16-bit ADC module and a power drive circuit were developed to convert the digital control signals output from the virtual model into analog signals to drive the actuators. Virtual interfaces corresponding to the physical interfaces were defined on the virtual simulation model side, establishing a one-to-one mapping relationship for parameters such as pressure, speed, and temperature. Subsequently, a data interaction and synchronization mechanism was built, using the OPCUA protocol as the data transmission protocol to meet the real-time requirements of aero-engine simulation, and employing AES128 encryption technology. The technology encrypts and protects core data such as pressure and temperature during transmission. It achieves time synchronization between the virtual model, physical test bench, and simulation platform using a dynamic time synchronization algorithm combined with the PTP precision clock protocol, achieving a synchronization accuracy of 1ms. Next, based on the characteristics of the aero-engine physical test bench, a high-precision virtual model is constructed. For the mechanical system, a multibody dynamics modeling method is used, considering factors such as turbine elastic deformation and bearing friction. For the electrical control system, an accurate circuit model is established, considering factors such as component nonlinearity and electromagnetic interference. Finally, using test data from the aero-engine physical test bench at idle, medium speed, and high speed, the gradient descent method is used to adjust parameters such as the fuel flow coefficient and turbine friction coefficient of the virtual model, completing parameter calibration and ensuring that the deviation between the simulation results of the virtual model and the test data from the physical test bench is controlled within 3%.

[0040] Comprehensive analysis and iterative optimization: During the virtual-real fusion simulation of the aero-engine, sensors and data acquisition modules acquire real-time performance indicators such as thrust-to-weight ratio, fuel consumption, response time, and reliability. Thresholds of 10 for thrust-to-weight ratio and 0.6 kg / (daN·h) for fuel consumption are set to trigger automatic alarms when these limits are exceeded. The platform's visualization module plots curves of thrust-to-weight ratio and fuel consumption over time, allowing for comparison of performance indicators under different operating conditions on the same chart, achieving multi-dimensional visualization analysis. The built-in verification rule library automatically verifies input data such as fuel flow rate and output turbine speed, identifying one abnormal fuel flow input. Simulation results show that the aero-engine… The fuel consumption under idling conditions was 0.62 kg / (daN·h), exceeding the design threshold. The design team recorded the value of this indicator, simulation time, fuel flow coefficient, and other relevant data. The analysis revealed that the root cause of the problem was the low atomization efficiency of the fuel supply unit. Subsequently, the atomization pressure parameter of the fuel supply unit in the SysML model was adjusted from 0.8 MPa to 0.9 MPa. After the adjustment, a virtual-real fusion simulation was performed again. The simulation results showed that the fuel consumption was reduced to 0.58 kg / (daN·h), which met the design requirements. At the same time, the thrust-to-weight ratio was increased to 10.2, meeting the development target. The process and results of this iteration were recorded in the platform database to provide a reference for the subsequent design of aero-engine modifications.

[0041] In this embodiment, through the technical solution of the present invention, the reuse rate of SysML models for aero-engines is increased from the original 20% to 85%, the cross-disciplinary design collaboration efficiency is increased by 60%, the matching degree between simulation results and actual systems is increased to over 97%, and the design optimization cycle is shortened from the original 15 days to 3 days, which greatly improves the development efficiency and quality of aero-engines.

[0042] Example 2: Application of SysML heterogeneous model integration for airborne equipment: This embodiment applies the SysML heterogeneous model integration method of the present invention to the development process of airborne navigation equipment. Airborne navigation equipment involves multiple disciplines such as electronics, communication, control, and software, and uses various modeling and simulation tools such as Rhapsody, Matlab / Simulink, AutoCAD, and ADAMS. Problems exist such as heterogeneous models, data incompatibility, and disconnect between virtual and real simulations. The specific implementation steps of this embodiment are as follows: SysML Model Analysis: First, the format specifications of the SysML model of the airborne navigation device generated by tools such as Rhapsody and Matlab / Simulink are clarified, adapting to Rhapsody's COM interface and Matlab / Simulink's MATLAB API, and configuring a general JSON file parsing interface; then, the original SysML model file of the airborne navigation device is read through the above interface, a 512KB data buffer pool and an anomaly capture mechanism are set to ensure the stability of data reading, the format and element integrity checks of the original data are performed, and 18 abnormal data points with missing constraints and port mapping errors are marked; then, according to SysML... The v2.0 general standard defines the mapping rules for blocks (such as satellite receiving modules, positioning calculation modules, and communication modules), ports (such as satellite signal input ports and positioning information output ports), relationships (such as data transmission relationships and control relationships), and their attributes and constraints in the airborne navigation equipment model. A Java conversion program is developed to convert the original model data into a general SysML model. During the conversion process, 18 marked abnormal data are cleaned and corrected, including adding 10 constraints and correcting 8 port mappings. Finally, the verified general model is stored in the platform model library in standard JSON format, version V1.0 is created, and dual backups are performed locally and in the cloud to achieve model version management.

[0043] Functional Unit Conversion: The functional modules of the parsed SysML model of the general airborne navigation equipment are identified and classified, identifying six major functional modules: satellite signal reception, positioning calculation, data communication, fault diagnosis, and power management. The input and output parameters of each functional unit are clarified. For example, the input of the satellite receiving module is a satellite radio frequency signal, and the output is a digital baseband signal. Based on the development indicators of the airborne navigation equipment (positioning accuracy ≤5m, cold start time ≤30s) and the requirements of the airborne installation architecture, functional unit conversion rules are formulated. The positioning calculation module and the fault diagnosis module are functionally associated. Fault self-checking logic is added to the positioning calculation module. The data communication module is split into two sub-units: wireless communication and wired communication, to adapt to the communication requirements of the airborne equipment. The input and output logic of each functional unit is adjusted to make the converted functional units compatible with the development requirements of the airborne navigation equipment.

[0044] Multi-source heterogeneous model integration: First, a unified modeling standard for heterogeneous models of airborne navigation equipment is established. For electronic models, the circuit component definition adopts the GB / T 14733 standard; for control models, the signal representation adopts the RS232 standard; and for mechanical models, the geometric dimension representation uses mm as the basic unit. Then, a middleware based on a distributed architecture is introduced. Its model management function completes the registration and querying of models such as satellite receiving modules and positioning calculation modules. The data conversion function converts ADAMS mechanical simulation data from txt format to the platform-compatible csv format. The communication coordination function manages OPC between models. UA communication connection was established; next, a dedicated format conversion tool was developed to convert the 3D model of the airborne navigation equipment shell created in AutoCAD into the IGES format recognizable by the simulation platform, and the positioning control system model created in Matlab / Simulink into the FMU format; a functional adaptation layer was designed to complete the conversion between satellite radio frequency signals and digital baseband signals, realizing signal interaction between various models; finally, the integrated model was verified, with unit tests verifying the independent functions of each functional module, achieving a 100% pass rate; integration tests checked the collaborative work between the models and corrected the signal packet loss problem between the data communication module and the positioning calculation module; comparative tests were conducted with the physical prototype of the airborne navigation equipment to verify the rationality of the simulation results, with the deviation between the simulation results and the physical prototype test results controlled within 4%, and the model parameters were optimized to address the deviation, completing the integration verification.

[0045] Virtual-Real Fusion Simulation: First, the virtual-real system interface of the airborne navigation equipment was designed. For the physical prototype's satellite signal receiver and GPS sensor, signal conditioning circuits and analog-to-digital conversion modules were designed to convert analog radio frequency signals into digital signals for transmission to the virtual model. For the physical prototype's display screen and alarm indicator lights, digital-to-analog conversion modules and drive circuits were developed to convert the positioning information and fault signals output by the virtual model into analog signals to drive the actuators. On the virtual simulation model, virtual interfaces corresponding to the physical interfaces were defined, establishing mapping relationships for parameters such as satellite signal strength, positioning accuracy, and cold start time. Subsequently, a data interaction and synchronization mechanism was built, using real-time Ethernet protocol as the data transmission protocol to meet the high real-time requirements of the airborne navigation equipment simulation. The requirements are as follows: AES128 encryption technology should be used to encrypt and protect the transmitted positioning data. Time synchronization should be achieved through a dynamic time synchronization algorithm combined with the PTP precision clock protocol, with a synchronization accuracy of 0.5ms. Next, based on the characteristics of the physical prototype, a high-precision virtual model should be constructed. An accurate circuit model of the electronic system should be established, considering the temperature drift characteristics and electromagnetic interference of components. A discrete event modeling method should be used for the control system, taking into account signal transmission delays. Finally, using test data from the physical prototype in different scenarios such as outdoor, indoor, and high-altitude environments, the satellite signal receiving sensitivity and positioning algorithm parameters of the virtual model should be adjusted using the gradient descent method to complete parameter calibration, ensuring that the deviation between the simulation results of the virtual model and the test data of the physical prototype is controlled within 2%.

[0046] Comprehensive analysis and iterative optimization: During the virtual-real fusion simulation of airborne navigation equipment, performance indicators such as positioning accuracy, cold start time, signal reception sensitivity, and fault alarm accuracy are acquired in real time through sensors and data acquisition modules. A positioning accuracy threshold of 5m and a cold start time threshold of 30s are set to achieve automatic alarms when these limits are exceeded. The platform's visualization module displays positioning accuracy curves and cold start time bar charts under different scenarios, enabling multi-dimensional visualization analysis. The built-in verification rule library automatically verifies the simulation data, identifying two instances of abnormal signal reception sensitivity. Simulation results show... The positioning accuracy of the airborne navigation equipment in indoor scenarios was 6.2m, exceeding the design threshold. The design team recorded relevant data and analyzed that the root cause of the problem was the low signal amplification factor of the satellite receiving module. Subsequently, they adjusted the amplification factor parameter of the satellite receiving module in the SysML model, increasing the amplification factor from 100x to 120x. After the adjustment, the simulation was re-run. The simulation results showed that the positioning accuracy in indoor scenarios decreased to 4.5m, the cold start time was shortened to 28s, and all performance indicators met the design requirements. The iteration process and results were recorded in the platform database to provide a reference for the subsequent upgrade design of airborne navigation equipment.

[0047] In this embodiment, through the technical solution of the present invention, the reuse rate of SysML models of airborne navigation equipment is increased from the original 15% to 80%, the cross-disciplinary design collaboration efficiency is increased by 55%, the matching degree between simulation results and actual systems is increased to over 98%, and the design optimization cycle is shortened from the original 10 days to 2 days, which greatly improves the development efficiency and product quality of airborne navigation equipment.

[0048] Example 3: Application of integrated SysML heterogeneous model of aircraft: This embodiment applies the SysML heterogeneous model integration method of the present invention to the overall development process of a small civil aircraft. This aircraft involves multiple professional fields such as mechanics, electrical engineering, control, navigation, and power, and employs more than ten modeling and simulation tools including MagicDraw, Rhapsody, Matlab / Simulink, CAD, CAE, and ADAMS. The heterogeneity of the models is a prominent issue, and data exchange and joint simulation cannot be achieved between the various professional models. The specific implementation steps of this embodiment are as follows: SysML Model Parsing: First, the format specifications of the aircraft SysML models generated by more than ten modeling tools are defined, and the dedicated APIs or parsing interfaces of each tool are adapted accordingly. Simultaneously, common XMI, JSON, and XML file parsing interfaces are configured to form a multi-interface parsing system. Then, the entire original SysML model file of the aircraft is read through these interfaces. A 2048KB data buffer pool and multi-layered anomaly capture mechanisms are set up to ensure the stability of reading large-scale model data. Format checks and element integrity checks are performed on the original data, marking 56 instances of abnormal data such as missing attributes, relational logic errors, and port mismatches. Next, based on SysML... The v2.0 general standard, combined with the characteristics of aircraft development, establishes full-element mapping rules for blocks (such as fuselage, wings, power system, navigation system), ports (such as power input ports, navigation signal ports), relationships (such as mechanical connections, data transmission, control commands), and their attributes and constraints. A C++ conversion program was developed to convert the original model data into a general SysML model. During the conversion process, 56 marked abnormal data were cleaned and corrected, including adding 28 missing attributes, correcting 15 relationship logic, and matching 13 ports. Finally, the verified general aviation aircraft SysML model is stored in the platform model library in the standard XMI format, creating version V1.0 and performing off-site multi-node backups to achieve version management and secure backup of large-scale models.

[0049] Functional Unit Conversion: The functional modules of the analyzed SysML model of the general aviation aircraft were identified and classified, resulting in 10 major functional modules, including fuselage structure, power supply, flight control, navigation and positioning, communication, and emergency response. The input and output parameters and internal logic of each functional unit were clarified. For example, the input of the power supply unit is fuel quantity and voltage, and the output is thrust and speed. Based on the development indicators of the small civil aircraft (maximum flight altitude 3000m, maximum flight speed 200km / h, endurance 4h) and overall architecture requirements, functional unit conversion rules were formulated. The flight control module and the navigation and positioning module were merged into a flight control and navigation unit, and the linkage logic between flight control and navigation was added. The emergency response module was associated with each functional module, and fault detection and emergency handling logic was added to each functional module. The power supply module was split into two sub-units: fuel power and electric auxiliary power, to adapt to the hybrid power requirements of the aircraft. The input and output logic of each functional unit was adjusted to ensure that the converted functional units are compatible with the overall development requirements of the aircraft.

[0050] Multi-source heterogeneous model integration: First, a unified modeling standard for heterogeneous aircraft models is established. For the mechanical domain model, the material properties and geometry adopt general aviation industry standards; for the control domain model, the command representation and signal transmission adopt the ISO 11898 standard; and for the power domain model, the performance parameters and calculation methods adopt GB / T standards. The system adopted the 25329 standard. Subsequently, a cloud-native architecture-based middleware was introduced. Its model management function was used to register, query, and schedule hundreds of sub-models across 10 major functional modules. Data conversion functionality enabled the mutual conversion of simulation data in different formats. Communication coordination functionality managed communication connections between large-scale models, ensuring data transmission stability. Next, a series of dedicated format conversion tools were developed to achieve format compatibility with CAD, CAE, and ADAMS tools, converting 3D geometric models to STL format and simulation models to FMU format to meet the integration requirements of a unified platform. A multi-layered functional adaptation layer was designed to complete signal conversion, parameter matching, and interaction between different professional models, resolving functional incompatibility issues between cross-domain models. Finally, the integrated large-scale heterogeneous model was validated. Unit testing verified the independent functionality of each sub-model, integration testing verified the collaborative work between functional modules, and comparative testing with a physical test platform for aircraft verified the rationality of the simulation results. The deviation between the simulation results and the physical test platform results was controlled within 5%. Model performance was optimized to address the linkage delay issue between power supply and flight control discovered during testing, completing the integration validation.

[0051] Virtual-Real Fusion Simulation: First, the virtual-real system interface of the aircraft is designed. Signal conditioning circuits and high-precision analog-to-digital conversion modules are designed for the speed, altitude, and attitude sensors of the physical test platform to convert analog signals into digital signals for transmission to the virtual model. For the propeller, servo motor, and landing gear of the physical test platform, analog-to-digital conversion modules and high-power drive circuits are developed to convert control commands output from the virtual model into analog signals to drive the actuators. Virtual interfaces corresponding to the physical interfaces are defined on the virtual simulation model side, establishing a one-to-one mapping relationship for core parameters such as flight altitude, speed, attitude, and thrust. Subsequently, a data interaction and synchronization mechanism is built, using real-time Ethernet protocol plus OPC. The UA protocol is used for data transmission via a combination of protocols to meet the high real-time and high reliability data interaction requirements of large-scale models. AES128 encryption technology is employed to encrypt and protect core data such as flight control commands and positioning data. A dynamic time synchronization algorithm combined with the PTP precision clock protocol achieves time synchronization between the physical test platform and the virtual simulation platform, with a synchronization accuracy of 1ms. Next, based on the characteristics of the physical test platform, a high-precision overall virtual model of the aircraft is constructed. Multibody dynamics modeling is used for the mechanical system, considering the elastic deformation of the fuselage and the aerodynamic characteristics of the wings. An accurate engine model is established for the power system, considering fuel consumption and thrust changes. A state machine modeling method is used for the control system, considering the adjustment logic of flight attitude. Finally, using test data from the physical test platform at different flight altitudes (low, medium, and high) and different flight states (uniform speed, acceleration, and turning), the aerodynamic coefficients, engine thrust coefficients, and flight control algorithm parameters of the virtual model are adjusted using the gradient descent method to complete parameter calibration, ensuring that the deviation between the simulation results of the virtual model and the test data of the physical test platform is controlled within 3%.

[0052] Comprehensive analysis and iterative optimization: During the virtual-real fusion simulation of the aircraft, core performance indicators such as maximum flight altitude, maximum flight speed, endurance, flight attitude stability, and fault emergency response time are acquired in real time through sensors and data acquisition modules. Maximum flight altitude thresholds of 3000m, maximum flight speed thresholds of 200km / h, and endurance time thresholds of 4h are set to achieve automatic alarms when these limits are exceeded. Through the platform's 3D correlation and visualization analysis module, the 3D geometric model of the aircraft is dynamically correlated with the SysML functional logic model and simulation data, displaying information such as flight attitude, thrust distribution, and fault location in real time in a 3D view, achieving three-dimensional visualization analysis. The built-in verification rule library is used to automatically verify large-scale simulation data, identifying three issues related to power supply and flight control linkage delays. Simulation results show... The aircraft's maximum flight altitude was 2800m, below the design threshold, and its endurance was 3.6 hours, also below the design requirements. The design team recorded relevant data and analyzed that the root cause was the low fuel utilization rate of the power supply unit and insufficient aerodynamic efficiency of the wing. Subsequently, they adjusted the fuel injection parameters of the power supply unit in the SysML model, increasing the fuel injection pressure from 1.0MPa to 1.1MPa, and simultaneously adjusted the wing airfoil parameters, increasing the wing's aspect ratio. After the adjustments, a new virtual-real fusion simulation was performed. The simulation results showed that the aircraft's maximum flight altitude increased to 3100m, the endurance extended to 4.2 hours, and the maximum flight speed reached 205km / h. All performance indicators met the design requirements. The iteration process and results were recorded in the platform database to provide a reference for the subsequent serial design of aircraft.

[0053] In this embodiment, through the technical solution of the present invention, the reuse rate of SysML models of small civil aircraft is increased from the original 10% to 75%, the cross-disciplinary design collaboration efficiency is increased by 70%, the matching degree between simulation results and actual systems is increased to over 97%, and the design optimization cycle is shortened from the original 30 days to 5 days. This effectively solves the model heterogeneity problem caused by multiple tools and multiple fields, and realizes the unified integration and high-precision joint simulation of the overall aircraft model.

[0054] Comparative Examples: To further verify the superiority of the technical solution of the present invention, three sets of comparative examples were set up to compare with the aero-engine application scenario of Embodiment 1 above. The comparative examples all adopted the model integration method of the prior art, as follows: Compared to Example 1, where no model ensemble method was used, and model preparation and simulation were performed manually: Different SysML models of aero-engines were manually organized and converted. The model splicing and simulation were completed manually, without virtual-real joint simulation, and the results were obtained only through virtual simulation. The results showed that the model organization took 7 days, 12 data errors occurred during the manual conversion process, the model reuse rate was only 20%, the cross-disciplinary collaboration efficiency was low, the deviation between the simulation results and the physical test bench reached 20%, the design optimization cycle was 15 days, and some performance indicators after optimization still did not meet the design requirements.

[0055] Compared to Example 2, which only uses some steps of the present invention, a closed-loop iteration is not formed: Using only the SysML model analysis and multi-source heterogeneous model integration steps of this invention, without functional unit conversion and virtual-real fusion test simulation, the analyzed model was directly subjected to virtual simulation and simple analysis. The results showed that the model reuse rate was increased to 45% and the cross-professional collaboration efficiency was increased by 20%. However, due to the lack of functional unit adaptation and virtual-real joint simulation, the simulation results deviated from the physical test bench by 12%. The design optimization cycle was 8 days, and the optimized fuel consumption index was still slightly higher than the design threshold.

[0056] Compare with Example 3, and perform model integration using a traditional single-toolchain integration platform: The TWC platform, designed for SysML modeling tools, was used for model integration. However, it could only manage models from single-type tools such as MagicDraw and Rhapsody, and could not integrate models from tools such as CAE, Matlab / Simulink. The results showed that the model reuse rate was 30%, the cross-disciplinary collaboration efficiency was improved by 15%, but cross-disciplinary joint simulation could not be achieved. Only virtual simulation of a single discipline could be completed, the simulation results deviated from the physical test bench by 18%, and the design optimization cycle was 12 days.

[0057] The comparison between the above three sets of comparative examples and Example 1 shows that the SysML heterogeneous model integration method of the present invention is significantly better than the existing methods in terms of model reuse rate, cross-disciplinary collaboration efficiency, simulation accuracy, and design optimization efficiency, and can fundamentally solve the integration problem of SysML heterogeneous models.

[0058] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The descriptions in the above embodiments and specification are merely illustrative of the principles of the invention. Various changes and modifications can be made to the present invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of protection claimed by the present invention. The scope of protection of the present invention is defined by the appended claims and their equivalents.

Claims

1. A SysML heterogeneous model integration method, characterized in that, Includes the following steps: S1, SysML model parsing, sequentially completes specification clarification and interface preparation, data reading and preliminary processing, mapping transformation and data verification, and model storage, converting SysML models generated by different modeling tools into models that conform to a common architecture and standards and building a model library, realizing version management and secure backup; S2. Functional Unit Conversion: Identify and classify the functional modules of the general SysML model, clarify the input, output and internal logic of each functional unit, formulate conversion rules based on the target system requirements and architecture, and complete the functional unit adaptation. S3, Multi-source heterogeneous model integration, achieves unified registration, scheduling and collaborative operation of cross-domain multi-format models through model standardization processing, middleware-based integration, model conversion and adaptation, integration verification and performance tuning; S4. Virtual-real fusion test simulation: Complete the interface design of virtual and real systems, build data interaction and synchronization mechanisms, construct and optimize virtual models and calibrate model parameters, and realize bidirectional real-time data interaction and high-precision joint simulation between physical systems and virtual simulation models; S5. Comprehensive analysis and iterative optimization: Through real-time monitoring of indicators, model optimization and adjustment, and closed-loop iterative optimization, the simulation results are analyzed from multiple dimensions and the SysML model is continuously optimized until the performance indicators meet the design requirements; data is shared among all steps and the results are synchronized in real time.

2. The SysML heterogeneous model integration method according to claim 1, characterized in that: In step S1, the specifications and interface preparation are defined as dedicated APIs or general file parsing interfaces that automatically adapt to modeling tools; data buffering and error capture mechanisms are set up when reading data; the format of the raw data is checked and abnormal data is marked in the preliminary processing stage; the mapping transformation covers the full element mapping of elements and attributes of blocks, ports, and relationships, as well as constraints; the data verification performs logical consistency and data integrity checks and cleans and corrects problematic data.

3. The SysML heterogeneous model integration method according to claim 1, characterized in that: In step S2, the functional unit conversion rules include merging, splitting and reorganizing functional modules. The adaptation method is to adjust the input and output logic of the functional units according to the architectural requirements of the target system so that the converted functional units match the technical indicators of the target system.

4. The SysML heterogeneous model integration method according to claim 1, characterized in that: In step S3, the middleware has model management, data conversion, and communication coordination functions, realizing model registration, query, scheduling, and format conversion and semantic parsing of different model data; model conversion and adaptation achieve format compatibility of models from different modeling tools by developing dedicated format conversion tools, and designing a functional adaptation layer to complete signal conversion and interaction between models.

5. The SysML heterogeneous model integration method according to claim 1, characterized in that: In step S3, integration verification and performance tuning include unit testing, integration testing, and comparative testing with the actual system. The accuracy and reliability of the model are verified through comprehensive testing, and the model performance is tuned to address the test issues.

6. The SysML heterogeneous model integration method according to claim 1, characterized in that: In step S4, the virtual-physical system interface includes a physical interface for the physical system and a virtual interface for the virtual simulation model. The physical interface includes a signal conditioning and analog-to-digital conversion module for sensors and a digital-to-analog conversion and drive circuit for actuators. A one-to-one parameter mapping relationship is established between the virtual interface and the physical interface. The virtual-physical data interaction adopts one or more of the following communication protocols: real-time Ethernet, OPC UA, bus, etc. The data transmission is protected by AES128 encryption technology.

7. The SysML heterogeneous model integration method according to claim 1, characterized in that: In step S4, the virtual model construction and optimization adopts multibody dynamics modeling for mechanical systems and establishes an accurate circuit model for electronic systems. The model parameter calibration is achieved by comparing the output of the virtual model with the test data of the actual system and adjusting the model parameters using the gradient descent method.

8. The SysML heterogeneous model integration method according to claim 1, characterized in that: In step S5, real-time monitoring of indicators acquires performance indicators such as reliability, efficiency, and response time through sensors and data acquisition modules, sets indicator thresholds, and implements automatic alarms when limits are exceeded. Data analysis uses visualization methods such as curves and charts, and at the same time, completes automated verification of performance indicators through a built-in verification rule base.

9. The SysML heterogeneous model integration method according to claim 1, characterized in that: In step S5, the comprehensive analysis adopts a three-dimensional lightweight model display method, which dynamically associates the three-dimensional geometric model of the system with the SysML functional logic model and simulation data to realize the three-dimensional visualization analysis of the system structure and the rapid location of faulty components.

10. The SysML heterogeneous model integration method according to any one of claims 1-9, characterized in that, This method is based on a unified data management platform, which includes a unified parsing and visualization editing module for multi-source SysML models, a multi-disciplinary simulation model scheduling engine module, a general real-time data interaction and closed-loop interface module, and a three-dimensional correlation visualization analysis module. All modules work together to achieve real-time data sharing and synchronization.

Images