An Exploration on Logical Ideas of System Architecture Modeling

CHEN Jie ,TAN Tianle

Shanghai Academy of Spaceflight Technology,Shanghai 201109

Shanghai Institute of Spaceflight Control Technology,Shanghai 201109

Abstract:China’s space technology has gradually improved from the early stages’ introduction,absorption and re-innovation based on backward design to independent innovation based on forward design.It is necessary to develop a new approach of systems engineering to improve the quality and efficiency of space systems design considering the large number of original design problems expected in the future.Adopting Model-Based Systems Engineering (MBSE)and Digital Twin method are important development initiatives in the field of modern engineering design.In the initial phase of system design,it is necessary to generate firm system architecture models based on the needs of stakeholders.The quality of the system design in this phase has a great impact on the detailed design and implementation for the subsequent system,and also plays an important role in the performance,development progress and cost of the whole system.Through the collaboration of cross-professional teams,modeling and model execution,comparing the model execution with expected results,MBSE has enabled digital model-level verification and validation before test verification and validation based on physical products,thus improving the design exactness,completeness and greatly reducing design errors or defects.This paper explores the logical ideas behind modeling of system architectures in order to promote the adoption of MBSE in the field of space systems.

Key words:Model-Based Systems Engineering (MBSE),system architecture modeling,logical ideas of working flow

1 INTRODUCTION

Major space systems are often complex engineering systems,involving multiple systems and hierarchical products such as launch vehicles,spacecraft and application systems,involving the orderly and collaborative participation of numerous organizations and professionals as well as multiple development phases and decades of coordinated work.How to ensure the success of the maiden launch and in-orbit operation of a system and how to improve the design quality,development efficiency and benefit of a system are always major issues to be considered for future complex space systems with originality and which there is a lack of experience to learn from,requiring continuously innovative development of new methods of systems engineering.

Systems engineering has played a very important role in China’s space engineering development.In summing up the experience of success and failure,systems engineering methods have been constantly developed and enriched.These acquired technologies and management approaches are now more focused on manufacturing,testing,and launch service technologies,as well as quality and program management.At present,the approach mainly relies on the proofreading and review of the design report by designers,or by a third party’s review and adjustment,thus the technical review ensures selecting the correct design and design systems correctly.The quality of the final design is influenced by the personnel’s sense of responsibility,experience and their mastery of information integrity.For products requiring process inspection and ground simulation test,final assembly inspection where only extensive test can find and correct the existing design errors,these tests and inspections need to be implemented after the realization of the product.Hence,the verification and validation (V&V) of the design requires a long period of time with high-cost,which is not conducive to improving the development efficiency or benefits.For products that cannot be inspected and tested in the simulated space environment on the ground,or the coverage of verification is insufficient,there are great technical risks in system development and the possibility of launch operation failure exists.

Model-Based Systems Engineering (MBSE) and the Digital Twin (DT) approach are important development initiatives in the field of modern complex engineering design.Based on the needs of the stakeholders and system use cases (UCs),MBSE continuously improves system requirements and model architecture elements through modeling and model implementation,and then derives a definitive hierarchical system architecture model.By comparing the result of model execution with the expected result of system operation,the digital model level V&V is implemented before the physical implementation and test V&V,thus improves the correctness and completeness of the design.DT technology is based on digital simulation technology,establishes accurate mapping twin relationships between physical entities and digital virtual entities,and then applies them in all stages of the system product design,eliminating possible defects of the physical systems before their implementation,and realizing an intelligent visual product design,optimization selection,virtual operation,virtual test,etc..The combination of MBSE and DT technologies can not only greatly improve the capability of system design V&V,but also can realize fault location and a maintenance strategy formulation during operation.This paper only discusses the subject of MBSE in the design stage,and analyzes the logical thinking behind the system architecture modeling,aiming to promote the better application of MBSE in the field of space engineering.

2 SYSTEMS ENGINEERING CONCEPT

2.1 Overview of Systems Engineering Methods

Systems engineering takes the system as the object and studies the engineering realization method.A system is an organic whole consisting of series of elements according to a certain structural relationship,centering on one or more prescribed goals,and it accomplishes the prescribed goals together through the interaction,interdependence and restriction of certain rules among all elements,between parts and the whole,and between internal and external environments.It can be expressed mathematically as:

Where,

S

is the whole system,

E

is all the elements {Ei}that constitute the system,and

R

is all interaction relations {Ri}of the system.System elements contain objects,which can be people,hardware,software,facilities,policies,documents or other representational items.System architecture refers to the composition of the system elements and the logical links that reflect the interaction,interdependence and constraints of system operation.The element composition and the linking relationships drive the operation of the system and enable the implement of the desired functions of the system.

The system mainly presents three kinds of characteristics.First is the integrity and purpose of the system ontology,which is the functional characteristics that the low-level elements do not exhibit,namely the emergence phenomenon of the whole system.Aiming at the overall optimization,system concepts always place the whole and the overall in the first place,dealing with the relationship between the whole and the parts from the overall perspective,emphasizing the subservience of parts to the whole.Second is the structure and hierarchy of the system composition.The elements of the system are diversified,and there is a certain structural static characteristic between the elements.Elements at different levels are presented in a tree-like hierarchy.The more levels the system contains,the more complex the system.Elements at the same level can be a distributed network,central network,series,bifurcation or exhibit a parallel link relationship.Third is the relevance and environmental adaptability of the system process,which is embodied in the interface relationship of material,energy and information,as well as the driving relationship of process or behavior,or the causal relationship following certain constraints.The system function can only be reflected in the interaction between the system and the environment,and the system should have certain adaptability to the environmental changes.

Systems engineering is a set of logical and interdisciplinary engineering technology realizations and management methods,which is applied to the whole life cycle process.The goal is to allocate elements reasonably,deal with all kinds of relationships,meet the expected goals of the system to the maximum extent,and pursue the optimal process of the system itself and that of the whole life cycle of the system.

The content involved in the systems engineering method can be described in the three-dimensional structure of Hall,namely the time dimension,the logical dimension and the professional dimension.In the time dimension,in order to ensure for the complex engineering system participation by many organizations and personnel,enabling the development cycle coordination and the technical status,development progress and quality control,we usually divide the life cycle of the system into multiple phases,classify the product status,and set milestone control nodes to carry out effective management and control.In the logical dimension,the V-shaped model is used to describe the whole system from the systems,sub-systems,and products,and describe the technical process and management process of the system from the requirements of analysis,logical function decomposition,design integration,to product implementation and integration verification and validation (I&VV).In the professional dimension,complex engineering systems are implemented by multiple enterprise organizations and professional teams,which are both supported and restricted by the organizational structure and industry norms.Figures 1 and 2 show the content of systems engineering work of a certain level of V-shaped model and the three-dimensional structure of the Hall system methodology captured in the two-dimensional plane based on the actual situation of China’s space industry.

Figure 1 Technical decomposition or integration and management activities at a certain system level

Figure 2 Plane unfolded model of Hall’s three-dimensional structure

2.2 System Architecture Modeling

Traditional Systems Engineering (TSE) adopts a document-centric systems engineering approach,and hence has an‘inherent defect’ in describing the system architecture.As documents are based on natural language,different people tend to have misunderstandings when expressing and transmitting the requirements for complex objects.With the expansion of the system scale and the increase of the personnel,organizations and levels involved in the system development,the problem becomes more prominent.

In October 2006,the International Council On Systems Engineering (INCOSE) in the Systems Engineering Vision 2020 formally proposedthe concept of MBSE,“Model-based systems engineering is the formalized application of modeling to support system requirements,design,analysis,verification and validation activities beginning in the conceptual design phase and continuing throughout development and later life cycle phases”.The advantage of MBSE is to ensure the consistency of information expression,it is based on a centralized model and requirements repository can support quasi-parallel work,facilitate knowledge accumulation and reuse,support early verification and validation.At present,MBSE is mainly used in the primary design phase of the project.

MBSE can describe,express and freeze the system concept formed in people’s mind through all kinds of model views,so as to facilitate the communication between all levels of personnel,and gradually deepen,refine and optimize the system design.MBSE describes system requirements,functional logic architecture,and physical architecture of the system through modeling methods (e.g.Harmony,OPM),visual modeling languages(e.g.SysML,UML),and modeling tools (e.g.MagicDraw,Rhapsody).Hence,driving the modeling and design activities based on the requirements of stakeholders to realize the verification and validation of requirements,functions and interfaces.With the help of relevant modeling tools software and the unified model repository,system designers at all levels can carry out collaborative design in different locations to continuously refine and improve the system design.With the help of models and data exchange standards,professional engineers of various disciplines can carry out work based on the system architecture models,extract models and parameters from the common repository,and use professional models and software tools of their disciplines to conduct professional modeling,simulation analysis,evaluation and optimization,as shown in Figure 3.

Figure 3 MBSE modeling,simulation analysis,evaluation and optimization

Figure 4 TSE comparison of MBSE design process

Figure 4 shows the main design process comparison between TSE and MBSE.The TSE system top-down design process can be summarized as four phases including the stakeholder needs survey,system requirements analysis,functional allocation and logical decomposition,and system design synthesis.The five loops include requirements verification,functional verification,design verification for functional architecture,design verification for the system requirements specification (SRS),and design verification for the stakeholder needs specification (SNS),as well as four kinds of output including SNS,SRS,system functional architecture,system physical architecture.

In order to support MBSE,INCOSE and the Object Management Group (OMG) proposed a Standard System Modeling Language (SysML) for systems engineering applications based on the Unified Modeling Language (UML).SysML is a subset of UML and extends the capabilities of UML.SysML is a graphical modeling language,which simplifies the physical system into graph modules and lines,and defines the meanings of various graph modules and lines.The combination of the whole graph can express the design intention of the system,which is easy for humans to understand.After being compiled by the computer,it can form data that the computer can understand.Through Extensible Markup Language Metadata Interchange (XMI) and systems engineering Data Exchange Standard (AP233),the data of system architecture model can be connected with professional fields through such models for mechanical,electrical,thermal software and so on.

SysML includes nine types of diagrams in three categories:requirement,structure and behavior,as shown in Figure 5.System design starts from the stakeholder needs,by defining system requirements and system use cases,module composition/connection relations/parameter constraints,behavior activities/sequences/states,repeated application of modeling and verification process,incremental iteration,hierarchical recursive work-flow,and depth to the lowest elements of the system.Integrating the models of all the elements at each level results in a complete system baseline architecture model.The system architecture models at different levels and dimensions can be used to carry out scheme trade-off studies,feasibility studies,alternative scheme demonstration and other work,and generate relevant documents automatically for relevant decision makers to use.

3 SYSTEM MODELING PREPARATION

Constructing the space system is to support the completion of specific missions,of which stakeholders have clear mission objectives and expectations of the system under development(SuD).Stakeholders may involve a large number of organizations and people.Investigating of the stakeholder needs,or mission analysis by a team of system developers and stakeholders,can identify such needs and operation expectations,which is the necessary preparatory work before system modeling.

Objectives of work:

To determine who is the stakeholder,to survey stakeholder’s ideas,conduct mission analysis with the stakeholder,define operation expectations and relationships with other systems of SuD,and define the Quality of Service(QoS) and key effectiveness evaluation indicators.

of work:

System developers can conduct stakeholder needs analysis in the form of interviews,seminars,and mission analysis.This process focuses on who,why,what,when,where,how and 8 kinds of constraints (5W1H8C),as shown in Table 1.

Output:

Word,Excel,Visio and other documents describe in words,items and graphics,and three categories of deliverables.

1) Expectation of operation and usage:Describes the expectation of operation and usage of the system in combination with the role orientation and use mode of the system in mission execution.

Figure 5 SysML diagrams relationship with UML diagrams

2) SNS:Describe stakeholder needs in the form of an item(STN,...,STN),describing the top level functionality and performance requirements of SuD,as well as the interface requirements between SuD and other external systems.

3) QoS:Performance,technology,cost,delivery time,reliability,safety,compliance,compatibility and other requirements;Mission execution environ-ment,life requirements,robustness/fault tolerance/redundancy requirements;Other professional engineering requirements,such as key performance indicators and productibility.

Table 1 5W1H8C

4 SYSTEM REQUIREMENTS ANALYSIS

The objective of the system requirements analysis is to map stakeholder needs into system requirements and system UCs,and to establish the traceability relationship between system requirements to stakeholder needs,and system requirements to system UCs.This ensures correspondence between stakeholder needs,system requirements,and system UCs.The subsequent functional logic analysis and system design synthesis will take the system UCs and system requirements as input to conduct the system architecture modeling and design.Through continuous improvement iteration and hierarchical recursion,the system UCs and system requirements will be improved to the lowest realizable unit,while the system UCs and SRS will be improved in the process.

The process of system requirements analysis is shown in Figure 6.

1) According to the results of mission analysis and stakeholder needs survey,the document System Requirements Specification (Draft) was sorted out in the form of items with the help of Word or Excel.System requirements items (SRS,...,SRS) in general have more entries than stakeholders needs (STN,...,STN);

2) Using requirement management tools (e.g.,DOORS)or MBSE modeling tools (e.g.,MagicDraw or Rhapsody),import SNS and SRS,and establish a traceability relationship between system requirements and stakeholder needs,compare the coverage between them,and improve the system requirements until full coverage.

3) According to system operational expectations,a set of use case diagrams were defined and drawn using system modeling tools.Establish a traceability relationship between system requirements to system UCs for each use case,and compare the coverage between them.For cases not covered,the system UCs need to be modified and improved until full coverage is achieved.

4) Exporting updated SNS,SRS and system UCs with an established relationship in the model/requirements repository.

5) Classify and sort the importance of system UCs.UCs with high importance should be guaranteed in the subsequent modeling design.UCs with low importance,if the system incurs a high cost,can be deleted under the premise of soliciting relevant opinions.

In fact system UC is a mode of work or a particular thread of operation that the system applies to perform a task.UCs do not reveal the internal structure of the system,only the relationships between the system and external actors.An external actor can be a person or another system outside of SuD.

Figure 6 System modeling preparation and system requirements analysis work-flow

5 FUNCTIONAL LOGIC ANALYSIS

The goal of system functional logic analysis is to transform each system UC into system functional modules and establish the logical correlation between functional modules through the process of system functional modeling and design.Among them,function modules carry functions required by the system to realize use cases,and logical association relations carry logical relations such as information drive or material/energy/information flow,serial/concurrent/condition judgment/transfer of activity or behavior processes between functional modules and between functional modules and external actors.

Functional logic analysis is based on the system UCs to carry out modeling and design,through the establishment of a system black-box (BB) model and the execution model,to confirm the consistency of the model execution results and the expected results,to verify the integrity and correctness of the model.The BB model only describes the correlation between the system functional modules and between modules and the external,but does not describe the internal structure and internal correlation of the functional modules.

MBSE starts from Use Case Diagram (uc) and uses Activity Diagram (act),Sequence Diagram (sd),Internal Block Diagram(ibd) and State machine Diagram (stm) to describe the functional blocks and logical association relations required by each system UC.Through a similar software debugging method,the BB model is executed step by step to verify the consistency between the execution result of the BB model and the expected system UC operation result,and ensure that every activity in the act and sd is executed.Through the ibd,the link ports and flow interfaces of functional modules are defined.Through the stm,state transition events and actor-driven behaviors are captured,and the results reflected by the system state transition and the sequence diagram are compared.Verify the satisfaction of system requirements by establishing a traceability relationship between UC attributes and system requirements.Finally,the functional modules derived from all the system UCs are combined with the logical association relations to obtain the overall black-box functional logical model of the system.

The black-box functional logic model of the system can be constructed in three orders,as shown in Figure 7.Among them,the second order is the commonly used order.During the functional logic analysis of all system UCs,some new requirements may be identified and some unusual condition branches or exceptions need to be addressed,and the system requirements specification or system UC may need to be updated.

Figure 7 Constructing work-flow for executable black-box functional logic model

To build the system functional logical model,the required inputs are SRS and System UCs.The output is the Activity Diagram,Sequence Diagram,Internal Block Diagram,State machine Diagram and system Interface Control Document (ICD)that describe each UC,as well as the updated System UCs and SRS.

6 SYSTEM DESIGN SYNTHESIS

System design synthesis is usually divided into two subphases:functional architecture trade-off analysis and physical architecture modeling and design.In the sub-phase of functional architecture trade-off analysis,various alternative schemes should be studied for each critical function identified.Through trade-off analysis,the optimal scheme can be selected,and finally,one or two system architecture schemes can be obtained by integrating each functional scheme.In the sub-phase of physical architecture modeling design,the white-box physical architecture model is established for the selected 1?2 system architecture schemes.Then decompose the refinement model until the physical implementation level.

The physical architecture models at all levels stored in the unified repository,combined with the data interface of the modeling tool,can lay a foundation for the subsequent modeling,simulation,optimization analysis,verification and evaluation at the professional level of multi-organizations and multi-specialties.

6.1 Functional Architecture Trade-off Analysis

The goal of functional architecture trade-off analysis is to determine the best scheme to realize the critical functions of the system under the constraints of system construction.Functional architecture trade-off analysis is widely used as a weighting method to evaluate candidate solutions that need to be evaluated multiple different metrics.Functional architecture trade-off analysis workflow is shown in Figure 8 (a) and as follows:

1) Identification and definition critical functions

CF

(

i=

1

,…

,

m

),the system may contain multiple critical functions.2) For each critical function,based on professional technical knowledge,identified possible technical solution

SA

(

i=

1

,…

,

m

;

j=

1

,…

,

n

).These solutions need to consider possible solutions for research and development,model development,flight applications,or commercial off the shelf (COTS).Critical function implementation technology solution selection,also need to consider technology maturity can reach level 5?6 constraint.3) A set of evaluation criteria

SC

(

i=

1

,…

,

m

;

k=

1

,…

,

p

)is extracted for each critical function.Evaluation criteria are usually based on stakeholder needs and the technical knowledge of the professional team.Usually,evaluation criteria include (performance,cost,time),or (function and performance requirements,QoS),or according to four categories of criteria (advanced performance,easy to use,system reliability,economic feasibility).4) Determine the weight of each evaluation criterion

Wt

(

i=

1

,…

,

m

;

k=

1

,…

,

p

) for each critical function

CF

.The weight,which is between [0,1],is assigned to each criterion according to its importance,and the sum of weight of all criteria should be 1.To determine the weight,qualitative and quantitative methods can be used,such as the Delphi method or other mathematical methods.5) For each evaluation criterion of critical function,define the utility function or utility curve,and determine the measurement of effectiveness

MoE

(

i=

1

,…

,

m

;

j=

1

,…

,

n

;

k=

1

,…

,

p

) value of each criterion of the candidate scheme.The utility function defined by each criterion is a dimensionless value,which is usually normalized as a value range [0,10].The utility value can also be proposed subjectively by professionals based on professional judgment.

7) The

SA

scheme corresponding to the maximum

TW

are selected as the best scheme of the critical function

CF

.8) Integrate the optimal critical function scheme

SA

into the system to obtain the overall optimal system function architecture scheme.

6.2 Physical Architecture Modeling and Design

The objective of the physical architecture modeling and design is to transform (1?2) system functional architecture schemes selected in the previous sub-phase into detailed physical architecture schemes.In this sub-phase,on the one hand,functions should be reorganized into corresponding physical product architectures.On the other hand,functional and non-functional requirements need to be assigned to the physical architecture level by level until the hardware and software configuration items are realizable.

The work-flow of the physical architecture modeling design is similar to the functional architecture analysis process,as shown in Figure 8(b).The physical architecture modeling and design process also starts from the system UCs and decomposes the UC realization into the system physical hardware or software architecture model.However,the model established is a white-box model,which needs to reveal the internal structure and relationships of subsystems,hardware components,and software configuration items.By reorganizing the black-box functional view and encapsulating it into the physical product,the black-box view is transformed into the white-box view,and the modeling and design process is completed continuously through iteration and hierarchical recursion.At this time,the nested white-box model view is formed.

Figure 8 Work-flow of system synthesis design

The final task is to integrate the system physical architecture model,which is a summary of all the UC realizations.The correctness and completeness of the physical architecture model of the system can be confirmed by the model execution,that is,by comparing the execution results step by step with the expected results.

In the process of model-driven system development,the final output is the baseline executable system physical architecture model,which contains system requirements specification,system use cases diagram,block definition diagram of each level,internal block diagram reflecting interface relation,parametric diagram reflecting parameter constraint relation,activity diagram,sequence diagram and state machine diagram of each level visualizing behavior,and package diagram reflecting nested model structure relation.The correctness is verified by execution of these model elements.

If the system,subsystem and product designers are from different organizations.The system designer delivers the architecture model of the system level,which is refined to the subsystem.The subsystem designer delivers the architecture model of the subsystem level,which is refined to the product.Finally,the system baseline architecture model is formed after integrating the modeling and design deliverables of designers at all levels.The baseline integrated system architecture model can be used as the basis of subsequent professional modeling and analysis.

6.3 Difficulties and Recommendations in Promoting MBSE Application

In order to promote the application of MBSE method in space system innovation,the difficulties and recommendations are as follows:

1) Designers at all levels who use MBSE need to have a deep understanding of systems engineering methods,SysML,MBSE software tools,and physical SuD.Therefore,designers are required to have higher quality and knowledge skills.

2) SysML diagram is quite flexible and does not strictly restrict in use in the development system.Promoting the use of SysML language,we can first use graphics as a sketch of the expression system,gradually cultivate the habit of using SysML;then in the project team or enterprise,the formation of the use of SysML description system specifications,accurate description of the system blueprint and system architecture code.

3) MBSE software tools for system requirements export,logical relationship test,model execution test and other processes,some can be automated,some need manual confirmation.Especially for the test of model execution results,we need to rely on manual step by step to verify the correctness,which is very cumbersome.For complex systems with complex logical processes,it is necessary to develop intelligent modeling tool software in the future,and there is also a need to develop Chinese independent software tools to promote the development of systems engineering methods.

7 CONCLUSIONS

By describing the logical process of MBSE system architecture modeling,the following conclusions and key points can be obtained:

1) Different from the traditional systems engineering based on text-driven,MBSE system design is a model-driven design process.While building a system architecture model,MBSE completes the system function and physical design.The system architecture model not only represents the composition of the functional (or physical) modules of the system,but also represents the logical relationship between the modules and the external actors.The system design starts from the highly generalized top-level stakeholder needs and operational expectations,enriches the system information in the process of architecture modeling and design,and complements the improved information to the system requirements or added system use cases.

2) During the requirement analysis phase,the emphasis is on ensuring the consistency among stakeholder requirements,system requirements and system use cases,which is realized by applying the traceability dependency relationship of modeling software tools.The subsequent modeling and design process is mainly based on system use cases.Therefore,building use cases covering all normal usage of the system is critical issue of this phase.Exceptions to use cases can be continuously identified and added during modeling and design processes.

3) In the functional logic analysis phase,it focuses on decomposing the functional modules needed by the system,and establishing the logical correlation between the functional modules and the external actors.The black box model,rather than the white box model,is used here to reserve the possibility for multi-scheme innovation realized by the functional module.To achieve the integrity of the system functional architecture model by ensuring the consistency of the model execution results with the expected system use case running results.

4) The system design synthesis phase is divided into two sub-phases.In the functional architecture trade-off analysis sub-phase,emphasis is placed on weighing multiple physical implementations of critical functions and optimizing schemes.The optimal selection scheme of multiple critical functions is integrated to obtain 1?2 system physical implementation architecture schemes.In the sub-phase of physical architecture modeling and design,on the one hand,the functional architecture is mapped to the physical product architecture,on the other hand,it is recursive to the realizable hardware and software configuration items.