MBSE Approach for Railway Digital Continuity

Railway system’s design, operation and maintenance require the involvement of several stakeholders such as system engineers, safety engineers, railway experts etc. Indeed, in the system design phases, actors and needs are raised and requirement engineers identify functional requirements, their allocation to stakeholders, flows and interfaces between physical objects. Each phase of the system development involves multi-disciplinary experts that share heterogeneous data and collaborate without an unambiguous terminological guide about exchanged concepts. Actual practices are mainly based on sharing specification and documents in textual form in order to communicate about system related choices. Furthermore, models are built independently and refers to different levels of Model-Driven Architecture (MDA), such as Computational Independent Model (CIM), Platform Independent Model (PIM) and Platform Specific Model (PSM) [1]. In the railway domain, different tools are used for each phase of the development process from the industrial needs specification to the prototype implementation. These tools does not generally allow a digital continuity in the information, model or data life-cycle. In order to ensure models traceability and to facilitate the communication between stakeholders, Model-Based System Engineering (MBSE) approaches are adopted in critical systems development, like railway systems [2]. These approaches allow a robust documentation of models and improve their reuse in different contexts. Nevertheless, MBSE does not allow to deal with semantic heterogeneity and to share a common terminology during the whole system development process. From this context, the following Research Question (RQ) is identified: How to ensure semantic interoperability in such multidisciplinary ecosystem, like the railway domain?

In this paper, we introduce a MBSE approach which is able to help decision making and to have an efficient collaboration between actors. This approach is based on the integration of ontology as a structured and shared representation which is maintainable during the system life-cycle [3]. This knowledge representation has been widely used in order to disambiguate different domains. In the next section, the articulation between knowledge engineering and system engineering is introduced with the aim to meet railway issues.

The proposed approach is illustrated by Fig. 1. It is based on three domains of interest:

This approach is structured in a top-down way from the abstract level which represents the need’s specification until the implementation level. Knowledge structuring using ontologies is proposed as a crucial step in the system development, especially in complex operational activities related to system performance evaluation. Indeed, an ontology is defined as “a formal, explicit specification of a shared conceptualisation” [4]. It is considered as the basis of the whole process since it disambiguates the domain terminology and provides a common and shared representation between all stakeholders.

Bild vergrößern

A domain ontology is formalised using the Web Ontology Language (OWL) and represents CIM level which is crucial to represent the domain view. In order to fulfil the critical needs of railway systems and to federate the railway data models representing heterogeneous subsystems of the railway infrastructure, an upper ontology, Unified Foundational Ontology (UFO) [5], is used. Upper ontologies provide foundational concepts and relations and their definition, such as space, process, situation, etc. [6]. Domain ontologies that are grounding in upper ontologies are more reusable since this practice facilitates the semantic alignment between heterogeneous knowledge domains. Conceptual models are considered as lightweight ontologies represented in Unified Modeling Language (UML) [7] and are a common part of both knowledge and system engineering. They are also used to be Business Models (BM) which represent the most pertinent concepts of the domain.

Using MBSE foundations, model-driven approaches and related standards, the conceptual model is then extended and refined in design model (PIM) to represent system features and attributes. Design models include several system properties as design constraints in the system architecture models for example. Therefore, these models are transformed into PSM to provide models in a specific language/format. At this level, the implementation/serialisation is performed to have a concrete prototype.

The whole process is supported by a documentation process which is required to maintain and manage changes traceability during the system design. Furthermore, the key aspect of this approach is the digital continuity between all system design phases since the ontology is the essence of the common semantic interpretation. This methodology is as generic as possible to ensure semantic interoperability in the design of the railway system assets. In the next section, the validation of this MBSE approach in the railway domain is presented.

In previous work, an ontological approach was proposed with the aim to integrate dysfunctional analysis into the system design process of railway systems [8]. Then, this approach is illustrated and validated by real railway accidents. The key contribution of this approach is to ensure a semantic link between the safety measures and safety requirements in the railway system. This approach helps the safety-decisions making process and enhances the collaboration and communication.

Several initiatives have been developed as railway reference models that represent the infrastructure, such as Rail System Model (RSM) [9], IFC Rail¹ and EULYNX² that are built independently in UML. However, there is a digital discontinuity and a lack of interoperability between these standards. The proposed MBSE approach is successfully used to deal with this semantic heterogeneity and to ensure digital continuity between the railway ontology and the system architecture. The developed railway domain ontology aims to avoid semantic ambiguities of railway concepts, such as Track, Signaling, Energy and Telecom and relations between them [10]. Then, this railway ontology is used to help decisions making in maintenance activities.

In order to ensure the digital continuity in the development of new systems with high performance, the proposed approach is validated and applied in the autonomous train development, namely the Autonomous Train Map (ATM) subsystem [11]. An ontology was developed in order to represent the ATM concepts and relations. Then, conceptual and design models were built, transformed in PSM and then serialised in code to be integrated in the final system.

The proposed top-down approach has powerful and generic capabilities to be applied in different fields and use-cases for the whole life-cycle phases from the design to operation and maintenance of systems. In the railway domain, this approach is used to ensure digital continuity. Indeed, the ontology layer is integrated as a knowledge basis to establish the semantic interoperability between railway actors. This MBSE approach improves collaboration and helps decisions-making process during the system life-cycle. It has been widely used and validated in the autonomous train development process and to ensure digital continuity in the railway domain.

In future work, we intend to adapt this approach for the development of railway digital twins (DT) with the integration of a digital representation layer, such as Building Information Modelling (BIM) [12]. Therefore, the integration of ontologies, Machine Learning (ML) algorithms and BIM in the DT development for the railway domain should meet scientific and industrial challenges related to data interoperability and integrity regarding the multiplicity of data sources. Finally, we plan to validate it by real use cases of predictive maintenance of track components using its BIM models.