Agile Model-Based System Engineering for multidisciplinary optimization in future vehicle development

he proposed research project aims at reinforcing IVECO research initiatives in the application of MBSE approaches to support the complex design of new vehicles. MBSE [1] is a methodology to enhance multidisciplinary design and simulation of complex systems and can be used to specify functions and behaviours of a system [2]. Model-based approaches have been incorporated into the safety and security analysis process to simplify the analysis process and improve the system design's efficiency and manageability [3]. Combining MBSE and MDAO (Multidisciplinary Design Analysis and Optimization) approaches for the benefits of systems engineers is still an open issue [4]. Top-down approaches are needed to ensure consistency and continuity from requirement level to behavioural level up to the functional and structural levels [5, 6]. The use of simulation is a specific part of an MBSE approach that is crucial in an efficient project management since it might be the source of high cost and skills needs but also a source of precious knowledge on the future system [7].

To achieve the full benefit of the MBSE approach the systems engineering community must also rely on external models that capture more sophisticated analysis across a wide variety of domains. These domain models include simulations that measure operational effectiveness, life cycle costing models, physics-based computational simulations, manufacturing models, etc. [8] propose an approach that integrates these external simulation models with the MBSE integrated model using statistical metamodels that act as surrogates to the simulations.

Surrogate models are simple analytical models that mimic the input/output behaviour of complex systems. Developing such models requires performing computationally expensive simulations at a set of carefully selected sample points. These models approximate the behaviour of the underlying complex simulations to a reasonable precision while also being computationally cheaper to evaluate. Surrogate models can thus be seen as a simple representation of a complex system with relaxed accuracy in a given domain. The trade-off between the accuracy and the computational time is an important consideration during the construction of these models [9]. Engineering Design via Surrogate Modelling guides to surrogate models and their use in engineering design [10]. System optimization to rapidly generate and assess new designs using interactive analysis and visualizations uses techniques such as surrogate modelling [11].

Design Space Exploration (DSE) is the process of finding a design for one or several solutions that best meet the desired design requirements, from a space of tentative design points. This exploration is naturally complex, as the search may involve tentative designs resultant from applying code transformations and compiler optimizations at various levels of abstraction and/or from the selection of specific values of parameters (e.g., configurations) or even the selection of algorithmic alternatives [12]. The design of system architectures often involves a combinatorial design-space made of technological and architectural choices. A complete or large exploration of this design space requires the use of a method to generate and evaluate design alternatives [13]. The design-space is explored by a procedure that uses beam search to promote fast convergence towards optimal or near-optimal solutions. Additionally, the optimality of the design solutions is improved by local searches which are performed after a first guess of the stacking sequences [14].