Application area: Mechatronic Systems.

Researcher in charge: Pouya Mahdavipour.

Supervisors: Prof. Törngren, Dr. Feng.

Host: Kungliga Tekniska Högskolan (KTH Royal Institute of Technology), Sweden.

Secondments: Philips Medical Systems, Netherlands.

Project Description and State of the Art

Industrial practices in model-driven engineering (MDE) of CPS, lack a systematic and cost-efficient approach to model integration and management. While MDE is often implemented for certain parts (e.g. mechanical, control and software engineering), dependencies and the overall architecting of the models across disciplines and aspects, are only rudimentarily addressed. This results in inconsistencies, difficulties to find and reuse information, and fragile tool integration. Product Lifecycle Management (PLM), Application Lifecycle Management (ALM) and domain-specific solutions are being proposed, but overall there is lack of methodology. The predominant industrial practice to CPS architecting often results in suboptimal designs since new functionalities are directly decomposed into part functions suitable for a specific technology and discipline, using the available platforms and components. There is thus a lack of system-level models and model integration that can support architecting at the level of CPS.

Expected Contributions

The design of CPS is inherently a multi-view problem because of the multiple concerns of interests including, e.g., dimension, power, weight, performance, safety, modularity and sustainability. This project focuses on developing a novel multi-view modelling framework for the holistic design optimization of CPS.

  • Multiview modelling and concurrent engineering of CPS: The objective is to develop a multi-view modelling framework, which explicitly formalizes various types of dependencies among heterogeneous views and supports dependency analysis using the graph theory. The analysis shall properly decompose a large problem into a group of smaller sub-problems and identify the critical kernel of the sub-problems.
  • Design space exploration techniques: The design of CPS involves choosing physical as well as cyber components and their synergetic integration. Since the number of possible alternatives is very large, there is a need for a systematic method to explore all possible solutions and find the optimal one. The searching and optimization include both architectural design and parameter optimization. The most effective design space exploration methods applicable to CPS, such as Constraint Satisfaction Problem (CSP), SAT solver, model checking, global optimization and etc., shall be studied.
  • Application of the developed theory and technologies to real-world case studies: Realistic and challenging case studies shall be investigated through collaborations with industrial partners. These case studies will be solved with both the proposed new methods and state-of-the-practice methods. The pros and cons of the proposed new methods will be summarized through quantitative comparison and analysis.