Use this URL to cite or link to this record in EThOS:
Title: Multilevel design for complex engineered systems
Author: Hambley, Chris J.
ISNI:       0000 0004 7655 2987
Awarding Body: University of Sheffield
Current Institution: University of Sheffield
Date of Award: 2018
Availability of Full Text:
Access from EThOS:
Access from Institution:
This thesis presents a multilevel design approach that uses formalized requirements to facilitate design synthesis techniques, such as optimization, at increasing levels of fidelity. The benefits are that verification is built into the design, guaranteeing requirements satisfaction. It also focuses the design effort higher up, spending more time considering what the system needs to do, and the attributes it should have. Specific examples of design synthesis techniques are developed in the main chapters, showing how they fit into the wider multilevel framework. Architecture optimization has been implemented at both high and low levels. High-level architectures are composed as a combination of physical means for achieving a set of functions. A multiobjective genetic algorithm is used to produce a set of Pareto-optimal solutions. Refinement to a single solution is then implemented using a customer-oriented approach. This produces solutions that the customer wants whilst reducing the need for iterative discussions with engineers. Low-level topology is represented as a graph with nodes (system components) and edges (interconnections between components). The topology requirements are formulated as constraints on the graph and synthesis is achieved via constrained optimization. The approach is applied to a turbofan oil system case study with two objectives: increasing controllability, via the addition of controllable valves, and minimising cost. The methodology provides benefits to system designers by selecting cheaper architectures with fewer valves when the need to control oil chambers separately is small. A simulation-based approach for performing control synthesis with signal temporal logic (STL) requirements is presented. The goal is to find control parameters that maximise the margin of satisfaction of the STL formulae. The quantitative semantics of STL are extended to a multiobjective formulation called multiSTL. In multiSTL each requirement margin is displayed on a parallel coordinates plot, which allows tradeoffs between different requirements to be analysed. This can also be used to highlight where relaxing of some requirements might yield better performance in other areas.
Supervisor: Kadirkamanathan, Visakan ; Jones, Bryn Sponsor: Not available
Qualification Name: Thesis (Ph.D.) Qualification Level: Doctoral
EThOS ID:  DOI: Not available