|CARLINO ANGELO||Cycle: XXXIV |
Section: Systems and Control
Tutor: DERCOLE FABIO
Advisor: CASTELLETTI ANDREA FRANCESCO Major Research topic
:Evolutionary multi-objective direct policy search for control of socio-climatic systemsAbstract:
The research project aims to develop a systems dynamics model to describe the interconnections among the economic and the climatic system, i.e. a simple integrated assessment model (IAM). The evolutionary multi-objective direct policy search (EMODPS) algorithm will be used to solve a multi-objective optimal control problem.
The DICE (Dynamic Integrated Climate Economy) model (Nordhaus, 1992) is recognized to be one of the most influential IAM and it is widely used as a support tool for policy makers and to assess future socio-climatic scenarios.
First we will rewrite the DICE model from a system dynamics perspective introducing stochastic disturbances to improve the problem formulation and its ability to represent the actual decision process.
The introduction of multiple objectives will avoid the aggregation into a single economic cost function of all the performance indicators of the system. This will be of fundamental importance as we will consider objectives that are difficult to convert into an economic value, e.g. temperature increase at the end of the century. The Pareto solutions will be more of interest for the policy makers as they make explicit the tradeoffs and the conflicts among the different objectives.
EMODPS will be used to find the Pareto optimal control policies which will undergo a robustness analysis to evaluate their reliability with respect to the wide range of uncertainties characterizing such models.