Portal de Periódicos da CAPES

Artigo Produção Nacional Revisado por pares

BIBTEX RIS

Mixture design of experiments on portfolio optimisation of power generation

2016; Institution of Engineering and Technology; Volume: 11; Issue: 2 Linguagem: Inglês

10.1049/iet-gtd.2016.0422

1751-8695

André Rodrigues Monticeli, Pedro Paulo Balestrassi, Antônio Carlos Zambroni de Souza, Rafael Coradi Leme, Anderson Paulo de Paiva,

Advanced Multi-Objective Optimization Algorithms

A methodology for obtaining an optimal portfolio for the generation of electricity at the lowest cost and risk is proposed. This methodology uses a mixture design of experiments (MDEs) as a strategy for building nonlinear models of risk and cost in portfolio optimisation for the generation of electricity. The result is compared with the traditional theory of Markowitz mean–variance (MVP). The following characteristics are also presented in this study: the seasonality and volatility of the time series were manipulated using moving windows and computational replicas in MDE; desirability functions were used to optimise multiple variables, leading to lower cost and risk; Shannon entropy index was used to handle better portfolio diversification. A case study based on the energy market of the state of California was used to illustrate the proposal. The results show that this methodology facilitates decision making.