The ACEWEM computational laboratory : an integrated agent-based and statistical modelling framework for experimental designs of repeated power auctions

This research work develops a novel framework for experimental designs of liberalised wholesale power markets, namely the Agent-based Computational Economics of Wholesale Electricity Market (ACEWEM) framework. The ACEWEM allows to further understand the effect of various market designs on market efficiency and to gain insights into market manipulation by electricity generators. The thesis describes a detailed market simulations whereby the strategies of power generators emerge as a result of a stochastic profit optimisation learning algorithm based upon the Generalized Additive Models for Location Scale and Shape statistical framework. The ACEWEM framework, which integrates the agent-based modelling paradigm with formal statistical methods to represent better real-world decision rules, is designed to be the foundation for large custom-purpose experimental studies inspired by computational learning. It makes a methodological contribution in the development of an expert computational laboratory for repeated power auctions with capacity and physical constraints. Furthermore, it contributes by developing a new computational learning algorithm. It integrates the reinforcement learning paradigm to engage past experience in decision making, with flexible statistical models adjust these decisions based on the vision of the future. In regard to policy contribution, this research work conducts a simulation study to identify whether high market prices can be ascribed to problems of market design and/or exercise of market power. Furthermore, the research work presents the detailed study of an abstract wholesale electricity market and real UK power market.

Name dc:type.qualificationname

phd

Level dc:type.qualificationlevel

doctoral

Grantor dc:publisher.institution

London Metropolitan University

Year dc:date.issued

2015

Author dc:creator

• Kiose, Daniil

• 330 Economics
• 380 Commerce, communications & transportation

Dc Identifier Grantnumber

N/A

OAI identifier oai:identifier

oai:repository.londonmet.ac.uk:1257