Ontologies for the interoperability of heterogeneous multi-agent systems in the scope of energy and power systems

Laburpena

The electricity sector, traditionally run by monopolies and powerful utilities, has undergone significant changes in the last decades. The most notable advances are an increased penetration of renewable energy sources (RES) and distributed generation, which have led to the adoption of the smart grids (SGs) paradigm and to the introduction of competitive approaches in wholesale and some retail electricity markets (EMs). SGs rapidly emerged from a widely accepted concept to reality. The intermittency of RES and their large-scale integration poses new constraints and challenges, strongly affecting EMs’ operations. The challenging environment of power and energy systems (PES) reinforces the need for studying, experimenting, and validating such competitive, dynamic, and complex operations and interactions. In this context, simulation, decision support, and intelligent management tools become essential to study different market mechanisms and the relationships among the involved stakeholders. To this end, the new generation of tools should be able to cope with the quick evolution of PES, providing participants with adequate means to adapt themselves, addressing new models and constraints, and their complex relationship with the technological and business developments. Multi-agent-based platforms are particularly well suited for analyzing complex interactions in dynamic systems, such as PES, due to their distributed and independent nature. The decomposition of complex tasks into simple assignments and the easy inclusion of new data and business models, constraints, types of players and operators, and their interactions are some of the main advantages of agent-based approaches. In this domain, several modeling tools have emerged to simulate, study, and solve problems of specific PES subdomains. However, there is a generalized limitation referring to the significant lack of interoperability between heterogeneous systems, which prevents from addressing the problem globally, considering all the relevant existing interrelationships. This is essential to enable players taking full advantage of the evolving opportunities. Thus, to accomplish such a complete framework while taking advantage of existing tools that allow the study of specific parts of the global problem, interoperability between these systems is required. Ontologies facilitate the interoperability between heterogeneous systems by giving semantic meaning to information exchanged between the various parties. The advantage lies in the fact that all those involved in a particular domain know them, understand, and agree with the conceptualization defined therein. There are, in the literature, several proposals for the use of ontologies within PES, encouraging their reuse and extension. However, most ontologies focus on a specific application scenario or a high-level abstraction of a PES subdomain. Moreover, there is considerable heterogeneity among these models, hardening their integration and adoption. It is essential to develop ontologies representing distinct knowledge sources to facilitate the interactions between entities of different natures, promoting interoperability between heterogeneous agent-based systems that enable solving specific PES problems. These gaps motivate the development of the research work of this Ph.D., which emerges to provide a solution for heterogeneous systems interoperability within PES. The several contributions of this work result in a society of multi-agent systems (MAS) for the simulation, study, decision support, operation, and intelligent management of PES. This MAS society addresses PES from the wholesale EM to the SG and consumer energy efficiency, taking advantage of existing simulation and decision support tools, complemented by newly developed ones, ensuring interoperability between them. It uses ontologies for knowledge representation in a common vocabulary, easing interoperability between the various systems. Furthermore, using ontologies and semantic web technologies allows the development of model agnostic tools for a flexible adaptation to new rules and constraints, promoting semantic reasoning for context-aware systems. The developed framework has been tested and validated against different contexts, considering both real-world operation and laboratory simulation environments, and under realistic scenarios using real and simulated data from heterogeneous sources acquired from databases or in real-time and represented in a common ground semantic model. The promising results achieved under realistic conditions support the thesis that ontologies contribute to increasing interoperability between heterogeneous tools directed to the study and management of PES, making it possible to address the problem globally.