Industrial complex systems demand the dynamic adaptation and optimization
of their operation to cope with operational and business changes. In order
to address such requirements and challenges, cyber-physical systems promotes
the development of intelligent production units and products. The realization of
such concepts requires, amongst others, advanced data analysis approaches,
capable to take advantage of increased availability of data, in order to overcome
the inherent dynamics of industrial environments, by providing more modular,
adaptable and responsiveness systems. In this context, this work introduces an
agent-based data analysis approach to support the supervisory and control levels
of industrial processes. It proposes to endow agents with data analysis capabilities
and cooperation strategies, enabling them to perform distributed data analysis and
dynamically improve their analysis capabilities, based on the aggregation of
shared knowledge. Some experiments have been performed in the context of an
electric micro grid to validate this approach.
