Multi-agent distributed model predictive control with fuzzy negotiation.

• A distributed model predictive control with fuzzy negotiation has been developed. • The controller has been developed using a multi-agent software platform. • The fuzzy negotiation allows for the inclusion of economic or operating criteria. • The validation has been performed in a simulation of the four coupled tanks system. In this work, a multi-agent distributed model predictive control (DMPC) including fuzzy negotiation has been developed. A novel fuzzy inference system is introduced as a negotiation technique between agents in a cooperative game algorithm, allowing for the consideration of economic criteria and process constraints within the negotiation process, providing an easier interpretation of the available knowledge. The fuzzy negotiation produces smoother control actions than where the negotiation is based only on costs evaluation, because both agents provide their best to generate the final control action. The results show good tracking and disturbance rejection in the case study proposed. The methodology has been implemented in a JAVA based platform with a friendly user interface to deploy the multi-agent system (MAS), and it has been validated in the water level control in a four coupled tanks system. [ABSTRACT FROM AUTHOR]