Development of a multi-agent system simulation platform for irrigation scheduling with case studies for garden irrigation

The adoption of irrigation control strategies, aimed at attaining the desired level of humidity for each plant type, can improve the costs and energy consumed in small-scale site-specific irrigation systems. The paper presents a knowledge-based and distributed framework that simulates the behaviour of an irrigation system and permits accurate determination of irrigation timing. Several agents, which represent the actors involved in this problem, coordinate their activities in order to evaluate different irrigation strategies. A common ontology shares the knowledge required in the agent-based framework, which can be tuned according to the particular circumstances of the field. The usefulness of the developed system is demonstrated in three case studies, in which the simulations performed by the system provide the answer to different questions (length of irrigation time, comparison of a fixed and a dynamic irrigation policy, and most efficient configuration of a garden). The system simulates the behaviour of the irrigation system for the possible solutions and finds the most efficient one in terms of water consumption. Although only at small areal scale, this paper shows how agent-based simulation techniques can be successfully used to solve agricultural problems.

► Irrigation scheduling policies should consider climatic variables and be dynamic.
► A multi-agent system composed of rule-based actors may exhibit a complex global behaviour.
► The proposed agent-based simulation framework is scalable and flexible enough to be adopted in a wide range of scenarios.