Fuzzy logic based automated irrigation control system optimized via neural networks

This research focused on application of fuzzy logic and neural networks to control an automatic irrigation system and to estimate evapotranspiration. A fuzzy logic system is able to simultaneously manage numerical data and linguistic statements, mimicking human reasoning. Fuzzy control basically refers to the control of processes through fuzzy linguistic descriptions. An artificial neural network has the ability to utilize examples taken from data and to organize the information into a model that relates the input and output variables. An automated irrigation control system was developed using the fuzzy logic approach as an alternative to improve the current management technologies applied in real-time irrigation scheduling and control. Data from an automated weather station and from soil moisture sensors were used as inputs to the system. A fuzzy logic estimation system was designed to map solar radiation and relative humidity into evapotranspiration. The fuzzy control system processes the estimated evapotranspiration and soil moisture contents at two soil depths using linguistic knowledge to produce a control output that is sent to a solenoid valve. The fuzzy irrigation control system was implemented and tested in 1997 using microirrigation under plastic mulch in bell peppers (Capsicum annuum L.). The fuzzy irrigation control system presented a very good performance in response to the climatic and soil moisture variations in a fully automated manner. The soil water potential was kept between –6.5 kPa and –15. 7 kPa, evidencing the feasibility of the application of fuzzy logic to automated irrigation control systems. Neural networks were then used to process the system input and output data for redefinition of membership functions, in order to optimize the system.