Simulation of garden water dispersal controller using fuzzy expert system / Ku Shairah Jazahanim

One of the most important elements in lawn maintenance is the moisture adequacy. For
this reason, irrigation, done by manual or automated sprinkler system, has been applied.
However, both systems may use excessive amount of water and the amount dispersed
may not be suitable for the moisture level of the lawn. Therefore, there is a need to
develop an irrigation system that can measure and monitor the soil moisture through data
acquired from the soil and also from the climatologic factors that will help to decide
when to water and how much water is needed. This research demonstrates the usage of
Fu2zy Logic in irrigation control system. The types of fijzzy inference that applied to the
irrigation system were Mandani with normal subsets, Mandani with hedges, Sugeno with
normal subsets and Sugeno with hedges. The first step applied is the element
identification involves in the irrigation process which was determined by consulting the
relevant expertise and literature. Once the actual rules and frizzy sets are determined, the
comparison of the conventional irrigation system with all four frizzy inference methods
was conducted with each other. The intention is to see which system is better in
optimizing water usage. Lastly, a simulation system was built to demonstrate the soil
moisture content of the lawn, the percentage pattern of soil moisture and daily data
involved in the system. This project was restricted to the Bermuda Turf grass
characteristic, the loam soil characteristic, 92.9m lawn sizes, pop-up spray head
sprinkler, evapotranspiration (ET); the climatology factor and the soil moisture reading
from tensiometer. This project is significant to the irrigation industry whereby, new
irrigation product can be produced using intelligent systems. From the comparison
made, it was shown that the frizzy expert irrigation system performed better based on the
lower annual average water usage for the whole year recorded. The most effective frizzy
inference method applied was the Mandani style with normal subsets which uses the
least amount of water. The result of this research shows promise for friture R&D in
intelligent irrigation system either for individual or enterprise level management of
plants.