| المستخلص: |
Predicting the growth rates of livestock is one of the interesting research areas because of the importance of livestock in countries economics. Because the field of research contains a lot of incomplete, unclear information and the absence of accurate statistics of livestock, fuzzy logic methods have been used. In this thesis, a fuzzy expert system is designed for predicting the growth rates of livestock (Sheep, Goat, Cow, Buffalo and Camel) in Thi-Qar province. The proposed system is built and developed by using Mamdani as fuzzy inference system, triangular membership function for fuzzification process, fuzzy rules to represent the knowledge base and center of area for defuzzification. Visual basic .net used to implement fuzzy inference engine and user interface while SQL server 2008 used to include the knowledge base. The knowledge base collected from experts and competent authority (Directorate of meteorological in Nasiriyah, the Veterinary Hospital in Nasiriyah and Directorate of Agriculture in the province of Thi-Qar). There are 11 main-knowledge which branch into sub-knowledge used to build rules. There is command knowledge and there is a specified knowledge for each livestock (Sheep, Goat, Cow, Buffalo and Camel).The collected data are to three years in (2013, 2014 and 2015) and the proposed system will find the rate of growth for each kind of livestock within two years (2013-2014 and 2014-2015) and then compare the results with the actual growth rates which are obtained from our proposed system with actual results by comparing year by year in actual world by the same years in our proposed system. There are differences in numbers of rules for each livestock kind, there are 832 fuzzy rules for sheep, 613 fuzzy rules for goat, 630 fuzzy rules for cow, 598 fuzzy rules for buffalo and 446 fuzzy rules for camel. Confusion matrix and relative error is used in this thesis for testing and evaluating the proposed system. The system testing shows that this fuzzy expert system has a very promising performance with an accuracy of 94%, 92% , 93% ,90% and 89% for sheep, goat, cow, buffalo and camel respectively.
|
