Performance analysis of a fuzzy decision support system for culling of dairy cows
R. Lacroix, Mark Strasser, Robert Kok, Kevin M. Wade · 1998
To investigate the use of fuzzy logic in decision-support systems for dairy cattle breeding, a prototype software system was developed. The objectives were to determine advantages and limitations of fuzzy logic for this type of application and to establish a methodological basis for the development ofmore complete decision-support systems in the future. The goal ofthe prototype decision-support system was to make culling decisions on the basis of monthly production data. During the development phase, three experts in the area ofanimal breeding were interviewed. The final version comprised three rule sets which considered a total of five input variables. The membership functions for most of the input variables were made herd-specific. Results showed that the use of fuzzy sets could increase the flexibility and adaptivity of rule-based expert systems. The same rule sets were appropriate under various scenarios (e.g., herds, regions, and breeds) with inferences being made specific to each context by adjusting the membership functions associated with the fuzzy sets. Results also showed that the inferences from fuzzy sets could be used as an alternative to methods currently used for within-herd cow rankings. The development of expert systems based on fuzzy logic seems relatively easy and such expert systems may require a smaller number of rules than traditional approaches to achieve similar output variations.