EURO-Online login
- New to EURO? Create an account
- I forgot my username and/or my password.
- Help with cookies
(important for IE8 users)
880. Comparative Study of Genetic and Differential Evolution Algorithms for Pareto Front Approximation in Dairy Feed Optimization
Invited abstract in session TA-12: OR in Livestock farming, stream OR in Agriculture and Forestry .
Tuesday, 8:30-10:00Room: 13 (building: 116)
Authors (first author is the speaker)
1. | Gastón Notte
|
UdelaR | |
2. | Pablo Chilibroste
|
Animal production and pastures, UdelaR | |
3. | Martin Pedemonte
|
UdelaR | |
4. | Hector Cancela
|
Facultad de Ingeniería, Universidad de la República |
Abstract
Evolutionary algorithms have emerged as potent solutions for addressing multi-objective optimization challenges across various practical scenarios. This study introduces a multi-objective dairy feed resource allocation optimization model and conducts a performance analysis of four methods to approximate the Pareto front; two genetic algorithms (GA: NSGA-II, SPEA-2) and two differential evolution algorithms (GDE-3, a variant of Pareto-based DE).
Experiments utilizing real-world data were performed to evaluate the algorithms based on criteria such as execution times, Pareto front solutions, and performance indicator values, at the same time looking for the best parametric combinations for each algorithm.
The results showed significant differences among the algorithms, both in their proficiency to approximate the Pareto front and in their execution times. SPEA-2 demonstrated superior results in terms of convergence, diversity, and cardinality, but suffered prolonged execution times. Future research could address this issue, combining SPEA-2 with other algorithms for efficient Pareto front approximations.
Regarding the parametric combinations, significant differences were observed. For both GA, the importance of using a low value in the mutation probability is highlighted. For the GDE-3, a low value in the population size (N) and a high value in the crossover probability (CR) stand out. Finally, for the Pareto-based DE, higher values for N and CR gave better results.
Keywords
- OR in Agriculture
- Combinatorial Optimization
- Algorithms
Status: accepted
Back to the list of papers