Una estrategia de descomposición de alta precisión para problemas con restricciones a gran escala

Decomposition strategies, such as cooperative coevolutionary (CC) algorithms, have proven effective in solving various large-scale constrained optimization problems. This approach decomposes a largescale constrained problem into simpler, smaller-scale subproblems without significantly affecting the performance of the optimization algorithm. The effectiveness of the approach depends greatly on the quality of this decomposition. An effective decomposition takes into account the interactions between the variables. However, recent strategies such as RDG and ERDG identify only additive interactions in the problem decomposition. As a result, these strategies tend to work well only for problems with this property. Nevertheless, in real-world problems, other types of separability besides additive separability may arise. This paper proposes a new decomposition strategy called Constrained Dual Differential Grouping (CDDG). The novelty of CDDG is that it can be used for both additive and multiplicative separable functions, which greatly increases the range of CC. To validate the effectiveness of CDDG, experiments and comparisons with state-of-the-art methods are conducted. The results demonstrate that CDDG can better adapt to the specific characteristics of each benchmark test problem.