Design of a novel multimodal optimization algorithm and its application in logistics optimization

Abstract

<abstract> <p>This study was purposed to design a multimodal continuous optimization algorithm based on a scheme agent to address the multidimensional complexity of optimization. An evolutionary sampling method of subarea exploration and multiple exploitations was developed by employing the scheme with variable population size so as to obtain higher optimization speed and accuracy. Second, the distribution plan was quantified into high-dimensional variable parameters based on the characteristics of logistics distribution optimization problems, and a high-dimensional discrete optimization model was constructed. Then, we identified and addressed the prominent issues and malignant virtual changes in the application of continuous algorithms to discrete problems. We have introduced a reasonable mutation mechanism during the optimization sampling process to mitigate this issue. Continuous real coordinate points were transformed across the neighborhood to standard discrete integer coordinate points by normalizing and logicizing the optimization sampling coordinates; also, the discretization of the continuous algorithm was realized. This approach could effectively prevent the algorithm from searching for targets in continuous optimization space, thereby fully reducing the complexity of the objective function distribution after conversion. The experiments showed that the transformed multimodal discrete optimization algorithm effectively addressed the optimization design problem of logistics distribution.</p> </abstract>