An Adaptive Memetic Algorithm for the Joint Allocation of Heterogeneous Stochastic Resources
针对复杂系统中异构随机资源的联合分配问题，建立了通用数学模型，并提出一种自适应模因算法，通过混合初始化、自适应替换和重启策略，在25个随机测试实例上优于现有方法。
This article investigates the joint allocation problem of stochastic resources (JASRs), which is widely found in complex systems. A general mathematical model for joint allocation of multiple heterogeneous stochastic resources is built, capturing the interdependencies between resources, quantity constraints, capability constraints, and strategy constraints of resources. An adaptive memetic algorithm (MA) is proposed for JASR, and the multipermutation encoding method is developed to denote assignment schemes of different resources. Multiple permutation-based operators are employed in the mutation and local search process under the genetic evolution framework and learning framework, respectively. Besides, a hybrid initialization method and an adaptive replacement strategy are put forward. Moreover, a restart strategy is developed to rebuild the population to maintain the genetic diversity. Twenty-five random test instances are produced to validate the effectiveness of the proposed MA. The results of computational experiments and the Wilcoxon rank-sum test demonstrate that these JASR instances can be well handled by the proposed adaptive MA, and the proposed MA is able to provide remarkably better decision schemes for the majority of the test instances than the prevailing solution methods.