This paper demonstrates an application of the improved crow search algorithm (I-CSA), which is a modified version of the crow search algorithm (CSA). CSA is a recent, nature-inspired meta-heuristic optimization algorithm. I-CSA differs from CSA by allowing application on a discrete problem, which is P-median and fortifying faster convergence to an optimal or near-optimal solution. Improvements are provided by local searches which support escaping from local optima or convergence to the optimal solution, elitism enhances the intensification through the utilization of nodes by selecting the most frequent centers that appeared in hiding better locations for local search, on the P-median problem. The application of the I-CSA is structured in three phases. In the first phase, parameters of I-CSA are analyzed and optimized using well-known data tests. The test datasets for the application of I-CSA on the P-median problem were retrieved from the OR-library to present the effectiveness and applicability of I-CSA. In the second phase, 40-pmed test problems from the library are solved using I-CSA and the results are compared with known optimal results and recorded results of other meta-heuristic approaches. In addition, Wilcoxon signed-rank test is applied in order to demonstrate the performance of I-CSA compared to well-known algorithms. The results of the proposed method demonstrated a faster convergence rate and better solution in most cases when compared with the standard CSA and other well-known meta-heuristic approaches. Finally, the proposed I-CSA approach is tested on a real-life case problem including 2121 nodes in Tunceli, Turkey. Obtaining the optimal results in a reasonable time indicates that the potential of the I-CSA is high and promising. In nutshell, this paper presents an improvement to CSA and evaluates its performance in a three-phase test procedure.

本文演示了改进的乌鸦搜索算法（I-CSA）的应用，它是乌鸦搜索算法（CSA）的改进版本。CSA是一种最新的，受自然启发的元启发式优化算法。I-CSA与CSA的不同之处在于，它允许在离散问题上进行应用，该问题为P中值，并加强了收敛到最佳或接近最佳解决方案的速度。局部搜索提供了改进，支持从局部最优解或收敛性逃脱到最佳解决方案，精英通过选择隐藏在更好的位置以进行局部搜索的最频繁出现的中心，在节点P上通过利用节点来增强集约化-中位数问题。I-CSA的应用分为三个阶段。在第一阶段，使用众所周知的数据测试来分析和优化I-CSA的参数。I-CSA在P上的应用的测试数据集从OR图书馆检索中位数问题，以展示I-CSA的有效性和适用性。在第二阶段，使用I-CSA解决了库中40个测试问题，并将结果与​​已知的最佳结果进行了比较，并记录了其他元启发式方法的结果。此外，为了证明I-CSA与知名算法相比的性能，采用了Wilcoxon符号秩检验。与标准CSA和其他众所周知的元启发式方法相比，该方法的结果在大多数情况下证明了更快的收敛速度和更好的解决方案。最后，针对土耳其Tunceli的2121个节点的实际案例问题，对提出的I-CSA方法进行了测试。在合理的时间内获得最佳结果表明，I-CSA的潜力很大并且很有前途。简而言之，本文提出了对CSA的改进，并在一个三相测试程序中评估了它的性能。