引入梯度下降的蚁群算法求解多约束服务质量路由

发布时间：2018-02-11 13:16

本文关键词： 服务质量路由蚁群算法梯度下降法信息素浓度收敛速度收敛结果算法稳定性　出处：《计算机应用》2017年03期 　论文类型：期刊论文

【摘要】：针对目前多数改进蚁群算法求解多约束服务质量路由(Qo SR)存在收敛速度慢、易陷入局部最优从而效率不高的问题,提出一种引入梯度下降的蚁群算法(ACAGD)。该算法将梯度下降法引入到蚁群的局部搜索中,结合残余信息素,综合决定蚂蚁的下一跳选择策略。蚁群不仅以一定概率按照信息素浓度搜索下一跳,还将以一定概率按照梯度下降法搜索下一跳,从而降低传统蚁群算法容易陷入局部最优的可能性。利用Waxman网络模型随机生成不同路由节点数量的网络拓扑进行仿真实验。实验结果表明,ACAGD相比其他改进蚁群算法,能够在收敛速度不受影响的情况下,取得综合代价相对较低的路由,且算法的稳定性较好。
[Abstract]:At present, most improved ant colony algorithms (ACA) have the problem of slow convergence, easy to fall into local optimum and low efficiency in solving multiple constrained QoS routing (QO SRs). In this paper, an ant colony algorithm with gradient descent is proposed. The gradient descent method is introduced into the local search of ant colony, and the residual pheromone is combined with the gradient descent algorithm. The ant colony not only searches for the next jump according to the pheromone concentration with a certain probability, but also searches the next jump with a certain probability according to the gradient descent method. In order to reduce the possibility that the traditional ant colony algorithm is prone to fall into local optimum, the network topology with different number of routing nodes is generated randomly by using Waxman network model. The experimental results show that the ant colony algorithm is better than other ant colony algorithms. Under the condition that the convergence rate is not affected, the route with relatively low synthesis cost can be obtained, and the stability of the algorithm is good.
【作者单位】：中山职业技术学院信息工程学院;广东科学技术职业学院计算机工程技术学院;东莞理工大学计算机学院;惠州学院教育技术中心;
【基金】：国家自然科学基金资助项目(61170193) 广东省工业高新技术领域科技计划项目(2013B010401036) 广东省自然科学基金资助项目(s2013010013432) 中山市社会公益科技研究项目(2016B2142)~~
【分类号】：TP18

【相似文献】