对大坝变形情况进行预测,明确大坝的实际状况是保证其长期安全稳定运行的关键之一,目前研究中普遍存在预测精度不足以满足实际需求的问题。为此,将长短时记忆网络(Long and Short-term Memory Network,LSTM)模型引入大坝变形预测的研究...对大坝变形情况进行预测,明确大坝的实际状况是保证其长期安全稳定运行的关键之一,目前研究中普遍存在预测精度不足以满足实际需求的问题。为此,将长短时记忆网络(Long and Short-term Memory Network,LSTM)模型引入大坝变形预测的研究,并利用自适应人工鱼群算法(Adaptive Artificial Fish School Algorithm,AAFSA)对模型的参数进行优化,以实际工程的数据对模型进行了实例验证,并将该模型与LSTM模型的性能进行对比,结果表明,优化后模型的平均绝对误差、平均相对误差、平均绝对百分比误差、均方根误差以及拟合度分别为0.225 9、0.031 6、0.289 2、0.054 7以及94.51%,即优化后的模型预测精度最高且误差最小,稳定性最好,从而为大坝的安全分析提供了新的借鉴。展开更多
由于工程中的复杂系统常常具有非线性的特点,因此寻找满足系统要求的最低成本成了复杂系统设计的难点。针对这一问题,文章对常规的人工鱼群算法(artificial fish school algorithm,AFSA)进行了双空间自适应嵌套式的改进,探讨了改进后的A...由于工程中的复杂系统常常具有非线性的特点,因此寻找满足系统要求的最低成本成了复杂系统设计的难点。针对这一问题,文章对常规的人工鱼群算法(artificial fish school algorithm,AFSA)进行了双空间自适应嵌套式的改进,探讨了改进后的AFSA算法在复杂系统寻优中的可行性,并对3个测试的复杂系统进行了分析计算;结果表明,与原算法相比,改进后的算法在提升寻优精确度与收敛速度方面有很好的效果。展开更多
Scalable video coding(SVC) is a powerful tool to solve the network heterogeneity and terminal diversity in video applications. However, in related works about the optimization of SVC-based video streaming over Softwar...Scalable video coding(SVC) is a powerful tool to solve the network heterogeneity and terminal diversity in video applications. However, in related works about the optimization of SVC-based video streaming over Software Defined Network(SDN), most of the them are focused either on the number of transmission layers or on the optimization of transmission path for specific layer. In this paper, we propose a noval optimization algorithm for SVC to dynamically adjust the number of layers and optimize the transmission paths simultaneously. We establish the problem model based on the 0/1 knapsack model, and then solve it with Artificial Fish Swarm Algorithm. Additionally, the simulations are carried out on the Mininet platform, which show that our approach can dynamically adjust the number of layers and select the optimal paths at the same time. As a result, it can achieve an effective allocation of network resources which mitigates the congestion and reduces the loss of non-SVC stream.展开更多
针对结构面产状常规分类方法存在的不足,提出一种新型的结构面分类算法.基于K-Means算法的结构面分类,将人工鱼群算法(artificial fish swarm algorithm,AFSA)与K-Means算法相结合,建立了AFSA-RSK结构面分类算法.利用鱼群算法强大的寻...针对结构面产状常规分类方法存在的不足,提出一种新型的结构面分类算法.基于K-Means算法的结构面分类,将人工鱼群算法(artificial fish swarm algorithm,AFSA)与K-Means算法相结合,建立了AFSA-RSK结构面分类算法.利用鱼群算法强大的寻优能力,代替K-Means算法对结构面产状聚心集进行搜寻,并通过K-Means算法进行聚类.聚类完成后,选择相应参数指标对聚类效果进行评价.针对存在的问题,对鱼群算法的步长和视野进行修正,提高寻找聚心集的精度,动态地调整了聚类过程.将改进后的AFSA-RSK算法与其他算法进行比较,结果表明在迭代速度、聚类精度以及内存占比上,改进后的AFSA-RSK算法都要更优,更适合在结构面分组方面的应用.展开更多
文摘对大坝变形情况进行预测,明确大坝的实际状况是保证其长期安全稳定运行的关键之一,目前研究中普遍存在预测精度不足以满足实际需求的问题。为此,将长短时记忆网络(Long and Short-term Memory Network,LSTM)模型引入大坝变形预测的研究,并利用自适应人工鱼群算法(Adaptive Artificial Fish School Algorithm,AAFSA)对模型的参数进行优化,以实际工程的数据对模型进行了实例验证,并将该模型与LSTM模型的性能进行对比,结果表明,优化后模型的平均绝对误差、平均相对误差、平均绝对百分比误差、均方根误差以及拟合度分别为0.225 9、0.031 6、0.289 2、0.054 7以及94.51%,即优化后的模型预测精度最高且误差最小,稳定性最好,从而为大坝的安全分析提供了新的借鉴。
文摘由于工程中的复杂系统常常具有非线性的特点,因此寻找满足系统要求的最低成本成了复杂系统设计的难点。针对这一问题,文章对常规的人工鱼群算法(artificial fish school algorithm,AFSA)进行了双空间自适应嵌套式的改进,探讨了改进后的AFSA算法在复杂系统寻优中的可行性,并对3个测试的复杂系统进行了分析计算;结果表明,与原算法相比,改进后的算法在提升寻优精确度与收敛速度方面有很好的效果。
文摘Scalable video coding(SVC) is a powerful tool to solve the network heterogeneity and terminal diversity in video applications. However, in related works about the optimization of SVC-based video streaming over Software Defined Network(SDN), most of the them are focused either on the number of transmission layers or on the optimization of transmission path for specific layer. In this paper, we propose a noval optimization algorithm for SVC to dynamically adjust the number of layers and optimize the transmission paths simultaneously. We establish the problem model based on the 0/1 knapsack model, and then solve it with Artificial Fish Swarm Algorithm. Additionally, the simulations are carried out on the Mininet platform, which show that our approach can dynamically adjust the number of layers and select the optimal paths at the same time. As a result, it can achieve an effective allocation of network resources which mitigates the congestion and reduces the loss of non-SVC stream.
文摘针对结构面产状常规分类方法存在的不足,提出一种新型的结构面分类算法.基于K-Means算法的结构面分类,将人工鱼群算法(artificial fish swarm algorithm,AFSA)与K-Means算法相结合,建立了AFSA-RSK结构面分类算法.利用鱼群算法强大的寻优能力,代替K-Means算法对结构面产状聚心集进行搜寻,并通过K-Means算法进行聚类.聚类完成后,选择相应参数指标对聚类效果进行评价.针对存在的问题,对鱼群算法的步长和视野进行修正,提高寻找聚心集的精度,动态地调整了聚类过程.将改进后的AFSA-RSK算法与其他算法进行比较,结果表明在迭代速度、聚类精度以及内存占比上,改进后的AFSA-RSK算法都要更优,更适合在结构面分组方面的应用.
