针对现有网络入侵检测方法的不足,提出了一种新的网络入侵检测方法——GATS-LSVM算法。该方法采用遗传算法(GA)与禁忌搜索(TS)相混合的搜索策略对特征子集空间进行随机搜索,利用提供的数据在无约束优化线性支持向量机(LSVM)上的分类错...针对现有网络入侵检测方法的不足,提出了一种新的网络入侵检测方法——GATS-LSVM算法。该方法采用遗传算法(GA)与禁忌搜索(TS)相混合的搜索策略对特征子集空间进行随机搜索,利用提供的数据在无约束优化线性支持向量机(LSVM)上的分类错误率作为特征子集的评估标准获取最优特征子集,从而有效地对入侵进行检测。大量基于著名的KDD Cup 1999数据集的实验表明,该新方法相对于其它一些传统的网络入侵检测方法,能在保证较高检测率的前提下,有效地降低误报率、入侵检测的计算复杂度和提高检测速度,能更适用于现实高速网络应用环境。展开更多
Some manufacturers replace traditional warehouses with shipping areas at the scattered plants for holding the finished products in order to reduce land usage and inventory cost.The limited storage capacity of such shi...Some manufacturers replace traditional warehouses with shipping areas at the scattered plants for holding the finished products in order to reduce land usage and inventory cost.The limited storage capacity of such shipping area leads to challenges of scheduling vehicles for pickup since the overflow of storage space is prohibited.A heuristic rule is developed for splitting the continuous arrival of inventory at a plant into a sequence of discrete tasks for pickup.In this way,the original problem can be converted into a multiple trip vehicle routing problem with time window(MTVRPTW).Subsequently,a modified tabu search(TS)algorithm is applied for deriving the schedule.Finally,an industry case of an electric apparatus manufacturer is studied to demonstrate and validate the developed optimization approach,and the results imply good performance of the developed tool.展开更多
局部放电(partialdischarge,PD)诊断与定位有助于在老化早期发现高压电力设备故障位置,对制定检修方案具有重要参考价值。目前常利用多个特高频(ultra high frequency,UHF)传感器组成传感器阵列,并定义三维坐标系对变压器进行局部放电...局部放电(partialdischarge,PD)诊断与定位有助于在老化早期发现高压电力设备故障位置,对制定检修方案具有重要参考价值。目前常利用多个特高频(ultra high frequency,UHF)传感器组成传感器阵列,并定义三维坐标系对变压器进行局部放电空间定位。该文研究基于能量积累法捕捉信号起始脉冲和基于到达时间差(timedifferenceofarrival,TDOA)算法实现定位的原理,构建以局部放电位置点坐标为未知数的非线性规划问题,并利用禁忌搜索-粒子群优化(tabu search particle swarm optimization,TS-PSO)算法进行最优解求解。该算法可以避免非线性方程组求解时不收敛、解不唯一以及最小二乘法对初值要求高等问题,既保证了求解的速度,又能保证解的唯一性与准确性。实验室测试和现场测试验证了定位结果的有效性。展开更多
文摘针对现有网络入侵检测方法的不足,提出了一种新的网络入侵检测方法——GATS-LSVM算法。该方法采用遗传算法(GA)与禁忌搜索(TS)相混合的搜索策略对特征子集空间进行随机搜索,利用提供的数据在无约束优化线性支持向量机(LSVM)上的分类错误率作为特征子集的评估标准获取最优特征子集,从而有效地对入侵进行检测。大量基于著名的KDD Cup 1999数据集的实验表明,该新方法相对于其它一些传统的网络入侵检测方法,能在保证较高检测率的前提下,有效地降低误报率、入侵检测的计算复杂度和提高检测速度,能更适用于现实高速网络应用环境。
基金National Natural Science Foundation of China(No.71271137)Natural Science Foundation of Shanghai,China(No.12ZR1415100)
文摘Some manufacturers replace traditional warehouses with shipping areas at the scattered plants for holding the finished products in order to reduce land usage and inventory cost.The limited storage capacity of such shipping area leads to challenges of scheduling vehicles for pickup since the overflow of storage space is prohibited.A heuristic rule is developed for splitting the continuous arrival of inventory at a plant into a sequence of discrete tasks for pickup.In this way,the original problem can be converted into a multiple trip vehicle routing problem with time window(MTVRPTW).Subsequently,a modified tabu search(TS)algorithm is applied for deriving the schedule.Finally,an industry case of an electric apparatus manufacturer is studied to demonstrate and validate the developed optimization approach,and the results imply good performance of the developed tool.
