基于SAPSO算法优化Elman神经网络的话务量预测

文章提出一种模拟退火(SA)与粒子群优化(PSO)算法相结合的算法来优化Elman神经网络权值和阈值。当PSO处于停滞状态时,利用粒子群优化算法的全局寻优性质,以及SA能跳出局部最优解的特性,在搜索到的最优位置处用模拟退火算法继续寻找最优解,并对具有动态递归性能的Elman神经网络进行学习训练,这样就能对忙时话务量进行预测。结果表明,与传统Elman神经网络和PSO-Elman神经网络相比,基于模拟退火粒子群算法训练的神经网络具有更高的预测精度和良好的自适应性。     

基于模拟退火菌群-RBF神经网络的甲醇净化CO_2含量软测量模型

为了解决菌群优化(BFO)算法易陷入局部最优,趋化操作中翻转方向不确定的问题,利用模拟退火(SA)算法在获得局部最优解的情况下能够以极大可能趋向于全局最优解的特点,提出模拟退火-菌群优化(SA-BFO)算法。同时,将改进后的算法用于优化RBF神经网络,建立基于甲醇净化CO2含量的软测量模型。仿真结果表明该模型具有更高的精度和准确性,对甲醇生产量的提高具有一定的贡献价值。     

基于函数复杂度的自适应模拟退火和禁忌搜索新算法

在求解多峰复杂函数的过程中,传统的模拟退火算法和禁忌搜索算法经常出现算法快速收敛于局部最优解、后期收敛速度变慢和搜索能力变差等问题.为解决这些问题,本文给出函数复杂度的定义,并提出基于函数复杂度的自适应模拟退火和禁忌搜索算法.该算法首先根据函数复杂度自适应调整步长控制参数,然后根据调整后步长求得函数的粗糙解,在此基础上再使用初始步长求得全局最优解.实验表明,该算法不仅可以跳出局部最优解的限制,并且减少了迭代次数,有效地提高了全局和局部搜索能力.     

针对摄像机矩阵估计的增强连续禁忌搜索方法

摄像机矩阵估计是机器视觉的一个重要问题。在2范数误差代价函数模型下,最小二乘法简单而有效,但因误差代价函数非凸,容易陷入局部最优。在无穷范数误差代价函数模型下,凸优化方法理论上可以获得全局最优,但计算效率较低,其计算耗时随着问题规模的增大而急剧增加。现代优化论中的增强连续禁忌搜索（Enhanced continu-ous taboo search,ECTS）方法具有逃离局部最优的优良性质,因此本文在2范数误差代价函数模型下提出一种针对摄像机矩阵估计的ECTS算法。在ECTS置信区间序列构造及最大置信区间选择环节,本文提出了一种非迭代的方法获取包含全局最优解的凸包。在增强禁忌搜索环节,本文提出了一种基于伪凸函数的候选解邻域构造方法。同时,给出了本文算法以概率1收敛于全局最优的理论证明。对虚拟场景和真实场景的实验结果表明本文算法可以快速获取摄像机矩阵估计的全局最优解。     

钛合金薄板激光焊接参数的多目标优化

激光焊接过程产生的焊斑熔深和热影响区宽度直接影响焊接质量。激光焊接过程复杂,影响因素众多,许多参数难以量化。本文以TC4钛合金薄板为实验样品进行脉冲激光焊接实验。利用两个径向基函数神经网络分别预测焊斑熔深和热影响区宽度。将上述两个径向基函数神经网络作为多目标优化算法的目标函数,以提高焊接熔深并减小热影响区宽度。通过模拟退火算法寻求多目标优化所得的非劣解集中的最优解。实验证明,该方法可有效平衡激光焊接过程的焊斑熔深和热影响区宽度。     

自适应遗传退火算法在薄膜椭偏测量中的应用

将自适应遗传模拟退火混合算法应用于薄膜椭偏测量的反演问题中.由于模拟退火算法的基本思想是跳出局部最优解而得到全局最优解,因此将模拟退火思想引入到遗传算法,遗传算法和模拟退火算法相结合,组建自适应遗传模拟退火算法,从而综合了全局优化和局部搜索的特点,并通过模拟计算,验证了此方法在薄膜椭偏测量问题中的可行性及有效性,为解决...     

基于MODPSO-SA算法的多AUV任务分配

提出一种基于Pareto解集的多目标模拟退火粒子群算法(MODPSO-SA),用于解决自主水下机器人(AUV)协同任务分配问题.为避免粒子群算法陷入局部最优,加入改进的模拟退火技术,形成一种新的多目标局部搜索策略.仿真结果表明,MODPSO-SA算法能够得出多组合理Pareto解集,可以有效解决多AUV任务分配问题.     

未知环境下一种移动机器人实时最优路径规划方法研究

 针对未知环境下移动机器人的安全路径规划问题,提出一种基于改进神经网络和模拟退火算法相结合的方法.神经网络表示机器人的工作空间,通过BP反向算法学习外部环境结构特征和信息表示,进而优化障碍物神经网络的连接权值,利用模拟退火算法搜寻代价函数的负梯度方向,采用组合探测器来减小模拟退火算法搜索区域和应用后退策略及设置虚拟目标点的方法处理局部路径规划中出现的陷阱问题.仿真验证此方法有效性和正确性.     

求解约束优化问题的混合粒子群算法

针对约束优化问题提出一种混合粒子群求解算法,该算法根据可行性规则,引入自适应惩罚函数,结合模拟退火算法,不断地寻找更优可行解,逐渐达到搜索全局最优解.通过对一些标准函数测试,计算机仿真结果表明,该方法是有效和可行的,且具有较高的计算精度,相比传统算法,最优解精度达到10-15.     

改进的多目标粒子群算法优化设计及应用

针对粒子群算法存在易陷入局部最优点的缺点,提出了一种改进的带变异算子的多目标粒子群优化算法。采用非支配排序策略和动态加权法选择最优粒子,引导种群飞行,提高帕累托(Pareto)最优解的多样性。与其他优化算法相比,该算法易于实现并且计算速度更快。通过计算Pareto前沿最优解设计最佳多层电磁吸收体,在吸收体的厚度与反射系数之间取得最佳折衷。通过对反射系数函数与吸收体厚度函数测试验证,该算法能够在保持优化解多样性的同时具有较好的收敛性。     

基于神经网络的S参数估计法

该文讨论了神经网络用于微波电路的一种新的训练算法,是基于误差反传算法与模拟退火算法相结合的。本算法可以解决BP网的局部最小问题。并且不用对神经网络模型的结构作任何改动。通过一个具体算例的计算验证了这种算法。     

Virtual network function deployment strategy based on improved genetic simulated annealing algorithm in MEC

In order to effectively improve the end-to-end service delay of the flow in multi-clusters coexisting mobile edge computing (MEC) network,a virtual network function deployment strategy based on improved genetic simulated annealing algorithm was proposed.The delay of mobile service flow was mathematically modeled through the open Jackson queuing network.After proving the NP attribute of this problem,a solution combining genetic algorithm and simulated annealing algorithm was proposed.In this strategy,the advance mapping mechanism avoids the possibility of network congestion,and the occurrence of local optima was avoided through using the methods of individual judgment and corrective genetic.Extensive simulation was set up to evaluate the effectiveness of the proposed strategy under different parameter settings,such as different volume of requests,different scale of service nodes,different number of MEC clusters,and logical link relationships between virtual network functions.Results show that this strategy can provide lower end-to-end services delay and better service experience for latency-sensitive mobile application.     

一种结合模拟退火和贪心策略的社团识别算法

为了提高复杂网络社团识别的精度和速度,文中结合模拟退火和贪心策略识别社团结构的优势,提出一种新的社团识别算法。该算法利用贪心策略引导模拟退火搜索最优解过程中单个结点的无规则盲目移动,消除了大量无效移动,在搜索到全局最优解的情况下,将搜索时间大幅缩减。实验表明,SAGA具有强大的搜索能力和较快的模拟退火执行速度,可获得较高的模块度,达到较为准确的社团分割,且具有一定的应用价值。     

Multiobjective genetic optimization of diagnostic classifiers withimplications for generating receiver operating characteristiccurves

It is well understood that binary classifiers have two implicit objective functions (sensitivity and specificity) describing their performance. Traditional methods of classifier training attempt to combine these two objective functions (or two analogous class performance measures) into one so that conventional scalar optimization techniques can be utilized. This involves incorporating a priori information into the aggregation method so that the resulting performance of the classifier is satisfactory for the task at hand. We have investigated the use of a niched Pareto multiobjective genetic algorithm (GA) for classifier optimization. With niched Pareto GA's, an objective vector is optimized instead of a scalar function, eliminating the need to aggregate classification objective functions. The niched Pareto GA returns a set of optimal solutions that are equivalent in the absence of any information regarding the preferences of the objectives. The a priori knowledge that was used for aggregating the objective functions in conventional classifier training can instead be applied post-optimization to select from one of the series of solutions returned from the multiobjective genetic optimization. We have applied this technique to train a linear classifier and an artificial neural network (ANN), using simulated datasets. The performances of the solutions returned from the multiobjective genetic optimization represent a series of optimal (sensitivity, specificity) pairs, which can be thought of as operating points on a receiver operating characteristic (ROC) curve. All possible ROC curves for a given dataset and classifier are less than or equal to the ROC curve generated by the niched Pareto genetic optimization.     

Topology design and bridge-capacity assignment for interconnecting token ring LANs: A simulated annealing approach

We address the problem of determining the topology and bridge-capacity assignments for a network connecting a number of token rings via source-routing bridges. The objective is to minimize the cost of bridge installations while meeting the network users' performance requirements. The problem is modeled as a mixed 0–1 integer program. A comparison is given between two solution algorithms: a simulated annealing algorithm using the flow-deviation algorithm for each routing subproblem, and a drop algorithm using the simplex method for the same subproblems to provide benchmark solutions. In the former algorithm, the routing subproblem is formulated as a nonlinear program with penalty functions to model node and link capacity constraints, and in the latter as a multicommodity flow model with the same capacity constraints. Computational results show that the simulated-annealing/flow-deviation algorithm produced substantially better solutions than the LP-based drop algorithm.     

Behavioral simulation of densely-connected analog cellular array processors for high-performance computing

The analog cellular neural network (CNN) model is a powerful parallel processing paradigm in solving many scientific and engineering problems. The network consists of densely-connected analog computing cells. Various applications can be accomplished by changing the local interconnection strengths, which are also called coefficient templates. The behavioral simulator could help designers not only gain insight on the system operations, but also optimize the hardware-software co-design characteristics. An unique feature of this simulator is the hardware annealing capability which provides an efficient method of finding globally optimal solutions. This paper first gives an overview of the cellular network paradigm, and then discusses the nonlinear integration techniques and related partition issues, previous work on the simulator and our own simulation environment. Selective simulation results are also presented at the end.     

A novel heuristic simulation‐optimization method for critical infrastructure in smart transportation systems

A simulation‐based optimization is a decision‐making tool that helps in identifying an optimal solution or a design for a system. An optimal solution and design are more meaningful if they enhance a smart system with sensing, computing, and monitoring capabilities with improved efficiency. In situations where testing the physical prototype is difficult, a computer‐based simulation and its optimization processes are helpful in providing low‐cost, speedy and lesser time‐ and resource‐consuming solutions. In this work, a comparative analysis of the proposed heuristic simulation‐optimization method for improving quality‐of‐service (QoS) is performed with generalized integrated optimization (a simulation approach based on genetic algorithms with evolutionary simulated annealing strategies having simplex search). In the proposed approach, feature‐based local (group) and global (network) formation processes are integrated with Internet of Things (IoT) based solutions for finding the optimum performance. Further, the simulated annealing method is applied for finding local and global optimum values supporting minimum traffic conditions. A small‐scale network of 50 to 100 nodes shows that genetic simulation optimization with multicriteria and multidimensional features performs better as compared to other simulation‐optimization approaches. Further, a minimum of 3.4% and a maximum of 16.2% improvement is observed in faster route identification for small‐scale IoT networks with simulation‐optimization constraints integrated model as compared to the traditional method. The proposed approach improves the critical infrastructure monitoring performance as compared to the generalized simulation‐optimization process in complex transportation scenarios with heavy traffic conditions. The communicational and computational‐cost complexities are least for the proposed approach.