一种离散型多目标粒子群优化算法

为获得更好的非劣前端,提出一种离散型多目标粒子群优化算法。该算法根据离散型多目标优化问题的特点,将种群分成多个子种群,在各个子种群中利用表现型共享的适应度函数选择每个子种群的最优粒子。通过多个最优粒子的引导,使整个种群分布更均匀,避免陷入局部最优,保证了解的多样性。实验表明了该算法的有效性。     

Multiple objective particle swarm optimization technique for economic load dispatch

INTRODUCTION The conventional economic load dispatch prob-lem of power generation involves allocation of power generation to different thermal units to minimize the operating cost subject to diverse equality and ine-quality constraints of the power system. This makes the economic load dispatch problem a large-scale highly non-linear constrained optimization problem. However, as a result of public awareness of envi-ronmental protection, diverse emission compliance strategies have emerged (…     

基于粒子群算法的太阳能电动汽车复合能源系统优化

太阳能电动汽车的复合能源系统优化匹配问题可以看成一个多目标优化问题,两个相互冲突的目标是极大化系统的峰值功率满足率和极小化系统的成本,前者关系到系统的可靠性后者涉及到样车能否量产,所以两个优化目标都很重要.本文提出了改进的粒子群算法优化配置太阳能电动汽车复合能源系统,这种改进的粒子群算法引进了遗传算法里的变异算子,并且打破常规算法里的加速因子为常数的惯例而使加速因子随时间改变.优化结果显示：改进的粒子群算法也能够很好地解决复合能源系统的多目标优化问题.     

随机加权的多目标混沌优化法

针对多目标优化问题,提出了一种变加权的多目标混沌优化方法,通过对多目标的随机加权处理,实现了算法在各个方向的搜索,能够找到不同方向的Pareto最优解。与混沌优化方法的结合使该方法不仅能够找到分布比较均匀的Pareto边界上的最优解,而且使用简单、方便。     

基于PSO的电力系统环境经济负荷调度

针对基于线性加权和处理成单目标优化问题的传统方法存在的缺陷,提出使用粒子群优化算法求解EELD多目标优化问题。该方法通过对粒子群算法个体极值和全局极值选取方式的改进,实现了对EELD多目标优化问题的非劣最优解集的搜索,为决策者提供了丰富的参考信息。在此基础上,应用模糊满意度方法求出的最优折衷解为调度运行人员提供了最佳调度折衷方案。最后,对一个三机系统进行了测试,并与线性加权人工神经网络法进行了比较分析,仿真结果验证了该方法的有效性。     

Multi-objective Optimization of Non-uniform Beam for Minimum Weight and Sound Radiation

A multi-objective optimization of non-uniform beams is presented for minimum radiated sound power and weight. The transfer matrix method is used to compute the structural and acoustic responses of a non-uniform beam accurately and efficiently. The multi-objective particle swarm optimization technique is applied to search the Pareto optimal solutions that represent various compromises between weight and sound radiation. Several constraints are imposed, which substantially reduce the volume fraction of feasible solutions in the design space. Two non-uniform beams with different boundary conditions are studied to demonstrate the multi-objective optimal designs of the structure.     

Non-Uniform Sediment Transport Downstream from Reservoir

The transport characters of non-uniform sediment before and after the reservoir operation are compared and analyzed. Field data from the Three Gorges Reservoir (TGR), Danjiangkou, and Sanmenxia reservoir indicate that the transport character of coarse particle is various in different reaches. Before the reservoir operation, the dramatic decrease of the hydraulic condition in some reaches is the main reason to cause the coarse particle deposited. After the reservoir operation, almost all grain sized sediment may be eroded, when the reach is eroded seriously. However, the transport character of the coarse particle will be consistent with what it was before, when the reach is not eroded seriously. The deposition amount will be less than its pre-dam value, and it will tend to decrease with the time going on. Erosion of fine particle results from its unsaturation. Long-distance erosion downstream from reservoir is caused by shortage of the supply from riverbed, especially the shortage of fine particle. Supported by the National Basic Research Program of China(“973” Program, No.2003CB415200). LI Yitian, born in 1957, male, Prof.     

一种有效的多目标混合遗传算法

设计了一种基于支配关系下的局部搜索方法，将此局部搜索方法嵌入到多目标遗传算法中，从而提出一种有效的求解多目标优化问题的混合遗传算法。为加速遗传算法在全局优化问题上的收敛性，发挥传统数值优化算法在计算速度与计算精度上的优势，在遗传算法中镶嵌一个多目标线搜索算子。线搜索算子与遗传算法中的选择算子、交叉算子和变异算子共同作用，使全局搜索和局部搜索都能够很好的实现。数值实验表明，该混合遗传算法能求得问题的数量更多、分布更广的Pareto最优解。     

面向全局优化的混沌粒子群遗传算法

结合遗传算法(GA)和粒子群算法(PSO)的优点以及混沌运动的特性,提出了混沌粒子群遗传算法(CPSO-GA),并使用五个高维非线性测试函数考察此算法的性能。在固定进化代数、所调用目标函数次数接近以及固定收敛精度三种情况下对算法进行数值试验,结果表明,与其他文献中提出的算法相比,CPSO-GA能100%地找到最优解,收敛效果及寻优能力好,并能有效摆脱局部极小点,且调用目标函数次数最少,大大降低了计算量。     

A pooled-neighbor swarm intelligence approach to optimal reactive power dispatch

INTRODUCTION The reactive power dispatch is aimed at mini- mizing the active power loss in the transmission network by allocating the reactive power generation under several security constraints. The reactive power dispatch problem has significant influence on secure and economic operation of power systems. The reac- tive power generation affects the overall generation cost via transmission loss. A procedure which allo- cates the reactive power generation so as to minimize the transmissio…     

基于QPSO的二级齿轮减速器的优化设计

基于量子粒子群算法理论,对二级齿轮减速器进行优化设计,并利用MATLAB软件对量子粒子群算法程序进行了设计验证。从实验结果来看,运用量子粒子群算法很好地实现了减小齿轮中心距的要求,收敛速度快,得到的结果令人较为满意。     

基于改进粒子群算法的PID参数优化研究

PID控制是典型的工业控制,其核心内容是PID参数优化。为解决参数优化时不能确保得到最佳性能且耗时问题,通过改进粒子群算法学习因子,研究基于相等随机因子粒子群算法的PID参数优化,将其与标准的粒子群算法及迭代次数线性变化的学习因子进行比较。仿真结果表明,该算法性能指标tr、ts、δ%分别为1.782、3.285、14.07%,两种对比算法的tr、ts、δ%分别为1.804、4.825、24.33%和1.802、4.135、16.56%,改进算法提高了PID参数的稳定性、收敛速度和搜索精度,性能指标更优。     

群智能算法及其应用

群智能算法是一种新兴的演化计算技术,它已经成功地被运用于许多领域。实践证明,群智能方法是一种能够解决许多全局优化问题的有效方法。本文介绍了群智能理论的产生和发展过程,并着力阐述群智能理论中的两个主要算法（粒子群算法和蚁群算法）的算法思想以及研究现状。     

Multi-objective aerodynamic optimization design of high-speed train head shape

To improve the aerodynamic performance of high-speed trains (HSTs) running in the open air, a multi-objective aerodynamic optimization design method for the head shape of a HST is proposed in this paper. A parametric model of the HST was established and seven design variables of the head shape were extracted. Sample points and their exact values of optimization objectives were obtained by an optimal Latin hypercube sampling (opt. LHS) plan and computational fluid dynamic (CFD) simulations, respectively. A Kriging surrogate model was constructed based on the sample points and their optimization objectives. Taking the total aerodynamic drag force and the aerodynamic lift force of the tail coach as the optimization objectives, the multi-objective aerodynamic optimization design was performed based on a non-dominated sorting genetic algorithm-II (NSGA-II) and the Kriging model. After optimization, a series of Pareto-optimal head shapes were obtained. An optimal head shape was selected from the Pareto-optimal head shapes, and the aerodynamic performance of the HST with the optimal head shape was compared with that of the original train in conditions with and without crosswinds. Compared with the original train, the total aerodynamic drag force is reduced by 2.61% and the lift force of the tail coach is reduced by 9.90% in conditions without crosswind. Moreover, the optimal train benefits from lower fluctuations in aerodynamic loads in crosswind conditions.     

贝叶斯网络结构粒子群优化学习算法

提出一种信息论结合粒子群优化的贝叶斯网络结构学习算法,将约束最大信息熵作为最高评分函数,对网络结构进行复杂度约束,设计了粒子位置和速度向量的操作方法,解决单纯利用KL距离进行搜索的缺陷。在网络结构的搜索空间相对较大的情况下,该优化算法能在较短的时间内收敛,获得更准确的网络结构。仿真实验结果表明,该算法在时间和精度上都具有较好的效果。     

一种跨邻域学习的改进粒子群优化算法

为了改善传统粒子群优化算法过早陷入局部最优解的缺点,进一步增强算法收敛性,通过使用一定范围内邻域最好位置lBest代替自身历史最好位置pBest进行速度与位置更新,以增强粒子跨邻域学习能力。使用整个群体中最好位置gBest进行速度与位置更新,可增强算法收敛性,且具有较好的全局搜索能力。在8个不同的单峰和多峰函数上系统地对3种算法进行测试与比较,实验结果表明,提出的跨邻域学习改进粒子群优化算法可避免粒子群陷入局部最优解,求解精度与算法收敛性都提升了15%以上。     

一种新型的求解约束优化问题的微粒群算法

本文提出了一种新的求解约束优化问题的微粒群算法。首先提出了三种构造初始微粒群的算法,然后给出了保证微粒在可行域内运动的混合微粒群算法。通过测试函数的对比分析,说明了该算法的有效性。     

基于改进PSO算法的电力系统无功优化

本文采用的无功优化数学模型虑了系统的实际约束条件和无功调节手段,能够准确直观地反映电力系统无功优化的实际问题。在此基础上对基于粒子群算法的无功优化作了深入的研究,尝试利用粒子群优化算法对无功优化问题进行求解。对常规粒子群算法的搜索策略进行了改进,对约束条件、离散变量都采取了一定的改进处理措施。以IEEE-6节点系统为例进行了仿真计算,验证了本文模型和算法的正确性、适用性和经济性。     

计及网损的电网连锁故障预防方法

针对电网连锁跳闸引起的停电事故以及降低电力系统的有功功率损耗,提出了一种计及网损的电网连锁故障预防方法。 该方法提出了网损指标和衡量电网安全稳定运行指标,综合构建了多目标优化模型。 该模型的优化目的在于确保均衡权重下安全裕度尽可能大,同时使网损尽可能小。 通过粒子群优化算法求解,在IEEE-39 节点系统上进行了算例分析。 仿真结果表明该优化模型不仅提高了电网安全运行的界限,也实现了降损节能。     

Diabetes diet nutrition optimization based on particle swarm optimization algorithm

Nutrition intervention is the key to prevent and control diabetes mellitus. Diabetic patients’ nutrition intervention requires reasonably control of the intakes of the three major nutrients. A diabetes diet nutrition optimization method based on particle swarm optimization (PSO) algorithm is proposed for individual nutritional intervention of diabetes. Firstly, a nutritional optimization mathematical model is established, which meets the demand of diabetic patients’ three major nutrients intake, and then an improved particle swarm optimization algorithm is applied to solve the model. To enhance the convergence speed and search ability of the standard particle swarm optimization, the adaptive inertia weight factor and the natural selection mechanism are introduced in PSO. The performance of the proposed method is verified through an experiment, and the attainment rates of three major nutrients in the designed recipes are close to 100? according to the experimental results, which indicates that it can greatly improve the precision of diabetes nutrition recipes