一种新型智能仿生类算法-蚁群算法

蚁群算法是一种新型智能仿生类算法,是受到蚂蚁在觅食过程中建立蚁巢到食物最短路径时的搜索机制启发而提出的一种算法。蚁群算法在求解一系列困难的组合优化问题上取得成效,成为解决TSP、VRP、QAP、JSP等典型问题的一种新型强有力算法。本文对蚁群算法理论研究的主要内容和方法、基于算法的改进等,进行了系统的总结与综述。     

混合智能算法及其在供水水库群优化调度中的应用

将遗传算法中的进化思想和蚁群算法中的群体智能技术有效地耦合,提出了一种基于两者的混合智能算法,应用于供水水库群系统的优化调度研究中。算法利用蚁群算法的并行性、正反馈性以及良好的全局寻优能力,避免搜索陷入局部最优,同时借鉴遗传算法的进化思想,利用杂交、变异算子来进行局部寻优,使其能快速搜索到全局最优点。在种群随机搜索过程中嵌入确定性的模式搜索,使得算法同时具有随机性和确定性。结合模拟退火思想,构造了罚因子处理约束条件,使该算法对水库优化调度问题以及其他优化问题具有一定的通用性。通过实例验证,并与大系统聚合分解经典算法进行比较,结果表明该算法是可行的和有效的。     

基于蚁群算法优化极限学习机模型的滑坡位移预测

采用高精度的优化算法对于提高滑坡位移预测模型的准确性具有重要意义,然而已有文献中很少对多种优化算法进行对比研究。以三峡库区的八字门滑坡为例,以极限学习机（ELM）理论为基础进行滑坡位移预测,同时运用多种算法对建立模型过程中的参数选择进行优化以期提高预测效果。为提高预测精度,以移动平均法为基础,将滑坡位移分解为趋势项和周期项,趋势项位移使用多项式函数进行预测,周期项位移使用MATLAB自编程序的极限学习机模型进行预测,两项预测值相加即可得到最终的累计位移预测值。计算结果表明：单一的ELM模型能够较为准确地预测具有阶跃式曲线的滑坡累计位移,预测结果的平均误差为23.5 mm,拟合优度为0.973。与粒子群算法和遗传算法相比,蚁群算法（ACO）在计算用时和优化效果上更优,蚁群算法优化极限学习机模型对位移的预测精度也最高,平均误差为10.1 mm,拟合优度为0.998,可在类似滑坡的位移预测研究中进行推广。     

改进蚁群算法及其在水利工程项目管理中的应用

为了更加有效解决水利工程项目管理中的多目标决策问题,提出了一种改进蚁群算法。该算法首先利用遗传算法的全局搜索能力将信息素初始化,然后在算法进行遍历过程中引入变异操作和交叉操作,提高算法的鲁棒性和有效性。水利工程项目多目标优化案例分析表明,较传统遗传算法和蚁群算法,本文提出的方法对于解的寻找速度更快,解的质量更高,该算法具有较高的全局寻优能力。该研究为水利工程项目管理多目标决策问题的解决提供了一种新的思路和方法。     

改进的蚁群算法在溶质运移参数识别问题中的应用

非饱和土壤水分和溶质运移参数(扩散率、导水率和水动力弥散系数)取值范围较大,往往跨越几个数量级。采用传统离散化蚁群算法求解此类问题,所需节点较多,这会造成算法收敛时间较长。该文在传统蚁群算法基础上,对蚂蚁搜索路径进行改进,改进后的蚁群算法寻优路径由参数精度位数(整数位和小数位)、参数个数以及0–9十个数字构成,并将路径解码公式修改为具有判别参数正负功能的解码公式。采用改进的连续蚁群算法对非饱和溶质运移参数识别优化模型进行求解。数值模拟表明相同迭代次数下改进的蚁群算法比传统蚁群算法耗时少,算法计算时间与迭代次数满足线性关系,含水率和溶质浓度实测值与计算值吻合较好、相关性较高。     

An Improved Continuous Ant Algorithm for Optimization of Water Resources Problems

Ant colony optimization was initially proposed for discrete search spaces while in continuous domains, discretization of the search space has been widely practiced. Attempts for direct extension of ant algorithms to continuous decision spaces are rapidly growing. This paper briefly reviews the central idea and mathematical representation of a recently proposed algorithm for continuous domains followed by further improvements in order to make the algorithm adaptive and more efficient in locating near optimal solutions. Performance of the proposed improved algorithm has been tested on few well-known benchmark problems as well as a real-world water resource optimization problem. The comparison of the results obtained by the present method with those of other ant-based algorithms emphasizes the robustness of the proposed algorithm in searching the continuous space more efficiently as locating the closest, among other ant methods, to the global optimal solution.     

仿生算法优化BP神经网络在降雨空间插值中的应用

人工神经网络能够充分挖掘已知样本中的规律,从而对未观测数据进行预测,可应用于降雨量空间插值计算中。在BP神经网络进行降雨空间插值的基础上,引入遗传、粒子群和蚁群3种仿生算法对BP神经网络初始权值和阈值进行优化,将优化后的BP神经网络应用于三峡区间流域年、月和日3个时间尺度的降雨空间插值中。结果表明:仿生算法对BP神经网络初始权值和阈值优化求解后,降低了BP神经网络陷入局部最小以及过拟合的风险,在插值过程中表现出较好的稳定性,取得了理想的插值结果。     

蚁群算法求解梯级电站短期优化调度改进研究

本文提出一种改进蚁群算法(Improved ant Colony Optimization Algorithm)求解梯级水库群短期优化调度问题。该算法的改进主要包括嵌入邻域搜索的单库轮换寻优、基于出力反推的初始解生成技术和约束优先的目标函数比较方法。以四川某中型流域梯级三级电站联合运行为背景,对蚁群算法和改进蚁群算法的求解质量和收敛性进行比较,实例验证表明,改进蚁群算法可以获得较好的优化调度结果。     

蚁群算法在工程项目工期—费用优化问题中的应用

论述了工期—费用优化问题的原理,分析了传统优化方法的优缺点。针对工期-费用这一连续空间优化问题,综合了基于网格划分策略的连续域蚁群算法和求解旅行商问题的基本蚁群算法的思想,构造了一种改进的蚁群算法。实例计算结果表明,该方法在求解工期费用优化问题方面是有效的。     

用于求解函数优化的蚁群算法设计

为了求解一般的函数优化，在标准蚁群算法的基础上，引入了遗传算法的编码方式，对蚁群算法的信息素更新进行改进。通过对几个经典测试函数的求解，证明了该算法的有效性。     

边坡临界滑动面搜索的奖惩蚁群算法研究

边坡滑动面搜索是边坡稳定计算中一项关键的问题,其实质为安全系数最小滑动路径的搜索问题,本文对采用路径搜索的蚁群算法引入奖惩策略,加大较优路径和普通路径上信息素的差异,分化各条路径上的信息素,以克服其收敛速度慢、早熟收敛的缺点。通过把边坡滑动面搜索模型离散化,采用奖惩蚁群算法解决滑动面搜索问题,提出了一种临界滑动面搜索的新方法。最后对一个简单边坡和复杂边坡的典型算例及一个大坝边坡工程的应用实例进行了计算,验证了新算法的有效性及其高效性。结果表明,无论是对简单边坡还是复杂边坡,本文算法都能以更快的速度搜索到结果更好的临界滑动面,工程应用效果良好。     

Partially and Fully Constrained Ant Algorithms for the Optimal Solution of Large Scale Reservoir Operation Problems

This paper presents a constrained formulation of the ant colony optimization algorithm (ACOA) for the optimization of large scale reservoir operation problems. ACO algorithms enjoy a unique feature namely incremental solution building capability. In ACO algorithms, each ant is required to make a decision at some points of the search space called decision points. If the constraints of the problem are of explicit type, then ants may be forced to satisfy the constraints when making decisions. This could be done via the provision of a tabu list for each ant at each decision point of the problem. This is very useful when attempting large scale optimization problem as it would lead to a considerable reduction of the search space size. Two different formulations namely partially constrained and fully constrained version of the proposed method are outlined here using Max-Min Ant System for the solution of reservoir operation problems. Two cases of simple and hydropower reservoir operation problems are considered with the storage volumes taken as the decision variables of the problems. In the partially constrained version of the algorithm, knowing the value of the storage volume at an arbitrary decision point, the continuity equation is used to provide a tabu list for the feasible options at the next decision point. The tabu list is designed such that commonly used box constraints for the release and storage volumes are simultaneously satisfied. In the second and fully constrained algorithm, the box constraints of storage volumes at each period are modified prior to the main calculation such that ants will not have any chance of making infeasible decision in the search process. The proposed methods are used to optimally solve the problem of simple and hydropower operation of “Dez” reservoir in Iran and the results are presented and compared with the conventional unconstrained ACO algorithm. The results indicate the ability of the proposed methods to optimally solve large scale reservoir operation problems where the conventional heuristic methods fail to even find a feasible solution.     

基于改进智能优化算法的投影寻踪模型在洪水评估中的应用研究

传统的洪水评估方法存在着评估等级离散和结果不易分辨等不足,如何更准确高效地解决洪水评估问题已成为研究领域的热点之一。以南京站的历史洪水及四川省历史洪水灾情为例,在改进智能优化算法的基础上,引入了基于智能优化算法的投影寻踪模型,并探讨该模型在洪水分类和洪灾等级评价中的应用。结果表明,人工蜂群和混合蛙跳这类新型智能优化算法具有简单、鲁棒、全局寻优和易于实现等特点,与广泛应用于水文界的SCE-UA、文献中的加速遗传等现代启发式算法相比,具有寻优速度更快、能力更强的优势,可为洪水分类和洪灾等级评价等相近领域研究提供新途径。     

边坡临界滑动面搜索的奖惩蚁群算法研究

边坡滑动面搜索是边坡稳定计算中一项关键的问题，其实质为安全系数最小的滑动路径的搜索问题，采用路径搜索算法蚁群算法是目前研究的热点。为了克服传统蚁群算法效率低、效果差的缺点，通过引入奖惩策略，加快较优路径和普通路径上信息素的差异，分化各条路径上的信息素，从而提出了一种奖惩蚁群算法。通过把边坡滑动面搜索问题的离散模型化，采用奖惩蚁群算法解决滑动面搜索问题，提出了一种临界滑动面搜索的新方法。通过一个简单边坡和复杂边坡的典型算例计算及一个大坝边坡的工程的应用实例应用，验证了新滑动面搜索算法的有效性及其高效性。结果表明，无论是对简单边坡还是复杂边坡，本文算法都可以能以更快的速度搜索得到效果更好的临界滑动面，且工程应用效果良好。     

混沌蚁群算法在水库优化调度中的应用

提出一种基于混沌优化算法和蚁群算法相结合的混合算法,在求解水库优化调度问题的方法。根据混沌变量的随机性和遍历性,利用混沌变量进行优化搜索,从而有效地克服了蚁群算法存在的效率低、易于演化停滞及陷入局部最优等问题。又利用蚁群算法信息素正反馈的优点,改善了混沌搜索的盲目性,提高了搜索的效率。通过实例计算,结果表明该算法具有效率高及较强的全局寻优能力。     

正弦余弦算法-投影寻踪水污染物总量分配模型

基于水资源禀赋条件、效率原则和尊重现状的原则,构建水污染物总量分配指标体系和水污染物分配投影寻踪(PP)模型。针对PP模型最佳投影方向难以确定的不足,利用正弦余弦算法(SCA)搜寻PP模型最佳投影方向,构建SCA-PP模型对云南省文山州壮族苗族自治州8县(市)水污染物控制总量进行分配。并通过6个典型测试函数对SCA算法进行仿真验证,仿真结果与蚁群优化(ACO)算法、模拟退火算法(SA)、文化算法(CA)、布谷鸟搜索(CS)算法和人工蜂群(ABC)算法进行对比。结果表明:(1)SCA算法寻优效果明显优于ACO、SA、CA、CS和ABC算法,具有模型简单、调节参数少、收敛速度快、寻优精度高、全局寻优能力强以及收敛稳定性与收敛可靠性好等特点。(2)SCA-PP模型水污染物控制总量分配结果符合区域经济社会发展和水污物染削减客观要求。模型及方法具有一定的可操作性和有效性,可为水污染物分配提供新的途径和方法。     

基于遗传算法的偏最小二乘回归模型在大坝安全监测中的应用

偏最小二乘回归能较好地解决自变量之间严重的相关性问题,遗传算法作为一种新的全局优化搜索方法,具有智能性搜索、并行式计算、鲁棒性强等优点.本文在偏最小二乘回归分析的基础上引入遗传算法,依靠其有效的自适应全局搜索优化功能,对偏回归模型中的回归系数进行重新评估,建立基于遗传算法的偏回归模型.实例分析表明:基于遗传算法的偏回归模型有良好的拟合效果和预测精度.     

Fuzzy Systems Tuned By Swarm Based Optimization Algorithms for Predicting Stream flow

River flow prediction is an important phenomenon in water resources for which different methods and perspective have been used. Using fuzzy system with black box perspective is one of them. Fuzzy systems have some parameters and properties that have to be determined. This is an optimization problem that can be solved by swarm optimization techniques among several techniques. Swarm optimization are developed by inspiring from the behavior of the animals living as swarm. The study presents two achievements fuzzy system that tuned by swarm optimization algorithms can be used for prediction of monthly mean streamflow and which swarm optimization algorithm is better than the others for tuning fuzzy systems. Three swarm optimization algorithms, hunter search, firefly, artificial bee colony are used in this study. These algorithms are compared with mean performance values and convergence speed. Monthly streamflow data of three stream gauging stations in Susurluk Basin are used for the case study. The results show, swarm optimization algorithms can be used for prediction of monthly mean streamflow and ABC algorithm has better performance values than other optimization algorithms.     

改进的实数编码加速遗传算法在暴雨强度公式参数优化中的应用

针对城市暴雨强度公式参数识别时,传统求解方法(如牛顿迭代法、高斯-牛顿法等)存在间接拟合,而优化算法(如实数编码加速遗传算法、蚁群算法等)存在随机性,盲目地在一个区间内寻优拟合精度不高等问题,本文将两种方法结合,提出一种改进的实数编码加速遗传算法(RAGA),为暴雨强度公式参数识别提供一种新途径。该方法将传统求解方法所求的可行解作为改进遗传算法的初始参数,通过在每次代际寻优时设置各参数廊道约束来改进RAGA以提高算法搜索效率,直至公式拟合精度无法提高为止。将该方法应用于国内多地暴雨强度公式参数识别中以评估算法的有效性,结果表明此方法实用可行、搜索效率较高,可以快速收敛到最优解。实例表明该方法在暴雨强度公式参数识别中是实用有效的。     

Optimization of Hydropower Reservoir Using Evolutionary Algorithms Coupled with Chaos

Over the past decade, several conventional optimization techniques had been developed for the optimization of complex water resources system. To overcome some of the drawbacks of conventional techniques, soft computing techniques were developed based on the principles of natural evolution. The major difference between the conventional optimization techniques and soft computing is that in the former case, the optimal solution is derived where as in the soft computing techniques, it is searched from a randomly generated population of possible solutions. The results of the evolutionary algorithm mainly depend on the randomly generated initial population that is arrived based on the probabilistic theory. Recent research findings proved that most of the water resources variables exhibit chaotic behavior, which is a projection depends upon the initial condition. In the present study, the chaos algorithm is coupled with evolutionary optimization algorithms such as genetic algorithm (GA) and differential evolution (DE) algorithm for generating the initial population and applied for maximizing the hydropower production from a reservoir. The results are then compared with conventional genetic algorithm and differential evolution algorithm. The results show that the chaotic differential evolution (CDE) algorithm performs better than other techniques in terms of total annual power production. This study also shows that the chaos algorithm has enriched the search of general optimization algorithm and thus may be used for optimizing complex non-linear water resources systems.