Support vector machine forecasting method improved by chaotic particle swarm optimization and its application

By adopting the chaotic searching to improve the global searching performance of the particle swarm optimization (PSO), and using the improved PSO to optimize the key parameters of the support vector machine (SVM) forecasting model, an improved SVM model named CPSO-SVM model was proposed. The new model was applied to predicting the short term load, and the improved effect of the new model was proved. The simulation results of the South China Power Market’s actual data show that the new method can effectively improve the forecast accuracy by 2.23% and 3.87%, respectively, compared with the PSO-SVM and SVM methods. Compared with that of the PSO-SVM and SVM methods, the time cost of the new model is only increased by 3.15 and 4.61 s, respectively, which indicates that the CPSO-SVM model gains significant improved effects. Foundation item: Project(70572090) supported by the National Natural Science Foundation of China     

混沌模拟退火粒子群优化算法研究及应用

针对粒子群优化算法容易陷入局部极值点、进化后期收敛速度慢、精度较差等缺点,提出混沌模拟退火粒子群优化(PSO)算法.引入混沌理论对粒子群优化算法的参数进行自适应调整,提高了算法的全局收敛性能|采用模拟退火(SA)算法,依据概率性的劣向转移,以一定概率接受劣解,使算法具有跳出局部最优而实现全局最优的能力.引入自适应温度衰变系数,使模拟退火算法能够根据当前环境自动调整搜索条件,从而提高算法的搜索效率.通过7个经典函数测试混沌模拟退火粒子群优化算法的性能,并将其应用于Job Shop调度问题.仿真实验结果表明,采用新算法有效地克服了停滞现象,增强了全局搜索能力,与遗传算法、粒子群优化算法相比寻优性能更佳.     

Improved PSO algorithm and its application

The mechanism of particle swarm optimization algorithm is studied, and one can draw the conclusion that the best particle found by the swarm falling into local minima is one of the main reasons for premature convergence. Therefore, an improved particle swarm optimization algorithm is proposed. This algorithm selects the best particle with roulette wheel selection method, so premature converging to local optima is avoided. At last, the improved particle swarm optimization algorithm is applied to optimization of time-sharing power supply for zinc electrolytic process. Simulation and practical results show that the global search ability of IPSO is improved greatly and optimization of time-sharing power supply for zinc electrolytic process can bring about outstanding economic benefit for plant. Foundation item: Project (2002CB312200) supported by the National Key Research 973 Program of China     

基于PSO-SVM算法的长微博贴图识别方法

微博由于字数的限制,当用户需要发较多内容时通常以附图的形式给出,识别包含文本内容贴图的长微博能够为微博研究提供更多有用的数据.在支持向量机（SVM）的基础上结合粒子群算法（PSO）提出了一种识别长微博贴图的PSO-SVM算法.该方法提取长微博贴图的颜色矩和灰度共生矩阵特征,然后利用PSO算法对SVM模型中的误差惩罚参数和核函数进行优化得到最佳分类模型,其最优参数将被用作长微博贴图和非长微博贴图进行分类.实验表明,与传统的基于网格搜索法优化的SVM算法相比,PSO-SVM算法对长微博贴图识别具有更高的准确率和召回率.     

A PSO-SVM Model for Short-Term Travel Time Prediction Based on Bluetooth Technology

The accurate prediction of travel time along roadway provides valuable traffic information for travelers and traffic managers. Aiming at short?term travel time forecasting on urban arterials, a predict...     

新型混合粒子群优化算法

针对粒子群算法易陷入局部极值、精度低等缺点,提出了一种基于模拟退火与混沌思想的新型粒子群优化算法(SA-CPSO).在该算法的初始阶段,对粒子位置进行混沌初始化,并引入模拟退火算法对每个粒子的适应度进行评价;在该算法运行过程中根据群体适应度方差对粒子群进行混沌更新;最后通过对几种经典函数的测试计算,结果表明,相对于标准粒子群算法,该新型混合算法提高了局部搜索能力和搜索精度,并有效避免了早熟现象的产生.     

PSO-SVM模型的构建与应用

为了使支持向量机(SVM)获得更好的分类效果,针对人为选择参数的随机性,提出了利用粒子群算法(PSO)进行参数自动选取的优化方法,构建了PSO-SVM模型.在个人信用评估中,通过对粒子适应度函数的设置来控制造成较大损失的第二类误判,应用结果表明:模型在训练和测试样本中的分类精度可以达到95%,第二类误判率分别仅为0.78%和2.02%.利用PSO对SVM中的参数进行优化,可以避免人为选择的随机性,并且在解决分类问题中表现出较好的稳健性.     

磁性材料成型烧结生产调度优化方法及应用

建立以最小化提前和拖期时间、最小化炉重偏差为目标的混合整数线性规划模型, 解决磁性材料成型-烧结两阶段生产调度问题. 提出一种混合粒子群优化算法(HPSO)进行模型的求解,该算法采用基于订单的编码方式. 针对粒子群算法易陷入局部最优, 在迭代过程中引入模拟退火思想. 改进粒子群算法的全局极值和个体极值选取方式, 使算法尽快收敛到非劣最优解. 生产现场实际数据仿真结果表明: 该混合粒子群算法无论在求解精度, 还是求解速度上均优于普通粒子群算法和遗传算法.     

基于改进粒子群优化算法和CRNN的多类SVM分类

为了提高支持向量机(SVM)在多类分类中的分类效果,提出了一种基于改进粒子群优化(IMPSO)算法和协作式递归神经网络(CRNN)的多类SVM分类方法(IMPSO_CRNN_SVM算法).首先引入自适应惯性权重及自适应粒子变异,以此改进粒子群优化算法(PSO)在优化SVM参数过程中存在的容易陷入局部最优和早熟等问题; 然后基于多类SVM设计一个CRNN,并利用随机分配的训练集对该网络进行训练并构建最终决策函数,从而实现多类数据的“一次性”分类.最后利用3种数据集和实际应用对IMPSO_CRNN_SVM算法进行验证,结果表明IMPSO_CRNN_SVM算法的分类精度优于未进行参数优化的传统SVM算法、基本PSO 进行SVM参数优化的算法和未进行PSO参数优化的基于CRNN的多类支持向量机算法,因此IMPSO_CRNN_SVM算法具有一定的实用性.     

基于PSO-SVM模型的城轨列车轮对尺寸预测

为了及时掌握轮对尺寸信息,从而保证列车正常运行,提出了建立粒子群优化支持向量机模型对轮对尺寸进行预测的方法.介绍了粒子群优化算法(PSO)及支持向量机(SVM)的相关概念,并利用粒子群优化算法能够实现快速全局优化的特点对支持向量机进行参数优化,解决了支持向量机参数选择盲目性的问题.以某城轨列车轮径值为研究对象,建立基于PSO-SVM的轮对尺寸预测模型,对轮径值进行预测分析.结果表明,轮径值预测相关度达到0.94,证实了建立的预测模型在轮对尺寸预测方面的可行性及有效性.     

双层规划模型的大规模UCAV编队队形优化

为解决复杂约束环境下大规模无人战斗机(UCAV)编队队形优化问题,提出基于双层规划模型的队形优化求解算法.以大规模UCAV编队空对地饱和打击作战场景为例,建立UCAV编队作战上层规划模型,通过采用离散粒子群-模拟退火(DPSO-SA)算法进行求解,得到执行每个任务的UCAV编号和最优队形;根据现有的编队作战队形库,建立编队中UCAV站位下层规划模型,通过采用遗传算法进行求解,得到UCAV在队形中的位置.仿真结果表明:在上层规划模型中引入改进模拟退火算法,可以解决离散粒子群算法易陷入局部极小值的问题;设计双层规划模型,可以解决DPSO-SA算法后期收敛速度慢的问题.相对于单层规划模型,双层规划模型求解大规模UCAV编队队形优化问题收敛速度更快,寻优效果更好.     

Application of a support vector machine for prediction of slope stability

Slope stability estimation is an engineering problem that involves several parameters. To address these problems, a hybrid model based on the combination of support vector machine(SVM) and particle swarm optimization(PSO) is proposed in this study to improve the forecasting performance. PSO was employed in selecting the appropriate SVM parameters to enhance the forecasting accuracy. Several important parameters, including the magnitude of unit weight, cohesion, angle of internal friction, slope angle, height, pore water pressure coefficient, were used as the input parameters, while the status of slope was the output parameter. The results show that the PSO-SVM is a powerful computational tool that can be used to predict the slope stability.     

基于PSO-SVM的离心式压缩机转速模型的研究

针对宁夏某石化公司离心式CO2压缩机透平转速预测难以实现问题,引入PSO-SVM回归模型对离心式压缩机透平转速进行预测.分析选取离心式压缩机透平转速作为模型因变量,通过相关分析从采集量中选取高相关度预测因子,运用粒子群算法选择模型最优参数,利用支持向量机的方法建立模型进行预测,与传统的SVM模型进行对比,该模型得到了良好效果,能够有效预测压缩机透平转速.     

求解背包问题的病毒协同进化粒子群算法

为提高粒子群算法的搜索性能,提出一种基于病毒进化理论的改进离散粒子群算法:病毒协同进化粒子群算法.在粒子群中引入生物病毒机制和宿主与病毒基于感染操作的思想,病毒采用与粒子等长的编码方式,执行反向代换、结合等操作,利用病毒的水平感染和垂直传播能力较好地维持个体的多样性和对解空间的局部搜索能力.通过解决背包问题对算法进行验证,仿真表明所提算法搜索性能优于遗传算法、模拟退火及标准粒子群等其他算法.该算法能有效求解背包问题等NP难题.     

基于粒子群和免疫克隆的模糊C-均值聚类

目的基于梯度下降的模糊聚类算法（FCM）选择最优解做改进,降低原FCM对初始值的敏感度,改进模糊C-均值算法,加快收敛速度,改善聚类的效果.方法该算法通过克隆选择改变粒子群优化算法（PSO）中群体的多样性,用PSO代替了FCM算法的基于梯度下降的迭代过程.结果算法具有很强的全局搜索能力,很大程度上避免了FCM算法易陷入局部极值的缺陷,同时也降低了FCM算法对初始值的敏感度.通过机器学习中的W ine和Iris数据对所提出的算法进行验证,取W ine样本数据为178个,条件属性为13,聚类类别数为3;Iris数据150个,条件属性个数为4,对这两类数据分别进行聚类分析,将试验结果与单纯的FCM和基于PSO的FCM比较,聚类的正确性有所提高.结论基于粒子群和免疫克隆的模糊C-均值聚类算法具有很强的全局搜索能力,提高了聚类的效果和效率.     

基于模拟退火PSO的电力系统无功优化

对粒子群优化算法方法进行改进,把模拟退火机制引入到粒子群优化算法方法中,提出了基于模拟退火粒子群优化PSOSA(PSO with Simulated Annealing)算法,通过适当选择种群大小、调整惯性权重系数ω和退火系数C,以温度的缓慢下降来控制粒子的寻优过程,提高了粒子群优化算法的全局收敛性,改善了粒子的局部搜索能力.建立了以网损最小为目标的电力系统无功优化模型.通过对IEEE-30系统的无功优化计算,结果表明,PSOSA算法具有更好的全局收敛性和良好的搜索能力.     

Identification of large-scale goaf instability in underground mine using particle swarm optimization and support vector machine

An approach which combines particle swarm optimization and support vector machine (PSO–SVM) is proposed to forecast large-scale goaf instability (LSGI). Firstly, influencing factors of goaf safety are analyzed, and following parameters were selected as evaluation indexes in the LSGI: uniaxial compressive strength (UCS) of rock, elastic modulus (E) of rock, rock quality designation (RQD), area ration of pillar (S_p), the ratio of width to height of the pillar (w/h), depth of ore body (H), volume of goaf (V), dip of ore body (α) and area of goaf (S_g). Then LSGI forecasting model by PSO-SVM was established according to the influencing factors. The performance of hybrid model (PSO + SVM = PSO–SVM) has been compared with the grid search method of support vector machine (GSM–SVM) model. The actual data of 40 goafs are applied to research the forecasting ability of the proposed method, and two cases of underground mine are also validated by the proposed model. The results indicated that the heuristic algorithm of PSO can speed up the SVM parameter optimization search, and the predictive ability of the PSO–SVM model with the RBF kernel function is acceptable and robust, which might hold a high potential to become a useful tool in goaf risky prediction research.     

基于IC-PSO和ISM的反馈控制算法设计

为了提高光纤偏振模色散(PMD)补偿系统的动态自适应补偿能力,提出了用改进粒子群优化(PSO)算法作为搜索算法,用改进单纯形法(ISM)作为跟踪算法的反馈控制算法设计方案,从而实现对PMD补偿单元的动态调整。在PSO算法中,引入免疫克隆(IC)原理提高了搜索算法的全局优化能力;对SM的反射操作和扩张操作进行改进,从而提高算法的收敛速度;用映射操作代替原有的顶点代换操作,从而使单纯形在迭代过程中不发生退化现象。实验结果证明了该算法用于PMD补偿系统的有效性和可行性。     

Hybrid optimization model of product concepts

Deficiencies of applying the simple genetic algorithm to generate concepts were specified. Based on analyzing conceptual design and the morphological matrix of an excavator, the hybrid optimization model of generating its concepts was proposed, viz. an improved adaptive genetic algorithm was applied to explore the excavator concepts in the searching space of conceptual design, and a neural network was used to evaluate the fitness of the population. The optimization of generating concepts was finished through the ＂evolution - evaluation＂ iteration. The results show that by using the hybrid optimization model, not only the fitness evaluation and constraint conditions are well processed, but also the search precision and convergence speed of the optimization process are greatly improved. An example is presented to demonstrate the advantages of the orooosed method and associated algorithms.     

一种动态分群带熵权的粒子群优化方法

为提高粒子群优化的求解性能,提出了一种动态分群带熵权的粒子群优化求解方法．该方法采用k的均值聚类获得子群总数,在子群粗搜索过程中充分利用其他粒子的熵信息,采用子群及其他子群搜索的最优解信息构建熵权以调整惯性权重,利用自身群粒子经过m次迭代时的优化信息构建熵权以调整本群的全局最优值．在子群精搜索过程中,利用各子群获得的最优解信息作为新群的初始设置,利用其他粒子的迭代信息构建熵权来调整全局最优值．采用传统的粒子群优化算法、其他文献中的方法以及新提出的方法分别对4个经典的测试函数进行对比实验,从获得解的最优值、平均值、标准差以及平均迭代数作对比,从而验证了该方法具有求解精度高以及优化求解迭代次数少等优点．