基于PSO-SVM算法的输电线路覆冰舞动预测模型EI北大核心CSCD

输电线路舞动往往会导致金具磨损、闪络、断线等电力事故,对电力系统的安全具有很大的负面影响。利用ANSYS软件模拟不同档距、风速等状态下覆冰四分裂导线在平均风与脉动风作用下的动态响应,进而根据模拟获得的数据集和PSO-SVM(particle swarm optimization-support vector machines)算法构建了四分裂导线覆冰舞动预警模型,将档距、风速、初始风攻角作为模型的输入,覆冰导线是否舞动作为输出。同时,为验证该预测模型的实用性及有效性,将PSO-SVM模型与其他智能算法如BP(back propagation)、支持向量机(support vector machine, SVM)、遗传算法优化支持向量机(genetic algorithm-optimization support vector, GA-SVM)模型的预测结果进行比较,结果表明PSO-SVM模型的预测结果精度更高,对输电线路覆冰舞动预警具有一定的参考意义。     

时序回归支持向量机在卫星振动预测中的应用

将支持向量机算法与时间序列原理相结合,可构造出基于时间序列的支持向量机模型。通过对大量遥感卫星振动数据进行分析,得出该卫星振动规律为有随机波动成分的简谐振动。应用时间序列的原理,动态更新模型训练集和预测集,构建基于时序回归的支持向量机在线预测模型。模型测试结果表明,这种方法可以比较准确有效地实现振动趋势的提前预测,为振动抑制措施的快速实现提供帮助。     

冷壁面霜层生长的支持向量机模型

针对结霜过程因具有明显的非线性特征,采用传统方法难以精确预测的问题。建立了基于支持向量机的冷壁面霜成生长的预测模型,应用实验数据对模型进行验证、评估,并与基于最小二乘法的非线性多元回归模型进行了对比、分析。结果表明,基于支持向量机的预测模型能够很好的解决非线性预测问题。在已建立的预测模型基础上,以霜层生长过程中传热率预测为例,分别在测试集中的自变量与因变量加入不同噪声信号对模型预测性能影响进行了研究。结果表明,基于支持向量机的模型具有良好的抗干扰能力。     

车内噪声声品质的支持向量机预测

对多元线性回归、神经网络和支持向量机的三个预测模型进行了研究。以车内噪声为例,建立了基于以上三种方法的车内噪声声品质预测模型,并采用留一法交叉检验作比较,所构建的支持向量机模型预测精度高于其他两种方法。实验结果同时也表明,支持向量计算法具有较强的稳健性和良好的泛化能力,能够用于车内噪声声品质的预测。     

遗传算法对RVM短期风速预测模型的多参数同步优化

该文利用混沌理论中的相空间重构方法,对基于相关向量机的风速预测模型的训练样本进行构建,然而通过混沌理论求出的相空间参数(嵌入维数E和时间延迟τ)往往不是预测模型的最优解。针对预测模型超参数优化问题,提出一种基于遗传算法的多参数优化方法,即对E、τ以及相关向量机核参数σ进行同步优化。该方法首先基于遗传算法搜索相关向量机预测模型参数(E、τ、σ)的全局最优解,进而建立预测模型;然后对待预测风速时间序列进行预测;最后以2组实际风速数据为例进行实验研究,并与对比模型方法(只优化参数σ)进行对比。结果表明:该文模型不仅具有较低的预测误差,而且可提高预测效率,缩短预测时间。     

基于遗传算法和最小二乘支持向量机的织物剪切性能预测

提出了一种基于最小二乘支持向量机的织物剪切性能预测模型,并且采用遗传算法进行最小二乘支持向量机的参数优化,将获得的样本进行归一化处理后,将其输入预测模型以得到预测结果.仿真结果表明,基于最小二乘支持向量机的预测模型比BP神经网络和线性回归方法具有更高的精度和范化能力.     

基于支持向量回归机的中央空调负荷预测模型

由于中央空调系统的时滞性、时变性、非线性和大惰性等特性,使得当前采用的中央空调负荷预测算法精度并不高,本文在江阴某楼宇空调系统节能改造项目的基础上,从中央空调系统的组成和特性出发,提出了基于支持向量回归机(Support Vector Regression SVR)理论的中央空调负荷预测模型。对项目楼宇历史负荷数据进行分析,分别采用SVR负荷预测模型和BP神经网络负荷预测模型进行了训练和预测。预测结果表明:基于SVR负荷预测模型较BP神经网络负荷预测模型精度更高,具有较强的实用性和可行性。     

输电线路工程全过程造价管理探讨

本文探讨了从项目决策阶段开始到竣工验收交付使用为止各个阶段输电线路工程造价的合理确定与有效控制,对于输电线路工程全过程造价管理具有指导意义。本文从"全过程造价管理"理论出发,结合现行输电线路工程造价构成特点,对输电线路工程全过程造价管理提出新思考。     

归一化小波熵与RVM的滚动轴承运行可靠度预测

针对传统可靠度评估和预测方法难以实现对正在服役中的单台机械设备进行可靠度评估和预测的问题。设计了一种基于归一化小波包信息熵与相关向量机的滚动轴承运行可靠度预测方法;该方法主要包括确立运行可靠度指标以及构建相关向量机预测模型,通过试验测取滚动轴承运行过程中的振动信号,利用小波包分解,提取反映滚动轴承运行状态的特征频带能量,基于信息熵理论建立运行可靠度指标;构建相关向量机预测模型,准确预测正在服役中的滚动轴承运行可靠度指标及其变化趋势。试验结果表明,采用归一化小波包信息熵与相关向量机的可靠度预测模型,能有效克服传统基于概率统计数据的平均可靠度计算问题,并且相关向量机的可靠度预测精度更高。     

基于移动客户数据流失预测模型的比较分析研究

分析常见的预测模型,结合移动客户流失数据的特征,针对移动客户流失问题,建立基于ICSVM算法支持向量机预测模型。测试结果表明,与传统预测模型相比,该模型对于大数据集和不平衡数据问题,能取得更好的客户流失预测效果。     

基于主成分分析与支持向量机的汽柴油需求预测

综合分析了影响汽柴油消费需求的关键因素,并针对其具有自相关性、复杂性、数据量大等特点,采用主成分分析法对样本数据进行降维处理,形成新的样本集。对支持向量机预测模型进行改进,在其基础之上引入时序动态因子,将上年的汽柴油需求历史数据作为时序反馈因子引入模型,从而形成新的动态反馈拟合模型,建立相应的需求预测模型。对1996~2012年的汽柴油需求预测进行实例研究,并将本文中所提方法的预测结果与灰色GM（1,1）模型、BP神经网络模型进行对比分析。结果表明本文中的主成分分析与改进支持向量机预测方法相对于GM（1,1）模型其预测误差均值分别降低了72.7%和74.86%,相对于BP神经网络其预测误差均值分别降低了81.3%和8166%,从而证明了此方法的有效性和优越性。     

Super-short-time wind power combination forecasting based on support vector machine optimized by EMD-SC and AGSA北大核心CSCD

Due to the randomness, volatility and relativity of the wind power, it brings great challenges to wind power integration. To improve the ultra short-term prediction accuracy of the wind power, a kind of method for predicting super-short-term wind power based on empirical mode decomposition (EMD) and spectral clustering (SC) and ameliorated gravitational search algorithm (AGSA) that could optimize the learning parameters of support vector machine (SVM) was put forward. Firstly, the raw data of the wind power was denoised by EMD to eliminate the irregular data; then the cluster analysis of the subsequences from EMD was carried out by SC, and SVM's model was optimized by applying AGSA to predict each subsequence respectively; finally the results of the subsequences were added together to get the ultimate predicted value. Taking one wind farm's actual data as an example, the simulation indicates that the proposed model can improve the accuracy and veracity when predicting wind power. Meanwhile, it also suggests the reasonability of this method. The method can forecast wind power accurately.     

基于EMD-SC和AGSA优化支持向量机的超短期风电功率组合预测

风电功率存在较大的随机性、波动性和相关性,这会对风电并网带来极大的挑战。为提高超短期风电功率预测精度,提出一种基于经验模态分解(empirical mode decomposition,EMD)、谱聚类(spectral clustering,SC)和改进型引力搜索算法(ameliorated gravitational search algorithm,AGSA)优化支持向量机(support vector machine,SVM)参数的超短期风电功率组合预测方法。首先通过经验模态分解对风电原始数据进行去噪处理,剔除不规则的数据;然后应用谱聚类对经验模态分解后的子序列进行聚类分析,再应用改进型引力搜索算法优化支持向量机模型对各个子序列进行预测;最后将各子序列的预测结果相加得到最终预测值。以某风电场的实际数据为算例,仿真研究表明,所提出的组合模型能够提高风电功率预测精度,且预测效果较好,同时也证明了所采用方法的合理性。该方法能够用于风电功率的精确预测。     

Intermittent wind generation in optimal power flow dispatching

A stochastic model of wind generation in an optimal power flow (OPF) dispatching program is presented. The model includes the error in wind power forecasts using a probability or relative frequency histogram. Compared with the deterministic OPF, the proposed model allows the coordination of wind and thermal power while accounting for (i) the expected penalty cost for not using all available wind power and (ii) the expected cost of calling up power reserves because of wind power shortage. The stochastic model is integrated in an extended conic quadratic OPF program in which wind driven generators are represented as induction machines. Simulation results are presented for cases where the forecasting error histogram is either derived from historical data or estimated by a bimodal normal distribution. The effect of the skewness of the error distribution on the optimal dispatch policy is studied.     

基于PSO-LSSVM算法的地下储气库洞室群爆破振动预测

超大断面小净距地下储气库洞室群开挖爆破工程中涉及到众多的影响因素,传统人工智能方法难以对爆破峰值振动速度准确预测。为了提高预测精度,引入粒子群算法,对传统的最小二乘支持向量机模型(LS-SVM)进行优化并建立粒子群最小二乘支持向量机爆破峰值振动速度预测模型(PSO-LSSVM)。以某地下储气库洞室群开挖爆破工程为研究对象,应用PSO-LSSVM模型,将PSO-LSSVM模型与LS-SVM模型、萨道夫斯基经验公式的预测结果进行对比,得到三种预测的结果平均绝对相对误差分别为:5.50%、8.56%、23.45%。由此可见,PSO-LSSVM模型的预测结果与实测数据拟合度更高,精确度更满足工程需求,可为多因素作用下类似工程爆破峰值振动速度预测提供借鉴。     

IP portfolios and evolution of biomedical additive manufacturing applications

Additive manufacturing (AM) or 3D printing includes techniques capable of manufacturing regular and irregular shapes for small batches of customized products. The ability to customize unusual shapes makes the process particularly suitable for prosthetic products used in biomedical applications. AM adoption in the field of biomedical applications (called bio-AM in this research) has seen significant growth over the last few years. This research develops an Intellectual Property (IP) analytical methodology to explore the portfolios and evolution of patents, as well as their relevance to Taiwan’s Ministry of Science and Technology (MOST) research projects in bio-AM domain. Specifically, global and domestic IP portfolios for bio-AM innovations are studied using the proposed method. First, the domain documents (of US patents and MOST projects) are collected from a global patent database and MOST project database. The key term frequency counts and technical clustering analysis of the collected documents are derived. The key terms and appearance frequencies in documents form the basis for document clustering and similarity analysis. The ontology of bio-AM is constructed based on the clustering results. Finally, the patents and projects in the adjusted clusters are subject to evolution analysis using concept lattice analysis. This research provides a computer supported IP evolution analysis system, based on the developed algorithms, for the decision support of IP and R&D strategic planning.     

Automated digital cause-and-effect diagrams to assist causal analysis in problem-solving: a data-driven approach

Causal analysis is an integral part of product quality problem-solving (QPS). Quality management within the manufacturing industry has generated a considerable amount of QPS data; while this implies a historical and extensive body of QPS experience, these valuable empirical data are not being fully utilised. Therefore, the current study proposes a method by which to mine know-why from historical empirical data, and it develops an approach for constructing digital cause-and-effect diagrams (CEDs). The K-means algorithm is first adopted to cluster the problems and causes. The random forest classifier is then selected to classify cause text into the main cause categories, which manifest as ‘rib branches’ in the CED. Based on the clustering and classification results, we obtain an abstract cause-and-effect diagram (ACED) and a detailed cause-and-effect diagram (DCED). We use the quality data of an automotive company to validate the method, and we additionally undertake a pilot run of the Fishbone Next system to demonstrate how users can obtain these two CEDs to support causal analysis in QPS. The results show that the proposed approach efficiently constructs a digital CED and thus provides quality management problem-solvers with decision support to derive the potential causes of problems, thereby improving the efficiency and effectiveness of their causal analysis initiatives.     

基于约束谱聚类的印刷套准状态识别

目的 针对实际生产中获取印刷标志图像标签成本较高的问题,研究基于约束谱聚类的印刷套准状态识别方法.方法 基于少量有标签的样本,建立样本之间的must-link约束和cannot-link约束,并进行约束扩展.计算印刷标志图像样本点欧式空间相似度矩阵,并根据扩展后约束关系修正,构建样本点的特征向量空间.采用K-means方法对样本点特征向量空间进行2类聚类,即印刷套准图像和印刷套不准图像.结果 文中方法在实验数据集上的最高印刷套准识别准确率为98.11％.文中方法(约束对数为30)的识别准确率优于无监督的谱聚类方法、朴素贝叶斯方法和决策树方法,文中方法与SVM方法的识别准确率接近.文中方法获取印刷标志图像标签的成本低于SVM方法,且模型建立和识别的时间也少于SVM方法.结论 文中方法以较少的获取印刷标志图像标签成本达到了较高的印刷套准识别准确率.     

Technological Trajectories in the Making: Two Case Studies from the Contemporary History of Wind Power

This paper traces the origins of two technological trajectories in the contemporary history of wind power technology: the American Smith‐Putnam Wind Turbine and the Danish Gedser Wind Turbine. Describing the two wind turbine projects in terms of their technical design characteristics, the professional background of the individuals involved, the organizational features of the technological knowledge production, and the historical context, the paper builds on the notion of technological trajectories in the making as a means of identifying emerging selection mechanisms for possible engineering problem solutions, scientific methods, material artifacts, and financial assessment techniques. Conceived during the Great Depression and World War II, respectively, both projects promoted the idea that wind power could make a cost‐effective contribution to the existing electric utility system. The two projects resulted in distinctive wind turbine innovations that paved the way for two technological trajectories in the contemporary history of wind power. Studying the emergence of technological trajectories, it is argued, requires the historian of technology to attend to well‐known features of technological design and its cultural context that in retrospect appear to be significant, but also to tackle the creation of novelty and the inevitability of technological uncertainties about which the logic of trajectories has little to say.     

Reactive power cost allocation by using a value-based approach

This study proposes a value-based approach to allocate the cost of reactive power production/absorption and the useful VAr reserve provided by the generators. Reactive power support can be segregated into two broad categories: the utilised component of the support and the reactive power reserve component for system security. The utilised reactive power capacity can be decomposed further into the support required for meeting the reactive power loads and that required to meet additional reactive power transmission losses for the MW load shipment. A value-based sensitivity approach has been utilised to compute utilisation factors (UFs) for allocation of the cost of reactive power production/absorption and reserve provision. This method provides transparency in determining the relative utilisation and supply of reactive power by sources to the customers. System operator (SO) can utilise this method for allocation of the cost incurred by the reactive power providers to the load serving entities. Case studies on five-bus test system, IEEE 24-bus reliability test system (RTS) and 75-bus Indian system demonstrate the proposed approach.