Pattern analysis and classification for security evaluation in power networks

Security evaluation is a major concern in real time operation of electric power networks, exhibiting behavioral patterns under abnormal conditions. Security assessment and evaluation can be viewed as a pattern analysis task identifying abnormal patterns of the power system behavior under highly loaded conditions. Traditional method of security evaluation are highly time consuming and infeasible for direct on-line implementation. This paper presents application of pattern directed inference system for static and transient security evaluation and classification. A straightforward and quick procedure called Sequential Forward Selection method is used for feature selection process. The classifier model in the pattern directed inference system is designed using different pattern classifier algorithms, viz., conventional, neural network and machine learning classifiers. Support Vector Machine (SVM), one of the popular machine learning classifier, is recognized as a suitable pattern classifier for security evaluation problem. The generalization performance of SVM classifier is greatly influenced by the proper setting of its parameters. This paper also addresses different heuristic optimization techniques used in the selection of SVM parameters. The design, development and performance of different classifiers for power system security classification are presented in detail. Simulation work is performed on standard New England 39-bus benchmark system and the feasibility of implementation of the proposed SVM based classifier system for on-line security evaluation is also discussed.     

基于图卷积网络的快速暂态安全评估方法

快速、可靠的电力系统动态安全评估能够显著提高电力系统运行方式优化调整的效率。针对电力系统暂态稳定预想事故扫描需要完成大量仿真、过于耗时的问题,提出了基于图卷积网络的快速动态安全分析方法。该方法基于电力系统的潮流特征和拓扑特征构建电力系统潮流特征图。利用图卷积方法对电力系统运行状态进行特征挖掘与特征学习,将动态安全评估问题建模为图上节点分类问题。所得模型在读取电网拓扑与潮流运行状态后,仅须完成一次前向计算即可同时给出预想事故集中多个预想事故的稳定性预测结果,无须依赖仿真波形或量测数据,实现快速暂态稳定预想事故扫描。IEEE39节点系统算例测试表明,算法设计正确、高效、准确率高,能够显著提高暂态稳定预想事故扫描的效率,实现快速动态安全评估。     

电力系统动态安全域的LS-SVM在线拟合法

提出了一种基于支持向量机的电力系统动态安全域在线拟合方法。支持向量机在解决非线性有限样本和高维识别方面有明显优势,但标准支持向量机在学习时需要求解复杂的二次规划问题,耗时较多。为此采用最小二乘支持向量机的二值分类算法,构造了二类和三类分类器对运行点的稳定状态进行判断,以最小二乘线性系统代替二次规划方法的标准支持向量机进行模式识别和函数估计,解决了大样本数据建模和运算速度慢的问题。同时采用回归算法构造稳定裕度拟合器,对系统既定故障下运行点的临界切除时间进行在线拟合并计算出稳定裕度。最后以EPRI-36节点模型为算例进行仿真计算,仿真结果表明该方法避免“维灾难”问题的同时,能更好地拟合动态安全域的边界,且进一步证明了该方法的有效性和准确性。     

Power transformer fault diagnosis based on dissolved gas analysis by support vector machine

This paper presents an intelligent fault classification approach to power transformer dissolved gas analysis (DGA). Support vector machine (SVM) is powerful for the problem with small sampling (small amounts of training data), nonlinear and high dimension (large amounts of input data). The standard IEC 60599 proposes two DGA methods which are the ratios and graphical representation. According the experimental data, for the same input data, these two methods give two different faults diagnosis results, what brings us to a problem. This paper investigates a novel extension method which consists in elaborating an input vector establishes by the combination of ratios and graphical representation to resolve this problem. SVM is applied to establish the power transformers faults classification and to choose the most appropriate gas signature between the DGA traditional methods and a novel extension method. The experimental data from Tunisian Company of Electricity and Gas (STEG) is used to illustrate the performance of proposed SVM models. Then, the multi-layer SVM classifier is trained with the training samples. Finally, the normal state and the six fault types of transformers are identified by the trained classifier. In comparison to the results obtained from the SVM, the proposed DGA method has been shown to possess superior performance in identifying the transformer fault type. The SVM approach is compared with other AI techniques (fuzzy logic, MLP and RBF neural network); the proposed method gives a good performance for transformers fault diagnosis. The test results indicate that the novel extension method and the SVM approach can significantly improve the diagnosis accuracies for power transformer fault classification.     

基于最小二乘支持向量机的电站锅炉空预器热点检测系统研究

回转式空气预热器是火力发电机组重要的换热设备。燃料的不完全燃烧以及低负荷或停炉后空预器内气体流速低造成散热条件变差等原因会引起空预器的再燃烧事故。论文利用最小二乘支持向量机这种新的机器学习工具，分别用两种核函数建立针对三对不同火情的判别模型，超平面参数通过交叉检验的方式确定。实验结果表明，支持向量机具有很好的分类和泛化能力。从两种核函数的ROC曲线可看出对于本问题选用RBF核函数相对于多项式核函数有更高的判别准确率。     

Supervised learning approach to online contingency screening and ranking in power systems

This paper proposes a supervised learning approach to fast and accurate power system security assessment and contingency analysis. The severity of the contingency is measured by two scalar performance indices (PIs): Voltage-reactive power Performance Index, PI_VQ and line MVA Performance Index, PI_MVA. In this paper, Feed-Forward Artificial Neural Network (FFNN) is employed that uses pattern recognition methodology for security assessment and contingency analysis. A feature selection technique based on the correlation coefficient has been employed to identify the inputs for the FFNN. The effectiveness of the proposed methodology is demonstrated on IEEE 39-bus New England system at different loading conditions corresponding to single line outage. The overall accuracy of the test results for unknown patterns highlights the suitability of the approach for online applications at Energy Management Center.     

A decision tree method for on-line steady state security assessment

In this paper the application of an inductive inference method to online steady state security assessment of a power system is proposed. For each contingency a number of decision rules in the form of a decision tree (DT) is built offline from a preclassified learning set consisting of operating points of the system. For the real time application of the method the DTs corresponding to the foreseen contingencies are searched online to provide optimal guidelines for preventive control of the system. The algorithm developed is applied to the steady state security assessment of a realistic model of the Hellenic interconnected power system comprising 240 busbars, 270 branches, 57 transformers and 30 equivalent generators     

电动车蓄电池荷电状态估计的神经网络方法

针对电动车蓄电池电能容量判别问题,将神经网络方法应用于电动车蓄电池荷电状态估计.对多种神经网络方法的估计性能进行了分析,包括多层感知器网络、径向基函数网络、线性支持向量机、使用MLP核函数的支持向量机、使用RBF核函数的支持向量机.实验结果表明:神经网络经过训练后,可以通过蓄电池的工作电压、工作电流和表面温度参数估计蓄电池的SOC实时值,其中多层感知器和支持向量机估计性能最好,同时,支持向量机较多层感知器有更高的噪声容忍能力.     

基于监督学习的非侵入式负荷监测算法比较

非侵入式负荷监测(NILM)能够在不干扰用户正常用电的情况下,低成本地实现用户用电设备类型的识别和用电负荷的分解,因此非常适用于家庭用户用电监测。大量智能电表在家庭用户中的安装为居民NILM提供了数据支撑,也使得居民NILM研究成为热点。基于家庭负荷稳态电流样本,采用负荷电流谐波系数作为负荷分类特征,建立了基于多层感知器(MLP)神经网络、k-近邻算法、逻辑回归、支持向量机的4种NILM分类模型,利用BLUED数据库对4种分类器进行训练和测试,对比分析其在识别精度、训练时间、识别速度和抗噪能力方面的表现,并对其在家庭负荷识别中的应用效果进行对比研究。结果表明,4种分类器中MLP神经网络具有总体最优的分类效果和计算性能,更适用于家庭用户负荷监测。     

On-line voltage stability assessment using radial basis function network model with reduced input features

In recent years, voltage instability has become a major threat for the operation of many power systems. This paper presents an artificial neural network (ANN)-based approach for on-line voltage security assessment. The proposed approach uses radial basis function (RBF) networks to estimate the voltage stability level of the system under contingency state. Maximum L-index of the load buses in the system is taken as the indicator of voltage stability. Pre-contingency state power flows are taken as the input to the neural network. The key feature of the proposed method is the use of dimensionality reduction techniques to improve the performance of the developed network. Mutual information based technique for feature selection is proposed to enhance overall design of neural network. The effectiveness of the proposed approach is demonstrated through voltage security assessment in IEEE 30-bus system and Indian practical 76 bus system under various operating conditions considering single and double line contingencies and is found to predict voltage stability index more accurate than feedforward neural networks trained by back propagation algorithm and AC load flow. Experimental results show that the proposed method reduces the training time and improves the generalization capability of the network than the multilayer perceptron networks.     

多输入特征融合的组合支持向量机电力系统暂态稳定评估

利用支持向量机(SVM)方法进行暂态稳定判别时，输入特征的选择是影响最终结果的最重要因素。传统启发式和试探式方法不能从根本上解决输入特征选择的问题。本文利用信息融合思想，在构造的具有不同输入特征的多组子分类器的基础上，对子分类器的结果在输出空间再进行信息融合，以提高分类准确率。文中从不同角度启发式的构造了4组不同的输入特征，构成四组弱分类器。以这四组弱分类器为子分类器，再构造一个融合SVM对几种子分类器的结果以回归方式进行融合，作为最终判别结果。IEEE39-BUS和IEEE145-BUS测试系统上进行的仿真表明，弱分类器的分类性能经过融合得到明显强化，融合后的结果比任何一种子分类器的结果以及一次包含所有输入特征的结果都更准确。该方法为在线快速进行暂态稳定计算提供了一条重要途径。     

Fast real power contingency ranking using counter propagation network: feature selection by neuro-fuzzy model

In deregulated operating regime power system security is an issue that needs due consideration from researchers in view of unbundling of generation and transmission. Real power contingency ranking is an integral part of security assessment. The objective of contingency screening and ranking is to quickly and accurately shortlist critical contingencies from a large list of credible contingencies and rank them according to their severity for further rigorous analysis. In the present work, modified counter propagation network (CPN) with neuro-fuzzy (NF) feature selector is used for real power contingency ranking of the transmission system. The CPN is trained to estimate the severity of a series of contingencies for given pre-contingencies line-flows. But for larger size system it becomes rather difficult to cope with the increased size of input pattern and network as well. And it adversely affected the performance of the network and computational overhead. The proposed NF feature selector prunes the size of input pattern by exploring the individual power of features to characterize/discriminate different clusters. The reduced set of discriminating inputs not only ensures saving in training time but also improves estimation accuracy and execution time and these are the deciding parameters in evaluating the performance of particular contingency ranking technique. The effectiveness of proposed approach is demonstrated on IEEE 30-bus test system and practical 75-bus Indian system.     

计及动作安全裕度的节点电压脆弱性评估

针对电力系统安全问题,定义了节点动作安全裕度指标,结合传统节点电压脆弱性评估指标,对预想事故下的节点电压脆弱性进行分析,建立了计及动作安全裕度的节点电压脆弱性评估指标。通过对IEEE14、IEEE30母线系统的仿真,结果表明节点事故动作安全裕度的引入部分改变了传统节点电压脆弱度的排序,与传统节点电压脆弱性评估结果相比,更全面考虑了事故发生概率以及事故保护作用和故障操作对节点脆弱性的影响,提高了不同事故下节点电压脆弱性的评估精度,验证了该方法的和合理性与有效性。     

基于历史数据聚类分析的暂态功角稳定故障筛选

大电网中有上千个暂态稳定故障,若对每个故障分别进行暂态评估,难以满足在线评估对时间的要求。为了满足电网在线暂态安全稳定评估快速性的要求,提出了一种基于电网运行历史数据聚类分析的暂态功角稳定故障筛选方法。基于历史数据中的电网运行方式和暂态功角稳定评估结果,提取关键特征量,通过计及稳定模式的矢量量化方法确定聚类数和初始聚类中心,采用K中心点算法对聚类中心进行优化。针对分类后暂态功角稳定的考察故障快速估算其暂态功角裕度,最后得到包含暂态功角失稳和估算裕度低于门槛值的故障组成的用于暂态稳定分析计算的严重故障集。通过对实际省级电网运行历史数据的聚类分析,验证了所述方法的有效性和实用性。     

基于DT法和粗糙集理论的接地网安全性状态评定

接地网是维护电力系统安全可靠运行的重要环节,对接地网安全状态的准确评估关系到电力系统能否安全稳定运行。针对接地网安全性分级仅凭经验确定,安全性评估只考虑单一特性参数是否满足规定限值的问题,提出了基于动态分级(DT)法和粗糙集理论的接地网安全状态综合评估方法。将DT法应用于接地网安全性分级评定,再根据相关导则规程等要求的安全限值对实测数据进行离散化处理,应用粗糙集理论计算出各个特性参数的综合权重,最后利用线性加权法得出接地网状态的综合评估结果。对工程实例的应用结果表明,该评估方法简单易行且合理有效。     

Dynamic preventive control for power system using transient stability assessment based on pattern recognition

To maintain power system security, the authors are developing an integrated security monitoring and control (ISMAC) system which consists of the three main functions: security monitoring, preventive control and emergency control. This paper focuses on the dynamic preventive control which deals with the transient stability immediately after the contingency has occurred. The proposed method is based on the transient stability assessment using the pattern recognition with two-dimensional feature space. Therefore, a preventive control strategy can be obtained rapidly. An index which represents the severity of the contingency quantitatively (security index) is defined by the distance from a linear decision surface which divides a feature plane into a stable and an unstable region. Further, this method has also the advantage that it is possible to consider the effect of the control devices or damping to some extent and specify the operator demand for stabilization flexibly. The effectiveness of the proposed method is ascertained through numerical examples for model power systems.     

Static security enhancement via optimal utilization ofthyristor-controlled series capacitors

This paper describes an approach for determining the most suitable locations for installing thyristor-controlled series capacitors (TCSCs) in order to eliminate line overloads under a single contingency. The proposed approach introduces an index called the single contingency sensitivity (SCS) index which is used to rank the system branches according to their suitability for installing a TCSC. Once the locations are determined, an optimization problem of finding the best settings for the installed TCSCs in case of a given set of contingencies, is formulated and solved. The objective of the optimization problem is to eliminate or minimize line overloads as well,as the unwanted loop flows under single contingencies. The proposed approach for locating and controlling TCSCs enhances the static security and reduces the power losses in a given power system. The IEEE 14-bus and 118-bus test systems are used to demonstrate the proposed approach     

基于HHT和特征离散化的电压暂降扰动源分类特征提取

特征提取是分类问题最关键的环节之一,针对电压暂降扰动源分类中分类特征的提取问题进行研究。首先基于希尔伯特—黄变换(HHT)和类别—属性关联程度最大化(CAIM)离散化方法提出了三种分类特征提取方案,然后分别在决策树(DT)、概率神经网络(PNN)和支持向量机(SVM)上进行了验证。仿真结果表明,基于HHT的特征提取方法可提取有效的电压暂降扰动源分类特征。而且特征的离散化处理可以在不降低分类精度的前提下,有效压缩训练样本集。同时增强分类算法的鲁棒性,对实现电压暂降扰动源的快速、准确识别具有重要的意义。     

Classification of disturbances in hybrid DG system using modular PNN and SVM

This paper presents the classification of islanding and power quality (PQ) disturbances in grid-connected distributed generation (DG) based hybrid power system. The penetration of DG influences the PQ levels in the distribution networks. Islanding disturbances are separated out from the PQ disturbances based on the selection of suitable threshold value, at the initial stage of classification process. Further, the power quality disturbances are automatically classified into distinct classes based on feature extraction using S-transform followed by training of two classifiers, namely, modular probabilistic neural network (MPNN) and support vector machines (SVMs). Five different types of disturbances are considered for the classification problem. The study reveals that S-transform (ST) in association with MPNN and SVM can effectively detect and classify islanding and PQ disturbances. The proposed methodology uses features instead of real data set and thereby reduces the data size to classify disturbance signal without losing its original property. The accuracy and reliability of proposed classifier is also tested on signals contaminated with noise and PQ disturbances caused due to wind speed variation on an experimental prototype set-up.     

A novel protection scheme for synchronous generator stator windings based on SVM

This paper proposes a novel scheme for detecting and classifying faults in stator windings of a synchronous generator (SG). The proposed scheme employs a new method for fault detection and classification based on Support Vector Machine (SVM). Two SVM classifiers are proposed. SVM1 is used to identify the fault occurrence in the system and SVM2 is used to determine whether the fault, if any, is internal or external. In this method, the detection and classification of faults are not affected by the fault type and location, pre-fault power, fault resistance or fault inception time. The proposed method increases the ability of detecting the ground faults near the neutral terminal of the stator windings for generators with high impedance grounding neutral point. The proposed scheme is compared with ANN-based method and gives faster response and better reliability for fault classification.