Methods to build neural networks for evaluation of state variable patterns

Evaluation of the system state is becoming imperative in supervising and controlling huge systems such as power generation and transmission systems and industrial plants. This paper discusses two methods of building the neural networks: one is based upon backpropagation, and the other is a new method proposed herein. Unlike backpropagation which forms the completely distributed activation structure of neurons, the proposed method enables self-organization of the modules which are activated selectively to an input pattern. The new method not only makes a biologically natural structure of neural network but also shortens considerably the learning time by localizing the links to be updated in the learning process.     

A global approach to the design of discrete-time cellular neural networks for associative memories

In this paper a global design method for associative memories using discrete-time cellular neural networks (DTCNNs) is presented. The proposed synthesis technique enables to realize associative memories with several advantageous features. First of all, grey-level as well as bipolar images can be stored. Moreover, the proposed approach generates networks with learning and forgetting capabilities. Finally, it is possible to design networks with any kind of predetermined interconnection structure. In particular, neighbourhoods without line crossings can be chosen, greatly simplifying the VLSI implementation of the designed DTCNNs. In the first part of this work a model of a multilevel threshold network is presented and a stability analysis is carried out using basic notions deriving from non-linear dynamical system theory. The synthesis procedure is then developed by means of a pseudoinversion technique, assuring learning and forgetting capabilities of the designed DTCNN. The use of a neighbourhood without line crossings is also discussed. Simulation results are reported to show the capability of the proposed approach.     

dynoNet: A neural network architecture for learning dynamical systems

This article introduces a network architecture, called dynoNet, utilizing linear dynamical operators as elementary building blocks. Owing to the dynamical nature of these blocks, dynoNet networks are tailored for sequence modeling and system identification purposes. The back-propagation behavior of the linear dynamical operator with respect to both its parameters and its input sequence is defined. This enables end-to-end training of structured networks containing linear dynamical operators and other differentiable units, exploiting existing deep learning software. Examples show the effectiveness of the proposed approach on well-known system identification benchmarks.     

组合蓄能独立式风力发电系统

提出了一种全新的采用组合蓄能方式的独立式风力发电系统.该风力发电系统采用电解水制氢与蓄电池相结合的组合蓄能方式,能够实现连续稳定自治运行且零排放、无污染,克服了采用单一蓄电池蓄能方式的系统或无法连续运行或需借助有污染的柴油机组辅助发电、或因采用泻荷装置而使系统效率过低等一系列问题.为优化系统的运行并准确估算系统元件的规...     

Hebbian learning rule restraining catastrophic forgetting in pulse neural network

In this paper, a Hebbian learning rule restraining “catastrophic forgetting” is proposed on a pulsed neural network (PNN) with leaky integrate‐and‐fire neurons. The strong point of this learning rule is that a learning of new pattern does not destroy past ones, and that an efficient use of synapses is enabled. First, in order to consider the function of the learning rule, a fundamental experiment is carried out. Next, to compare the performance between the proposed learning rule and conventional ones on the application, simulation experiments are examined using autonomous behavior robots which are forced to learn concurrently two different environments. The results of the experiments show that the proposed learning rule clearly restrains “catastrophic forgetting” and enables working of more efficient than conventional PNN learning. © 2005 Wiley Periodicals, Inc. Electr Eng Jpn, 151(3): 50–60, 2005; Published online in Wiley InterScience ( www.interscience.wiley.com ). DOI 10.1002/eej.10343     

基于多模型融合集成学习的智能变电站二次设备状态评估

为准确评估智能变电站二次设备运行状态,建立了二次设备状态评估指标体系,并结合多种机器学习算法的差异性,提出了基于多模型融合集成学习的二次设备状态评估法。该方法采用双层结构,上层中利用划分好的数据对数个基学习器进行k折验证,下层中利用全连接级联神经网络融合多个基学习器,并采用改进的列文伯格-马夸尔特算法训练该神经网络加速模型收敛。实例分析表明,所提出的方法可以准确地评估二次设备的运行状态,并为智能变电站系统和二次设备的检修工作提供指导意见。     

面向调度应用的输电线路故障定位优化方案研究

结合电网现有的故障信息采集设备现状,提出利用波形匹配原理,对故障时的电网测量值与仿真波形进行匹配,决定故障发生位置的相量匹配故障定位新算法;并提出一种面向电网调度系统的电网故障定位优化方案,该方案适用于任意结构类型的电网,通过结合电网地理信息系统的可视化应用,使调度员可以更加直观和全局地判断故障位置,利用PSACD的仿真结果表明方案核心算法的有效性和精确性。     

基于迭代学习控制的永磁直线同步电机伺服系统

针对永磁直线同步电机伺服系统,采用迭代学习控制策略来实现参考位置信号的跟踪控制。详细分析了迭代学习ILC伺服控制器的模型结构,并给出伺服控制器的迭代学习更新法则。采用高性能DS1103控制器作为控制核心,搭建永磁直线伺服系统。实验结果表明,迭代学习控制下的永磁直线伺服系统具有准确的位置跟踪能力,对外部扰动具有很强的鲁棒性。     

基于神经网络多模型自适应切换控制研究

针对传统自适应控制和现有多模型自适应切换控制理论和方法中存在的问题，提出面向复杂系统的神经网络多模型自适应切换控制方法。采用最近邻聚类学习算法对样本分类，并利用RBF神经网络的学习能力和非线性逼近能力进行离线建模。同时，引入动态模型库技术来动态建立多模型。系统运行时在线实时检测系统当前状态，若超出现有各子模型所构成的状态空间，则利用在线神经网络学习新状态并建立新模型加入动态模型库中，从而改善动态系统的暂态响应和增强系统的鲁棒性。计算机仿真结果，表明了该算法的有效性。     

Intelligent browser-based systems to assist Internet users

New client-based systems that filter Web pages, infer user learning styles, and recommend relevant pages are described. The systems provide easy, structured, focused, and controlled access to the Internet. A first system, called iLessons, is embedded within Microsoft Internet Explorer 6 and provides teachers with tools to create lesson Web pages, define zones of the Internet that can be accessed during a lesson, and enforce these settings in a set of computers. A second system enables students to investigate and collaborate using the Internet. The system filters Web pages based on the relevance of their contents and assists students by inferring their learning style (active or reflective) and by recommending pages found by fellow students based on page relevancy, student learning style, and state of mind measured by activity.     

Collaborative web-based experimentation in flexible engineering education

The Ecole Polytechnique Federale de Lausanne (EPFL), Lausanne, Switzerland, is deploying a flexible learning scheme for selected pilot courses in engineering education. In such a scheme, traditional lectures and written exercises are combined with additional Web-based learning resources. The main objective of this initiative is to sustain the evolution from traditional teaching to active learning and to better integrate the increasing number of educational resources available online. In engineering education, a key activity to sustain the learning process is hands-on experimentation carried out using either simulation tools or real equipment. This paper describes how a collaborative Web-based experimentation environment has been introduced at the EPFL for providing more flexibility to students performing laboratory experiments in automatic control, biomechanics, and fluid mechanics. It particularly describes the eJournal, a Web service integrated in the proposed learning environment that enables the collection and sharing of preparatory notes and experimental results with both peers and teaching assistants.     

基于深度字典学习的输电线路故障分类方法

针对当前输电线路故障分类识别方法存在的阈值整定复杂、人工智能算法可解释性不足等问题,提出了一种基于深度字典学习的输电线路故障分类方法。该方法利用稀疏性约束驱动字典自动提取样本中的故障特征,同时深度字典结构使得所提取的故障特征具有较好的层次性和物理含义,符合人对故障的直观认识,一定程度上解决了数据驱动型方法可解释性不足的问题。最后,通过PSCAD/EMTDC仿真验证了所提方法的有效性。     

基于时空耦合特性的充电站运行状态预测

提出一种基于时空耦合特性和深度学习模型的充电站运行状态预测方法。首先,基于充电站历史运行数据和所在区域的交通通行速度数据集,利用k-means聚类方法将充电站划分为不同类型,分析充电站运行状态在时间上的特性;建立单个充电站的"偏移量-交通-时间"三维矩阵模型,深度挖掘充电站运行状态与周边交通状况在时间和空间上的耦合相关性。其次,将充电站状态与交通状况的时间滞后相关特性进行空间重构,利用卷积神经网络进行特征提取,通过长短期记忆网络进行时间序列预测,构建基于Keras深度学习框架的充电站运行状态多步预测模型。最后,以20个充电站的真实运行数据进行验证,并与多种预测算法进行对比,结果表明,所提方法具有较高的预测精度。     

Intelligent control of induction servo motor drive via wavelet neural network

This study presents an intelligent control system for an induction servo motor drive to track periodic commands using a wavelet neural network (WNN). With the field orientation mechanism, the dynamic behavior of the induction servo motor drive system is rather similar to a linear system. However, the uncertainties, such as mechanical parametric variation, external disturbance, unstructured uncertainty due to nonideal field orientation in transient state, and unmodelled dynamics in practical applications influence the control performance. Therefore, an intelligent control system that is an on-line trained WNN controller with adaptive learning rates is proposed to control the rotor position of the induction servo motor drive. The adaptive learning rates are derived in the sense of discrete-type Lyapunov stability theorem, so that the convergence of the tracking error can be guaranteed in the closed-loop system. In the whole design process, the strict constrained conditions and prior knowledge of the controlled plant are not necessary according to the powerful learning ability of the intelligent control system. With the proposed intelligent control system, the controlled induction servo motor drive possesses the advantages of good tracking control performance and robustness to uncertainties under wide operating ranges. The effectiveness of the proposed control scheme is verified by both simulated and experimental results.     

A New Algorithm for Predicting Out-of-step Using Rotor Speed-acceleration Based on Phasor Measurement Units (PMU) Data

Transient instability is one of the major threats to system security which can cause out-of-step condition. Out-of-step condition can result in mechanical and thermal damage to generators. Therefore, in the case of out-of-step, early detection and disconnection of the generator from grid is essential. In this article, using generator rotor speed-acceleration (ω ? α) data obtained from phasor measurement units (PMU) measurements, a new algorithm for predicting out-of-step condition for generators is proposed. The trend of the movement of the (ω ? α) locus curve in the plane provides a measure for predicting and detecting out-of-step status. The predictive ability of this method enables early tripping of the unstable generator, thereby avoiding hazard damage. The proposed algorithm is examined using an IEEE 39 bus system. The simulation results demonstrate the ability of the proposed algorithm for correct prediction of various unstable power swing conditions, with sufficient early prediction compared with the actual instability time.     

Implementation of Emotional Controller for Interior Permanent-Magnet Synchronous Motor Drive

This paper presents for the first time the real-time implementation of an emotional controller for interior permanent-magnet synchronous motor (IPMSM) drives. The proposed novel controller is called brain emotional learning based intelligent controller (BELBIC). The utilization of BELBIC is based on an emotion processing mechanism in the brain and is essentially an action that is based on sensory inputs and emotional cues. The emotional learning occurs mainly in the amygdala. The amygdala is mathematically modeled, and BELBIC is introduced. This type of controller is insensitive to noise and variance of the parameters. The controller is successfully implemented in real time using a digital signal processor board ds1102 for a laboratory 1-hp IPMSM. The results show superior control characteristics, particularly very fast response, simple implementation, and robustness with respect to disturbances and manufacturing imperfections. The proposed method enables the designer to shape the response in accordance with the multiple objectives of choices.      

基于深度强化学习的居民实时自治最优能量管理策略

随着居民分布式资源的普及,如何考虑用户多类型设备的运行特性,满足实时自治能量管理需求以达到用户侧经济性最优成为亟待解决的课题。传统基于模型的最优化方法在模型精准构建和应对多重不确定性等方面存在局限性,为此提出一种无模型的基于深度强化学习的实时自治能量管理优化方法。首先,对用户设备进行分类,采用统一的三元组描述其运行特性,并确定相应的能量管理动作;接着,采用长短期记忆神经网络提取环境状态中多源时序数据的未来走势;进而,基于近端策略优化算法,赋能在多维连续-离散混合的动作空间中高效学习最优能量管理策略,在最小化用电成本的同时提升策略对不确定性的适应性;最后,通过实际情境对比现有方法的优化决策效果,验证所提方法的有效性。     

基于迭代学习的配料电机振幅控制

针对具有较强重复性的工业配料称量过程,提出采用迭代学习算法对电振机的振幅进行控制。设计了电振机振幅控制的迭代学习控制器,并分析了控制算法的收敛性。该方法控制器的设计不需要系统的模型信息,利用配料过程的重复性通过学习可以实现给料速度的精确控制。仿真实验由电振机实验装置与MATLAB仿真软件平台组成,仿真结果表明,与PID控制相比ILC控制器不但可以获得较好的跟踪效果,而且还可以有效抑制负载扰动的作用,具有较强鲁棒性。该方法计算量较小、便于实现,适合工业实际控制系统的应用。     

基于模糊神经网络再励学习控制器设计及其在倒车模型中的应用

考虑到神经网络学习算法的特点，给出了一种基于再励学习的自组织模糊CPN。它结合了模糊自组织CPN和再励算法的优点，在控制过程中在线调整网络结构以及对网络参数学习，学习效率高，控制结构简单。可以不要求受控对象的学习模型，实现在线控制，应用在倒车模型中仿真结果展示了所设计系统的良好控制性能。     

基于大数据分析的电力信息系统安全状态监测技术研究

信息系统作为电力物联网建设的核心关键部分,一旦电力信息系统受到威胁,将影响整个电网的安全稳定运行.目前基于电力信息系统的安全防护策略主要集中于传统的保护和检测方法上.然而,许多威胁发生在很短的时间内,很容易被忽略,无法及时发现.这些威胁通常会对电力信息系统造成巨大影响,干扰其正常运行.针对这一问题,文中提出了一种基于大数据分析的电力信息系统安全状态监测机制.基于模糊聚类,有效评估网络运行情况.同时提出了一种博弈理论和机器学习相结合网态势感知模型,有效降低电力信息系统的安全运维风险.仿真结果验证了所提安全状态分析策略的有效性和实用性.