Presidential Faculty Fellows

使用自适应识别技术和人工神经网络开发智能诊断方法来检测电力系统设备的早期故障。我们将继续利用人工神经网络对电力系统设备进行诊断。以前开发的方法和经验将得到广泛检验。故障诊断的一项重要任务是确定最具代表性的设备状态参数，以及最可能导致特定故障或性能恶化的原因。这些将通过对人工神经网络中单个或一组参数的敏感性测试来实现。尽管基于人工神经网络的诊断系统是依赖于问题的，但一些关于输入特征选择、人工神经网络架构和训练算法的一般规则将被开发出来，作为类似应用的指导方针。利用基于卫星的相量测量技术开发动态应急分析和控制算法。建立实时动态模型，以研究实时跟踪电力系统动态的可能性。建立电力系统谐波噪声表征的研究计划，开发综合概率谐波流算法，考虑谐波电压和电流的非线性依赖，以及谐波源运行模式和开关状态的随机变化。更多的努力将致力于开发算法，以识别谐波源基于有限的测量多个电路节点。用于教授困难概念的计算机动画包将被扩展，更复杂的领域问题和概念将用动画来说明。多媒体工具将逐步加入到传统的课堂教学中。电力工程核心课程材料将被更新和修订，以便为今天的电气工程毕业生建立一个理想的知识组合。电能质量教学与研究实验室正在筹建中，实验室设备正在研制中。计划为谐波测量增加新的实验室设备。这将使我们能够开始数据收集和分析。此外，本课程亦会发展实验室实验，以供学生在新的电能品质课程中使用。＊＊＊

9453422 Liu Develop intelligent diagnostic methodologies to detect incipient failures in power system equipment using adaptive recognition techniques and artificial neural networks. We will continue the work of power system equipment diagnosis using artificial neural networks. The previously developed methods and experiences will be widely tested. An important task for failure diagnosis is to determine the most representative parameters of equipment status, and the most likely cause of a particular failure or performance deterioration. These will be achieved through sensitivity tests of individual or a group of parameters in ANNs. Although ANN- based diagnosis systems are problem-dependent, some general rules on input feature selection, ANN architecture, and training algorithms will be developed for use as guidelines in similar applications. Develop dynamic contingency analysis and control algorithms using Satellite-Based Phasor Measurement Technology. Develop real-time dynamic models in order to investigate the possibility of tracking power system dynamics in real-time. Establish research programs in characterizing harmonic noise in power systems by developing an integrated probabilistic harmonic flow algorithm that considers the nonlinear dependence of harmonic voltages and currents, and random variations of operation modes and switching status of harmonic sources. More effort will be devoted to develop algorithms for identifying harmonic sources based on limited measurements for multiple circuit nodes. The computer animation packages for teaching difficult concepts will be expanded and more complex field problems and concepts will be illustrated with animation. Multimedia tools will be gradually added to the conventional classroom teaching. The power engineering core course materials will be updated and revised in order to establish an ideal knowledge mix for today's Electrical Engineering graduates. The Power Quality Teaching and Research Lab are being planned and lab equipment are being researched. It is planned that new lab equipment will be added for harmonic measurements. This will allow us to begin data gathering and analysis. Also, lab experiments will be developed for students to use in the new power quality course. ***