Free-Model Based Intelligent Control of Power Plants and Power Systems

基于自由模型的发电厂和电力系统智能控制

• 批准号: 9705105
• 负责人: Kwang Lee
• 金额: $ 23.68万
• 依托单位: Pennsylvania State Univ University Park
• 依托单位国家: 美国
• 项目类别: Standard Grant
• 财政年份: 1998
• 资助国家: 美国
• 起止时间: 1998-01-01 至 2001-12-31
• 项目状态: 已结题
• 来源: https://www.nsf.gov/awardsearch/showAward?AWD_ID=9705105&HistoricalAwards=false
• 关键词: Free; Model; Based; Intelligent; Control

ECS-9705105 Lee This project is to develop a Free-Model Based Intelligent Control (FMBIC) system for power plants and power systems. Automation of normal power plant operation is currently achieved in a simple and direct manner using single-loop feedback and feedforward controls based on linearized models of plant subsystems. In this environment, controlling a system is very difficult when there are plant or controller failures or a significant variation in operating conditions. The proposed free-model based intelligent control approach offers an alternative in achieving highly autonomous power plants and power systems. Current main streams for intelligent control are fuzzy logic and artificial neural networks. Both attempt to emulate human control schemes by incorporating human decision making and learning ability. However, they alone can not emulate humans perfectly. Therefore, it is proposed to develop a new approach based on the "free model" concept, which will help in modeling human behavior more closely. The free model is defined only with input and output data obtained from the system and consequently, no mathematical model is required for control purposes. Thus, it can be used for any system which is complex, highly nonlinear, and susceptible to disturbances, such as power plants and power systems. By representing a system in terms of the free model, human control schemes will be modeled in five different areas: (1) human learning ability will be emulated on the premise of the free model using neural networks and fuzzy systems; (2) a supervisory control system will be developed to emulate human supervisory ability using a bank of controllers in parallel; (3) a decentralized control scheme will be developed by partitioning a complex system into a number of smaller subsystems which are coupled by interaction variables; (4) a predictive control scheme will be developed by adding an optimizing feedforward control; and (5) a fault-accommodating control system will be devel oped by using the free-model based neuro or fuzzy identifier. This research an the Free-Model Based Intelligent Control (FMBIC) will be conducted at Penn State University as a regular project. In order to enrich the project and enhance the progress of the research, it is also proposed to add an international dimension by collaborating with Seoul National University (SNU), Seoul, Korea through the U.S.-Korea Cooperative Research Program. This will enable the Principal Investigator and a graduate student to visit SNU and exchange research results and experiences with the Korean counterpart. The theoretical and implementation aspects of this collaboration will be of benefit both to Penn State University and Seoul National University, as well as to the U.S. and Korea power industries, in general. The anticipated scientific and practical benefits are: (1) sharing of laboratory facilities and theoretical developments in the Free-Model Based Intelligent Control (FMBIC) system; (2) joining of complementary skills: Penn State's experience in power plant intelligent distributed control and SNU's experience in power system intelligent control; (3) providing new insights and improvements to power engineering education in the U.S. and Korea; and (4) development of intelligent control technology providing improved efficiency, economy, and safety of power plants and power systems in the face of future deregulation and plant construction difficulties in both countries.

该项目是为发电厂和电力系统开发基于自由模型的智能控制（FMBIC）系统。目前，利用基于电厂子系统线性化模型的单环反馈和前馈控制，以简单直接的方式实现电厂正常运行的自动化。在这种环境下，当设备或控制器发生故障或运行条件发生重大变化时，控制系统是非常困难的。提出的基于自由模型的智能控制方法为实现高度自治的电厂和电力系统提供了另一种选择。目前智能控制的主流是模糊逻辑和人工神经网络。两者都试图通过结合人类的决策和学习能力来模仿人类的控制方案。然而，它们本身并不能完美地模仿人类。因此，提出了一种基于“自由模型”概念的新方法，这将有助于更紧密地建模人类行为。自由模型仅使用从系统获得的输入和输出数据来定义，因此，不需要用于控制目的的数学模型。因此，它可以用于任何复杂的、高度非线性的、易受干扰的系统，如发电厂和电力系统。通过用自由模型来表示一个系统，人类控制方案将在五个不同的领域建模：(1)在自由模型的前提下，使用神经网络和模糊系统来模拟人类的学习能力；(2)将开发一个监控系统，使用一组并行控制器来模拟人类的监督能力；(3)将一个复杂的系统划分为若干较小的子系统，这些子系统通过相互作用变量耦合，从而形成分散的控制方案；(4)加入最优前馈控制，形成预测控制方案；(5)采用基于自由模型的神经或模糊辨识器，建立自适应控制系统。这项基于自由模型的智能控制（FMBIC）的研究将作为一个常规项目在宾夕法尼亚州立大学进行。为了丰富项目内容和提高研究进展，还提议通过美韩合作研究项目与韩国首尔国立大学（SNU）合作，增加国际维度。这将使首席研究员和一名研究生访问首尔大学，并与韩国同行交流研究成果和经验。此次合作的理论和实施方面将有利于宾夕法尼亚州立大学和首尔国立大学，以及美国和韩国的电力行业。预期的科学和实际效益是：(1)共享实验室设施和基于自由模型的智能控制（FMBIC）系统的理论发展；(2)互补技能的加入：宾夕法尼亚州立大学在电厂智能分布式控制方面的经验与首尔国立大学在电力系统智能控制方面的经验；(3)为美国和韩国的电力工程教育提供新的见解和改进；(4)发展智能控制技术，提高电厂和电力系统的效率、经济性和安全性，以应对未来两国放松管制和电厂建设的困难。