Regularized Learning Enabled Monitoring and Control for Wind Power Systems

风电系统的常规学习监控和控制

基本信息

批准号：
1362513
负责人：
Eunshin Byon
金额：
$ 32.5万
依托单位：
Regents of the University of Michigan - Ann Arbor
依托单位国家：
美国
项目类别：
Standard Grant
财政年份：
2014
资助国家：
美国
起止时间：
2014-05-01 至 2018-04-30
项目状态：
已结题

来源：
https://www.nsf.gov/awardsearch/showAward?AWD_ID=1362513&HistoricalAwards=false
关键词：
Regularized Learning Enabled Monitoring Control

项目摘要

The objective of this project is to develop new monitoring and control strategies for enhancing wind turbine reliability so that operations and maintenance costs of wind energy can be reduced. In wind power systems, the "wind input"-to-"turbine response" relationship is nonstationary, due to both internal (e.g., system's degradation) and external (e.g., surface contamination on blades) changes. This nonstationary dependency causes significant technological challenges in managing the health and performance of wind turbines. This project will develop a new regularized learning method to characterize the time-varying dependency among system variables so that changes in a turbine system can be tracked and predicted. Subsequently, a statistical monitoring method with adaptive control limits will be devised to signal the occurrence of anomalies. Based on the results from the regularized learning process, an adaptive control strategy will be developed to mitigate excessive and undesired mechanical stresses on turbine subsystems in an effort to prevent or slow the deterioration process. A new wireless structural health monitoring (SHM) system supporting real-time, embedded data processing will be advanced to track the behavior, performance and health of operational turbines. The outcomes of this research will facilitate the wind industry's smooth transition from using rudimentary diagnosis and control techniques to the use of sophisticated and integrative monitoring and control technologies. The new monitoring method will enable timely detection of anomalies while reducing the false alarms. Optimally determined control parameters will balance between power production and stress levels in an effort to extend a turbine?s service life. While using wind turbines as the primary application target, the methodology is applicable to other engineering systems subject to dynamic operating conditions including civil infrastructure systems. This project will contribute toward the preparation of a future workforce in the field of renewable energy and sustainability through an array of mechanisms including the integration of under-represented students in the STEM field into renewable energy research, opportunities for students to interact with national laboratories, and to be engaged with other domestic and international research groups.

该项目的目的是制定新的监控和控制策略，以提高风力涡轮机的可靠性，以便可以降低风能的运营和维护成本。在风能系统中，由于内部（例如系统的降解）和外部（例如，叶片上的表面污染）变化，“风输入” - 对“涡轮响应”的关系是非平稳的。这种非平稳的依赖性在管理风力涡轮机的健康和性能方面引起了重大的技术挑战。该项目将开发一种新的正规学习方法，以表征系统变量之间的时间变化依赖关系，以便可以跟踪和预测涡轮系统的变化。随后，将设计具有自适应控制限制的统计监测方法，以发出异常的发生。根据正规学习过程的结果，将制定一种自适应控制策略，以减轻涡轮副系统上的过度和不希望的机械应力，以防止或减慢恶化过程。一个新的无线结构健康监测（SHM）系统，支持实时，嵌入式数据处理的系统，以跟踪操作涡轮机的行为，性能和健康。这项研究的结果将有助于风力行业从使用基本诊断和控制技术到使用复杂和综合监测和控制技术的平稳过渡。新的监视方法将在减少错误警报的同时及时检测异常。最佳确定的控制参数将在功率产生和压力水平之间平衡，以延长涡轮机的使用寿命。在使用风力涡轮机作为主要应用程序目标时，该方法适用于其他工程系统，受到包括民用基础设施系统在内的动态操作条件。该项目将通过一系列机制在可再生能源和可持续性领域的未来劳动力方面做出贡献，包括将不足的学生在STEM领域的融合到可再生能源研究中，这是学生与国家实验室互动的机会，并与其他国内和国际研究小组互动。