Neural Network Observers for Tracking Synchronous Machine Parameters and Incipient Failure Detection

俄亥俄州立大学、亚利桑那州立大学和亚利桑那州公共服务公司提议开展合作研究，开发同步发电机的在线参数识别和早期故障检测技术。该程序将基于开发的基于神经网络的观测器来跟踪动态扰动下的电机阻尼器电流、参数、励磁绕组与定子匝数比，并识别电机参数、电机中性电流和扭矩角的特征。发电机关键指标的在线参数跟踪和签名识别的影响将对在线检测和识别励磁绕组退化的早期故障产生重大影响。例如，亚利桑那州公共服务公司Four-Corner单元被迫停电的经济成本约为每天20万美元。早期故障的在线检测将有助于发电机有序停机检修。预计该项目将影响大型同步发电机原有的检修计划。此外，在接下来的几十年里，由于几乎没有新的输电，美国的电力系统将面临大规模电力传输能力的瓶颈。随着大规模电力传输和第三方接入的规模不断扩大，提高现有设施的利用率将需要更密切地审查系统稳定性要求。对机器参数的跟踪将有助于更准确的稳定性研究。因此，拟议的研究对于提高输电能力和推迟新建设的需要，使电力系统的运行更接近稳定极限，具有相当重要的经济影响。

ECS-9722844 Keyhani The Ohio - State University, Arizona State University and Arizona Public Service Company propose to undertake a collaborative research effort in development of technology for on-line parameter identification and incipient failure detection of synchronous generators. The procedure will be based on the development of neural network based observers to track the machine damper currents, parameters, field winding to-stator turns ratio in dynamic disturbances and identify the signatures of the machine parameters, the machine neutral current and torque angle. The impact of on-line parameter tracking and signature identification of the generator key indicators will have significant impact on-line incipient failure detection and identification of field winding degradation. For example, the economic cost of a forced outage of The Arizona Public Service company Four-Corner unit is approximately two hundred thousand dollars a day. The on-line detection of incipient failure will facilitate the orderly shutdown of generators for repair. It is expected that the project will impact the original maintenance scheduling, of large synchronous generators. Furthermore, in the next decades, the power systems of the United States will face bottlenecks of bulk power transfer capabilities because of the virtual absence of new transmission. The increased utilization of existing facilities to cop with growing magnitude of bulk power transfers and third-party access will require a closer look at system stability requirements. The tracking of the machine parameters will facilitate more accurate stability studies. Therefore, the economic impact of the proposed research is quite significant in permitting operation of power systems much closer to the stability limit by increasing the transfer capabilities and postponing the need for new construction