Jidong Chai

Wide-Area-Measurement System Development at the Distribution Level: An FNET/GridEye Example

Summary form only given. Electric power grid wide-area monitoring system (WAMS) have been extended from the transmission to distribution level. As the first WAMS deployed at the distribution level, the frequency monitoring network FNET/GridEye uses GPS-time-synchronized monitors called frequency disturbance recorders (FDRs) to capture dynamic grid behaviors. In this paper, the latest developments of monitor design and the state-of-the-art data analytics applications of FNET/GridEye are introduced. Its innovations and uniqueness are also discussed. Thanks to its low cost, easy installation and multi-functionalities, FNET/GridEye works as a cost-effective situational awareness tool for power grid operators and pioneers the development of WAMS in electric power grids.

Distributed Data Analytics Platform for Wide-Area Synchrophasor Measurement Systems

As synchrophasor data start to play a significant role in power system operation and dynamic study, data processing and data analysis capability are critical to wide-area measurement systems (WAMSs). The frequency monitoring network (FNET/GridEye) is a WAMS network that collects data from hundreds of frequency disturbance recorders at the distribution level. The previous FNET/GridEye data center is limited by its data storage capability and computation power. Targeting scalability, extensibility, concurrency, and robustness, a distributed data analytics platform is proposed in this paper to process large volume, high velocity dataset. A variety of real-time and non-real-time synchrophasor data analytics applications are hosted by this platform. The computation load is shared with balance by multiple nodes of the analytics cluster, and big data analytics tools such as Apache Spark are adopted to manage large volume data and to boost the data processing speed. Future data analytics applications can be easily developed and plugged into the system with simple configuration.

Real-time power system electromechanical mode estimation implementation and visualization utilizing synchrophasor data

Wide area measurement systems (WAMS) could provide enough information to estimate the electromechanical modes which characterize the dynamic properties of the power grid. Utilizing the measured synchrophasor data from frequency monitoring network (FNET/GridEye), this paper proposes an approach to estimate the oscillation frequency and damping regardless of ring-down or ambient condition in real-time environment. The empirical mode decomposition (EMD) is used to detrend the frequency signal, and an auto-regressing moving-average (ARMA) model is used to describe the time series data and the modified yuler walker (MYW) algorithm is used to estimate the AR parameters. Additionally, this paper describes the details of the implementation of a real-time monitoring website showing the estimated dominant frequency and damping from the four interconnections in North America power grid.

Application of wide area power system measurement for digital authentication

The Electric Network Frequency (ENF) analysis of a digital recording is a two-step process. First, an ENF extraction algorithm is applied to the recording and the second step is comparing the extracted target sequence to subsequences contained in a grid reference database. In this paper, a new comparison algorithm is created that can remove the effects of oscillator errors from comparisons. Another study is to have a better understanding of the phenomenology by which sinusoidal grid signals find their way onto digital recordings. In addition, a study that uses both frequency and phase angle is done to determine whether tampering has occurred in digital recordings. Besides being able to determine the recording time, a study that distinguishes the frequency from different locations is performed to possibly determine the location of the recording.

Measurement-based power system dynamic model reductions

Interconnected power systems experienced a significant increase in size and complexity. It is computationally burdensome to represent the entire system in detail to conduct power system analysis. Therefore, the model of the study system must be retained in detail while the external system can be reduced using system reduction techniques. This paper proposes a measurement-based dynamic equivalent in order to increase both model accuracy and simulation speed. The proposed method uses a set of measurements at the boundary nodes between the study area and external area for model parameter identification. Case studies demonstrate that the measurement-based technique can capture the main system behaviors accurately and improve computational efficiency.

Comparison of MIMO system identification methods for electromechanical oscillation damping estimation

The paper presents two groups of measurement based multi-inputs, multi-outputs (MIMO) system identification approaches, which are subspace state-space identification and transfer function identification. The purpose is to compare the performance and features of these methods in terms of electromechanical oscillation damping estimation. The selected methods are validated by using ambient data and ring-down data in the 16-machine 68-bus Northeast Power Coordinating Council (NPCC) system. Meanwhile, the advantages and limitations of technologies are described.

An ensemble solar power output forecasting model through statistical learning of historical weather dataset

Due to its economical and environmental benefits to society and industry, integrating solar power is continuously growing in many utilities and Independent System Operators (ISOs). However, the intermittent nature of the renewable energy brings new challenges to the system operators. One key to resolve this problem is to have a ubiquitously efficient solar power output forecasting system, so as to help enhance system reliability, improve power quality, achieve better generation scheduling and formulate superior bidding strategies in market. This paper proposes an ensemble learning method to forecast solar power output, combining the state-of-art statistical learning methods. The performance of the model is evaluated through comparing with a benchmark with different metrics, and the numerical results validate the effectiveness of the model.