Center for Information Technology Research in the Interest of Society (CITRIS)

With the unprecedented growth of renewable resources, electric vehicles, and controllable loads, power system has been incorporating increasing amount of unconventional generations and loads. As a consequence, significant dynamic and stochastic power flow are introduced into distribution network, requiring high resolution monitoring technology and agile decision support techniques for system diagnosis and control. In this paper, we discuss the application of micro-synchrophasor measurement unit (µPMU) for power distribution network monitoring, and we propose a novel data-driven method, namely Ensembles of Bundle Classifier (EBC), for event detection. The main idea is: multiple classifiers are learned each with a short slot of µPMU measurement generated by a single event. Then their decisions are combined with a “winner-takes-all” scheme. This framework naturally resolves the challenging issue of heterogeneity in the high resolution µPMU data, and significantly outperforms classic data-driven event detection methods. In this paper, the proposed framework is tested on an actual distribution network with µPMUs, and is compared to other state-of-the-art methods. The result justifies the effectiveness of EBC as a promising tool to improve the security and reliability of distribution network.

Optimization of distributed power assets is a powerful tool that has the potential to assist utility efforts to ensure customer voltages are within pre-defined tolerances and to improve distribution system operations. While convex relaxations of Optimal Power Flow (OPF) problems have been proposed for both balanced and unbalanced networks, these approaches do not provide universal convexity guarantees and scale inefficiently as network size and the number of constraints increase. In balanced networks, a linearized model of power flow, the LinDistFlow model, has been successfully employed to solve approximate OPF problems quickly and with high degrees of accuracy. In this work, an extension of the LinDistFlow model is proposed for unbalanced distribution systems, and is subsequently used to formulate an approximate unbalanced OPF problem that uses VAR assets for voltage balancing and regulation. Simulation results on the IEEE 13 node test feeder demonstrate the ability of the unbalanced LinDistFlow model to perform voltage regulation and balance system voltages.

Power system has been incorporating increasing amount of unconventional generations and loads such as renewable resources, electric vehicles, and controllable loads. The induced short term and stochastic power flow requires high resolution monitoring technology and agile decision support techniques for system diagnosis and control. In this paper, we discuss the application of micro-phasor measurement unit (μPMU) for power distribution network monitoring, and study learning based data-driven methods for abnormal event detection. We first resolve the challenging problem of information representation for the multiple streams of high resolution μPMU data, by proposing a pooling-picking scheme. With that, a kernel Principle Component Analysis (kPCA) is adopted to build statistical models for nominal state and detect possible anomalies. To distinguish event types, we propose a novel discriminative method that only requires partial expert knowledge for training. Finally, our methods are tested on an actual distribution network with μPMUs, and the results justifies the effectiveness of the data driven event detection framework, as well as its potentials to serve as one of the core algorithms to ensure power system security and reliability.

Modern society relies heavily upon complex and widespread electric grids. In recent years, advanced sensors, intelligent automation, communication networks, and information technologies (IT) have been integrated into the electric grid to enhance its performance and efficiency. Integrating these new technologies has resulted in more interconnections and interdependencies between the physical and cyber components of the grid. Natural disasters and man-made perturbations have begun to threaten grid integrity more often. Urban infrastructure networks are highly reliant on the electric grid and consequently, the vulnerability of infrastructure networks to electric grid outages is becoming a major global concern. In order to minimize the economic, social, and political impacts of large-scale power system outages, the grid must be resilient in addition of being robust and reliable. The concept of a power system’s cyber-physical resilience centers around maintaining critical functionality of the system backbone in the presence of unexpected extreme disturbances. Resilience is a multidimensional property of the electric grid; it requires managing disturbances originating from physical component failures, cyber component malfunctions, and human attacks. In the electric grid community, there is not a clear and universally accepted definition of cyber-physical resilience. This paper focuses on the definition of resilience for the electric grid and reviews key concepts related to system resilience. This paper aims to advance the field not only by adding cyber-physical resilience concepts to power systems vocabulary, but also by proposing a new way of thinking about grid operation with unexpected extreme disturbances and hazards and leveraging distributed energy resources. The concepts of service availability and quality are not new, but many recognize the need of resilience in maintaining essential services to critical loads, for example to allow home refrigerators to operate for food conservation in the aftermath of a hurricane landfall. By providing a comprehensive definition of power system resilience, this paper paves the way for creating appropriate and effective resilience standards and metrics.

Power flow analysis is one of the tools that is required in most of the distribution system studies. An important characteristic of distribution systems is the load unbalance in the phases and a three-phase power flow analysis is needed. In this paper, a three-phase linear power flow (3LPF) formulation is derived based on the fact that in a typical distribution system, voltage angles and magnitudes vary within relatively narrow boundaries. The accuracy of the proposed 3LPF is verified using several test cases. Potential applications of the proposed method are in distribution systems state estimation and volt-VAR optimization.

The increase in high-precision, high-sample-rate telemetry timeseries poses a problem for existing timeseries databases which can neither cope with the throughput demands of these streams nor provide the necessary primitives for effective analysis of them. We present a novel abstraction for telemetry timeseries data and a data structure for providing this abstraction: a timepartitioning version-annotated copy-on-write tree. An implementation in Go is shown to outperform existing solutions, demonstrating a throughput of 53 million inserted values per second and 119 million queried values per second on a four-node cluster. The system achieves a 2.9x compression ratio and satisfies statistical queries spanning a year of data in under 200ms, as demonstrated on a year-long production deployment storing 2.1 trillion data points. The principles and design of this database are generally applicable to a large variety of timeseries types and represent a significant advance in the development of technology for the Internet of Things.

The introduction and deployment of cheap, high precision, high-sample-rate next-generation synchrophasors en-masse in both the transmission and distribution tier – while invaluable for event diagnosis, situational awareness and capacity planning – poses a problem for existing methods of phasor data analysis and storage.

Addressing this, we present the design andimplementation of a novel architecture for synchrophasor data analysis on distributed commodity hardware. At the core is a newfeature-rich timeseries store, BTrDB.

Capable of sustained writes and reads in excess of 16 million points per second per cluster node, advanced query functionality and highly efficient storage, this database enables novel analysis and visualization techniques.

Leveraging this, a distillate framework has been developed that enables agile development of scalable analysis pipelines with strict guarantees on result integrity despite asynchronous changes in data or out of order arrival. Finally, the system is evaluated in a pilot deployment, archiving more than 216 billion raw datapoints and 515 billion derived datapoints from 13 devices in just 3.9TB.

We show that the system is capable of scaling to handle complex analytics and storage for tens of thousands of next-generation synchrophasors on off-the-shelf servers.

This paper proposes a data-driven approach to detect the switching actions and topology transitions in distribution networks. It is based on the real time analysis of time-series voltages measurements. The analysis approach draws on data from high-precision phasor measurement units (μPMUs or synchrophasors) for distribution networks. The key fact is that time-series measurement data taken from the distribution network has specific patterns representing state transitions such as topology changes. The proposed algorithm is based on comparison of actual voltage measurements with a library of signatures derived from the possible topologies simulation. The IEEE 33-bus model is used for the algorithm validation.

Network topology in distribution networks is often unknown, because most switches are not equipped with measurement devices and communication links. However, knowledge about the actual topology is critical for safe and reliable grid operation. This paper proposes a voting-based topology detection method based on micro-synchrophasor measurements. The minimal difference between measured and calculated voltage angle or voltage magnitude, respectively, indicates the actual topology. Micro-synchrophasors or micro-Phasor Measurement Units (μPMU) are high-precision devices that can measure voltage angle differences on the order of ten millidegrees. This accuracy is important for distribution networks due to the smaller angle differences as compared to transmission networks. For this paper, a microgrid test bed is implemented in MATLAB with simulated measurements from μPMUs as well as SCADA measurement devices. The results show that topologies can be detected with high accuracy. Additionally, topology detection by voltage angle shows better results than detection by voltage magnitude.