Opportunistic Hybrid Communications Systems for Distributed PV Coordination

Objective 1

Develop a novel publish-subscribe pattern-based communications network in order to better suit the needs of the monitoring and control of distributed PV generators.

Publish-subscribe networks offer a number of benefits for distributed PV systems, including better scalability than traditional client-server operation, and a loose coupling that allows for flexibility with changing system topology. The content of the events published by distributed generators should follow a standardized format to guarantee interoperability. The Common Information Model (CIM) is a standard developed by the electric power industry, and adopted by the International Electrotechnical Commission (IEC), that facilitates information exchange in electrical networks. CIM defines an object-oriented model that contains a complex class structure and allows for information exchange among different sectors of the electricity system, including: energy management systems, distribution management systems, and energy markets. CIM's object-oriented approach accommodates for the addition of new components, which will be necessary in the ever-evolving electricity landscape. Fortunately, the hierarchical structure of CIM can also facilitate the design of content filter aggregation algorithms in the publish-subscribe network.

Objective 2

Develop transformative, novel decentralized state estimation and prediction schemes with dimensionality reduction algorithms that are resilient to measurement outliers and missing measurements.

State estimation and prediction schemes are a critical component of the proposed communications framework because they can greatly reduce the number of measurements (and hence communications) necessary, as well as providing estimates where no data is currently available. Big data analytics only become necessary if the prediction algorithm demands do not scale well with the ability to efficiently store, transfer, and access large amounts of PV and network data. The use of robust machine learning algorithms will enable large-scale application of the state estimation techniques with computational bottlenecks minimized. It is not only the scale of the data that can give rise to formidable computational complexity, but faster than real-time computation is also critically important for enhancing the value of prediction. Hence, we will also pursue a sparse system representation as well as distributed algorithms to facilitate real-time execution of the information exchange.

Objective 3

Rigorous validation of the communications systems developed through hardware-in-the-loop testing and communications systems coupled with integrated distribution-transmission grid joint simulation.

A two-pronged validation approach is planned for the proposed communications system in order to ensure that it is ready for pilot-scale demonstration and deployment. First, hardware-in-the-loop testing will ensure that the system architecture, middleware, and algorithmic layers can operate with the necessary speed to fulfill system response time goals for both normal and contingency event operational states. This testing will also assure that the interoperability design goals are met through the physical connection with real PV inverters. The second means of system testing is through a combination with the Integrated Grid Modeling System (IGMS) software developed at NREL for combined distribution-transmission system simulation. This testing will ensure that the scalability goals are met, and will provide a platform for certifying not only that the communications systems function as designed but tests implications of communications system behavior on the physical power system.