Abstract Details

<< Back to Schedule

10/14/2015  |   3:15 PM - 4:00 PM   |  Atlantic II

A Data-Driven Approach to Analyze the Spatial and Temporal Variations in the Power Distribution Grid

With the development of Distributed Automation Systems (DAS) and Geographic Information Systems (GIS), it is becoming an emerging topic to integrate the technology from these two fields. Based on this integrated intelligent system, this work presents a data-driven method for the analysis of spatial and temporal features of the power grid load, taking advantaging of the massive history data stored in power company’s databases, including history on-line sampling data and GIS information. Our project aims at providing a stable and healthy distribution grid by optimizing its configuration through the analysis of past sampling data. For example, a feeder line with minimum fluctuation is considered as steady and robust. Same thing would happen to a distribution grid. Generally, the less fluctuated a distribution grid is, the more controllability and longer lifespan it will have. Nowadays, DAS applications in distribution grid have already achieved remote and instant operation monitoring and controlling of the topological structure, making optimizing network configuration in large distribution systems possible. Most distribution grids in cities have periodic load characteristics, yearly and daily. In this project, we took usage reads from remote terminal units every 15 minutes. That allows us to make plans for a year, a season or a special day, such as Christmas Eve and Chinese New Year. For example, we can use two three-month data from last two springs to optimize the configuration plan in this spring. On the other hand, as small-scale renewable energy is developing quickly, more and more small-scale power sources are widely applied and directly connected to the distribution network. However, these small-scale power sources usually have more random fluctuation and will change the network configuration, thus providing an unpredictable factor to the whole distribution network. Therefore, small adjustments will be made according to past-day or past-week reads. Our analysis method starts by sifting the trend components sifted out from the daily load curves at every spatial point in the DAS. Auto-regressive Moving Average Module (ref.) is used to smooth the data. The temporal probability distribution of the resulting trend components is analyzed transversely (same time in different days) and longitudinally (different time at same day). Finally, the correlation between load temporal distributions at different points in the grid is computed, taking into consideration the spatial relationships between these points. Our analysis results in obtaining the spatial-temporal distribution features of power grid load can be obtained. This approach was applied to a typical feeder line of a large-scale city distribution grid in a major Chinese city. This method can provide useful references to optimize the power grid system for better planning, operation and intelligent control, which we plan to address as in our future work.

Presentation:
This presentation has not yet been uploaded.

Handouts:
No handouts have been uploaded.

Qiwei Zheng (Primary Presenter,Author), University of Connecticut, qiwei.zheng@uconn.edu;
Qiwei Zheng is an undergraduate student perusing dual B.S. degree in Computer Science and Geography at the University of Connecticut. Expected graduation time: May 2016. Research interests: data visualization, spatial analysis, big-data.

Peng Xiao (Co-Author), University of Connecticut, peng.xiao@uconn.edu;
Peng Xiao received the B.S. degree in automation and M.S. degree in electrical engineering from Northwestern Polytechnical University, Xi'an, China, in 2011 and 2014, respectively. Now he is working towards Ph.D. degree in Computer Science and Engineering Department in University of Connecticut. His research interest includes: sensor network security, design of secure and energy-efficient underwater sensor nodes, computer architecture.

2013 Sponsors: IEEE and IEEE Computer Society