Phasor Measurement Units (PMUs) are nowadays considered as the most important measuring devices in the future of power networks. In order to monitor the energy security and specifically to provide a wide-area grid visibility to the operator, the PMUs have seen significant use in North America. In the United States alone, the number of PMUs is increased from 166 networked devices in 2010 to 1,043 in 2014.
This thesis aims to address the issue of the process of using PMUs measures to achieve their full potential during their integration in the operational environment of transmission power system security. Indeed at PMU measures qualities such as accuracy, latency and visibility, there is the large volume of real-time data to know manage. To this end, based on phasor measurements, a systematic procedure development of security monitoring models in three teps is proposed.
This process is in the context of a statistical learning framework. The analyzed data set is extracted from the database EMS/SCADA of Hydro-Québec (HQ), and spread over four years, at an interval of one minute. First, following a statistical study of the database, we identified the voltage phase angles as predictors of stability and consequently, defined the limits of the phase angles where the transmission power system appears stable. Second, with random forests (RF) algorithm in the statistical learning framework, we established forecast models security margins and power transfers with defined predictors. Then, a comprehensive validation study was performed. Third, with a generalized linear model (GLM) algorithm, we developed forecast models of voltages phase angles from day-ahead to hour-ahead time frames. The mapping of these voltages phase angles predicted on models established by the FA, allowed us to anticipate the computation of new stability limit values from day-ahead to hour-ahead time frames.
Our first contribution is mainly focused on the use of a real database of HQ’s network for design and validation of our models. Our second contribution is the use of data mining techniques to propose a solution for the rapid assessment of the limits of dynamic stability. Our third contribution is the identification of the angular deviations as power flow predictors as well as associated dynamic stability limits. Furthermore, the fourth contribution is related to the design of a new approach reflecting real dynamic behavior of the transmission network from Synchrophasors data. Finally, from an economic viewpoint, this thesis contributes to the improvement of the conventional approach based on the simulation of the monitoring of the dynamic security with synchronized measurements inputs.
Our work provides many research perspectives, namely: generalize the used approach to mesh networks or not; introduce new features to the angular deviations (reactive reserves, active and reactive power injections, critics ’bars, etc.); combine RF with fuzzy logic in order to improve the prediction of phase angles voltage; and finally conduct a similar study to identify other defected models.
| Date | 31 Mar 2016 |
|---|
| Original language | French |
|---|
| Awarding Institution | - École de technologie supérieure
|
|---|
| Supervisor | Louis-A. Dessaint (Supervisor) & Innocent Kamwa (Co-supervisor) |
|---|
Kaci, A. (Author),
Dessaint, L.-A. (Supervisor) & Kamwa, I. (Co-supervisor),
31 Mar 2016Student thesis: Doctoral thesis › Doctorate in Engineering: Engineering