Loading Events

« All Events

  • This event has passed.

Ramtin Khalili’s PhD Dissertation Defense

December 5, 2022 @ 10:00 am - 11:00 am


State estimation is a critical application in energy management systems. Due to the increased penetration of inverter-based resources, installed advanced infrastructure at all voltage levels, and unconventional loads like electric vehicle charging stations, a three-phase state estimator formulation is essential. The first issue is the convoluted formulation and modeling techniques that are required in three-phase systems studies. Moreover, the size of network matrices expanded, which makes the analysis computationally costly. This dissertation addresses this by proposing a new decoupled state estimation method. The idea is to exploit the linearity of measurement equations, decompose the three-phase coupled equations into three independent modal measurement equations, perform the state estimation independently for each mode, and finally reconstruct the three-phase quantities. This method is applicable to both radial and meshed three-phase networks. Furthermore, multi-phase structures can be handled by the new estimator, which makes the approach practical when monitoring mixed-phase feeder sections is of interest.

While utilities are investing in expanding the grid and installing more PMUs, there might not be enough PMUs to make the network observable in all networks, especially at lower voltage levels. So, PMU-based linear state estimators are not always feasible. On the other hand, SCADA measurements are available with adequate redundancy in most networks. However, SCADA-based state estimation is nonlinear, which brings various problems like divergence issues and significant CPU times. The computational complexity will be even worse if the three-phase state estimation is formulated based on SCADA measurements due to their nonlinear nature, which makes modal decoupling impossible. So, a new linear formulation has been proposed for both the positive-sequence and three-phase networks based on conventional measurements. This approach converts the nonlinear recursive problem into an iterative linear state estimation problem.

The inherent assumption in most of the state estimators is a perfect network model. However, network parameter errors are susceptible to errors that can bias the state estimation solution. This can deceive the existing bad data tools as parameter errors appear as if multiple interacting measurement errors occur locally. So, a two-stage method is proposed for parameter error identification and correction for large three-phase networks. A systematic PMU placement strategy is also proposed to ensure the detectability of parameter errors. The benefits of multi-area state estimation are demonstrated for the deregulated power grids for monitoring the local and boundary areas. It has also shown promising results in increasing the efficiency of state estimation using a distributed framework. Parameter and measurement errors can remain undetected as a result of weakened measurement redundancy on the boundaries. However, boundary errors in the area boundaries will be detected due to measurement consolidation at the coordination level.

Prof. Ali Abur (Advisor)
Prof. Bahram Shafai
Prof. Mahshid Amirabadi