IEEE Power & Energy Magazine - September/October 2015 - 66

Substation state estimation
provides an effective alternative by
prefiltering bad data.

employed to realize generator parameter identification. The
input PMU curves are the generator terminal and excitation
voltages, with the objective of minimizing the error between
the simulated generator current and the current measured by
the PMU. The identified generator parameters include synchronous reactance, transient reactance, subtransient reactance for direct and quadrature axes, and their corresponding
time constants. The moment of inertia can also be identified.

Monitoring and Analyzing
Real-Time Power System Dynamics
Substation State Estimation
Traditional state estimation (SE) is implemented in the electrical power control center using measurements remotely
collected from the supervisory control and data acquisition
(SCADA) system. Identifying topology errors is a major
challenge for traditional SE.
Substation state estimation (SSE) provides an effective
alternative by prefiltering bad data using a detailed model and
redundant local measurements. In SSE, the zero impedance
subnetwork, with circuit breakers (CBs) as zero impedance
branches, is handled at each voltage level separately. In traditional SE, the topology must be determined first to build
nodal equations using Ohm's law. With SSE, only the analog
measurements from the PMU and SCADA systems are used
to first solve the local SE for each voltage level using Kirchhoff's current law. The estimated power flows across each CB
are then used to check for potentially bad CB status data. In
this process, the detected bad analog and topology data are

1.05%
Bad Analog Data Rate
Bad Digital Data Rate

1.00%
0.80%
0.60%
0.40%
0.20%
0.00%

0.41%

0.08%
0.04%

Before SSE

After SSE

figure 6. A comparison of bad analog and digital data
rates before and after SSE.
66

ieee power & energy magazine

decoupled, making them minimally susceptible to the measurement errors of CB status. Bad topology data can be further
identified using methods such as a hypothesis test.
With a small problem size, the complete utilization of
fast-sampled PMU measurements becomes possible. This
benefits SSE in terms of accuracy and reliability. The
dynamic transformation of topology, power flow, and phasor
changes can be captured with the PMU. A three-phase system imbalance can also be monitored through three-phase
substation modeling.
Statistical analysis via Monte Carlo simulation shows that
the rates of bad analog and digital data can be significantly
reduced by SSE (Figure 6). The local SSE results are sent to the
control center to further enhance the reliability of the global SE.
In China, the SSE method was initially implemented in
four 500-kV experimental substations in the power system
of East China. Recently, it has also been applied to other
power companies' substations. A field test reveals that data
quality can be greatly improved with SSE using hybrid
measurements from the WAMS/SCADA system. The average computing time for SSE is 20 ms, which is feasible for
real-time application.

Identifying and Assessing
Low-Frequency Oscillation
Low frequency oscillation (LFO) is always a stability threat
for China's interconnected power grids. A comprehensive
solution for monitoring and assessing LFO stability has been
employed in a number of the country's power dispatching
centers. The solution includes three functions:
✔ real-time warning regarding harmful oscillation
based on ringdown signal identification
✔ early warning about small oscillation based on statistics
✔ identification of oscillation modes and mode shapes
based on the power systems' ambient data.
With this solution, weakly damped oscillation modes can be
detected early, and dispatchers can make effective decisions
based on the identified information.
The Prony algorithm is the basic method for detecting and identifying oscillation based on ringdown signals
after disturbances. This algorithm has been implemented
in many dispatching systems. A number of advanced algorithms such as the Hilbert-Huang transform and estimation of signal parameters via rotational invariance techniques have also been tested in the provincial power grids
of Zhejiang and Sichuan.
september/october 2015

Table of Contents for the Digital Edition of IEEE Power & Energy Magazine - September/October 2015