What Is State Estimation? SCADA, EMS and the State Vector
- Part 1 — What Is State Estimation? SCADA, EMS and the State Vector ← you are here
- Part 2 — The Measurement Model: z = h(x) + e
- Part 3 — Weighted Least Squares (WLS) State Estimation
- Part 4 — Building the Jacobian Matrix
- Part 5 — Observability Analysis and Measurement Placement
- Part 6 — Bad Data Detection and Identification
- Part 7 — Topology Error Processing
- Part 8 — Linear and DC State Estimation
- Part 9 — PMU-Based and Hybrid State Estimation
- Part 10 — PowerFactory + Python Implementation (IEEE 14-Bus)
Every power system operator needs to know, right now, what is happening in the network — the voltage at every bus, the power flowing through every line, the reactive dispatch of every generator. Without this information there can be no automatic protection, no economic dispatch, no contingency analysis, and no safe real-time operation.
The problem is that you can never measure all of these quantities directly and accurately at the same time. Sensors have noise. Communications fail. Meters drift. Voltage angles cannot be measured at all by traditional instruments. State estimation is the mathematical process that takes a large, noisy, redundant, incomplete set of real-time measurements and computes the best possible estimate of the system state.
1. The Power System State Vector
The state of an AC power system is the minimal set of variables from which every other electrical quantity can be computed using the network equations. For an n-bus system with bus 1 as the reference (θ1 = 0 by definition):
This gives 2n − 1 state variables — n−1 voltage angles plus n voltage magnitudes. Once x is known, every line flow, power injection, transformer loading, and reactive power balance can be calculated from the standard power flow equations.
| State variable | Symbol | Units | Count (n-bus system) |
|---|---|---|---|
| Voltage angles (all buses except reference) | θi | rad or deg | n − 1 |
| Voltage magnitudes (all buses) | |Vi| | pu | n |
| Total | 2n − 1 |
Traditional instrument transformers measure RMS magnitudes of voltage and current, not phasors. Phasor angle information is encoded in the power flow equations and can only be inferred from those — or measured directly by PMUs with GPS time-stamping (Part 9).
2. Measurement Types in SCADA
Traditional SCADA collects analogue measurements from Remote Terminal Units (RTUs) at substations every 2–10 seconds. The measurement types available to a state estimator are:
| Type | Symbol | Relates to h(x) | Typical σ (pu) |
|---|---|---|---|
| Active power flow | Pij | Nonlinear in θ, |V| | 0.008–0.02 |
| Reactive power flow | Qij | Nonlinear in θ, |V| | 0.008–0.02 |
| Active power injection | Pi | Nonlinear in θ, |V| | 0.008–0.02 |
| Reactive power injection | Qi | Nonlinear in θ, |V| | 0.008–0.02 |
| Voltage magnitude | |Vi| | Linear (= state variable) | 0.002–0.005 |
| Current magnitude | |Iij| | Nonlinear | 0.010 |
| PMU voltage phasor | Vi∠θi | Linear (phasor) | ~0.001 |
| PMU current phasor | Iij∠φij | Linear (phasor) | ~0.001 |
Traditional SCADA measures magnitudes of power and voltage but not angles. Voltage angles can only be inferred through power flow measurements or measured directly by PMUs. This asymmetry is the fundamental reason state estimation cannot be replaced by simple meter reading.
3. The SCADA and EMS Architecture
State estimation sits inside the Energy Management System (EMS) at the transmission system operator’s control centre. The data flow from field to application:
- Field measurements: CTs and VTs at substations measure currents and voltages; transducers compute P, Q, |V|
- RTUs / IEDs: convert analogue signals to digital; transmit via IEC 60870-5, DNP3, or IEC 61850
- SCADA front-end: receives data, performs validity checks, stores in real-time database every 2–10 s
- Network Topology Processor: converts breaker status to bus-branch network model
- State Estimator: runs every 1–5 minutes; outputs estimated voltage magnitudes and angles at every bus
- Downstream applications: Optimal Power Flow, Security Assessment, Contingency Analysis, Voltage Stability Monitoring — all start from the estimated state
Optimal power flow, N-1 contingency analysis, and voltage stability monitoring all require a physically consistent, converged system state as their starting point. Raw SCADA data is noisy and incomplete — state estimation filters the noise and fills the gaps that all downstream applications need.
4. Observability
A power system is observable if the available measurements are sufficient to uniquely determine all 2n−1 state variables. The necessary condition is:
But this is only necessary, not sufficient. Measurements must also be well-placed: if an entire region of the network has no measurements, it forms an unobservable island regardless of how many measurements exist elsewhere.
The redundancy ratio ρ = m/(2n−1) measures measurement richness. Typical transmission systems operate at ρ ≈ 2–3. This redundancy is essential for bad data detection (Part 6).
| Situation | Redundancy ρ | Effect on SE |
|---|---|---|
| Barely observable | ρ = 1.0 | SE runs; cannot detect bad data |
| Normal operation | ρ = 2–3 | Reliable SE; bad data detectable |
| High redundancy | ρ > 3 | Very robust; multiple bad data detectable |
| Unobservable island | ρ < 1 locally | SE fails or unreliable |
5. State Estimation vs. Load Flow
| Property | Load Flow | State Estimation |
|---|---|---|
| Input | Scheduled P, Q, |V| at buses | Noisy redundant real-time measurements |
| Output | Exact solution given perfect inputs | Statistical best estimate |
| System size | Square: exactly 2n−1 equations | Overdetermined: m > 2n−1 |
| Handles noise | No | Yes — noise averaged over redundant measurements |
| Handles bad data | No | Yes — chi-squared test and LNR (Part 6) |
| Real-time use | Planning, offline studies | Online every 1–5 min in EMS |
| Algorithm | Newton-Raphson | Gauss-Newton WLS (Part 3) |
6. The SE Problem — Preview
The measurement model (derived fully in Part 2):
State estimation finds the x that minimises the weighted sum of squared residuals:
where W = R−1 and R = diag(σ12, …, σm2). The solution — the Weighted Least Squares (WLS) estimator — is derived in Part 3.
7. Series Roadmap
| Part | Topic | What you will gain |
|---|---|---|
| 1 (this article) | Introduction, SCADA, EMS, observability | Why SE exists and what it produces |
| 2 | Measurement model z = h(x) + e | Write h(x) for any measurement type |
| 3 | Weighted Least Squares derivation | Implement the core WLS algorithm |
| 4 | Building the Jacobian matrix H | Compute H analytically for all measurement types |
| 5 | Observability and measurement placement | Diagnose and fix observability problems |
| 6 | Bad data detection and identification | Detect and remove faulty measurements |
| 7 | Topology error processing | Handle network model errors |
| 8 | Linear and DC state estimation | Use fast approximate SE methods |
| 9 | PMU-based and hybrid SE | Integrate synchrophasor measurements |
| 10 | PowerFactory + Python implementation | Full working code on IEEE 14-bus |
References
- A. Abur and A. G. Exposito, Power System State Estimation: Theory and Implementation, Marcel Dekker, 2004.
- A. Monticelli, State Estimation in Electric Power Systems: A Generalized Approach, Kluwer Academic, 1999.
- F. C. Schweppe and J. Wildes, “Power system static-state estimation, Parts I, II, III,” IEEE Trans. Power Apparatus and Systems, vol. PAS-89, pp. 120–135, Jan. 1970.
- P. Kundur, Power System Stability and Control, McGraw-Hill, 1994 — Chapter 12 covers EMS architecture.