Distribution State Estimator (DSE)

Unlike transmission systems, distribution feeders are inherently unbalanced. A DSE must model each phase (A, B, C) independently, including the neutral conductor and ground. The state vector includes complex voltages for every phase at every bus. The measurement model handles single-phase, two-phase, and three-phase measurements natively, accurately capturing mutual coupling and asymmetrical loading.

Distribution networks are chronically under-instrumented. A DSE achieves observability by injecting pseudo-measurements—statistically derived estimates for unmonitored loads. These are generated from:

• Historical smart meter data and load profiles
• Transformer capacity ratings and service type
• Time-of-day and weather-adjusted scaling factors High-variance pseudo-measurements are weighted appropriately in the estimation process to avoid corrupting the solution.

Gross measurement errors from faulty sensors, communication noise, or stuck meters must be identified and rejected. A DSE employs statistical tests on the measurement residuals:

• Chi-squared test on the weighted sum of squared residuals
• Largest normalized residual test to identify individual bad measurements
• Iterative re-weighting to suppress outliers without manual intervention This ensures the estimated state is not corrupted by a single failed sensor.

The DSE must dynamically adapt to switching operations. A topology processor analyzes the real-time status of switches, breakers, and fuses to build the bus-branch model. The estimator then maps measurements to the correct electrical nodes. This enables the DSE to function correctly through feeder reconfiguration, fault isolation, and service restoration events without manual model updates.

A modern DSE fuses heterogeneous data streams with varying latency and accuracy:

• SCADA: Slow-scan (2-5 sec) voltage, current, and power flow measurements
• AMI: Asynchronous, 15-minute interval voltage and energy data from endpoints
• PMUs: High-speed (30-120 samples/sec) synchronized phasor data
• DER telemetry: Smart inverter status and reactive power output The estimator aligns these temporally and weights them by their respective uncertainty.

Before estimation, the DSE must determine if the available measurement set renders the network algebraically observable. If not, it identifies observable islands and strategically places additional pseudo-measurements to restore observability. This process uses the rank of the measurement Jacobian matrix and ensures a unique solution exists for the entire feeder model.

The foundational mathematical engine behind most DSE implementations. WLS minimizes the sum of weighted squared residuals between measured values and estimated states.

• Assigns higher confidence weights to accurate sensors (e.g., PMUs) and lower weights to pseudo-measurements
• Iteratively solves the gain matrix equation until convergence
• Provides a maximum likelihood estimate under the assumption of Gaussian measurement noise

A critical post-processing function that prevents corrupted sensor readings from poisoning the state estimate.

• Uses the Chi-squared test on the sum of squared measurement residuals to detect gross errors
• Applies normalized residual tests to pinpoint the specific faulty measurement
• Enables automatic removal or correction of anomalous data before it feeds into Volt-VAR Optimization

A topological algorithm that determines whether the available measurement set is sufficient to uniquely estimate all bus voltage phasors.

• Identifies observable islands where state estimation is possible
• Detects critical measurements whose loss renders the system unobservable
• Guides optimal meter placement by quantifying numerical observability via the rank of the measurement Jacobian

The process of synthesizing virtual measurements to supplement scarce real-time sensor data and achieve network observability.

• Derives load allocation values from historical AMI data, transformer ratings, or customer class profiles
• Incorporates renewable generation forecasts as pseudo-injections for unmonitored solar PV systems
• Assigns high error variance to pseudo-measurements to reflect their lower confidence compared to physical sensors

The incorporation of high-resolution, time-synchronized synchrophasor data into the DSE to dramatically improve accuracy.

• Provides sub-cycle resolution voltage and current phasors with precise GPS timestamps
• Enables hybrid state estimation that fuses traditional SCADA scans with streaming PMU data
• Improves estimation of dynamic states like generator rotor angles during transient events

An advanced DSE formulation that models each phase conductor independently rather than assuming balanced conditions.

• Captures asymmetrical loading common in distribution feeders with single-phase laterals
• Estimates mutual coupling effects between phase conductors and neutral return paths
• Provides per-phase voltage estimates essential for detecting unbalanced voltage violations and optimizing three-phase capacitor bank control