⚡ electrical engineering

Geometric Time-Domain Identification of Three-Phase Load Equivalents from Terminal Measurements

This paper presents a geometric time-domain method that identifies three-phase load equivalents from terminal voltage and current measurements by interpreting waveforms as trajectories in Euclidean signal spaces, extending single-phase formulations to multi-wire systems while incorporating passivity constraints and robustly handling data insufficiency through explicit identifiability tests.

Original authors: Francisco G. Montoya, Francisco de Leon, Francisco M. Arrabal-Campos, Alfredo Alcayde

Published 2026-06-08

📖 4 min read☕ Coffee break read

CC BY 4.0

Original authors: Francisco G. Montoya, Francisco de Leon, Francisco M. Arrabal-Campos, Alfredo Alcayde

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

Imagine you are a detective trying to figure out what's inside a locked black box. You can't open the box, but you can see the electricity flowing in and out of it. You have a voltmeter and an ammeter, and you watch how the voltage and current wiggle and dance over time.

This paper presents a new, clever way to solve that mystery for three-phase power systems (the kind used to run factories and big buildings). Instead of guessing or using complex frequency charts, the authors use geometry to "see" the invisible parts inside the box.

Here is how the method works, broken down into simple concepts:

1. The "Dance Floor" Analogy

Think of the electricity flowing through the wires not as numbers on a screen, but as a dancer moving across a floor.

In a simple system, the dancer might just move in a straight line.
In a complex system with resistors, inductors, and capacitors, the dancer moves in a specific, twisting 3D shape.

The authors say: "If we can map the exact shape of this dance, we can mathematically reverse-engineer the dancer's moves to figure out what kind of shoes (resistors), springs (inductors), or balloons (capacitors) they are wearing."

2. The Three-Wire Puzzle

The paper focuses on a tricky situation called a three-wire system.

The Easy Case (Four-Wire): Imagine a house with a neutral wire. You can measure the voltage of each wire against the ground. It's like having a map with a clear "North."
The Hard Case (Three-Wire): In many industrial setups, there is no neutral wire. You only have the three live wires. It's like trying to navigate a ship without a compass or a map, only knowing the distance between the three masts. The math gets messy because the wires are all tied together by strict rules (Kirchhoff's laws).

The authors developed a special geometric trick to solve this "no-compass" puzzle. They proved that even without a neutral wire, you can still figure out the internal components if you look at the shape of the voltage and current waves correctly.

3. The "Snapshot" Method (Time Windows)

You can't just look at one single instant of electricity; it's too noisy and chaotic. Instead, the method takes short snapshots (called "windows") of the data, lasting about half a second to two seconds.

The Camera: The camera takes a picture of the "dance" during that short window.
The Filter: It checks if the dance is interesting enough. If the dancer is just standing still or moving in a boring, repetitive circle (like a pure sine wave), the camera says, "Not enough information here." It refuses to guess.
The Calculation: If the dance is complex enough (full of twists and turns), the camera uses a matrix equation (a fancy spreadsheet of math) to calculate the exact values of the resistors, inductors, and capacitors inside the box.

4. The "Energy Bill" Check

This is the most important safety feature of the method.
Once the computer guesses the values of the parts inside the box, it runs a reality check:

"Does the energy going in match the energy being used up (heat) plus the energy being stored (magnetic/electric fields)?"
If the math adds up perfectly, the guess is accepted.
If the energy doesn't balance (like a bank account that doesn't add up), the method rejects the guess.

This prevents the system from making up a fake answer just because the numbers looked close. It ensures the answer is physically possible.

5. What Happens When Things Go Wrong?

The paper tested this method with "dirty" data—simulating real-world problems like:

Static on the line (Noise): The method handles this well, as long as the "dance" is complex enough.
Delayed Sensors: If the sensor reporting the current is slightly slow (like a laggy video call), the method detects this. It doesn't try to hide the error; instead, it flags the result as "unreliable" because the timing is off.

The Bottom Line

This paper doesn't just say, "Here is a formula." It says, "Here is a way to look at electricity, take a geometric snapshot, and if the data is good, give you a physical circuit diagram that matches the real-world energy flow. If the data is bad or the timing is off, it will tell you, 'I don't know,' rather than giving you a wrong answer."

It turns the invisible, chaotic flow of electricity into a clear, auditable picture of what's happening inside a three-phase load, even when you can't see the neutral wire.

Technical Summary: Geometric Time-Domain Identification of Three-Phase Load Equivalents

Problem Statement
Modern power systems require real-time load parameters for protection, monitoring, and adaptive control. While existing methods estimate aggregate static or dynamic parameters from events or phasor-domain quantities, they often fail to recover element-level energetic equivalents directly from instantaneous terminal waveforms. A specific challenge arises in three-wire systems where no neutral voltage is measured. In these cases, branch variables are hidden behind line-to-line voltages and constrained line currents ( $i_a + i_b + i_c = 0$ ), making it difficult to distinguish between delta and wye equivalents or to determine if the measured data contains sufficient information to support a chosen physical model.

Methodology
The paper proposes a geometric time-domain method that interprets measured voltage and current waveforms as trajectories in Euclidean signal spaces. The core methodology extends a previously published single-phase geometric identification formulation to three- and four-wire systems.

Geometric Formulation: The method constructs a "spacor" (space vector) from measured variables and their required derivatives or integrals (e.g., $v, \dot{v}, \int v$ ). Kirchhoff's Current and Voltage Laws (KCL/KVL) confine these trajectories to lower-dimensional subspaces within a higher-dimensional Euclidean space.
Parameter Recovery: Equivalent parameters ( $R, L, C$ ) are recovered by comparing the geometry of the measured trajectory (the osculating subspace) with the model subspace. In the symbolic domain, this yields closed-form quotients of blade coefficients from the osculating hypervolume.
Sampled-Data Implementation: To handle real-world data, the theory is implemented as a windowed matrix estimator.
- Windowing: Data is processed in finite local windows (typically 0.5–2 cycles).
- Derivative Reconstruction: Derivatives and primitives are computed using harmonic projection, FIR filters, or regularized finite differences to avoid high-order pointwise differentiation noise.
- Normalization: Matrix equations are normalized to remove unit-dependent scaling effects.
Diagnostic Layers: The method incorporates explicit checks before accepting a solution:
- Identifiability: Tests the geometric rank and condition number of the local trajectory to ensure the window contains enough information (e.g., sufficient harmonics or transients) to resolve the chosen model order.
- Energy-Admissibility: A physical validation gate that verifies the identified equivalent closes the instantaneous terminal energy balance ( $p_{PCC} = p_{dissipated} + \dot{W}_{stored}$ ).

Key Contributions
Relative to prior single-phase work, this paper introduces:

Unified Three-Phase Formulation: A geometric framework for both wye and delta connections in three- and four-wire systems, specifically addressing the coupled constraints of three-wire systems without a measured neutral.
Observability Analysis: A topology-dependent analysis identifying which branch forms are directly recoverable. It proves that series-delta and parallel-wye configurations are structurally singular in three-wire systems due to the lack of absolute phase potentials or internal branch currents.
Windowed Matrix Estimator: A numerical realization of the geometric construction using finite windows, harmonic-projection derivatives, and explicit rank/conditioning diagnostics.
Energy-Admissibility Gate: A criterion that rejects numerically fitted equivalents if they fail to satisfy the instantaneous energy conservation law of the assumed topology.
Low-Information Screening: A mechanism to report windows with insufficient geometric rank (e.g., pure single-frequency excitation for high-order models) as low-confidence evidence rather than forcing a fit.

Results
The method was validated through numerical simulations using clean data, measurement noise (up to 40 dB SNR), 12-bit quantization, gain mismatches, and sensor delays.

Accuracy: In controlled, noise-free cases, the windowed estimator reproduces analytical branch constants to machine precision. Under realistic conditions (60 dB noise, 0.5–2 cycle windows), recovered $R, L,$ and $C$ values remain within a 5% engineering band.
Robustness: The method successfully identifies informative trajectories while exposing structurally degenerate cases.
Failure Modes: The energy-admissibility gate effectively flags coherent timing biases. For instance, a 50 µs current-sensor delay caused parameter errors to rise to ~37%, which the method correctly identified as a synchronization issue rather than a reliable equivalent.
Validation: Fifteen of twenty-one practical-window configurations in the stress study met high-confidence criteria (coverage $\ge$ 80%, relative error $\le$ 5%, energy residual $\le$ 5%).

Significance and Claims
The paper claims that its primary value lies in determining an equivalent circuit that, for an assumed admissible topology, closes the instantaneous terminal energy balance of the measured load. The identified $R, L,$ and $C$ elements form a "physically structured equivalent" rather than a black-box fit.

The authors emphasize that the method does not claim to uniquely determine the internal energy localization of a load (as multiple topologies can satisfy terminal data), but rather provides a physically consistent equivalent for a chosen topology. The significance of the work is the introduction of a "confidence gate": the estimator either returns an auditable, energy-consistent equivalent or declares the window uninformative, thereby preventing the deployment of unreliable parameters in protection or control systems. The work is presented as a step toward element-level energy accounting, distinct from system-level state estimation.