Module se 

State estimation: recovering the most likely grid state from noisy, redundant, partial measurements.

Power flow answers “given the injections, what are the voltages”. State estimation answers the operational question instead: given a set of measurements that disagree with each other, none of them exact and not enough of them to determine the state directly, what state best explains them.

This module holds the pieces both planned estimator methods share. The measurement set comes from crate::measurement; what lives here is the measurement model — the functions h(x) that say what each measurement would read in a given state.

Why h(x) is not just “read it off the network”

Three of power-grid-model’s sensor types describe quantities that gridoxide models structurally rather than as appliances, so their measurement functions are not the bus injection despite PGM presenting them that way:

PGM sensor	PGM’s model	gridoxide’s model	h(x) here
source	appliance at the node	virtual slack bus behind an impedance branch	that branch’s flow
shunt	appliance at the node	Y-bus diagonal entry	`-
node	sum of all appliances	—	bus injection plus the two above

Treating any of them as the plain bus injection is wrong by construction, not by a sign: gridoxide’s power_injections at a source node is zero by KCL, because the source’s power arrives through a branch that is part of the network. tests/measurement_residual_test.rs is what established this — it reported a 63-sigma disagreement on exactly that quantity.

Modules

bad_data

Detecting and identifying measurements that disagree with the rest.

batch

Batched state estimation: one topology and one measurement structure, many readings, estimated across cores.

case

A network, a starting state and a measurement set that agree with each other — what an estimate is run on.

constraints

Zero-injection buses as hard equality constraints.

functional

Every quantity the estimator measures, as one shape.

iterative

Iterative-linear state estimation: power-grid-model’s default method.

jacobian

The measurement Jacobian H and the gain matrix G = HᵀWH.

nr

Weighted-least-squares state estimation by Gauss-Newton.

observability

Observability analysis: which parts of the state the measurements actually determine.

Structs

MeasurementModel

One resolved CurrentFunctional per measurement, in input order.

SeNetwork

Everything h(x) needs about the network, beyond the state itself.

Functions

measurement_functions

Evaluates h(x) for every measurement in measurements, in order.

measurement_functions_with

measurement_functions against an already-resolved model.