Glossary · Control theory and process control
State observer
Also known as: Observer, State estimator, Luenberger observer
German: Zustandsbeobachter
In control engineering, a state observer is an algorithm that estimates internal states of a system that are not measured directly, by running a model of the system in parallel and correcting it with the difference between measured and predicted outputs.
- Control theory
In one sentence
A state observer estimates unmeasured states by running a system model in parallel and correcting it with measured outputs.
Example
A sensorless drive uses a state observer to estimate rotor speed and position from motor currents and voltages instead of an encoder.
How it applies
- Engineering: Observers enable state feedback without sensors for every state, and they support fault detection by comparing model and measurement. They require an adequate State-space model and observability of the states from the available measurements.
- Operation: Estimated values can drift when the model no longer matches the real system, for example after mechanical changes or at operating points outside the model range.
- Documentation: Mark estimated values as estimates in HMI texts, data exports and manuals, and state the conditions under which they are valid. For safety-relevant uses, an estimated value needs its own analysis; it is not equivalent to a measured one.
State observer vs. Kalman filter
The Kalman filter is a specific observer that chooses its correction gains optimally from assumed noise statistics. A general (Luenberger) observer uses gains chosen by the designer.