State observers are algorithms from control engineering that reconstruct non-measurable variables (states) from known input variables (e.g. manipulated variables or measurable disturbance variables) and output variables (measured variables) of an observed reference system. Observers are used, for example, in state controllers to reconstruct non-measurable state variables, in discrete-time controls
...
where the measured variable cannot be updated in every cycle or in measurement technology as a substitute for technically or economically impossible measurements.
Observers can also be used, however, to monitor operation using individual measured variables, for example, and to derive the operating state from the comparison of different measured variables and only report a fault in the event that conspicuous values are measured. By using state observers, systems with fewer direct sensors can be monitored, as they can reconstruct the internal state of the system from the available input and output data.
In terms of data sufficiency, this means that with a state observer, less data may be needed to monitor or control the state of a system, as the observer can infer data from the internal state of the system that does not need to be measured directly. This can be particularly useful when direct measurements are expensive, difficult or inaccurate. The state observer can make optimal use of the available data to obtain accurate estimates of the system state.