To detect disturbances in signal processing a common step is to extract signal characteristics to analyze them, among these characteristics it is recommended to use the signal energy. Why energy is recommended over other features that can be extracted, such as statistical parameters of the signal or frequency components?
In short, energy is not a good feature per se. Disturbance enhancement sometimes requires preprocessing, and might not be resistant to outliers. The notion of disturbance in your question could be made more precise.
However, since the energy squares amplitudes of signal values, and is generally integrated over several samples, it has the ability to emphasize disturbances whose amplitude dynamics vary from the "normal" ones.
Aside, within the scope of automated detection, energies are quantities that are relatively easy to minimize: least-squares are common techniques in signal processing.
Yet I would not consider energy as a good feature in general. Depending on the need, other nonlinear operators can be used, as can be seen in the family of Teager Kaiser operators, see for instance: Signal processing using the Teager Energy Operator and other nonlinear operators.