Load alteration fault detection and reconstruction in power networks modelled in semi-explicit differential algebraic equation form

In this paper a sliding mode observer is proposed to detect and reconstruct load alteration failures in a power network. The power network, specifically a 9-bus IEEE benchmark model, is modelled as a semi-explicit class of differential algebraic equations (DAEs). Only the phase angle of the generators, a subset of the differential states of the DAE, are assumed to be measured. The objective is to estimate simultaneously the phase angles, the frequencies of the generators (differential states), the phase angles of the load bus tensions (algebraic states) of the power network, and any exogenous signals representing disturbances and failures. The faults introduced in this paper fall into the class of load altering attacks which affect the algebraic states of the DAEs. The observer is designed based on a recently proposed multivariable super-twisting structure.