Realistic simulation tests for LOUD dynamic smart grid system change detection

Previous research has justified the use of a state space model for describing smart grid system dynamics. A fault or intrusion in the grid can be found by recognizing a change in system matrices of the state space model. A new hypothesis testing-based approach using the Locally Optimum Unknown Direction (LOUD) test has been proposed to detect possible changes. Previous work has shown that the LOUD test makes reliable change detection decisions with less observed data than the well accepted Generalized Likelihood Ratio (GLR) test. Here we employ realistic simulations of the IEEE 14-bus system to more fully evaluate the LOUD test. We investigate the fitness of the linear state space model. We compare the performance of the LOUD test with the optimum ideal likelihood ratio test. We also look at the influence of the sampling rate on the detection performance.