Change detection in smart grids using errors in variables models

We consider fault detection through apparent changes in the bus susceptance parameters of modern power grids. We formulate the problem using a linear errors-invariables model and derive its corresponding generalized likelihood ratio (GLRT) based on the total least squares (TLS) methodology. Next, we propose a competing detection technique based on the recently proposed total maximum likelihood (TML) framework. We derive the so called TML-GLRT, and show that it can be interpreted as a regularized TLS-GLRT. Numerical simulations in a noisy smart grid setting illustrate the advantages of TML-GLRT over TLS-GLRT with no additional computational costs.