Residual generation in linear stochastic systems-a polynomial approach

A polynomial design algorithm for innovation/residual generation for fault diagnosis is given. The class of systems considered is linear stochastic state-space and descriptor systems in both continuous and discrete time. A new class of residual generators based on stochastic models is defined and design algorithms are derived. The algorithms is based on standard operations such as null-space computation and spectral factorization of polynomial matrices for which numerically reliable implementations is readily available