The multi-state progressive damage of a structure is introduced into the change point detection problem: a generalized Hidden Markov Model (HMM) with multiple states is developed. Applications to Structural Health Monitoring (SHM) are explored. Demonstrate development of suitable likelihood ratio for simple progressive damage case, as well as generalized to arbitrary HMM. Computational costs are limited by exploiting structure of this model, and further limited by “windowing” technique. Demonstrations of algorithm are shown, with use of multi-damage state response from a structural frame in a test site.