Online diagnosis of brick interior fault based on spectrum analysis and multi-principal component model

In this paper, a method to online detect and diagnose brick interior different types of faults is proposed based on striking sound. Apply spectrum analysis to striking sound signals of bricks normal and with different types of faults, pick out characteristic frequencies of every spectrum to form data matrixes, then apply principal component analysis to build a normal and several different faults principal component models and get thresholds of statistics, next take these models to monitor striking sound signals of being-detected bricks, compute corresponding statistics, judge bricks interior defective or not and fault type by comparing with the thresholds when online detection. Experiment results show that the method is effective.