This study aims for the development of suitable fault detection and isolation technique for industrial machines. The study focuses on the development of a unified platform that considers the traits of reliable information from models, multidomain features and optimally incorporates the capabilities of intelligent processing (i.e. handling uncertainties and data unavailability). Considering the effectiveness of significant, well-established statistical and spectral features towards fault diagnosis, in this study a hybrid fault detection approach is presented that uses signal processing methods to extract multi-perspective feature knowledge of the system and make use of this low-dimensional, significant feature knowledge for intelligent classification based on artificial immune system. Furthermore, a generalized system identification approach is investigated which incorporates adaptive thresholding based fault detection and fault severity indices based fault isolation. Finally, a comprehensive hybrid fault detection and isolation method is proposed which is inspired by multilayer biological immune system. The developed hybrid fault diagnosis technique opts for the parallel operation of model-based fault diagnostics and intelligent fault detection. The proposed hybrid method carefully analyze the data-parallel operation of both methods and further incorporates a comprehensive information resolution mechanism to ensure improved and reliable fault diagnostics. In contrast to the conventional methods, the developed hybrid fault diagnosis scheme offers an improved reliability, safety and added efficiency in industrial machines.