Intelligent fault detection and diagnostic (iFDD) is a technology with growing interest and importance in engineering. The developments of several mathematical modeling and pragmatic techniques have facilitated the R/D with better approaches to improve the iFDD technology from biological to electronics fields. These techniques are applied from the component level to the system level. During the last years, many research efforts have been presented in the field of fault detection and diagnosis. This dissertation presents the research in intelligent fault detection and diagnostic for low-power electrical systems, which are often found in household for daily use of electrical appliances. The research in iFDD utilizes data obtained from sensors and sensor network in a typical home electrical system to prevent hazardous conditions during abnormal operation. The diagnostic systems presented in this dissertation include the model-based and signal-based approaches. In the model-based approach, the physical model of the system is employed in the analysis of fault detection and diagnosis. In the signal-based approach, methodology of pattern recognition and fingerprint analysis is used to construct the iFDD model. Wavelet method has been employed to reduce the redundancy in a sensor network. The fingerprint of sampled signals under controlled faults of serial and parallel arcing are drawn fro m the coefficients of the wavelet decomposition to establish the relationship between the signature and the target faults. Experimental results and analysis are presented to illustrate the principle and applications of the iFDD technique.