Abstract: In recent years there have been widespread interests in Sweep Frequency Response Analysis (SFRA) of transformer. SFRA is popular because without dismantle of transformer we can detect the different faults. SFRA is used for assessment of the mechanical condition of the transformer core and winding. It provides an early indication of the growing problems. This method uses a low voltage sinusoidal signal which contains a sweep of frequencies covering a range between 10 Hz and 2 MHz. This signal is applied at one end and measured on the other end of the winding with one phase at a time and the impedance of the transformer winding over a wide range of frequencies is measured. This method is based on a comparison approach with respect to reference data. It is advisable to have reference tests on the new transformer. Faults which can change either the winding capacitance or the winding inductance are detectable. It is desirable to monitor the mechanical condition of the transformers periodically during their service life to provide an early warning of faults in transformer. In classical method result interpretation of SFRA to diagnosis the fault in power transformer depends upon the experience of human experts. Sometime a high degree of inconsistency and ambiguity have been observed regarding interpretation of result by different human experts, so there is a need to develop expert system for SFRA. In this thesis expert system programming has been developed using suitable MATLAB code. It has been successfully applied to various power transformers.