In the motor current signal, the characteristic frequency of broken rotor bar (BRB) fault is modulated by the supply frequency and it decreases with the decrease of the load, resulting it to be easily buried under light load conditions. Teager-Kaiser energy operator (TKEO) has shown better performance to detect BRB faults than classical methods, such as envelope and spectral analysis. However, the original definition of TKEO leads to its result lack of physical meanings and the causal processing in TKEO can lead to phase distortion and non-ideal filter characteristics. Therefore, this paper proposes a normalized frequency domain energy operator (FDEO) for the BRB fault diagnosis, which does not require causal processing and calculates multiple differentiations in the frequency domain with equal accuracy in one operation. Furthermore, normalized FDEO removes the influence of the supply frequency followed by spectral analysis to extract fault features. The mathematical model of induction motor under healthy and faulty condition are studied in this article. Then, the proposed approach is experimentally validated with seeded one and two BRB faults operating under various load conditions. To verify the effectiveness, the results are compared with TKEO, envelope and spectral analysis. It was found that the proposed method provides slightly obvious fault features with respect to TKEO, especially when the IMs run under light load conditions with two BRB faults. Index Terms-Broken rotor bar, Induction motor, Motor current signature analysis, Frequency domain energy operator, Fault diagnosis. NOMENCLATURE () instantaneous amplitude the amplitude of the fault frequency for fault case the amplitude of the fault frequency for normal case the modulation index Envsq[ ()] the squared envelope of () ℱ Fourier transform ℱ −1 inverse Fourier transform BRB fault characteristic frequency supply frequency rotating frequency the amplitude of the supply current