Induction motor is an extremely non-linear system and it poses a great challenge for the fault detection scheme due to large and complex data processing. Induction motor fault can lead to huge losses and excessive downtimes with regards to maintenance and production. An external fault like unbalanced voltage supply could be very sever and result in excessive losses, mechanical oscillations, overvoltage, and interference with control electronics. Detection of an abnormality like unbalanced voltage supply is a challenging task in the interaction of electrical motor and the power grid. In this paper, a new method is presented to diagnose unbalanced voltage supply at the incipient stage. In the proposed method, first of all, a new approach is introduced to formulate the total number of winding turns associated with a particular slot. After the formulation, the unbalanced voltage supply is diagnosed through signal processing, symbolic time series analysis and D-Markov state transition. The proposed method also distinguishes motor operation under balanced and unbalanced voltage supply. In the proposed work, hardware setup is designed for experimental verification. For validation of the method, an experimental setup is designed to justify and distinguish the motor operation under balanced and unbalanced voltage supply. © 2016 IEEE.