In the real world, the state-of-power estimation in applications where lithium-ion batteries are used plays a vital role to understand the performance conduction of the battery or battery packs. But, correlation between state-of-power and state-of-charge is particularly nonlinear. The state-of-charge usually reduces, due to the available power which can descend immediately, especially near the lower and upper extremes of state-of-charge. So due to both the state-of-charge and different operating conditions the state-of-power can be affected making it difficult to establish a straightforward correlation. The key motivation to carryout this work was to estimate the state-of-power of li ion batteries using state-of-charge and incremental open-circuit voltage by using machine learning techniques as these methods were not completely explored earlier on. In this research work, the major contributions include design of an equivalent circuit model of second order which is developed based upon state-of-charge estimation and open-circuit voltage to tune the parameters of 2RC model using the real-world data. Here, in this work least-squares fitting procedure with random forest regressor machine learning model is used so that the error is minimized. The parameters are estimated by simulating the equivalent circuit model to predict the battery’s voltage response under different current profiles. By including two RC components in the circuit, the second order equivalent circuit model has better recorded the slow and fast transient behavior of the battery. This method makes it more precise for applications that involve rapid acceleration, regenerative braking, and load changes in electric vehicles.