Processor in the Loop Implementation of State of Charge Estimation Strategies for Electric Vehicle Applications

In light of the recent emergence of Vehicle To Grid (V2G) technology, electric vehicles (EVs) are no longer viewed as just transportation tools. They could rather serve as energy sources available at disposal of the electrical grid for ancillary services provision. As a result, an accurate estimation of their battery state of charge (SOC) is now more crucial than ever. Knowing that the choice of the appropriate SOC estimation strategy must consider the computational aspects of each approach, in this paper we investigate the implementation of two advanced SOC estimation strategies; The Feedforward Neural Network (FFNN) and Adaptive Gain Sliding Mode Observer (AGSMO). To verify the performances of both strategies, Processor In the Loop (PIL) implementations were conducted using an STM32F429ZI discovery board. The obtained experimental results prove that both algorithms perform well in battery SOC estimation. However, due to its slight edge in terms of precision, we recommend the AGSMO over the FFNN for the targeted application