An Iterative Learning Control Method for Non-Repetitive Electric Vehicle Battery Discharging

In this paper, an iterative learning control (ILC) problem of electric vehicle (EV) battery state-of-charge (SoC) profile is exploited, because of the identical driving routes daily. Due to the variations on departure and return times as well as traffic conditions, the daily SoC profile is not the same, but could be slightly different from one working day to the other. This paper, therefore, proposes a novel ILC algorithm to deal with such daily-varying characteristic of battery SoC profiles, namely a quadratic ILC (Q-ILC) law with an iteration-varying weighting matrix for the control input update. It is then proved that the proposed ILC law is rapidly converged after just one iteration, under suitable selections of the Q-ILC weighting matrices. Numerical simulations are carried out, whose results validate the superior performance of the proposed ILC algorithm. As such, the proposed ILC scheme has the great potential for rapidly achieving a desired, predicted SoC profile of the EV battery in the next day, and hence, would help save the EV energy consumption.