An evolutionary online sequential extreme learning machine for maximum power point tracking and control in multi-photovoltaic microgrid system

In this paper, a new non-iterative Maximum Power Point tracking (MPPT) scheme is proposed for Photovoltaic (PV) based utility grid interactive microgrid application. A multiple PV based two stage conversion unit is studied while considering worst grid scenario by UL 1741 standards. An evolutionary approach called Hybrid Firefly Algorithm based Improved Online Sequential Extreme Learning Machine (HFA-IOSELM) is proposed for the MPPT study. The tracking error is significantly reduced with the proposed Improved Online Sequential Extreme Learning Machine (IOSELM) due to the dynamic clustering of input data (based on solar irradiation). Hybrid Firefly Algorithm (HFA) is used to reduce the randomness in the input weights of the IOSELM. To ensure improvement in both DC side (duty cycle) and AC side, i.e. VSC Phase locked loop (PLL), a MPPT based reference calculation procedure is proposed. VSC pulse width modulation (PWM) is sensitive to erroneous reference value during operational uncertainty and hence impact of proposed dynamic control reference calculation is validated with a nonlinear Lyapunov Finite Time Sliding Mode Control (LFSMC) scheme. MATLAB, TMS320 C6713 test bench simulation is considered for validation of effectiveness, where improvement in power quality (reduction in erroneous reference calculation, dead time, lower order harmonics) as well as enhanced stability margin (VSC closed loop PLL) are ensured.