Real-time Harmonic Filtering in Grid Interfaced Photovoltaic System Based on Adaptive NLMLS

In this paper, an adaptive normalized least mean logarithmic square (NLMLS) framework based proposal is described for real-time harmonic filtering in grid-interfaced photovoltaic (PV) system. In the devised controller, the NLMLS exhibits logarithmic characteristic resulting in mathematically tractable adaptive structure. Practically, an adaptive square function enables online learning of weight identification thereby ensuring elegant performance under interferences in solar insolation and load. The normalized/regularized feature is fundamentally responsible for eliminating the dependency of input harmonically polluted load currents on the weight estimation phenomenon. This way, adaptive NLMLS detects the fundamental weight component of the load current accurately and ensures guaranteed online harmonic support to address the various power quality (PQ) challenges [i.e., grid current harmonic emission and balancing, reactive power support, and power factor correction]. A straightforward operating principle of the proposed adaptive NLMLS is exposed and its efficacy is examined through MATLAB/Simulink under steady-state and transients scenarios. Proceeding further, the robustness of the proposed adaptive NLMLS is verified and compared with prevalent state-of-the-arts least-square algorithms using a dSPACE-1104 board. In summary, it is believed that this proposal is a step towards achieving comparable convergence performance and minimum steady-state misadjustment when exposed to disturbances in grid voltage and load current.