Modelling and Tracking of the Global Maximum Power Point in Shaded Solar PV Systems Using Computational Intelligence
Permanent link to this Item
Link to Journal
Solar Photovoltaic (PV) systems are renewable energy sources that are environmentally friendly and are now widely used as a source of power generation. The power produced by solar PV varies with temperature, solar irradiance and load. This variation is nonlinear and it is difficult to predict how much power will be produced by the solar PV system. When the solar panel is directly coupled to the load, the power delivered is not optimal unless the load is properly matched to the PV system. In the case of a matched load the variation of irradiance and temperature will change this matching so a maximum peak power point tracking is therefore necessary for maximum efficiency. The complete PV system with a maximum power point tracking (MPPT) includes the solar panel array, MPPT algorithm and a DC-DC converter topology. Each subsystem is modelled and simulated in MATLAB/Simulink environment. The components are then combined with a DC resistive load to assess the overall performance when the PV panels are subjected to different weather conditions. The PV panel is modelled based on the Shockley diode equation and is used to predict the electrical characteristic curves under different irradiances and temperatures. In this dissertation, five MPPT algorithms were investigated. These algorithms include the standard Perturb and Observe (PnO), Incremental conductance (IC), Fuzzy Logic (FL), Particle Swarm Optimisation (PSO) and the Firefly Optimisation (FA). The algorithms are tested under different weather conditions including partial shading. The Particle Swarm and Firefly algorithm performed relatively the same and were chosen to be the best under all test conditions as they were the most efficient and were able to track the global maximum power point under partial shading. The PnO and IC performed well under static and varying irradiance, the PnO was seen to lose track of the MPP under rapid increasing irradiance. The PnO was tested under partial shaded conditions and it was seen that it is not reliable under these conditions. The Fuzzy logic performed better than the PnO and IC but was not as good as the PSO and FA. Since the fuzzy logic requires extensive tuning to converge it was not tested under partial shaded conditions. A DC-DC boost converter interface study between a DC source and the DC load are performed. This includes the steady state and dynamic analysis of the Boost converter. The converter is linearised about its steady state operating point and the transfer function is obtained using the state space averaged model. The simulation results of the complete PV system show that PSO and Firefly algorithm provided the best results under all weather conditions compared to other algorithms. They provided less oscillations at steady state, high efficiency in tracking (99%), quick convergence time at maximum power point and where able to track global power under partial shaded weather conditions for all partial shaded patterns. The Fuzzy logic performed well for what it was tested for which are static irradiance and rapid varying irradiance. The PnO and IC also performed relatively well but showed a lot of ringing at steady state. The PnO failed to track the MPP at certain instances under rapid increasing irradiance and the IC was shown to be unstable at low irradiance. The PnO was not reliable in tracking the global maximum power point under partial shaded conditions as it converged at local maximum power points for some partial shaded patterns.
Sagonda, A. 2019. Modelling and Tracking of the Global Maximum Power Point in Shaded Solar PV Systems Using Computational Intelligence.