Experimental Evaluation of MPPT algorithms: A Comparative Study

Abstract

Photovoltaic (PV) energy is among the most used renewable sources. Grid-connected PV systems should yield as much energy as possible. However, external influencers such as irradiance and temperature impose a non-linear characteristic of the PV system, which hinder its operation at the maximum power point. Additionally, other factors, such as shading or internal degradation, can change this characteristic by making local maximums appear, which makes it difficult to extract the maximum available power. There are several techniques for maximum power point tracking (MPPT) and very diverse algorithms for this purpose. There are also some published works with comparative studies. However, in most of these works, the comparison is based on a literature review or on simulation. An experimental evaluation of MPPT techniques, from the simplest to the most complex, remains relevant. Thus, this paper presents an experimental analysis of five MPPT algorithms: two of the simplest and widely used (Perturb & Observe and Incremental Conductance) and three of the most complex (Fuzzy Logic Controller, Kalman Filter and Particle Swarm Optimization). The experimental tests were carried out under real test conditions, using Simulink and the dSPACE 1103 real-time controller board. The results show that the five MPPT algorithms are able to track the MPP with a difference of less than 2% in their efficiency under normal operating conditions. This difference increases under shadow effect. The PSO algorithm was the only one able to find the global MPP under the effect of partial shading.