Intelligent Maximum Power Point Tracking of Solar PV Systems

Abstract

The project involves the evaluation of the use of an intelligent technique in maximum power point tracking of Solar PV systems. Evaluation is made in reference to a technique using Perturb and Observe based MPPT algorithm. The power point tracking mechanism is designed to ensure maximum power transfer from the PV module to the PV system. Systems operating without utilisation of this tool are seen to operate in a non-optimal nature owing to the variation of the solar irradiance conditions causing shift in operating point from the set one.

As a result, the system is seen to vary in performance with the irradiance conditions only performing optimally when the present conditions are in line with the set operating point. This project entails evaluation of two MPPT techniques i.e the suggested intelligent technique, An Adaptive Neural Fuzzy Inference System with its non-linear parameters further optimised by an intelligent technique i.e Particle Sawarm Optimisation.

This algorithm used by swarms of birds during migration is utilised as the particles are seen to adjust accordingly towards the optimal position basing on each particle’s best optimum and best global optimum for the swarm. The optimisation and adjustment of these values for each particle is seen to continue till the optimal position at which point, the particles are seen to obtain zero velocity. These features of the optimisation tool are desirable for use in the power tracking method as the zero velocity obtained at the optimal point ensures lack of oscillations at the operating point once it has been obtained.

Additionally, the use of an algorithm that is not derivative based ensures that the tracking algorithm doesn’t get stuck in a local optima during conditions when the PV module operates in conditions of partial shading evaluation is done basing on these features in addition to power output in comparison to the ideal time and response time of the Tracking model. The MPPT model is modelled in MATLAB Simulink with a buck boost converter used so as to alter the apparent impedance of the PV module to achieve the impedance matching. The intelligent model is then tested and validated to adjust this duty cycle after training the model on a data set.