Particle swarm optimization based adaptive neuro-fuzzy interference system for MPPT control of a three phase grid connected photovoltaic system

Abstract

There is a significant demand for renewable energy systems to ensure sustainable and environmentally friendly living. Shifting from the use of fossil fuels to renewable energy sources will decrease both the reliance on these fossil fuels and associated pollution. By decreasing greenhouse gas emissions from energy generation, human induced global warming and environment destruction will slow. According to Environment and Climate Change Canada, Canada has committed, along with other leading countries in greenhouse gas emissions, to maintain the total global temperature increase below 2°C. Solar photovoltaic (PV) energy systems are of particular interest due to the availability and portability of such systems. According to Natural Resources Canada (NRCan), Canada’s reliance on solar energy as a renewable energy source is rapidly growing. NRCan claims that Canada’s total quantity of installed solar energy reached 1834 MW in 2014. Therefore, the main objective of this thesis is to design an intelligent controller-based efficient solar energy conversion system in order to meet the growing demand for clean energy. Solar energy systems consist of a PV cell array (solar panel) that uses the sunlight to generate direct current (DC) power. To maximize efficiency of the energy conversion, a novel maximum power point tracking (MPPT) algorithm is developed to deliver maximum power from the PV panel to the load. The conversion system must be designed to transfer maximum power regardless of the intensity of the sunlight and size of the load. Due to this requirement, the buck boost converter with an intelligent controller generating its control signal is the ideal solution. The converter is able to both step up and step down the input hence transferring maximum possible power at all times. Intelligent algorithms do not need exact mathematical models of the system and can handle any nonlinearity of the system. As an intelligent controller, a neuro-fuzzy controller (NFC), specifically an adaptive neuro-fuzzy inference system (ANFIS), will be developed to generate the control signal for the DC-DC converter while coping with variable weather conditions. A hybrid training algorithm is developed that implements particle swarm optimization to train nonlinear system parameters and the least squares estimator to train the linear parameters. The power at the output of the DC-DC converter can be either stored directly in batteries or converted to alternating current (AC) power. For simulation purposes of this thesis, the DC power available at the output of the converter is fed into a three phase, two level voltage source inverter that is controlled using proportional-integral controllers to control the d and q axis output voltages. Three phase output with constant amplitude and constant frequency is required to connect the system to the grid. The AC inverter output is filtered with an L filter and is interfaced with the grid to achieve effective grid connection. Simulations of the proposed energy conversion system and the proposed ANFIS training algorithm are completed in MATLAB/Simulink. The simulation results prove the effectiveness of the designed ANFIS and proposed training algorithm as the ANFIS-based MPPT controller is able to extract maximum power from the solar panel for varying irradiance conditions. The simulations further prove that grid connection is possible while obtaining three phase output voltage and current with low total harmonic distortion. The real-time implementation of the system is performed using the dSPACE DS1104 development board for communication to and from Simulink running on a PC. The proposed ANFIS-based MPPT controller and the proposed training algorithm are verified in real-time for a wide range of irradiance condition and changes in load. As determined by real-time implementation, however, the grid connection poses a significant challenge due to unknown factors in the Centennial building at Lakehead University as well as a lack of funds preventing the purchase of vital equipment. As such, stand-alone mode of operation is attempted in which the output of the buck boost converter is connected to a resistive load. The real time results prove the efficacy of the proposed ANFIS-based control and training algorithms.