Maximum power point tracking using artificial neural network for photovoltaic standalone system

Solar energy has drawn much attention in recent years because of high demand for green energy resources. Electrical power can be generated by using semiconductors in photovoltaic (PV) cells to convert solar irradiance into DC current. Each PV module has its own optimum point at which the power delivered from the PV is at its maximum value. Since the initial cost for using PV is high, it is essential to make the PV module to work at its maximum power point. Thus, an algorithm named as maximum power point tracking (MPPT) has been introduced. These algorithms by controlling the duty cycle of a converter which is inserted between the PV module and the load make the PV to work at its maximum power point (MPP). Since the characteristics of PV module are dependent on atmospheric conditions of solar irradiance and cell temperature, conventional MPPT methods fail to find the MPP under rapidly changing of solar irradiance. Artificial intelligence methods have drawn much attraction in recent years due to their capability of handling uncertainty and nonlinearity conditions. In this work, an improved MPPT using Artificial Neural Network (ANN) has been presented. The control unit is comprised of two stages where at the first stage, ANN finds the voltage and current at which the maximum power is delivered, and at the second stage, another algorithm by developing the mathematical equation in related to input impedance, output impedance and duty cycle of the boost converter, tracks the MPP independent from the load, under changing condition of solar irradiance and cell temperature.The overall system consists of a PV module, a DC-DC boost converter, a control system and a resistive load. Also, a digital signal processor is used to generate the pulse width modulation signals for the driver of the converter. The proposed MPPT system is simulated using MATLAB. The results are compared with the results of the perturbation and observation (P&O) method under low and high solar irradiances; and slowly and rapidly changing of solar irradiance. Furthermore, the results of the proposed method are compared with results of the previous ANN MPPTs in two aspects of ANN outputs, and PV MPPT performance. The simulation and experimental results show that for both high and low solar irradiances, the proposed ANN method has smaller trackingtime, less power oscillation at steady-state, and higher efficiency than P&O MPPT with different step- sizes. Simulation results for different loads of 20 Ω, 33 Ω, and 40 Ω show that the proposed MPPT has efficiency between 99.96-100%, for different irradiances between 300-1000 W/m2 . In term of ANN output, the percentage error between the expected power and power predicted from ANN in this work is 0-0.119 %, which is more accurate than the previous ANN MPPT works with error percentage of 0.05- 3.66 %. In term of MPPT performance, the proposed MPPT has efficiency of 99.97% for low irradiance of 200 W/m2 and temperature of 31.9OC, which shows better performance as compared to ANN MPPT using PI controller which has efficiency of about 84% for low irradiance. As conclusion, the proposed ANN MPPT has high precision in finding the optimum points, as compared to previous ANN works. Furthermore, it tracks the MPP independent from the load, with high efficiency as compared to P&O with differentstep sizes and ANN MPPT using PI controller.

Preview Text FK 2014 76IR.pdf Download (1MB) | Preview

Actions (login required)

View Item