| With the rapid development of society, energy problem has become a big international puzzle for all human beings. Meanwhile, new energy resources such as solar energy with almost infinite reserve, wide distribution range, clean and non-polluting and huge development potential and other advantages, has been highly valued by universities and research institutions at home and abroad.The aim of maximum power point tracking is to output solar energy, which conversed by photovoltaic cells, at the maximum power. Just because of it's great importance, it needs to be researched and discussed deeply. There are several ways of MPPT, all of them have their advantages and disadvantages. Among the disadvantages like, the problems of wasting more time to reach the steady state and greater fluctuation at steady state may be very serious. So, to solve these problems, a new method of MPPT is proposed. It is called MPPT based on neural network prediction model. Also, harmony search algorithm was used to optimize neural network model, so a high-precision prediction model is established.First of all, the prediction model will predict the voltage of MPP or nearby MPP. Then, PID controller can turn the working voltage of PV cells to the predicted voltage, also called reference voltage, quickly. At last, it is time to go to P&O with small step so as to track the MPP. Also, MPPT prediction model is established on the MATLAB platform.To establish the prediction model, the data of PV cells is necessary. So, PV cell data acquisition device based on Boost converter is designed. Lots of available data is acquired by experiment. Simulation research based on P&O and the proposed method is executed to compare the MPPT effect of different methods. The results show that the proposed method can save more time, improve MPPT accuracy, and lower the power loss effectively.At the same time, the MPPT control strategy proposed in this paper is implemented in hardware, which proved the feasibility of this method. |
PDF Full Text Request