The Application Of Neural Network Control In Solar Inverter Power Supply

For its own unique advantages of solar power, it has become one of the most ideal new energy. Photovoltaic power generation is the main form of solar energy utilization and Inverter power supply is an important component of the system and the performance of that affects the whole system performance. The application of advanced control algorithms is the important means of improving the performance of an inverter. In this paper, based on the strong ability of approaching to any nonlinear of neural network algorithm, exploring how let the control method used in the inverter. The simulation and experiment show that the method makes a good result in the inverter control and optimizes the inverter performance in a greater degree.Firstly, this paper describes the background and meanings of using solar energy as well as the feature and classification of photovoltaic power generation systems. Introduced the current situation and development of Inverter technology and inverter power electronics technology in the inverter power and analyzed two key technologies of affecting Inverter performance: PWM soft-switching technology and digital control techniques. This led to the main task of this paper, achieving the control of inverter with a digital control platform based on neural network algorithm.Secondly, this paper designs the inverter power system integrally, analyzes the main circuit topology and mathematical model of the inverter and introduces the basic principle of inverter sine pulse width modulation technology, which is used widely, and it's controlling and sampling methods. The advantages and disadvantages of several conventional control methods of the inverter were discussed, also the basic principles of neural network control and application was focused on. It details the BP neural network and its improved and the method of BP network design. On this basis, using two control methods of the direct control of BP neural network and BP-PID compound control and also using MATLAB software to simulate the training program, compared to the BP-PID control and conventional PID control and results show that superiority of the neural network control for nonlinear systems.Finally, based on the simulation, 500W single-phase inverter power experimental board using the TMS320F2812 as the main controller parts was designed, detailing the hardware and software design and the experimental results are shown finally.