Research On The Control Method Of Inverter In Microgrid Based On Neural Network

With the global energy crisis, the advancement of energy conservation and emission reduction strategic plan, and the continuous improvement of user requirements for power supply quality, countries have begun to study microgrid. In the practical operation of microgrid, as the interconnection interface between microsource and microgrid, inverter is a key part to affect power quality, so its control technology is very important. This thesis focuses on the control technology of the microsource inverter in islanding mode and grid-connected mode of microgrid.Firstly, this thesis introduces the structure and the working principle of the microgrid with a variety of microsources, and details the system-level control strategies of microgrid and the control method of microsource inverter. Then, in view of the microgrid islanding and grid-connected mode, this thesis uses the traditional PID control to design the control system, including the selection of the filter parameters and the isolation transformer, and the design of the PI control parameters, respectively. According to the selection of parameters, the simulation models under the two modes are built, and the simulation results are analyzed.Since the inverter itself runs in switch state, has strong nonlinear and time-varying uncertainty, and also is influenced by the nonlinear elements, its accurate model is often difficult to establish. In addition, the existence of intermittent and uncertainty of the microsource power output in the microgrid, the existence of a large amount of non-linear loads that can bring serious harmonic pollution, and all of these factors have adverse effects on the modeling of inverter. While the design of traditional controller is based on the mathematical model of the controlled object, and once the parameters are established, it is very difficult to taking into account the requirements of performance index under different working environment. So using traditional control methods can not achieve the desired control effect.In view of the above problem, this thesis adopts neural network control. The neural network control doesn’t depend on the precise model, has good fault tolerance, and has the advantages of self-learning and self-adaptive. And this thesis uses it to improve the performance of the inverter, so as to improve its control precision. Firstly, in view of the microgrid islanding and grid-connected mode, this thesis designs the neural network control system, including the design of reference model, neural network controller and neural network identifier. Secondly, in order to verify the rationality of the control method, this thesis establishes the corresponding simulation model. Finally, a contrast analysis of simulation results is made between the neural network control and the traditional PID control so as to verify the effectiveness and superiority of the proposed control method.