Research On Control Strategy Of Quasi-Z Source Inverter Based On Neural Network

With the escalating problems of fossil energy scarcity and environmental pollution worldwide,the development and utilization of renewable energy is receiving increasing attention.Compared with conventional inverters,quasi-Z-source inverters utilize a single-stage structure to achieve voltage rise and fall and allow bridge arms to pass through,making them more flexible and reliable when applied to distributed energy generation.Because the quasi-Z source inverter is affected by its own structural nonlinearity,the traditional linear control methods cannot meet the control demand well,while the neural network has good generalization ability and can solve the control problem of highly nonlinear controlled objects.Therefore,this paper investigates the quasi-Z-source inverter based on neural networks.Firstly,the research status and application fields of quasi-Z source inverter are introduced,the topology and operation principle of quasi-Z source inverter are analyzed,the small signal model is established by using the state space averaging method,and then the transfer function from the disturbance quantity to the state variable is derived.The modulation principles of different boost modulation methods are discussed in detail,and the third harmonic injection boost modulation method has better gain and freedom through comparative performance analysis.Secondly,the quasi-Z source inverter control loop is designed based on the conventional PID control strategy,and the control parameters are rectified.The neural network characteristics are briefly introduced,the structure and characteristics of BP neural network and RBF neural network models are analyzed,and then the neural network algorithm is combined with PID control to adaptively adjust the PID control parameters by using the nonlinear mapping capability of neural network.The iterative adjustment process of the neural network algorithm is also analyzed in detail,and a simulation circuit model is built using MATLAB/Simulink to compare and analyze the performance of the quasi-Z source inverter under different kinds of control strategies.Finally,the design indexes are given,the hardware circuit is designed,and the principles of device selection and parameter calculation methods are given.The software program is designed with TMS320F28335 DSP as the main controller,and an experimental test verification platform is built to test the performance of the quasi-Z-source inverter under different control strategies when the input voltage changes abruptly or the load fluctuates.The experimental results show that the proposed RBF neural network PID control strategy has better control performance than the traditional PID control strategy and BP neural network PID control strategy,which can adaptively optimize and adjust control parameters,effectively improve the dynamic and static response performance of the quasi-Z source inverter to adapt to the requirements of different environments.