Research On The Operation And Control For The Microgrid Based On Neural Network Algorithm

With the rapid development of the world economy, it sharply increase in demandfor energy for human. However, conventional energy reserves are limited, fossil fuelsis facing depletion. Therefore, optimizing energy structure, the development ofrenewable energy has becomed an inevitable choice to solve the energy problem.Microgrid using renewable energy, nearby power supply, improve power quality,improving power supply reliability and energy efficiency.improve reliability andimprove energy efficiency. High quality and reliable electricity supply is the guaranteeof economic development, microgrid is of great significance for utilizing the structureof power grid, satisfying the increasingly safe and reliable and diversified demands ofusers for electricity and reducing pollution,etc.Microgrid operates in islanded mode when fault occurs in main grid. Microgridmust maintain voltage and frequency by its own regulating ability. Because thenatural changes power has characteristics that nonlinear, time-varying and intermittentin the micro-grid systems with new energy distributed power supply, so the traditionalPID control is often difficult to achieve the desired control effect. To overcomeshortcomings of U/f control for microgrid in disconnected operation, based on BPneural network and RBF neural network algorithm, an improved voltage and currentdouble-loop controller with a PID structure is proposed in this article. The PIDparameters are adaptive adjusted using BP neural network earning algorithm orJacobian information identification algorithm. The PID control with adaptive adjustedparameters contributes to reduce variations of voltage and frequency in isolation dueto the fluctuation of DG and load, which can improve the performance of thecontrolled system. The simulation demonstrates that the PID controller with voltageand current double-loop has good adaptability and anti-disturbance, which helps tomaintain stable operation of microgrid.Taking microgrid as the research object, this paper proposed that adopting theMatlab/Simulink software to build the simulation model can simulation analysis ofthe traditional microgrid control mode and operational control. at the same time, useof Matlab M file to write Jacobian information identification algorithm and BP neuralnetwork algorithm, called the M file by S-Function module, and then embedded in theSimulink model of the whole microgrid to test the microgrid operational control. Thesimulation results show that compared with the conventional PI control, neural network PID control can achieve a more ideal control effect.