| Due to the increasing living standards of people,the proportion of family vehicles has expanded rapidly,resulting in excessive demand for fossil energy such as oil.This not only exacerbates the rate of consumption of fossil energy,but also contributes to excessive global carbon dioxide emissions.The problem of global warming caused by excessive carbon emissions cannot be ignored.Therefore,in September 2020,the President proposed the concept of carbon peak and carbon neutrality for the first time at the 75 th session of the United Nations General Assembly.Under the background of energy conservation and emission reduction,the transportation industry,which accounts for nearly 60% of oil consumption,urgently needs to transform to new energy and green transportation.Although new energy vehicles have developed rapidly in China,their penetration rate in China is still very low.The main reason for this is that a series of new energy power generation dominated by solar power generation have some key technical problems that have not been solved.Solar power generation has the characteristics of poor flexibility,non-scheduling,non-linear output power,and poor input power quality due to large changes in light intensity and temperature,which affect the use of subsequent electric energy.Moreover,the DC-DC converter currently used is generally a Buck-Boost converter,which has low conversion efficiency and is easy to cause energy waste.Therefore,this paper aims to use the inductor-inductor-capacitor(LLC)topology to replace the traditional DC-DC converter,and to study a more efficient control strategy to improve the power quality and solar energy utilization of photovoltaic power generation by introducing artificial intelligence algorithms.This paper mainly studies the topology of photovoltaic DC-DC converter and its control strategy.Firstly,the working principle and static characteristics of the LLC resonant converter are analyzed,and then the main parameters of the resonant converter are designed according to the actual photovoltaic module parameters and the voltage requirements of the subsequent circuit.Secondly,the particle swarm optimization neural network algorithm(PSO-NN)is used to improve the maximum power point tracking control(MPPT)strategy.Write code in Matlab,after training the feature parameters,save the training results as a two-dimensional array.Through PSIM modeling and data calling,the LLC resonant converter can quickly track the voltage of the maximum power point of the PV module,and the simulation results show that the time to track the maximum power point is only 0.3ms,and the error is not more than 0.3%.Thirdly,the digital signal processor TMS320F28335 is used as the control core to design the hardware circuit,and GaN HEMT is used instead of Si MOSFET to reduce switching losses and increase power density.Through device selection,the models of two different switches were determined,and after the loss comparison was performed using LTspice modeling,the EMI simulation test of the main circuit was carried out and the filter was designed to meet the national general electromagnetic compatibility standards.Finally,the experimental platform was built,and the working characteristics of photovoltaic modules were simulated using the PSW160-7.2 programmable DC power supply in PTS-3000,and the CV mode in the programmable DC electronic load(PEL-2040)simulated the battery.Experiments show that the system has good anti-interference ability under different light intensity and temperature within a certain range. |
