摘要
针对柴油发电机组-三相不控整流器-DC/DC变换器-脉冲负载系统中柴油发电机组的输出特性,提出仿真模型与试验电源输出电压(流)波形契合度的评价指标,以判断模型的仿真程度;提出了基于反向(BP)传播神经网络算法的同步发电机励磁电压输出动态限幅方法,应用于脉冲负载下柴油发电机组的模型优化。试验验证表明:27组算例中,初始仿真模型有18组波形实时契合度值不足90%,优化仿真模型27组值均大于90%。说明本文提出的优化方法使仿真模型比初始模型更加有效,可应用于后续柴油发电机组带脉冲负载系统的研究。
Aiming at the output characteristics of diesel generator set with pulse load,an index is proposed to evaluate the consistency of output voltage(current)waveform to judge the simulation degree of the model.Also,a dynamic limiting method for synchronous generator excitation voltage output based on BP neural network algorithm is proposed,and it is applied to the model optimization of diesel generator set.Experimental results show that among the 27 sets of examples,18 groups’real-time waveform proximity(RWP)values are less than 90%in the original simulation but in the optimized simulation,the RWP values of all the test groups are higher than 90%,indicating that the optimized simulation model is more effective,and can be applied to further research on diesel generator set with pulse load.
作者
师萌
杨艺斌
杨定富
黄克峰
何凯
吴振
Shi Meng;Yang Yibin;Yang Dingfu;Huang Kefeng;He Kai;Wu Zhen(Unit 32214 of PLA,Nanjing 211113,China;Unit 96784 of PLA,Luoyang 471300,China;College of Defense Engineering,Army Engineering University,Nanjing 210007,China)
出处
《强激光与粒子束》
CAS
CSCD
北大核心
2024年第5期156-162,共7页
High Power Laser and Particle Beams
关键词
模型优化
脉冲负载
反向传播神经网络算法
波形实时契合度
动态限幅方法
model optimization
pulse load
back propagation neural network algorithm
real-time waveform proximity
dynamic limiting method
