摘要
近十年来,为提高齿轮传动的承载能力、延长寿命、减小体积等,许多专家学者致力于齿轮传动的优化设计、胶合承载能力计算、可靠性设计等工作。为克服和改进传统的BP算法的不足,发挥神经网络和遗传算法的优势,提出一种基于遗传算法的神经网络二次训练方法.将遗传算法应用于神经网络的权值训练中,并用神经网络二次训练得到最终结果,降低了齿轮设计计算时间,是一种比较有效的方法。
In the past ten years, many experts have aimed at optimization design of gear transmission & counting its bonding - load - carrying capacity in order to improve its carrying capacity, increase service life & reduce volume etc. To overcome and improve traditional BP & display the advantages of NN and inherited algorithm, the essay has put forward NN quadric training method, applied inherited algorithm to NN weight and got the result from training. This has shortened gear designing time.
出处
《计算技术与自动化》
2007年第2期10-12,共3页
Computing Technology and Automation
基金
全国教育科学"十五"规划教育部规划课题(课题项目号:FJB030842)