摘要
为了克服目前排气消声器常见设计方法中固有的建模假定引起的设计误差,尝试将BP网络应用于排气消声器设计;利用正交试验和排气噪声数字信号处理获得了1/3倍频程声压级和柴油机燃油消耗率作为输入、结构参数作为输出的人工神经网络的学习样本;通过MATLAB中人工神经网络工具箱完成了网络的训练;实例验证了BP网络应用于排气消声器设计是完全可行的;同时,通过训练后网络的具体应用,展示了BP网络在消声器设计方面的优越性.
BP network was applied to the design of diesel engine exhaust muffler to reduce the intrinsic error caused by the assumption on which the muffler's acoustic model was based. By means of orthogonal planning and digital signal processing, the learningsamples of artificial neural networks were obtained, which regarded the sound pressure level of 1/3 octavebands and the specific fuel consumption of diesel engine as inputs, the structural parameters of muffler as outputs. Through the artificial neural network toolbox of MATLAB, the training of BP network was finished. The application of BP network to muffler designs was feasible, which had been proved by the practical examples, meanwhile the application of the trainednetwork showed that BP network has many advantages in muffler designs.
出处
《福建农林大学学报（自然科学版）》
CSCD
北大核心
2003年第3期394-397,共4页
Journal of Fujian Agriculture and Forestry University:Natural Science Edition
