摘要
在不同的燃烧状况下同时测量缸盖表面振动信号和缸内压力信号,通过对两信号的分析得到与缸内燃烧过程密切相关的振动信号的频谱范围,据此设计了FIR低通滤波器,并对振动信号进行滤波处理。通过分析滤波后的振动信号与缸内压力信号可知,缸盖表面振动信号同缸内压力信号在时域上具有密切联系。建立了BP和RBF神经网络,并用同样的训练样本进行训练,训练的结果表明,RBF神经网络可以在更短的训练时间内,获得更小的均方误差。用同样的测试样本对神经网络进行检验的结果表明,RBF神经网络重构的缸内压力波形更逼近于实际波形。
The cylinder pressure and vibration of cylinder head surface were measured when diesel engine runs under different operating conditions. Frequency domain of vibration signal related to combustion process was found out by analyzing cylinder pressure signal and cylinder head vibration signal. A FIR low-pass filter was designed according to the frequency domain and it was used to filter the vibration signal. Comparison of the filtered vibration signal and cylinder pressure signal show that this two signals has a close connection with each other in time domain. BP and RBF neural networks were designed and trained with same training data. Training results show that RBF neural network can obtain smaller MSE in shorter time. The neural networks are verified with the same test data and test results show that RBF neural networks have a better performance.
出处
《内燃机工程》
EI
CAS
CSCD
北大核心
2008年第2期76-80,共5页
Chinese Internal Combustion Engine Engineering
关键词
内燃机
缸内压力信号
振动信号
时域信号
神经网络
IC engine
cylinder pressure signal
vibration signal
time domain signal
neural network