摘要
根据样本空间的内积特性，在输入样本中引入快变参数的导数项，提出了一种无需迭代学习的内积神经网络。以某型航空发动机的机载记录数据为例，对发动机进行了建模，计算表明该模型自学习速度快，计算简便，抗干扰能力强，且精度较高。
According to the inner product feature of sample space this paper takes the differential of fast changing parameters into input samples and puts forward a new neural network not to require to iterate learn, taking it for example recording data of the record system of a type of aeroengine and taking out identification to it. It represents that the main feature of the manner result in that self learning velocity is faster, accuracy is higher, the ability of anti disturbance is stronger and the maintenance work is fewer.
出处
《机械科学与技术》
CSCD
北大核心
1997年第6期1111-1114,共4页
Mechanical Science and Technology for Aerospace Engineering
