摘要
软测量模型广泛应用于替代实物传感器预测与估计不可在线测量的变量。从利用前向神经网络建立复杂的软测量模型的问题出发,从空间理论的角度揭示了前向神经网络以及各层的数学模型,并提出了一种新型的网络设计方法。最后,通过上述算法建立了某型无人机飞行仿真系统中的角速度软测量模型。
Soft-sensing model is widely applied to substitute for sensors and predict or estimate on-line unmeasured variables. The problem of design of feedforward neural network is discussed from the soft-sensing model viewpoint; The mathematic model of every layer of feedforward neural networks is described as viewed from space theory,and a new method for design is presented.Soft-sensing model of the angular speed in the flight simulation system of one unmanned aircraftis is realized by this method.
出处
《传感器技术》
CSCD
北大核心
2004年第4期73-76,共4页
Journal of Transducer Technology
基金
航空基金资助项目(01C52015)
关键词
前向神经网络
软测量
核函数
数学模型
feedforward neural network
soft measurement
kernel functions
