论文部分内容阅读
提出基于神经网络的软测量技术的一般框架,并针对一个实际的催化裂化装置,采用多层前向网络和推广随机逼近算法对粗汽油干点的软测量进行了研究.理论分析和模拟表明,该模型可以很好地描述实际对象特性.
A general framework of soft sensor technology based on neural network is put forward, and a soft catalytic cracking device with multi-layer forward network and generalized stochastic approximation algorithm is used to study the soft measurement of crude gasoline dry point. Theoretical analysis and simulation show that, The model can describe the actual object characteristics well.