胶料粘度的PCA-RBF多网络软测量研究

Multi-network Soft Sensor Based on PCA-RBF for Viscosity

针对传统混炼过程胶料粘度预测存在的辅助变量信息冗余大及冗余导致模型结构的复杂化和泛化能力的降低提出先进行主元分析;同时由于对象的复杂性,传统的建立单一模型的思想存在着弊端,从而提出基于RBF神经网络建立多网络模型思想并进行仿真。

A large amount of redundancy information among the secondary variables, which wash＂ t be processed during the traditional modeling for rubber viscosity. In order to simplify the structure and improve the generalization of model, PCA is introduced in this paper. Because of the complexity of object, the traditional method based on a single model was not the best. This paper presents a multi-network soft sensor based on RBF neural network and simulates it.