Soft sensor of outlet acetylene concentration in acetylene hydrogenation reactor based on multiple neural network structure

Based on the idea of combining models to improve prediction accuracy and robustness, this paper uses FCM to separate a whole training data set into several clusters with different centers. Each subset is trained by BP neural network. The degrees of membership are used for combining these models to obtain the final result. It has higher approaching precision and better generalization capability than the BP neural network. The result is satisfying when it is used in the soft sensing of outlet concentration of acetylene hydrogenation reactor. Practice has proved that this method is worthy of further application.