Soft Sensing of Polypropylene Melt Index Based on Improved RBF Network

In view of the problem in the practical application of radial basis function (RBF) network, such as the number of nodes in the hidden layer and the parameters (w, c, and σ) are hard to determine, an improved particle swarm optimization (PSO) algorithm which makes use of the advantages of PSO algorithm and genetic algorithm (GA) is proposed, and then optimize the RBF network model with the new algorithm for soft sensing of polypropylene melt index. Simulation results show that the new method can improve the network´s training speed and predictive precision effectively, and is more suitable for soft sensing of polypropylene melt index.