Abstract:
Hot metal desulfurization pretreatment is a very sophisticated reaction which is not only diverse but also non-linear.A model identification method based on genetic RBF neural network to predict desulfurizer quantity is put forward,which can perfectly resolve the problem of random selection of RBF cluster center number and sample data clustering.In order to ensure structure of neural network to fit with continuous incremental data set,a method for dealing with incremental data is presented,which is applied to amend the parameters of neural network.Then the request of continuous production is satisfied.Finally the testing results are given,showing that after adopting the algorithm,the error of result is less than before and end-point hitting ratio satisfies to ninety percent,indicating that the algorithm has the engineering practicability.