Application of projection pursuit in gas emission predictio

In view of problem of data information lost and low prediction accuracy caused by common descending dimension methods for gas monitoring data in gas emission prediction, high dimensional sample data is transformed into one-dimensional projection data by combining principle of projection pursuit with differential evolution algorithm, and a gas emission prediction model which adopts sum of sine function is established by use of Matlab curve fitting toolbox. The experimental results show that the model has higher prediction precision and operability.