Soft Sensor Measurement Research on Resolution Ratio of Cement Kiln Tail Based on IPSO-SVR

In order to realize the real-time on-line detection on resolution ratio of cement kiln tail, by means of the advantages from soft sensor measurement technology in which process parameters cannot be directly measured at present, a support vector regression (SVR) optimized by the improved particle swarm optimization (IPSO) algorithm is proposed. Based on the idea of adaptive weight, this algorithm overcomes the shortcomings of PSO which is prone to show premature convergence and poor local search ability, and improves its global search ability and local improvement ability. The parameters of SVR machine are optimized, and then soft sensor measurement model of resolution ratio of cement kiln tail is established. The simulation is compared with those based on the cross validation method and PSO, the results show that the IPSO-SVR algorithm has better ability of modeling, prediction and generalization, and the average relative error of prediction is 0.75%, so it can be further applied to the product real-time on-line detection on large-scale industries such as cement production.