LS_SVM prediction model based on VPRS and its application on alloy powder production

Aiming at the characteristics of vagueness and uncertainty of information existed in the alloy powder production, a LS_SVM prediction model based on VPRS is presented in this paper, in which variable precision rough set is analyzed with set pair situation and mixture kernels function is used as a modeling tool. Firstly, an initial decision table is constructed after information pretreatment, and the cupidity algorithm is used to reduce the redundant embedding and variables to acquire the reduced sample space. And then, the reduced result is input into LS_SVM model to identify and optimize the key variables or parameters. The simulation results show that the model has better performance than that of which based on gauss kernels and has fine generalization performance and high prediction precision.