Prediction of the vertical roller mill key parameters based on optimized ESN algorithm

Abstract

The prediction of some key parameters in the operational process of vertical roller mill is very important for its safety and reliability. Because the vertical mill works too long in poor conditions, its key operating parameters are nonlinear and time-varying and the traditional prediction methods are difficult to achieve high accuracy. This paper selects the upper shell vibration signal of the vertical mill for the research. The signal processing has three steps. The first step is to remove the noise of the signal by using ensemble empirical mode decomposition (EEMD); the second step is to introduce the echo state network (ESN), which is a very suitable method for predicting in nonlinear chaotic system; the last step is to optimize the key parameters of the reserve pool in echo state network using fruit fly optimization algorithm (FOA) and the echo state network prediction model based on fruit fly optimization algorithm is put forward. The simulation results show the improved prediction model has higher prediction accuracy compared with the method without optimization, which proves the improved prediction model is suitable for the prediction on the upper shell vibration signal of the vertical mill.