Hematite grinding particle size soft measurement method based on multi-dimensional input time lag

Abstract

The application discloses a hematite grinding granularity soft measurement method based on multi-dimensional input time lag, and comprises the following steps: based on a typical one-stage grinding process of hematite grinding, a multi-dimensional time lag stochastic configuration network model is constructed to represent dynamic characteristics thereof; in view of the multi-dimensional time lag in the hematite grinding process, a time lag term in the system is modeled in combination with a posterior probability density of the time lag, and a noise characteristic is modeled by using a Gaussian distribution, so that a stochastic configuration network model under the multi-dimensional time lag system is obtained; an expectation maximization algorithm is used to jointly optimize multi-dimensional time lag parameters and network output weights, and an optimal parameter combination of the model is determined through iterative optimization, so that an optimized hematite grinding granularity time lag soft measurement model is obtained; and mill feed quantity, mill inlet water quantity and classification machine overflow mass concentration data are input into the trained hematite grinding granularity time lag soft measurement model, and target grinding granularity data is output.

Images