A semi-supervised soft measurement method for concentration of diethylbenzene in ethylbenzene rectification

Abstract

The application discloses a semi-supervised soft measurement method for the concentration of diethylbenzene in ethylbenzene rectification, comprising the following steps: obtaining the ethylbenzene distillation process variables in the ethylbenzene column with labels and without labels, calculating the mixed mutual information for feature selection to obtain auxiliary variables of soft measurement; performing data padding and time series decomposition on each auxiliary variable to obtain fluctuation items and trend items of the auxiliary variables; constructing a double-channel prediction network, including a fluctuation item prediction channel and a trend item prediction channel, and finally performing regression prediction on the two channels through a full connection layer, and superimposing the outputs of the two channels to obtain the predicted value of the diethylbenzene concentration; performing network training, and constructing a semi-supervised loss function for the samples with labels and without labels respectively; inputting the auxiliary variables after time series decomposition into the trained network to obtain the soft measurement result of the diethylbenzene concentration. According to the application, the accurate soft measurement of the diethylbenzene concentration in the ethylbenzene rectification process can be realized.

Images