Soft sensor method for aviation fuel dry point in atmospheric distillation column based on dynamic moving window least squares support vector machine

Abstract

The invention discloses a soft measurement method for aviation coal dry point of an atmospheric rectification tower based on a dynamic moving window least squares support vector machine. The method disclosed in the invention selects the relevant operation and state parameters of the atmospheric rectification tower as the input of the model , the dry point of jet fuel to be predicted is taken as the output of the model, the historical operation data of the rectification tower is selected as the initial training sample, and the initial model of dry point of jet fuel is established by using the least squares support vector machine method. In addition, based on the analysis of the time-varying characteristics of the atmospheric distillation column, a dynamic moving window-based update strategy of sample deletion and sample addition is proposed, and two modes of sample deletion and sample addition are used to implement incrementally. Parameter solving and model updating.

Description

technical field [0001] The present invention relates to the technical field of detection of dry point of aviation kerosene (hereinafter referred to as aviation kerosene) quality index in atmospheric distillation tower distillates of oil refining enterprises, and in particular to an atmospheric distillation tower aviation kerosene based on dynamic moving window least squares support vector machine Dry point soft measurement method. Background technique [0002] Usually, the detection problem of the industrial process is solved by developing a new type of process measuring instrument to realize the direct online measurement of the process parameters in the form of hardware. However, there are always some important variables that cannot be detected in real time in the industrial production process. For example, when measuring quality indicators such as the quality parameters of aviation fuel products in atmospheric distillation towers in oil refineries, there are currently no a...

AI Technical Summary

Problems solved by technology

However, in actual working conditions, the production environment is generally in dynamic changes, and the inherent static modeling method cannot or is difficult to adapt to the changes in the sample data of the production process, resulting in the soft sensor software not being used normally or the estimation effect is not good under time-varying conditions.

Images