Automatic updating method for petrochemical device agent model

Abstract

The invention relates to an automatic updating method for a petrochemical device agent model, which comprises the following steps: firstly, acquiring DCS, LIMS, MES and laboratory analysis and detection data from related departments of a refinery plant, performing steady-state analysis and screening on the acquired data, and performing data processing and arrangement according to a timestamp to obtain basic data which can be used for establishing a plan optimization agent model; then, establishing a mathematical optimization model by using the processed data, and calculating Base and Delta coefficients of a processing device agent model; in the model, the model segmentation point is used as a variable, and the optimal segmentation point is obtained by solving the optimization model. According to the method, an enterprise device production database and a material detection analysis database can be communicated, the updating frequency and the updating criterion of the plan optimization device model are set, and the plan optimization device model is updated without human intervention.

Description

technical field [0001] The invention relates to the technical fields of petroleum refining and chemical production, in particular to an automatic update method of a petrochemical device agent model. Background technique [0002] Petrochemical enterprises must rationally plan production, broaden their own profit margins, and enhance market competitiveness. In the past, a large number of relatively mature planning and optimization commercial software have emerged, such as PIMS, APS, GRPIMS and GIOPIMS, etc. These softwares have been fully applied in the daily production planning and scheduling of petrochemical enterprises. The above-mentioned software can help enterprises optimize the crude oil suitable for their own processing by establishing accurate mathematical models for each processing device in the factory and optimize the solution, and optimize operations for multiple devices in multiple processing schemes, so as to meet different requirements. Under the premise of pr...

AI Technical Summary

Problems solved by technology

[0012] (2) At present, when refining and chemical enterprises calculate the product yield or distribution of the benchmark of the device, they often select a group of working conditions as the benchmark working condition through simple human experience discrimination. In this process, on the one hand, the operator may simultaneously use The steady-state data and unsteady-state data of the device are included, and most of the data are unsteady-state data of the device

Unsteady state data cannot describe the real operating conditions of the device, and should not be included in the statistical scope as valid data, and need to be excluded; choice, often has a greater impact on the outcome

[0013] (3) It is usually difficult to find at least two groups of feeding conditions in which a single property changes for a real plant in operation

[0015] The above problems are directly related to the quality of the proxy model of the processing device, which in turn affects the results and reference value of the entire planning optimization model

However, it can be seen that in the current actual production process, it often relies more on the experience of relevant technical personnel to deal with these problems. Even the commercial software currently on the market cannot solve related problems, resulting in the processing equipment used in refineries. The accuracy of the surrogate model is greatly reduced

For example, if the selection of segmentation points is not reasonable enough, it may lead to a large number of outliers in a certain segment, which will lead to unreasonable Base and Delta coefficients obtained, and the prediction results of the proxy model of the processing device are different from the actual production. The large gap will eventually lead to poor referenceability or even infeasibility of the plan optimization results.

Images