Optimized utilization and interpretability of process data with data-driven model

Abstract

Data-driven model has been widely used in process industry; the process data in complex process industry has timeliness, collinearity and correlation, which is difficult to explain. This paper optimizes the use of process data based on models and data, and explains its significance in the process. The combination of model and data not only guarantees the generality of analysis, but also promotes the real-time nature of data. The characteristics of the extracted data are used to explain the performance and working conditions in complex industries; adding the traditional mechanism model to the data analysis can speed up the training cost and generalization ability of the data. The data extraction and model verification prove the feasibility of the proposed method.