A Study Of Wide-sense Steady State Identification Of Chemical Production Data Based On Double-thresholds

In order to improve the accuracy of the model,the most parameters of process model must be estimated from the measured data.Therefore,the measured data is the foundation and starting for many technical works,its reliability and accuracy not only affects the quality of the results of the optimization model,but also estimation of the model parameters.The application of measurement data are in many aspects,for example,the data required in the system test and process design is always expected to be obtained in steady-state.It is hoped that the material and heat balance is calculated in the process data of steady-state.On another occasion,it is hoped that the data will be obtained when the process conditions will be changed,such as dynamic process control.Therefore,steady state detection(identification)has become an essential part of data reconciliation,model adjust,system optimization and advanced control.From the actual,the concept of wide-sense steady state is proposed because of the phenomenon of wide-sense steady state is always true and universal in the process industry.Compared with other models,the model of chemical production in the process is not so strict for the requirements of the steady state,as long as the true value has not changed too much.As a result,the existing steady-state identification methods limit the identification of the steady-state data under the generalized conditions.This paper mainly deals with filtering of measurement data and wide-sense steady-state identification.Firstly,the filtering of measurement data is an important part of process automation,and it is also a potential problem to be solved before the implementation of steady state detection.In this paper,a novel filtering algorithm is developed to achieve a good filtering effect based on the traditional first-order filtering algorithm.Secondly,focusing on making a detailed study of the steady-state detection method based on adaptive polynomial and F-test method,and the advantages and disadvantages of two kinds of steady state detection method is proved through examples,and has established the wide-sense steady-state detective method based on double-thresholds;finally,through the simulation experiment proves the method of all.In order to highlight the practicability of the wide-sense steady-state identification method based on double thresholds,it is applied to the experimental data of ethanol water distillation column.Finally,the detailed process of modeling and simulation of ethanol water distillation column is described.Through the output data of the device,the various kinds of data filter and double threshold wide-sense steady-state detection are validated.This research can lay a solid foundation for data coordination,model optimization,fault diagnosis,advanced control and so on.