A Hybrid Statistical Model Based Approach To Prosess Operating Performance Assessment

Process operating safety and smoothness are two crucial issues in process operations, which are influenced by process disturbance and human operations along with variations of process characteristics. Therefore, the assessment of operating performance on both safety and smoothness is of important practical significance for process operations. Motivated by these observations, the following research work is conducted accordingly.Firstly, concerning the cross-correlation and autocorrelation of process variables, a novel hybrid statistical model combining PCA (Principle Component Analysis) and time series analysis is presented, in which, the PCA model is employed to eliminate the cross-correlation while the time series analysis is used to deal with the autocorrelation.Subsequently, taking advantage of the elimination of the cross-correlation and autocorrelation based on the hybrid statistical model, the T2 and Q statistics are generated from the PCA model to measure the safety operating performance, while the smoothness indices are synthesized from the moving average model before the global performance indices are built according to the contribution rates of the principal components. Additionally, the hybrid model can be updated while disturbances are detected so as to attenuate the influence of data drift on the accuracy of the data-driven model.At last, the proposed method is applied to Tennessee Eastman (TE) process. Experimental results show that a satisfactory assessment result of the process operation can be achieved, which demonstrates the feasibility and effectiveness of the proposed online operating performance assessment strategy.In conclusion, the study shows that the PCA and time series analysis based hybrid statistical model can be used to created safety and smoothness operating performance indices, as well as to assess the process operating performance effectively, expecting promising applications potential.