The Application Of Methods Of Steady State Data Rectification In The Separation Of Olefin

The accurate and reliable process measurement data are the foundation of monitoring production conditions and optimizing production process for chemical industries.In the practical production,nevertheless,process measurement data are always disturbed by many factors,such as measurement equipment,measurement environment and so on.So they have difficulty in meeting the conservation of mass and other constrains.What is more,they may not reflect the state of a system,which sets an obstacle for realizing informationization of chemical industries.By a series of algorithms and knowledge of statistics,the content of data rectification includes detection and elimination of gross errors in data,reducing the overall influence of random errors,supplementing unmeasured data and raising the accuracy of data so that the validity of data is guaranteed.Based on olefin separation process,the dissertation studied the strategies of steady data rectification.On the part of gross error detection,it introduced some methods for simple problems and a NT-MT algorithm for complex problems.As for the NT-MT algorithm,three drawbacks that fail to detect gross errors in some circumstances were found.In order to tackle the drawbacks,an improved NT-MT algorithm was raised.Two calculation cases were presented to prove the mentioned drawbacks.Also,they confirmed the improved NT-MT algorithm make up the drawbacks.On the part of data reconciliation,the principle of data reconciliation was explained.In terms of bilinear data reconciliation problems,two algorithms,Crowe's projection matrix method and independent flows method,were illustrated and compared.The results from a calculation case showed that Crowe's projection matrix method is better than the other.However,a problem was found in the method.So the dissertation presented an improved Crowe's projection method,which prevented the problem happening.On the part of practical application,the dissertation used the improved NT-MT algorithm and the improved Crowe's projection method to preform data rectification and replenish unmeasured data.The rectified data met material balance equation and raised the overall accuracy of the data.By data rectification,the dissertation provided foundation of improving and optimizing production process for chemical industries.The improved NT-MT algorithm and the improved Crowe's projection method suggested by the dissertation can perform data rectification for steady state,which laid foundation for optimizing production process of chemical industries.