Spectroscopic Data Based Kinetic Parameter Estimation

The chemical industry was developed by the needs of human life and production,and played an important role in the historical industrial revolution and the contemporary new technological revolution.At present,fine chemicals have become the main development direction of the chemical industry in all the countries.There is a wide variety of products and it is difficult to find the substitutes.The production process has a long flowsheet with various reaction units and complicated materials.It is necessary to have a in-depth research on reaction kinetics for the requirement of strong ability to control the process.Kinetics reaction is a study of chemical reaction processes and chemical reactions is observed from a dynamic perspective.It studies the time required for the transformation of the reaction system and the microscopic processes involved.In this thesis,reasonable models parameters are estimated by studying the estimation approaches for reaction kinetics.Chemical reaction process can be reasonably predicted and effectively controlled,which helps to enhance the main reaction and suppress the side reactions.Hence,the yield of the target product and the purity of the target product can be increased.Also,the consumption of chemical raw materials and the output of by-products can be decreased.The quality of the product is improved.At the same time,it is possible to avoid dangerous situations that may occur during the production of chemical industrial products.The research content of the thesis mainly includes the following three aspects:(1)The basic concepts and basic knowledge of reaction kinetics are introduced,and the basic principles of dynamics are applied to simple reactions.Comparing the classic curve resolution(CCR)and the traditional curve fitting(TCF),CCR has better performance when the spectral overlap is severe.The adaptive step finite difference method was combined with CCR and verified four simulation cases.The experimental results show that CCR based on adaptive step size has higher computational efficiency than CCR with fixed step size.(2)Noise during spectral data sampling is unavoidable,which causes systematic errors.Unbiased estimation method is employed to reduce the bias.The concepts of point estimation and unbiasedness are introduced.A CCR method based on unbiased estimation is proposed and verified four simulation cases.The results show that the unbiased estimation can effectively reduce the influence of sampling noise and make the result closer to the true value.(3)CCR uses the Newton-Gauss-Levenberg/Marquardt algorithm,which is a gradient algorithm that tends to fall into local optimal solutions.Aiming at this problem,combining Monte Carlo methods with CCR to realize the global optimization.It is found that the method takes a long time in the simulation process.Therefore,the MC-CCR method based on the tabu strategy is proposed,which is verified by four simulation cases.Results show that the method has achieved good results.The poor initial value does not affect the search for the global optimal solution,and the method reduces the calculation time successfully.Finally,the main work done in this paper is summarized,and some research ideas and methods are provided for further research on the estimation of reaction kinetic parameters.