Soft Sensing Model Of 4-CBA Content In PTA Oxidation Process

PTA oxidation process is an important chemical reaction process in petrochemical production,and its reaction product is an important chemical raw material in the production of polyester products.4-CBA is the main by-product in the oxidation process.Because of the harsh reaction conditions,the reaction mechanism and the reaction process of PTA oxidation process,the soft sensing technique is used to predict the reaction process in real time.Soft sensor uses some measurable variables to predict the uncertain variables.In this paper,the PTA oxidation process is studied.The 4-CBA content is taken as the research object,and the soft sensing model is established by AdaBoost algorithm.This paper will study the following several aspects.AdaBoost algorithm is a kind of combination algorithm,which combines a set of weak learners with different training intensity into a strong learner.In this paper,BP neural network and support vector machine(SVM)are selected as weak learners.In order to solve the weak training problem of AdaBoost algorithm,using the method of dual threshold to update the sample weight,reduce the influence of larger sample error of weak learner,using roulette method to sample resampling,and to prove the feasibility of the improved algorithm by nonlinear function fitting.The soft measurement model of 4-CBA content in PTA oxidation process,using BP neural network and support vector machine as a weak learner,with improved AdaBoost algorithm to do is a soft measurement model for strong learning,through to the4-CBA content of MATLAB simulation training.The model is compared with a single weak learner model and an improved AdaBoost algorithm.The results show that the improved AdaBoost algorithm has higher prediction accuracy in these models.