Polypropylene melt index hybrid modeling method based on dynamic error compensation mechanism

Abstract

The invention discloses a polypropylene melt index hybrid modeling method based on a dynamic error compensation mechanism. The method comprises the following steps: firstly dividing an original sampledata set into a training sample set and a test sample set, establishing a mechanism model by adopting the training set, establishing a plurality of new training sample subsets by adopting a random resampling method, and training each subset to establish a base learner set of an Elman neural network; performing clustering analysis, and selecting individuals with good performance in each cluster asselective integration subsets; finally, using the parallel hybrid modeling method of the dynamic compensation mechanism for testing a sample set, judging whether a mechanism model training sample andcorresponding parameters need to be updated according to the similarity principle or not through errors, and obtaining a corresponding polypropylene melt index prediction value. The method is high inprediction precision and high in reliability, the nonlinear dynamic change trend of the melt index in the propylene polymerization production process can be well tracked according to the obtained soft measurement result, and the effective technical support is provided for operation optimization and quality control in the polypropylene production process.

Description

technical field [0001] The invention belongs to the field of research and application of soft-sensing methods in the polypropylene production process, and in particular relates to a polypropylene melt index mixing modeling method based on a dynamic error compensation mechanism. Background technique [0002] With the increasing complexity of industrial process objects, in order to control product quality, it is difficult for conventional industrial monitoring technology to achieve satisfactory results. The inspection of many off-line samples is not only time-consuming and expensive, but also easily leads to control lag, and it is difficult to guarantee product quality and production safety. Therefore, it is of great significance to introduce soft sensing technology into the chemical production process. [0003] In the interdisciplinary application of chemical processes and other disciplines, the accuracy of the mechanism model based on chemical objects may deviate from the a...

AI Technical Summary

Problems solved by technology

[0005] In order to overcome the shortcomings of the existing polypropylene melt index soft sensor model that cannot guarantee the prediction accuracy in the global scope and the model stability is not strong, the present invention provides a dynamic error compensation mechanism based on high prediction accuracy and strong model stability Polypropylene Melt Index Hybrid Modeling Method

Images