The Study On Complicated Batch Process Model With Intelligent Identification Method

In the past decades, the chemical process industry has changed enormously. The batch process has been paid attention again. Batch process is a kind of process that gains finite production following job order by treating finite materials in one or more devices.It is difficult to build a model for batch process using common methods because of the characteristics of non-linear, time-delay and indetermination. Intelligent system theory is based on many kinds of new knowledge, and intelligent identification theory belongs to it. The fuzzy system theory, neural network theory and evolve calculation theory in intelligent control theory provide model methods for batch process plants.Based on researching traditional identification methods and those of intelligent control, the temperature plant of omethoate synthesis batch process will be identified. By comprehend and analysis the technology of it, the main factors that influences temperature and the laws of temperature variable are found. Input and output data for identification is made up of history data that has been pretreated from spot.The important part of this paper is model identification with neural network method. After reaching the features of BP network, L-M learning algorithm is adopted. With this algorithm, the training is faster than ever and the ability of net generalization is better. According to technology, there is some limitation using simple neural network to build model. So the stage model is presented. That is, separate the temperature into some stages in accordance with technology and establish the BP model for every stage. BP net realizes the static process identification. For dynamic process, recurrent neural network is reasonable. In this paper, the recurrent neural network consists of BP net added TDL blocks. Put the temperature data through a TDL block and make it as a new input data of neural network. In addition, the models established using all kinds of methods are put into control system to simulate the practice spot situation.Fuzzy T-S system identification whose parameters are optimized with GA is researched. In this part, generalized T-S fuzzy model is introduced, so it is feasible for finding the best parameters. With this method, the generalized T-S fuzzy model of omethoate synthesis batch process plant is built.Otherwise, this paper talks about the transformation from monomethyl amine fluxin simulation to open status of valve in practice. In order to realize it, curve-fitting technology is adopted. In a result, the flux curve and the experience equation are obtained.