Research On Soft Sensor Of The Thermosyphon Reboiler Considering Deterioration Of Equipment Performance

Thermosyphon reboiler is an important component of a distillation system.In order to provide a good initial value for distillation column control,data-based models for reboiler duty is studied as follows:(1)The variables of the reboiler are analyzed on the basis of the mechanism model.Five key variables are selected as the input and output variables of the data model.(2)The relationship between the data model and the influencing factors was experimented by using the support vector machine.The influence of different influencing factors on the soft sensor of heat duty was investigated.(3)BP neural network is used to simulate the reboiler.The soft sensor of the reboiler duty was established by using neural network method.The influence of all factors on the soft sensor was investigated.(4)According to the time-varying characteristics of the production process and the requirement of the prediction reliability,the neural network model is improved by moving window method to make the model fit the reboiler running state as much as possible.Under the premise of ensuring the prediction accuracy,the model update frequency is reduced and the calculation is reduced.The results show that the method of neural network and moving window method can be used to establish the soft sensor of the thermosyphon reboiler duty,which can provide experience and technical support for on-line inspection of time-varying production process.