Application of Feed Forward and Recurrent Neural Network Topologies for the Modeling and Identification of Binary Distillation Column

CATEGORY: DISTILLATION COLUMNS
IETE Journal of Research, 2013, Volume 59 Issue 2, Pages 167-175, DOI: 10.4103/0377-2063.113038
Application of Feed Forward and Recurrent Neural Network Topologies for the Modeling and Identification of Binary Distillation Column
Amit Kumar Singh, Barjeev Tyagi, Vishal Kumar
Department of Electrical Engineering, Indian Institute of Technology Roorkee, Roorkee, India
Abstract
This paper presents identification of artificial neural network model of a Binary Distillation Column (BDC). In this paper, the two most common topologies of artificial neural networks in the area of control are introduced: Feed forward neural network and recurrent neural networks. The training of neural network has been performed by the data set acquired from real 9-tray continuous BDC setup available in laboratory. The network model is composed of two layers. A hyperbolic tangent sigmoid function and a pure linear function have been utilized as activation functions in the first and the second layers, respectively. The developed neural network model has been validated by an extensive data set of practical data received from real BDC setup.
Distillation columns are widely used in chemical processes, such as crude oil refinery and hydrocarbon processing industries. The control of the overhead and bottom compositions in a binary distillation column (BDC) using reflux and steam flow rates has shown to be a particularly difficult problem because the product quality cannot be measured economically on line. This is because the instrumentation is either very expensive and/or measurement lags and sampling delays make impossible to design an effective control system. A solution to this problem is the use of secondary measurements which replaces a mathematical model of the process with the input/output relationship-based model to predict the product quality. Neural network can be used as a model identification technique to fulfill this purpose.
Artificial neural networks (ANNs) are very effective for modeling and control applications. The ANN-based approach has some significant advantages over conventional methods such as adaptive learning ability and nonlinear mapping ability, since it is more flexible and easy to be implemented in practice. A number of applications of NNs to process control problems have been reported. Piovoso et al.  have compared NN to other modeling approaches for Internal Model Control (IMC), global linearization, and generic model Control. Seaborg and co-workers have used radial basis function NN for nonlinear control and they have applied their approaches to simulated systems
This paper presents two neural network identification topologies, i.e., feed forward neural network (FFNN) and recurrent neural network (RNN) for the model identification of binary continuous 9-tray distillation column. The output results of developed models have been validated and compared with the experimentally acquired results by the operation of laboratory setup of BDC. A brief discussion is given on the laboratory setup of BDC in Section 2. Section 3 presents the neural network-based algorithm for the model identification. This section also gives the introduction to the considered neural network topologies and their application for the model identification of distillation column. In this section, training of neural network model of BDC is also discussed. Section 4 contains simulation and results. The work has been summarized and concluded in Section 5.
Free Full Text Source: http://www.jr.ietejournals.org/article.asp?issn=0377-2063;year=2013;volume=59;issue=2;spage=167;epage=175;aulast=Singh