ANN based soft sensor model for reactive distillation column

To cope up with the strict emission and product quality standards the process industries are stern for the maintenance of the product quality and waste emission. For process industries, online hardware analyzers are costly alternate for analyzing the product quality, they are switching to online soft sensors which are cheap and efficient in use. In this work, soft sensor based on artificial neural network (ANN) has been developed. The motive of the sensor is to measure the immeasurable primary variables of the reactive distillation column i.e. product concentration, using the data of easily measurable secondary variables i.e. tray temperatures. ANN soft sensor has been trained using Levenberg-Marquardt algorithm. The predictions of the soft sensor are compared with the MATLAB mathematical model outputs.