Control of brown stock washing process in a paper mill using ANN Strategy

Brown Stock Washing (BSW) is a complex multi-variable continuous process where there is a need to control variables that are not directly or instantaneously controllable. Most of the conventional control systems for BSW process utilize individual PID loops to control several process variables. The set points for each loop are determined by process experts. Artificial Neural Networks provide a powerful tool for the process control and overcomes the problems of the prior art, including manual control and statistical control. In the present work, Artificial Neural Network model was developed using Single Perceptron Algorithm and Back Propagation Algorithm. The process variables were identified and their values were generated using known mathematical relationships for supervised learning and for unsupervised learning for which the range of values were taken from M/s Star Paper Mills Ltd, Saharanpur, India. Then the neural network was trained from the set of values generated through random reference.