Application of Artificial Neural Network for modelling of the ABS latex coagulation processes

فاقد چكيده

This paper presents an application of Artificial Neural Network (ANN) to modeling of acrylonitrilebutadiene- styrene (ABS) latex coagulation processes. For this purpose, some important parameters (concentration of acid, temperature, agitation speed, time and ratio of latex to acid) which were effective on particle-size distribution of coagulated resin were selected. The effect of main process parameter on the decreasing of fine particle(less than 100nm) was evaluated. The laser light scattering method was used for determination of fine particle. The obtained results were used for construction of ANN base model for predicting of the present of fine particle in product from initial process condition in ABS latex coagulation. The ANN structure with five inputs and one outputs and one hidden layer is trained to produce forecast actual g-ABS fine particles. The comparison between real and predicted data show better performance based on root mean square error of correlation (RMSEC) criterion. This approach could be applicable for prediction of fine particle and result in decreasing g-ABS loss in ABS plant and finally decrease waste water pretreatment costs.