In this study, a model for the relation between magnetic filter (MF) performance and time was developed by artificial neural network (ANN). The ANN model is includes a hidden layer. The input parameters are concentration of magnetic particles in output of MF and time. The output parameter is MF performance. The estimation performance of ANN is evaluated by using sum of squared errors (SSE), correlation coefficient (R2 ) and mean relative errors (MRE). The ANN model resulted in a good regression analysis for test data set in which the R2 is 0.999857, SSE is 0.000132 and MRE is 1.004543. The regression coefficient shows that ANN approach with high level of accuracy can be considered as an alternative and practical technique to estimate performance parameters for MFs. The model enables us to estimate the variable characteristics of filter performance and time used in the cleaning process of industrial liquid. These estimated results provide solutions to be used for the optimization and control of magnetic filtration process and also new filter designs.
Ismail SARITAS, Ilker Ali OZKAN, Saadetdin HERDEM, (2012), An Artificial Neural Network Model for Magnetic Filter Performance, 1 st Taibah University International Conference on Computing and Information Technology (ICCIT2012), Vol 1., pp. 50-55, March 12-14, 2012, Al-Madinah Al-Munawwarah, SAUDI ARABIA.