Modelling Methods HAO TO ONG UN Input and output parameters Wheel speed Roughness Grinding force Model Removal rate Wear of belt Model construction Wear of belt,which is quantified by the working time of the belt,is contained in the model to improve the predicting accuracy. Wheel speed and grinding force are set with four levels to carry out experiment. MT333 Materials Applications Practice Prof.XiaoQi Chen C.15 ANN modelling ·Black-box approach. AO TONG U Use neural network to represent input-output relationship. Use available to train NN 35 -Predicting value of roughness/um Predicting value of loss/g A-True value of roughness/um 30 -True value of loss/g 25 20 10 Output of ANN modeling MT333 Materials Applications&Practice Prof.XiaoQi Chen C.16MT333 Materials Applications C.15 & Practice Prof. XiaoQi Chen Model Wheel speed Grinding force Wear of belt Roughness Removal rate Model construction Input and output parameters • Wear of belt, which is quantified by the working time of the belt, is contained in the model to improve the predicting accuracy. • Wheel speed and grinding force are set with four levels to carry out experiment. Modelling Methods MT333 Materials Applications C.16 & Practice Prof. XiaoQi Chen 2 4 6 8 10 12 5 10 15 20 25 30 35 Predicting value of roughness/um Predicting value of loss/g True value of roughness/um True value of loss/g Output of ANN modeling ANN modelling • Black-box approach. • Use neural network to represent input-output relationship. • Use available to train NN