Prediction of workpiece dynamic motion using an optimized artificial neural network

Abstract

A machining fixture is an element used to hold the workpiece in the desired position and orientation during machining. The overall machining error in a workpiece is a result of different sources of errors in a workpiece–fixture system. One among them is the motion of the workpiece under the action of cutting forces. Evaluation of this dynamic motion is essential for the determination of the overall machining error. Most commonly, the finite element method is employed to compute the workpiece dynamic motion. During optimization of fixture layout, a large number of layouts are generated and the workpiece dynamic motion must be computed for each of the layouts. In such cases, use of the finite element method is prohibitive because of the long computation time required. Also, the results of the finite element analysis are susceptible to different parameters used in the analysis. Hence, an alternate and efficient methodology is necessary to determine the workpiece displacement for a given fixture layout. This article proposes a method of using an artificial neural network for the prediction of workpiece dynamic motion. Different layouts are obtained using a modular fixture and actual machining is performed on the workpiece. For each layout, the workpiece dynamic motion is computed at select datum points and an artificial neural network is trained with these data. To achieve better prediction capability of the artificial neural network and minimize different forms of errors in training and generalization, critical parameters of the artificial neural network are optimized using a genetic algorithm. Then, this optimized network is employed to predict the workpiece dynamic motion for any arbitrary layout. Results show that the optimized artificial neural network is capable of predicting the workpiece dynamic motion with acceptable accuracy (maximum absolute relative error 9.71%). This method, hence, can serve as an economical means of computing the overall machining error during optimization of fixture layouts.