Y. Kang

Integrated “CAE” strategies for the design of machine tool spindle-bearing systems

This paper adopts both static and dynamic analyses to examine the necessary integrated procedures for the design of spindle-bearing systems, with modeling and analysis of these systems being based on the finite element method. The study further addresses the sub-structure procedures and the application of a commercial package dynamic solver. Within this study, the effects of design parameters on static and dynamic performance of spindle-bearing systems are analyzed in order to establish the requirement for design modifications, and the paper proposes a number of examples, along with a set of guidelines for the design of machine tool spindles.

An adaptive control using multiple neural networks for the position control in hydraulic servo system

A model following adaptive control based on neural network for the electro-hydraulic servo system (EHSS) subjected to varied load is proposed. This proposed control utilizes multiple neural networks including a neural controller, a neural emulator and a neural tuner. The neural controller with specialized learning architecture utilizes a linear combination of error and the errors derivative to approximate the back propagation error for weights update. The neural tuner is designed to adjust the parameters of the linear combination. The neural emulator is used to approximate the Jacobian of plant. The control of the hydraulic servo actuator is investigated by simulation and experiment, and a favorable model-following characteristic is achieved.

An Accuracy Improvement for Balancing Crankshafts

Due to measurement errors, the final accuracy of rotor balancing may not be satisfied. This study is based on a modified influence coefficient method associated with multi-plane technique for the improvement of accuracy in balancing crankshafts. This method extends the conventional influence coefficient method, in which two trial masses in one balancing plane are employed, to one utilizing three trial masses in one plane. On the basis of three trial runs, the balancing accuracy can then be improved by the optimization of influence coefficient matrices resulting from the minimization of measurement errors. The feasibility of this modified approach is carried out by the verification of accuracy improvement in experiments, balancing two crankshafts.

An integration method for detecting chaotic responses of nonlinear systems subjected to double excitations

Abstract A methodology designed for identifying chaos of the nonlinear systems subjected to double excitations is proposed. Based on simulations in this study, it is shown by bifurcation diagram that method of Poincare sections, the conventional chaos-observing method, fails to pinpoint the onset of chaotic motions with the nonlinear systems subjected to double excitations. To remedy this problem, “ K s integration method” is proposed, which integrates the distance between trajectories and origin in phase plane over an excitation period and designates the obtained integration values as K s s to take the roles of the sampling points derived by Poincare sections in constructing bifurcation diagram. This “ K s integration method” is shown capable of providing valuable information in bifurcation diagram such that the parameter range leading to chaos can be easily decided and the number of distinguishable time-domain responses can be determined.

Gear Fault Diagnosis by Using Wavelet Neural Networks

Fault diagnosis in gear train system is important in order to transmitting power effectively. The artificial intelligent such as neural network is widely used in fault diagnosis and already substituted for traditional methods such as kurtosis method, time analysis and so on. The symptoms of vibration signals in frequency domains have been used as inputs to the neural network and diagnosis results are obtained by network computation. This study presents gear fault diagnosis by using wavelet neural networks (WNN) and Morlet wavelet is used as the activation function in hidden layer of back-propagation neural networks (BPNN). Furthermore, the diagnosis results are compared within both methods of WNN and BPNN in four gear cases.

Thermal deformation prediction in machine tools by using neural network

Thermal deformation is a nonlinear dynamic phenomenon and is one of the significant factors for the accuracy of machine tools. In this study, a dynamic feed-forward neural network model is built to predict the thermal deformation of machine tool. The temperatures and thermal deformations data at present and past sampling time interval are used train the proposed neural model. Thus, it can model dynamic and the nonlinear relationship between input and output data pairs. According to the comparison results, the proposed neural model can obtain better predictive accuracy than that of some other neural model.

An optimal estimation for neural network by using genetic algorithm for the prediction of thermal deformation in machine tools

Thermal deformations cause 40-70% error during the manufacturing process for the machine tools. In order to improve the accuracy of the machine tools, this study proposes a hybrid model, which predicts thermal deformation by combining an ARIMA and a feed-forward neural network (FNN) models. The genetic algorithm (GA) method is used to optimize this prediction model. The GA is used to search the optimal normalization coefficients, number of ARMA outputs and number of hidden neurons of FNN. It can reduce the network size and improve the propagation accuracy. In this study, comparisons between conventional FNN and the proposed hybrid model with or without using GA. The compared results show that the proposed hybrid model has better accuracy than the conventional FNN model and most accurate can be obtained by the proposed hybrid using GA. The predicted results, the hybrid model with GA can reduce the thermal deformation to 2 /spl mu/m.