B.Y. Lee

Modeling the surface roughness and cutting force for turning

Abstract In this paper, an abductive network is adopted to construct a prediction model for surface roughness and cutting force. This network is composed of a number of functional nodes, which are self-configured to form an optimal network hierarchy by using a predicted square error (PSE) criterion. Once the process parameters (cutting speed, feed rate and depth of cut) are given, the surface roughness and cutting force can be predicted by this network. To verify the accuracy of the abductive network, regression analysis has been adopted in the paper to develop a second prediction model for surface roughness and cutting force. Comparison of the two models indicates that the prediction model developed by the abductive network is more accurate than that by regression analysis. Experimental results are provided to confirm the effectiveness of this approach.

Surface roughness inspection by computer vision in turning operations

The use of computer vision techniques to inspect surface roughness of a workpiece under a variation of turning operations has been reported in this paper. The surface image of the workpiece is first acquired using a digital camera and then the feature of the surface image is extracted. A polynomial network using a self-organizing adaptive modeling method is applied to constructing the relationships between the feature of the surface image and the actual surface roughness under a variation of turning operations. As a result, the surface roughness of the turned part can be predicted with reasonable accuracy if the image of the turned surface and turning conditions are given.

Cutting-parameter selection for maximizing production rate or minimizing production cost in multistage turning operations

Abstract In this paper, an investigation of optimal cutting parameters for maximizing production rate or minimizing production cost in multistage turning operations is reported. A machining model is constructed based on a polynomial network. The polynomial network can learn the relationships between cutting parameters (cutting speed, feed rate, and depth of cut) and cutting performance (surface roughness, cutting force, and tool life) through a self-organizing adaptive modeling technique. Once the geometric model for machined parts and various time and cost components of the turning operation are given, an optimization algorithm using a sequential quadratic programming method is then applied to the polynomial network for determining optimal cutting parameters. The optimal cutting parameters are subjected to an objective function of maximum production rate or minimum production cost with the constraints of a permissible limit of surface roughness and cutting force and a feasible range of cutting parameters.

Modeling of the process damping force in chatter vibration

Abstract This paper presents a new mechanistic model to study the complex and highly nonlinear process damping force in chatter vibration. In the developed model, a feedforward neural network is used to model cutting force components. The process damping force due to the interface between the tool flank and machined surface is estimated through the calculation of the volume of the work material displaced by the tool flank. To properly calculate the volume of the displaced work material, the vibration of the tool relative to the workpiece is solved using the equations of motion iteratively until a convergence criterion is satisfied. The study has shown that the developed model is much better than previous models in the analysis of dynamic behaviors of the nonlinear process damping force in chatter vibration.

Modeling and optimization of drilling process

In this paper, the use of an abductive network for modeling drilling processes is first described. The abductive network is composed of a number of functional nodes, these nodes being self-organized to form an optimal network architecture by using a predicted squared error (PSE) criterion. Once the process parameters (drill diameter, cutting speed and feedrate) are given, the drilling performance (tool life, metal removal rate, thrust force and torque) can be predicted by this developed network. A simulated annealing optimization algorithm with a performance index is then applied to the developed network when searching for the optimal process parameters. Experimental results are provided to confirm the effectiveness of this approach.

The optimal cutting-parameter selection of production cost in HSM for SKD61 tool steels

The main purpose of this study was to construct an investigation of optimal cutting parameters for minimizing production cost on the rough machining of high speed milling operation. A machining model is constructed based on a polynomial network. The polynomial network can learn the relationships between cutting parameters (cutting speed, feed per tooth, and axial depth of cut) and tool life through a self-organizing technique. Once the material removal volume for machined parts and various time and cost components of the high speed milling operations are given, an optimization algorithm using a simulated annealing method is then applied to the polynomial network for determining optimal cutting parameters. The optimal cutting parameters are subjected to an objective function of minimum production cost with the feasible range of cutting parameters.

The model of surface roughness inspection by vision system in turning

Abstract The paper presents a system for measuring surface roughness of turned parts through computer vision system. The images of specimens grabbed by the computer vision system could be treated by some techniques to get the features of image texture (major peak frequency F1, principal component magnitude squared value F2, and the standard deviation of grey level STD). These features could be used as input data of a abductive network. Using the trained abductive network, the experimental result had shown that the surface roughness of turned parts measured by computer vision system over a wide range of turning conditions could be got with a reasonable accuracy compared with those measured by traditional stylus method. Comparing with the stylus method, the constructed computer vision system is a useful method for measuring the surface roughness with faster, lower price, and lower environment noise in manufacturing process.

An analytical model of chatter vibration in metal cutting

Abstract A new analytical model of chatter vibration in metal cutting is presented. The basic cutting mechanics adopted in the model is derived from a predictive machining theory based on a shear zone model of chip formation. A feature of this model is that variations of the undeformed chip thickness and rake angle due to the machine tool vibration are taken into account in determining the cutting forces and the forces are then coupled with the equations of motion to solve for the vibrational amplitudes with iterative techniques. Non-linearities in dynamic cutting processes caused by the effects of tool disengagement from the cut and cutting process damping are also included in the model. It is shown that the proposed model can be applied to make predictions for the suppression of chatter vibration by a change of tool geometries.

In-process prediction of corner wear in drilling operations

Abstract The paper presents an in-process prediction of corner wear in drilling operations by means of a polynomial network. The polynomial network is composed of a number of functional nodes and well organized to form an optimal network architecture using an algorithm for synthesis of polynomial networks (ASPNs). Thrust force or torque in drilling operations has been correlated with corner wear in this study. It has been shown that the thrust force is better than the torque as the sensing signal for the in-process prediction of corner wear. Experimental results have shown that the corner wear over a wide range of drilling conditions can be predicted with a reasonable accuracy if the cutting speed, feed rate, drill diameter, and thrust force are given.

An abductive network for predicting tool life in drilling

This paper presents an abductive network for predicting tool life in drilling operations. The abductive network is composed of a number of functional nodes. These functional nodes are well organized to form an optimal network architecture by using a predicted squared error criterion. Once the drill diameter, cutting speed and feedrate are given, tool life can be predicted based on the developed network. Experimental results have shown that the abductive network can be effectively used to predict drill life under varying cutting conditions and the prediction error of drill life is less than 10%.