graduate student researcher, UCLA My research focus is on mechatronics and control
Use of controls and robotics to increase productivity and avoid vibrations in machining
My research studies the use of industrial robot manipulators to act as active steady rest in machining thin-walled parts to prevent machining chatter instability. Machining is the process of material removal from more commonly metal parts to render desired shapes or sizes. Thin walled parts are light weight and popular in aerospace and automotive industries. Regenerative chatter which causes unwanted vibrations in the machining system is a bigger problem with thin-walled parts due to their higher flexibility. Chatter causes tool wear and damage, bad surface finish on the machined part and high levels of noise in the work place. To avoid chatter lower speeds and depths of cut are usually chosen which reduce the productivity of the plant. Using robot manipulators, which are already available in machining facilities and are used for other purposes, is a low cost solution to the chatter problem. In my work feedback control is applied to the robotic steady rest in contrast to the other more popular active control method where the cutting tool is controlled. My approach leaves the tool side rigid which in turn helps with chatter avoidance. The entire dynamic system model is characterized by delay differential equations with time varying interactions between the tool side and work side dynamics. Semi-discretization, finite element modeling, and model reduction are applied to obtain a lower order plant model for the feedback control design. My results demonstrate the effectiveness of the proposed modeling and control approach. Robot manipulator control coupled with the machining system leads to higher material removal rates with no chatter.
Abstract: Regenerative chatter is self-excited vibration which may occur during the machining process. Chatter will inhibit the improvement of productivity, produce poor surface finish, even damage the tool or machine and so on. Simultaneously, cutters with variable pitch angle or non-uniform helix angle are helpful for avoiding the regenerative chatter. So, the stability prediction for the cutter, especially for the cutter with non-uniform pitch and helix angle has become more and more important for current high performance machining process. In this paper, in order to combine the effect of the cutter’s variable-helix angle’s effect, the mathematical model with multi-delays of the dynamic machining process is firstly constructed. In this model, the cutter is divided into a series of cutting elements along the cutter axis. Then the model of the chatter for non-uniform helix tools can be deduced after all cutting elements’ dynamic equations are obtained. Subsequently, every period of the rotational spindle is divided by a series of knots which are not equally distributed for the sake of computing this model. After that, a method is proposed to predict the stability boundaries. Finally, four groups of examples are conducted to verify the validity of the proposed method. And the first group of the example is 1-dof dynamic system, the radial cutting depth is very small and two types of cutters with uniform and non-uniform pitch angles are chosen. The second section is made up of 2-dof dynamic systems utilizing three types of cutters, they are the cutter with uniform pitch and zero helix angle, the cutter with variable pitch and zero helix angle as well as the cutter with variable pitch and non-zero helix angle. Another group concludes two cutters with non-uniform helix angle for 1-dof dynamic machining system. The last group is composed by an experiment with thin-wall part. The comparing results show that the curves computed using the presented model agree closely to that in literatures and experiment, which proves that the model is correct and the stability lobes can be predicted with high accuracy whatever the dynamic system with the radial cutting depth is small or large, the dynamic system is 1-dof or 2-dof and the cutter’s pitch angle of the dynamic system is variable or uniform.
Pub.: 18 Feb '16, Pinned: 03 Jul '17
Abstract: Depending on the machining process chatter might occur at an eigenfrequency of the machine's structure. Electrodynamic proof-mass actuators can be attached to the structure in order to mitigate chatter. This paper gives an overview of different existing control strategies for active damping and compares them with one another. First, the control strategies were implemented and tested in a coupled simulation model. Then, the simulation results were validated by modal tests. For a sample process the analytically predicted depths of cut were finally verified in cutting tests.
Pub.: 30 May '16, Pinned: 03 Jul '17
Abstract: Chatter happens due to the low rigidity of the machining system, and chatter identification is essential for improving the surface quality and productivity. More understanding on the dynamic characteristics of the machining system would help identify chatter more efficiently. This paper proposes a method for detecting chatter in the milling process of the thin-walled workpiece by monitoring and analyzing the cutting torque. A varying chatter detection threshold based on the workpiece geometries, tool path, and dynamic characteristics is utilized to identify chatter. A high-frequency filter and a median filter are utilized for signal denoising. The method is experimentally validated, which proved that it could predict chatter efficiently at its early infancy. The method could be a supplementary to other chatter identification methods.
Pub.: 02 Aug '16, Pinned: 03 Jul '17
Abstract: In this paper, the well-established problem of self-excited vibrations in machining is revisited to include the nonlinearity of process damping at the tool and workpiece interface. Machining dynamics is modeled using a time-delayed system with nonlinear damping, and the method of averaging is used to obtain the amplitude of the resulting limit cycles. As a result, an analytical relationship is presented to establish the stability charts corresponding with arbitrary limit cycles in machining systems. The presented analytical solutions are verified using experiments and numerical solutions.
Pub.: 18 Jan '17, Pinned: 28 Jun '17
Abstract: This article extends a recent publication [MSSP (2016), 176−196] by developing a Rayleigh-Ritz model of a thin cylindrical shell to predict its response subject to concurrent active and passive damping treatments. These take the form of piezoelectric patches and a distributed cardboard liner, since the effects of such combined treatments are yet to be investigated. Furthermore, prior literature typically considers only the “bimorph” active patch configuration (with patches on the inner and outer shell surfaces), which is not feasible with an interior passive liner treatment. Therefore, a novel configuration—termed as “unimorph”—is proposed and included in the model. Experiments are performed on a shell with active patches (under harmonic excitation from 200 to 2000 Hz) in both the bimorph and unimorph configurations to provide evidence for the analytical model predictions. The proposed model is then employed to assess competing control system designs by examining local vs. global control schemes as well as considering several alternate active patch locations, both with and without the passive damping. Non-dimensional performance metrics are devised to facilitate comparisons of vibration attenuation among different designs. Finally, insertion loss values are measured under single-frequency excitation to evaluate several vibration control designs, and to compare the effects of alternate damping treatments.
Pub.: 28 Dec '16, Pinned: 28 Jun '17
Abstract: The impulse excitation of mechanism causes transient vibration. In order to achieve adaptive transient vibration control, a method which can exactly model the response need to be proposed. This paper presents an analytical model to obtain the response of the primary system attached with dynamic vibration absorber (DVA) under impulse excitation. The impulse excitation which can be divided into single-impulse excitation and multi-impulse excitation is simplified as sinusoidal wave to establish the analytical model. To decouple the differential governing equations, a transform matrix is applied to convert the response from the physical coordinate to model coordinate. Therefore, the analytical response in the physical coordinate can be obtained by inverse transformation. The numerical Runge-Kutta method and experimental tests have demonstrated the effectiveness of the analytical model proposed. The wavelet of the response indicates that the transient vibration consists of components with multiple frequencies, and it shows that the modeling results coincide with the experiments. The optimizing simulations based on genetic algorithm and experimental tests demonstrate that the transient vibration of the primary system can be decreased by changing the stiffness of the DVA. The results presented in this paper are the foundations for us to develop the adaptive transient vibration absorber in the future.
Pub.: 29 Jan '17, Pinned: 28 Jun '17