Modelling, fault detection and active vibration control of rotor dynamic systems with applications in machine tools
Loading...
Date
2020-05
Authors
Vashisht, Rajiv Kumar
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
Rotors are extensively used to transfer energy from one part to other parts of a mechanical system. During the service life of the rotors, they are subjected to different faults like the rotor crack, coupling misalignment, rotor bow, rotor-stator rubbing and unbalance. One of the important aspects of these faults is their interconnection. Presence of one fault can accelerate the growth of the other. If these faults are not identified timely, costly breakdown of the rotor system can occur. The presence of these faults changes the overall nature of the system dynamics from the linear to nonlinear one. This aspect can be used to identify the presence of these faults. Various system components like ball bearings and journal bearings also make the overall dynamics of a healthy rotor system nonlinear. Hence, it is not easy to detect the presence of faults based on the vibration signal. Transient response-based crack detection technique has been developed and verified using simulation techniques.
Rotor dynamics with transverse crack, coupling miss-alignment and rotor-stator rubbing can be represented using ordinary differential equations with time-periodic coefficients and intermittently varying coefficients. A new fault detection strategy based on short time Fourier transformation has been developed to detect the presence of crack and rotor stator rubbing even if the both coexist simultaneously. Due to the presence of unbalance, rotor systems with constant spinning speed are subjected to steady-state vibrations that can accelerate the crack growth or chances of rotor/stator rubbing action. During changes of the spinning speed, transient vibrations corresponding to resonance frequencies are also generated. To compensate both types of vibrations, a hybrid controller is successfully developed. Electromagnetic actuators are used to generate required control forces. From the spectrum of the control signal, rotor faults like transverse crack and coupling miss-alignment can be efficiently diagnosed.
Applications of rotor dynamics can also be seen in machine tools. The spindle and ball screw drives are important examples of their applications. In the spindle, transverse and torsional vibrations are produced. However, in the ball screw drives, axial vibrations are generated. For precision control of the drives, a hybrid controller is proposed to increase command tracking efficiency and reduce the dynamic deflection of the carriage during metal cutting process. Experimental verification of the proposed controller has been done to verify its effectiveness. The dynamics of cutting tools is represented in the form of time-delay systems. Different time-domain algorithms are used to identify stable regions of machining operations. To improve the depth of cut, the active chatter control of turning and boring operations is carried out. Experimental results verify the effectiveness of the proposed active chatter control strategies. Based on the relative motion between the workpiece and cutting tool (during boring operation), cutting operations can be divided into different categories. The dynamics of such systems is represented using delay differential and time-periodic delay differential equations. Feasibility of active chatter control for each category is analyzed in detail. The effect of eccentricity and workpiece ovality on the dynamics of boring operations is thoroughly investigated. Linear Parameter Varying controllers are proposed to compensate for the changing dynamics of the system due to continuous metal removal in turning operations. Fractional-order proportional derivative controllers are designed using response optimization techniques and implemented in boring operations. Experimental results verify that these controllers save the control energy in active chatter control applications.
Machine learning techniques are used for online chatter detections in milling operations. Various chatter indicators are proposed. Improved dynamic equations are proposed for milling operations to find the behaviour of vibrations during chatter conditions. Simulated data are used to train Convolutional Neural Networks that are then used to detect chatter in an online manner. Experimental results verify the effectiveness of the proposed technique
Description
Keywords
Rotor dynamics, machine tools, chatter detection , deep learning