This paper analyzes the causes of servo motor vibration and summarizes the main vibration suppression methods from a large number of documents. It analyzes and compares the suppression methods for torsional vibration and torque pulsation from the perspectives of dynamic indicators and system structure. The paper points out the problems existing in current vibration suppression methods and provides an outlook on the development of vibration suppression methods.
With the development of servo drive technology, servo systems are widely used in industrial robots, CNC machine tools, etc. The advancement of industrial robots and CNC machine tools depends on machining accuracy. When a servo motor drives machinery, the rigidity and structure of the machinery, clearance, and axis center offset can all cause motor vibration. Motor vibration leads to a decrease in equipment accuracy. If the accuracy of industrial robots and CNC machine tools is affected by motor vibration, it will affect the quality of the machined equipment. From this perspective, research on motor vibration suppression is essential. Motor vibration suppression is an important branch of motor control, and research on motor vibration suppression is also of great significance to promoting motor control. Research on control technology aimed at vibration suppression has developed rapidly, especially on torsional vibration caused by low rigidity of mechanical systems, and many solutions have been proposed. Traditional PID control has been a common method for the past 50 years. Subsequently, control methods combining observers, notch filters, and filters with PID control have emerged to improve traditional PID control. To date, control methods combining adaptive algorithms, fuzzy control theory, sliding mode variable structure, H∞ control theory, neural networks, and genetic algorithms have been proposed, as well as compensation methods incorporating Fast Fourier Transform (FFT). In addition, the state-space methodology, a method for analyzing system dynamics proposed in the early 1860s, has only recently begun to be applied in industry. In recent years, a control method with high suppression of response nonlinearity has also been proposed.
Electric motor vibration is classified into mechanical vibration and electromagnetic vibration. Torsional vibration and torque pulsation are two typical types of vibration, and many scholars have studied them. This paper summarizes a large number of papers from the past five years and a small number from five years ago, focusing on the analysis of methods for suppressing torsional vibration and torque pulsation. It conducts a longitudinal comparison of similar methods proposed in the literature to identify the differences and applicable objects between methods. A horizontal comparison is also made of suppression methods for the same type of vibration in terms of system complexity, dynamics, robustness, and other characteristics to understand the advantages and disadvantages of various vibration suppression methods. Finally, it provides an outlook on the development of vibration suppression methods.
1. Methods for suppressing torsional vibration
Torsional vibration is prevalent in double-mass systems or double-inertial systems. The presence of torsional vibration affects the system's accuracy. Many researchers have studied the suppression of torsional vibration, resulting in numerous control methods, such as PID-based control, observer-based control, filter and notch filter-based control, H∞ control, resonance ratio control, and intelligent control. Vibration suppression methods have evolved to include composite control methods that combine multiple techniques.
1.1 Analysis of typical combined methods for suppressing torsional vibration
Each control strategy has its own drawbacks in suppressing torsional vibration. For example, PID control alone requires retuning when the input changes, resulting in poor adaptability. Similarly, notch filters alone are effective in suppressing high-frequency vibrations. To improve vibration suppression methods and make them more adaptive and robust, a combined control method that integrates multiple vibration suppression methods can effectively achieve this goal. However, this also introduces new problems. Several typical combined control methods will be analyzed below.
1.1.1 Observer-based composite vibration suppression method
In a dual-inertial system, the ratio of the anti-resonance frequency to the resonance frequency is related to the inertia ratio, which significantly impacts the performance of vibration suppression control. Based on this, Liangsong Huang et al. proposed an adjustable inertia ratio control strategy based on a disturbance observer design. This strategy aims to suppress torsional vibration by adjusting the inertia ratio to an optimized level. This method is simple and easy to implement, with low system complexity, enabling the servo system to operate smoothly and stably. However, this method makes numerous assumptions during the transformation of the formula, the validity of which remains to be discussed. Even assuming these assumptions hold true, the system's scalability is relatively low.
WenLi proposed a method for vibration suppression in a dual-inertial system using a fractional-order disturbance observer and a neuron-based PI fuzzy controller. The fractional-order disturbance observer is used to obtain disturbance estimates and generate compensation signals, while the neuron-based PI fuzzy controller is used to implement outer-loop control. The advantage of using a fractional-order disturbance observer is that, by introducing fractional computation, the Q-filter is extended from the integer domain to the real domain, allowing for a wider trade-off between appropriate robustness and vibration suppression, thus enhancing the system's robustness.
1.1.2 Composite Vibration Suppression Method Based on Notch Filter
Filters and notch filters play an important role in vibration suppression of dual-inertial systems. Lü Jin et al. proposed a method of adding a notch filter to a PI controller to filter out harmonic components in the velocity, thereby suppressing torsional vibration.
Stonecheng et al. proposed a method to improve the bandwidth of the velocity control loop using an adaptive notch filter. This structure is based on the control structure of the adaptive IIR notch filter technology, which solves the shortcomings of the adaptive IIR notch filter method, which can only eliminate high-frequency vibrations and has a large computational load, and makes it applicable to the suppression of low-frequency vibrations.
Adaptive notch filters can automatically identify vibration frequencies and adjust their parameters to achieve automatic suppression, exhibiting strong adaptability.
1.1.3 Analysis of Composite Control Method Based on H∞ Control Theory
Controllers designed based on H∞ control theory can maintain system characteristics under perturbation, and this advantage makes them widely used in vibration suppression. For example, Shigeo Mofimoto et al. proposed a two-degree-of-freedom controller design based on H∞ controllers to improve the trajectory characteristics of velocity commands. Higher-order H∞ controllers can be applied to practical systems. Compared with traditional PI speed control, this method has significantly improved robustness and dynamic response, and can effectively suppress torque rotation and disturbances.
JianFU designed a system combining an observer and an H∞ filter. He also proposed a CCHFLo (Compensative Control Based on H∞ Filter) control method. This method effectively suppresses vibrations and is robust; however, the system is relatively complex.
Teresa Orlowska-Kowalska and Krzysztof Szabat proposed another control method combining H∞ control theory with neurofuzzy logic. This method uses adaptive neurofuzzy sliding velocity control, which can effectively suppress torsional vibrations, has a wide range of system parameters, strong robustness, and good dynamic response.
In addition, Jianhuiwang et al. proposed a new predictive control model based on optimal control theory. This model can make the dynamic response of the system close to the reference input, and the system dynamic curve is smooth. However, the reference input used by the authors is very small, limiting its applicability and requiring further improvement. Another method utilizing predictive models is the vibration suppression achieved by using the anticipatory observation function of the TurboPMAC motion controller, proposed by Niu Zhigang et al. Anticipatory observation is essentially a control method with a variable interpolation period, independent of the kinematic model of the specific mechanism.
1.2 Comparison of methods for suppressing torsional vibration
In summary, traditional PI control systems are relatively simple, but their effect on reducing vibration is not significant. In the process of using PID control, once the parameters are tuned, they cannot be changed, resulting in poor robustness. Filters and notch filters can suppress high-frequency vibrations, automatically identify vibration frequencies, and automatically adjust parameters, exhibiting strong adaptability and robustness, but they have a significant impact on the system's dynamics. Resonance ratio control (RRC) has good suppression capabilities when the resonant frequency and anti-resonant frequency values are large. Intelligent control methods are complex but possess excellent adaptability and robustness. H∞ control can effectively control the robustness of the system, but this method comes at the cost of sacrificing the system's dynamics.
