Abstract
Low-resolution encoders are normally used as speed sensors in many traction drive systems, such as electric trains, electric vehicles, etc. For example, 60-pulse-per-revolution pulse generators are used to detect speed for electric railway vehicles and low-resolution Hall-effect sensors are commonly used in many electric vehicles. Fig. 1 illustrates the sensors. The signals from these kinds of sensors are pulses, which are then used to calculated speed and acceleration. The speed and the acceleration, therefore, are updated only when the next pulse is detected. As a result, the speed and the acceleration cannot be achieved precisely, especially at low speed. Nowadays, high-frequency digital sampling processors (DSP) are commonly used as controllers. This allows us to precisely control the system. However, it is very difficult to realize in the mentioned systems since the low-resolution sensors are installed. Fig. 2 illustrates the condition at low speed. The period between two consecutive pulses, T1, is longer than the control period, T2. It implies that precise control cannot be achieved even though the high-frequency DSPs are used. To solve this problem, an instantaneous speed observer was proposed to estimate the speed between two consecutive encoder’s pulses. The instantaneous speed observer is a specific discrete-time observer, which estimates the speed at every sampling time and corrects the estimated speed when the next pulse is detected. As a result, the precise speed is achieved at every sampling time. The instantaneous speed observer can also be used to improve the performance of linear scales for linear drives. Its concept is also useful for visual servo, where the sampling rate of the CCD camera is much slower than the sampling rate of the controller. However, the model of the instantaneous speed observer is very specific and limited to motor drive, and its application cannot be extended for the higher order plant such as robot manipulators. For traction control, one of the difficulties is that it deals with a wide speed range. The authors also applied the instantaneous speed observer for speed estimation of train speed. To apply the instantaneous speed observer, which is designed in discrete-time domain, it is necessary to change the observer gains according to the speed to prevent the noise problem at high speed and instability problem at low speed. The observer gain was tuned by simply considering only the relation of pole locations on sand zplanes. As a result, the system can operate stably in a Improvement of Performance and Stability of a Drive System with a Low-Resolution Position Sensor by Multirate Sampling Observer
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have