Velocity Estimation of Motion Systems Based on Low-Resolution Encoders

Abstract

This paper proposes a new approach to estimate the velocity of mechanical system in the case where the optical incremental encoder is used as the position sensor. First, the actual angular position is reconstructed via moving horizon polynomial fitting method by taking account of quantization feature and the plant dynamics. Then, the reconstruction signal is applied to a classical observer to obtain the velocity estimation. Its robustness against the position sensor resolution and the degree of the polynomial is discussed by some numerical examples. Experiments with very low-resolution encoder in low speed range also confirm its effectiveness.

Proposal of Position Reconstruction with Polynomial Fitting Approach for Precise Motion Control

“…Since the true system output y should be smooth and band limited due to the system dynamics, it is possible to get more precise information on y[k] from the past quantized measurement series y q [i] (i ≤ k). In the next section, we utilize the moving horizon polynomial fitting approach (Zhu and Sugie, 2012) and an observer technique to reconstruct the position measurements.…”

“…Since the model-based methods share the feature of providing good estimation performance and polynomial fitting methods can achieve a smooth reconstruction signal, a polynomial fitting approach based on 1 -norm regularization and observer is exploited to reconstruct the real position signal. Part of the idea is also considered in (Zhu and Sugie, 2012) for estimating the velocity information from the quantized measurements. However, the quantization effects on position control, such as the limit cycle oscillation phenomenon, are not studied.…”

“…However, the quantization effects on position control, such as the limit cycle oscillation phenomenon, are not studied. Moreover, in order to obtain a good velocity estimation, the input disturbance is assumed to be zero-mean in (Zhu and Sugie, 2012), which may not suitable in practical situations. In this paper, we also take the disturbance into account to make the reconstruction approach more applicable.…”

Smooth Output Reconstruction for Linear Systems with Quantized Measurements

“…Based on the above observation, Zhu et al. proposed a method to estimate the velocity of motion systems based on low‐resolution encoders . The method combines a curve fitting approach with the observer‐based approaches.…”

“…The purpose of this paper is to extend the method proposed in to output reconstruction of general linear systems while overcoming the above drawbacks. More precisely, the contribution of this paper is as follows: (i) A new method for general linear systems is given to reconstruct the smooth output from the quantized measurement in the presence of (non‐zero mean) system disturbances based on . The method to reconstruct the output uses moving horizon polynomial curve fitting, where ℓ 1 ‐norm regularization is adopted to determine the appropriate polynomial curve and the information of the system models is exploited.…”