FMCW雷達測距與測速信號處理研究

Abstract

近年來FMCW雷達被廣泛利用於各種工業用途，在各不同應用領域需要各別的信號處理演算法，本研究將著力於兩個主題：一是如何提高FMCW液位計之高度量測精度，另一則是提升側向單雷達車輛偵測器之測速與車長量測精度。 本研究首先探討FMCW 雷達測距與測速原理後，提出一個高精準度頻率估計的方法-梯度搜尋法，文中除了說明梯度搜尋法原理，並分析此方法之誤差來源與推導出誤差邊界的方程式，以及在不同訊雜比下，與是否加窗函數下與 Cramer Rao 邊界進行比較。文中也會藉由模擬與硬體實作方式與其它文獻中的方法進行量測精度上的比較。經由此梯度搜尋法提高頻率估測精準度從而提高FMCW雷達距離量測精度，在距離20公尺範圍內測距精度可達±5 mm。 在另一研究領域-利用側向單雷達車輛偵測器提高量測車速與車長準確度，本研究也提出一個不同於目前側向車輛偵測器之安裝概念，利用斜視角並配合2D FFT演算法，實驗結果顯示量測速度與估計車長準確度分別可達±4 km/h與 ±1 m，即使在目標物速度小於5km/h時仍可辨識諸如行人或自行車等慢速移動物體。

The frequency modulated continuous wave (FMCW) radar is widely used in the industrial applications. However, in a specific application it requires a specific signal processing algorithm. In this research, two applications are focused, one is the high accuracy range measurement in a FMCW level gauge, and the other one is estimating vehicle speed and length using a side-looking single-radar Vehicle Detector (VD). Estimating the frequency of a single-tone signal is an essential problem in a FMCW level gauge. This research presents a frequency estimation algorithm called the gradient search method (GSM) using the derivative of discrete Fourier transform (DFT). The analytical boundaries of the proposed method for different signal-to-noise ratios in the conditions of with the rectangular window and with the Hann window are derived. This research compares the most appropriate algorithms available in the literature. Simulation and experimental results show that the proposed algorithm provides superior performance than previous methods. This research also presents a side-looking single-beam microwave vehicle detector system for estimating speed and length of multiple vehicles. The proposed vehicle detector system is equipped with a FMCW radar using a squint angle. The associated 2D range-Doppler Fourier processing can extract the range and speed information of each vehicle using a single-beam FMCW radar. The simulation and experimental results show accurate estimations of vehicle speed and length. The measurement errors of speed and length were approximately ±4 km/h and ±1 m respectively. The proposed method has excellent detection capability for small and slow moving targets, such as bikes and pedestrians, at speeds down to 5 km/h.