驼峰测速雷达信号的小波滤波分析

Analysis on the Wavelet Filter of the Radar Signal for Speed Measurement in Hump

针对卡尔曼滤波方法存在的缺点,研究采用小波滤波方法进行驼峰测速雷达信号滤波。小波滤波的基本原理是对信号小波变换后的小波系数进行非线性处理,然后重构信号,滤除信号中的噪声。根据雷达信号的特点,初步选用Haar小波和二阶Dauhechies(db2)小波、3层分解、通用阈值和半软阈值算法,进行离线试验及分析。根据离线试验的滤波效果,确定选用二阶Daubechies(db2)小波、3层分解和半软阈值算法进行雷达信号滤波。利用离线试验选定的小波和算法,对采集的雷达信号进行实时滤波仿真,仿真结果与离线试验结果基本一致。将小波滤波方法与卡尔曼滤波方法对比可知,小波滤波能有效地滤除噪声、提高信噪比、减少均方差,滤波效果比较理想。因此,采用小波滤波方法进行驼峰测速雷达信号滤波,可以获得更准确的车速。

Since there are disadvantages existed in Kalman filter method, wavelet filter method is developed to filter the radar signal for speed measurement in hump. The basic principle of wavelet filter is to nonlinearly process the wavelet coefficient of the signals after wavelet transform, and reconstruct the signals afterwards to filter the noise in the signals. According to the characteristics of the radar signal, Haar wavelet, second-order Dauhechies （db2） wavelet, three-layer decomposition, universal threshold and semisoft threshold algorithm are preliminarily selected to carry through off-line tests and analyses. Based on the filter effect of the tests, second-order Dauhechies （db2） wavelet, three-layer decomposition and semisoft threshold algorithm are finally chosen for filtering radar signals. The collected radar signals are realtime filtered and simulated by means of the wavelet and algorithm selected from off-line tests. Simulation results basically agree with off-line test results. Through the comparison between wavelet filter method and Kalman filter method, it shows that the wavelet filter method can effectively filter the noise, improve the signal to noise ratio and reduce the mean square deviation. The filter effect is relatively ideal. Accordingly, to filter the radar signals for speed measurement in hump by wavelet method can obtain the train speed with more accuracy.