搭建了针对车轮轮速信号的卡尔曼滤波器,分析了不准确的观测噪声统计特征对该卡尔曼滤波器性能的影响,在此基础上利用小波变换事先估计出了轮速信号的噪声方差值。试验结果表明,该估计算法能适应轮速信号噪声方差的变化,且计算量小,解决了在未知观测噪声条件下卡尔曼滤波的失效问题。
Firstly a Kalman filter is constructed for wheel speed signal. Then the effects of the inaccuracy of measured noise statistic characteristic on the filter performance are analyzed. And on this basis, wavelet transform (WT) is used to estimate the noise variance of wheel speed signal. Test results show that this estimation method can adapt to variation of wheel speed signal noise variance, and needs less calculation, solves the failure of Kalman filter under the condition of unknown noise conditions.