【摘要】 测量仪表获得的信号带有噪声 ,使内燃机关键参数的计算比较困难 ,已有的FIR和IIR数字滤波器的去噪效果较差 ,且使内燃机的瞬时转速曲线产生了相移 ,难以确定速度曲线局部极值点的位置 .为去除信号中的噪声 ,提出了一种基于多分辨率分析的相关算法 ,首先对原始信号进行小波变换 ,根据规划后的相关函数 ,仅选择对应于信号的小波变换的模极大值 ,将信号和噪声分离开来 ,然后再重建信号 .将该方法用于R410 2型内燃机的瞬时转速曲线的信号滤波中 ,并与IIR滤波后的结果进行比较 .可以看出 ,相关算法不仅使滤波后的曲线更加平滑 ,还避免了IIR滤波中存在的相位滞后 .这表明相关滤波法在精确地重建原信号的同时 ,又能较理想地去除噪声的干扰更多还原

【Abstract】 The measured signals are accompanied with n oi ses, leading to the difficulties in the calculations of the key parameters of ga s engine. Furthermore, the noise filtering performance is poor for both FIR and IIR digital filters, causi n g a phase lag in the transient speed curve, which makes it difficult to determin e the position of the local extreme speed value. Based on the multiresolution an alysis of wavelet transform, a correlation algorithm is proposed to remove noise s from signals. According to the correlation functions, the signals are differen t iated from noises by selecting the wavelet transform maxima corresponding to the singularities. A`denoise d signal’is constructed afterwards. The algorithm is applied in the signal filte ring of transient speed for R4102 type gas engine, with comparative filtering experiment conducted by using HR filter. The results indicate that the curve af t er filtering via the proposed algorithm is smoother without the phase lag which occurs in case of HR filter, revealing its excellent filtering performance.更多还原