摘要
提出了一种基于稀疏信号分解的多阶分数阶傅里叶变换(Fractional Fourier Transform,FRFT)自适应滤波方法,用于分离加减速过程啮合频率包络调制信号,提取微弱故障特征。首先提出基于两级步长FRFT确定基函数来改进多尺度线调频基稀疏信号分解方法,然后根据分解信号将分析信号分成具有较好LFM特性的信号段,采用确定基函数时保留的最佳阶次和分数阶域聚集点对各段信号进行单阶FRFT滤波,实现多阶FRFT自适应滤波。采用该方法对变速器加减速过程振动信号进行滤波解调分析。试验结果表明:基于两级步长FRFT确定基函数,速度快、精度高、抗干扰能力强;该滤波方法计算效率高,不需要选择和设置复杂滤波器,解决了信号频率呈曲线变化时,单阶FRFT滤波失效和多阶FRFT滤波各阶次难以确定的问题,能有效剥离出啮合频率包络调制信号,滤波分量的解调谱能有效提取出早期齿轮故障微弱特征。
A multi-order Fractional Fourier Transform （FRFT） adaptive filter based on sparse signal decomposition （SSD-based MOFAF） is proposed,and is applied to extract the weak fault feature by separating the AM-FM signal of meshing frequency from the accelerating-decelerating process.First,the method of ascertaining the basic function by two-step FRFT is proposed to improve the sparse signal decomposition method based on multi-scale chirplet,then the signal is divided into many sections,in which the LFM property of signal is good,according to the time domains corresponding to the decomposed signal,and the signal in each section is filtered by single-order FRFT adaptive filter with the best order and fractional concentrating position reserved as the basic function is ascertained,so the SSD-based MOFAF is carried out.The AM-FM signal of meshing frequency is filtered by the SSD-based MOFAF,and the filtered signal is analyzed by demodulation.Experimental results show that the basic function is ascertained rapidly,precisely and robustly by two-step FRFT,the SSD-based MOFAF is efficient and not necessary to select and set the filter and its parameters,and is able to solve the problem of the invalidation of single-order FRFT filter and the difficulty of order ascertain of multi-order FRFT filter.The AM-FM signal of meshing frequency is filtered well by SSD-based MOFAF and the weak fault feature of early fault is extracted effectively by demodulation of the filtered signal.
出处
《振动工程学报》
EI
CSCD
北大核心
2013年第5期771-778,共8页
Journal of Vibration Engineering
基金
总后勤部预研资助项目(AS407C001)
