基于变参数域和短时高斯线性调频基的自适应信号分解算法

摘要

基于高斯线性调频基的参数化时频分析方法由于具有很高的时域和频域分辨能力，而被广泛应用于非线性非稳态信号的分解和特征提取中，但其巨大的计算量常常让工程人员望而生畏。因此结合变参数域和短时傅里叶变换的方法提出了一种改进的短时高斯线性调频基自适应信号分解算法，将四参数优化问题转化成窄带范围的两参数优化问题，提高了参数化时频分析的时效性。利用改进算法对四原子组合的非线性解析信号和动检列车轴箱振动加速度信号进行分解，结果表明该方法能有效消除交叉项干扰，时频分辨率高，而且具有计算量小，速度快的优点，对分析动检列车轴箱振动与轮轨短波冲击有实际意义。%The parametric time-frequency analysis method based on Gaussian chirplet function has the best time-frequency resolution,so,it is widely used in non-linear and non-stationary signal decomposition and feature extraction. But it needs a large amount of computation.A reformed short time Gaussian chirplet signal decomposition algorithm based on variable parameters domain method and short time Fourier transform (STFT)was proposed.Taking as an example,it tranfers a four parameters optimization problem to two parameters one in a narrow range and improve the efficiency of computation.The reformed algorithm was used to decompose a four atoms non-linear analytic signal and the vibration accelation signal of a high speed comprehensive inspection train's axle box.The results show the algorithm can avoid the cross-term's interferer and achieve fast computation.It can be applied to analyze the vibration of axle box and the wheel-rail shortwave shock.