提出一种基于峭度能量的谱相关分析方法.它利用每个循环频率切片的峭度值来衡量该循环频率的调制能力,并以此作为加权因子进行循环频率的能量累积,最终实现故障特征的有效提取.相对于传统的谱相关分析方法,本文方法降低了信号中多倍频谐波对故障特征频率的干扰,能更清晰准确地提取出故障频率特征.利用传统的谱相关分析方法、本文方法和共振解调三种方法对仿真信号、低速重载试验台的滚动轴承外圈故障信号进行分析,证明了本文方法的有效性. A spectral correlation analysis method based on kurtosis energy is proposed, which uses the kurtosis value of each frequency slice to measure the modulation ability of the cyclic frequency and uses it as a weighting factor to accumulate the energy of the cyclic frequency and finally realize the fault feature Compared with the traditional spectral correlation analysis method, the proposed method reduces the interference of multifrequency harmonics in the signal to the fault characteristic frequency and can extract the fault frequency characteristics more clearly and accurately.Using the traditional spectral correlation analysis method, The method of this paper and three methods of resonance demodulation are used to analyze the fault signal of the rolling bearing outer ring of the simulated signal and the low-speed and heavy-duty test rig, which proves the effectiveness of the method in this paper.