提出一种分维数和子波重构相结合的混沌信号与噪声分离方法，与以往出现的子波滤波方法和经典滤波方法相比较，该方法有如下特点：利用子波变换计算带观测噪声混沌信号的真实分数维，进而根据观测序列维数与真实维数的差值控制子波重构的参数，构造出混沌与噪声分离的自适应滤波器结构。仿真实验说明了该算法的有效性。 This paper proposes a method of separating chaos signal and noise by combining fractal dimension and wavelet reconstruction. Compared with the classical wavelet filtering method and classical filtering method, this method has the following characteristics: Using wavelet transform to calculate the chaotic signal with the observed noises The real fractal dimension of the signal is obtained, and then the parameters of the wavelet reconstruction are controlled according to the difference between the dimension of the observation sequence and the real dimension, and the structure of the adaptive filter separating the chaos from the noise is constructed. Simulation experiments show the effectiveness of the algorithm.