In broadband power line carrier communication in new energy grid connection, phase distortion and time delay distortion may occur when signals of different frequencies propagate in the power line. The existence of noise will affect the clarity and stability of the signal, and a series of problems such as packet loss or error. In order to reduce the influence of noise interference on broadband power line carrier communication in new energy gridconnected, this paper proposes an improved compressed sensing denoising method based on new energy grid-connected power line communication. This method takes the compressed sensing theory as the core, compresses and collects the field signal and transmits it to the terminal sparse reconstruction, and filters out the noise at the same time. The algorithm uses spectral decomposition to increase the column independence of the observation matrix. At the same time, the optimized H-Toeplitz observation matrix design method is used to reduce the cross-correlation between the observation matrix and the sparse base. The observation matrix that strictly satisfies the RIP condition is constructed instead of the Gaussian matrix, which improves the robustness of the method. The experimental results show that the signal-to-noise ratio (SNR) of the optimized observation matrix is 5.45 dB higher than that of the Gaussian observation matrix, and the root mean square error (RMSE) is reduced by about 32.67 % on average. The algorithm in this paper has better denoising effect. It can be adjusted and optimized more flexibly for colored background noise, effectively reduce the noise components in the signal, improve the signal-to-noise ratio, and make the signal clearer and easier to identify and analyze. It ensures that the signal is more stable in different environments, better resists the influence of noise and interference, improves the robustness of the system and the reliability of data transmission.