Abstract: The installation of wireless sensor networks in substations has become an inevitable trend in the development of smart grids. Aiming at the interference problem caused by partial discharge impulse noise in the substation to LoRa wireless communication, the principle of LoRa modulation and demodulation was first analyzed. On this basis, a LoRa simulation model was established and its accuracy was verified according to the demodulated signal-to-noise ratio under Gaussian white noise. Secondly, the Middleton Class A impulse noise model was established, and the accuracy of the measured partial discharge signal of the Middleton Class A impulse noise of specific parameters was verified by mathematical statistical methods. Finally, the error rate curve was used to analyze the difference in LoRa communication performance between a noise environment and a Middleton Class A impulse noise environment with specific parameters. The research results show that LoRa has excellent anti-noise ability, allowing data reception even when the signal-to-noise ratio is negative; the measured partial discharge signal of a 500 kV substation can be expressed as Middleton Class A impulse noise with specific parameters. Compared with Gaussian noise, the impulse noise will cause loss of signal-to-noise ratio to the LoRa system, resulting in a decrease in system communication performance. The research results can provide a theoretical reference for the deployment of LoRa wireless communication networks in substations.