Abstract: The current status of rail transit turnout health evaluation is analyzed, an overall development methodology and direction of turnout health evaluation is proposed. Based on the analysis of influencing factors of turnout health, the power curve of switch machine is selected as the evaluation carrier. Then, the architecture of artificial intelligence algorithm convolutional neural network, its application mechanism in turnout fault analysis and judgment are elaborated. The construction path of turnout health evaluation system based on artificial intelligence is proposed.