With the rapid development of urbanization in China, there are more and more bridge projects, among which bridge health monitoring is a very important project, and its safety is related to people’s immediate interests. This article selects appropriate neural network structures, such as deep neural networks, recurrent neural networks, etc., and uses them as inputs to train the model, adjust weights and biases, and achieve accurate learning and expression of bridge health status characteristics. Neural network models can be used to quickly calculate and analyze real-time data received, achieving the goal of real-time monitoring of the health status of bridge structures. The temperature at 08:00 on January 1, 2023 is 10 °C, the vibration frequency is 1.2 Hz, the displacement is 0.5 mm, and the crack width is 0.1 mm; The temperature from January 1, 2023 to 10:00 is 12 °C, the vibration frequency is 1.1 Hz, the displacement is 0.4 mm, and the crack width is 0.1 mm. The research results of this article cam provide new ideas for bridge structure health monitoring based on neural networks.