This paper proposes a visual intelligent model for monitoring abnormal travel flows based on deep learning. The model makes use of convolutional neural network network (CNN) to study and forecast the abnormal tourist flow data, and transforms the tourist flow data into two-dimensional image form through data transformation. In the process of model training, we use a variety of techniques to improve the performance and generalization ability of the model, it includes using convolution kernel to convolution image, using ReLU activation function to increase the nonlinear expression ability of the model, and using pool layer to reduce the dimension of data. In addition, we also use data enhancement and regularization techniques to increase the diversity of training data and reduce the over-fitting problem of the model. Finally, through the visualization display technology, we will display the abnormal flow data in the form of thermal diagram, line diagram, so as to facilitate tourism managers to understand and deal with the abnormal situation in time. The experimental results show that the model can effectively monitor the abnormal flow of tourism, and has high accuracy and generalization ability.