In order to capture the hidden structure of the network and reduce the dependence of community detection models on the choice of initial parameters, this paper proposes a novel community detection model——Hierarchical Graph Clustering Network with Probability Distribution, HGCPD. Firstly, utilize graph convolutional networks to learn and cache feature representations of nodes in the graph. Secondly, introduce a hierarchical clustering method based on node pair similarity probability to recursively construct community structures at different levels. Finally, explore the problem of model hyperparameter optimization and design Bayesian optimization methods to automatically adjust parameters, thereby improving the efficiency of the model. Experiments on multiple network datasets of different scales have shown that the HGCPD model is superior to mainstream methods in terms of the accuracy and effectiveness of community detection, and the model""s interpretability has been verified through visualization.