Safety is the primary prerequisite for the development of a city. Underground pipelines (UP) carry the development of a city. The research on the SM methods of UP has become a hot topic. This paper combines machine learning (ML) and parameter statistics (PS) to study and analyze the new method of underground pipeline SM. This paper briefly analyzes the layout principle of urban UP and the SM system of the spatial layout of UP, discusses the safety evaluation indexes of UP, combines ML and PS, and studies the new method of SM with bending stress as the most critical safety evaluation index of pipelines; At last, the experimental results show that the new method of merging ML and parameter statistical security monitoring proposed in this paper is effective and feasible.