A corrosion depth prediction model based on the electrochemical corrosion mechanism was proposed in order to address the limitations of existing steel pipe corrosion depth prediction models, which do not consider the electrochemical corrosion mechanism and are difficult to generalize to different application scenarios. Dissolved oxygen mass concentration and water temperature monitoring data were used as inputs to predict the average corrosion depth of the pipe wall. A chemo-mechanical model was proposed combined with the corrosion depth prediction model, an empirical formula for corrosion depth-mechanical performance degradation and the Chaboche constitutive model in order to estimate the uniaxial stress-strain curve of the corroded pipe material. This uniaxial stress-strain curve can serve as a basis for the safety assessment of pipeline structures throughout their entire operational lifespan. A case study of an operational water supply steel pipe validated the rationality and feasibility of the model by using real-time monitoring data.