To optimize the allocation of limited maintenance funds for channel infrastructure and provide a scientific basis for the formulation of building maintenance budgets, statistics are conducted on improvement building technical condition categories from 2017 to 2021, and the number of improvement buildings requiring maintenance next year is predicted through expectation and quadratic regression. The spatio-temporal distribution characteristics of improvement building service status are analyzed, and a novel method of predicting future technical condition categories in each jurisdiction is proposed. The results show that the temporal distribution of technical condition categories is relatively stable, with little change from year to year, while the spatial distribution is greatly influenced by the specific environment of different reaches, with prominent proportions of 2nd and 3rd category buildings in the upper reach and prominent proportions of the 1st category buildings in the middle and lower reaches. Additionally, the prediction accuracy of the expectation method is strongly influenced by the fluctuation degree of the time series sample, but the quadratic regression method is less affected.