With the deepening of new urbanization, local government debt risk in China is accumulating and amplifying. This paper first constructs local government debt risk index system and utilizes factor analysis to calculate the overall debt risk interval. Secondly, this paper builds a micro cash flow model based on genetic algorithm to estimate the reasonable scale of future borrowing under the cash flow constraint and risk index constraints. The objective is to achieve the optimal disposable revenue of local government. The model not only estimates the total borrowing scale, but also forms the different strategies of issuing bonds at different times and with different maturities. At the same time, we can change specific parameters (such as land transfer revenue) to analyse the variation of the future debt scale. Finally, this paper carries on an empirical study of city $X$ to testify the methods and models, which provides scientific management of local government debt in the process of new urbanization.