The invention discloses an initial solution optimization method based on dynamic programming in distribution network cable variable relation identification. The method belongs to the technical field of electric power, and comprises the following steps: firstly, for I distribution transformers and J distribution lines, based on energy conservation constraints and topology constraints, establishinga linear integer programming model for distribution line change relationship identification, the linear integer programming model comprising a to-be-solved distribution line change relationship incidence matrix; enabling the line change relationship between the distribution line and all undistributed distribution transformers to be equivalent to a 0-1 knapsack problem model one by one; based on dynamic planning, solving the 0-1 knapsack problem models corresponding to the distribution lines one by one, and performing backtracking in sequence to obtain a distribution network cable variable relationship incidence matrix. According to the method, the initial solution is optimized, so that the optimization speed of the subsequent distribution network line variable relationship intelligent identification precision or approximate algorithm is increased, and the time required for identification is greatly reduced.