The large scale capacitated arc routing problem(LSCARP) is a combinatorial optimization problem and has a wide range of applications. The divide and conquer strategy is one of the effective methods to solve the LSCARP. In order to use the divide and conquer strategy to obtain better decomposition results, an improved route cutting operator is proposed to solve the LSCARP. The proposed operator can automatically identify the path with poor shape in the path set and carry out targeted processing on it. In order to achieve better decomposition by reorganizing the divided paths in the iteration, it is beneficial for the algorithm to jump out of the local optimum and obtain a smaller final cost. In addition, since the structure of the LSCARP may affect the final effect of the algorithm, an adaptive dataset detection operator is designed, which can allocate parameters according to the relationship between the task edge and non-task edge in order to improve the decomposition quality. Finally, the above two operators are used in the SHAiD algorithm. The effectiveness of the proposed algorithm is evaluated by compared with the stat-of-the-art-algorithms.