Dynamic compaction replacement depth estimation method based on machine learning

The invention discloses a machine learning-based dynamic compaction replacement depth estimation method, relates to the technical field of machine learning, and effectively and accurately determines the dynamic compaction replacement depth. Various data of a plurality of tamping scenes are collected, corresponding historical tamping event data are generated, a tamping orthogonal table and a tamping rule model are established according to the historical tamping event data, and a model orthogonal experiment is performed on the tamping rule model according to the event orthogonal table. Then, a corresponding dynamic compaction replacement depth estimation formula is generated, a plurality of pieces of real-time compaction scene data are collected and input into the compaction rule model and the dynamic compaction replacement depth estimation formula, then, the compaction rule model outputs a compaction decision, and the real-time compaction scene is subjected to compaction according to the compaction decision. A