Interlocking point prediction method in high-speed rail graded broken stone vibration compaction process

The invention provides an interlocking point prediction method in a high-speed rail graded broken stone vibration compaction process, which comprises the following steps of: S1, fitting physical and mechanical characteristics in the high-speed rail graded broken stone compaction process by adopting a hyperbolic model, calculating and constructing original data, and determining an interlocking point Tlp; s2, acquiring data, and analyzing Tlp master control characteristics by adopting a grey correlation analysis (GRA) algorithm; s3, establishing an IPSO-ML mixed Tlp prediction model based on an improved particle swarm IPSO algorithm and a machine learning ML algorithm; s4, adopting an interpretability method SHAP to interpret and analyze the optimal Tlp prediction model, and determining the optimal Tlp of the vibration compaction process; the compaction quality is controlled by predicting the interlocking point Tlp, the problems of uncertain compaction time, single evaluation index and the like existing in eval