A Review of Research Progress on Intelligent Subgrade Compaction Methods Based on Vibration Signals

The limitations of conventional spot-sampling-based quality inspection in road construction can be effectively addressed by Intelligent Compaction (IC) technology. Acceleration signals acquired from the steel drums of vibratory rollers during the compaction process are widely used for real-time compaction monitoring. The signal-based evaluation framework is the core focus of the present work. First, fundamental mechanisms of vibratory compaction are elucidated. The theory of compaction energy transfer based on stress waves is further elaborated, thus establishing a theoretical foundation for compaction quality assessment based on acceleration signals. Second, a systematic overview of a dedicated signal processing workflow is presented, which comprises three sequential core procedures: signal acquisition, filtering, and time–frequency transformation. Third, Intelligent Compaction Measurement Values (ICMVs) derived from time-domain, frequency-domain, and time–frequency-domain analyses are classified, with their underlying calculation principles clarified. Finally, a comparative analysis of the predictive performance of three types of compaction control indicator prediction models, namely linear regression, traditional machine learning, and deep learning, is performed. Their applicable scenarios are evaluated. Meanwhile, the system architecture and core operational logic of an Internet of Things (IoT)-enabled IC system are introduced. Based on the above systematic analysis, prospective research directions are proposed, which are intended to provide theoretical references and technical guidance for the engineering application of IC technology.