Quality-Oriented Smart Sensing and Sensing-to-Control Pathways Across the Mechanized Forage Production Chain: A Review

Abstract

The increasing automation of mechanized forage production has improved operational efficiency, but achieving consistent product quality remains challenging because material conditions evolve continuously across sequential operations. Current control strategies primarily focus on machine-level objectives, such as load regulation, throughput stabilization, and actuator response, while the linkage between operational decisions and final forage quality is still insufficiently established. Machinery-, sensor-, and stage-centered reviews chiefly catalog technologies and prediction performance, without consistently testing whether observations are representative, transferable, timely, and assigned to actionable material. This critical narrative review integrates recent evidence identified through Web of Science Core Collection and CNKI searches, supplemented by citation tracing of foundational studies. It presents a state-centered framework for smart sensing and sensing-to-control pathways throughout the mechanized forage production chain. Key intermediate states, including moisture distribution, leaf integrity, windrow geometry and linear density, mass flow, particle characteristics, bale density, and pore structure, are analyzed as the links between mechanical actions and downstream outcomes, such as drying performance, material loss, fermentation, heating risk, and storage stability. Sensing technologies based on electrical, mechanical, optical, microwave, weighing, and spectroscopic principles are evaluated from the perspectives of measurement representativeness, transferability, latency, uncertainty, and operational validity rather than prediction accuracy alone. Prediction accuracy describes agreement with a reference, whereas decision utility reflects whether a valid, representative, and timely estimate can change an operational decision and reduce downstream quality risk. The reviewed evidence indicates that practical quality-oriented automation requires rapid proxy-state feedback for immediate machine regulation combined with batch-linked quality measurements for delayed verification and model updating. Future progress depends on establishing causal machine–quality relationships, transferable sensing models, cross-stage data continuity, and production-scale validation.