Data-Processing and Signal Analysis Methods for Operational Status Monitoring and Sensor–Actuator Anomaly Detection for Smart Vertical Farms

Reliable operation of smart vertical farms requires coordinated sensing, actuation, communication, and control, yet sensor and actuator faults can compromise environmental regulation. This structured narrative review examined signal-processing methods for operational monitoring, anomaly detection, and fault isolation in vertical farms and related controlled-environment systems. The PRISMA-style diagram reports 410 identified records and 78 cited sources. A four-tier framework distinguished direct vertical-farm evidence, adjacent controlled-environment studies, transferable engineering research, and foundational methods. Statistical monitoring and change detection were suitable for slowly varying environmental and nutrient measurements, whereas frequency and time-frequency analyses were more appropriate for electrical, vibration, and acoustic signals. Sensor redundancy, correlation analysis, and model-based residuals supported fault isolation by comparing actuator commands with measured responses. Available evidence was stronger for hydroponic sensor faults than for progressive actuator degradation in commercial vertical farms. Crop development, photoperiod, irrigation, and operating-state changes could also produce anomalies without equipment failure. Reliable diagnosis therefore requires synchronized signals, contextual metadata, state-specific baselines, and independent fault confirmation. Evaluation should consider false alarms, detection delay, isolation accuracy, computational demand, and generalization across facilities. Future priorities include benchmark datasets, multimodal diagnosis, edge implementation, digital-twin-supported evaluation, and safe fault-tolerant control under varying crop stages, equipment conditions, and control schedules.