Study on smart farm actuator complex state-based failure timing prediction

This paper focuses on the fact that malfunctions in smart farm actuators (such as openers, pumps, and fans) significantly impact agricultural productivity. It proposes a condition-based fault timing prediction method to detect faults in advance and optimize maintenance schedules. First, dedicated measuring boards and software were developed to collect normal and abnormal data for voltage, current, vibration, noise, and temperature under different actuator operating states (OPEN and CLOSE). Next, statistical techniques (IQR and Z-Score) were combined with the Isolation Forest algorithm in a hybrid approach to effectively detect anomalies. By distinguishing sensor characteristics based on each operating mode (OPEN or CLOSE), the method enables accurate fault-time prediction using threshold values. The proposed algorithm can be applied to a variety of actuators and environmental conditions, and through further expansion of data diversity and adoption of a distributed architecture, it has the potential to evolve into a robust predictive maintenance solution.