Abnormal Operation Detection of Automated Orchard Irrigation System Actuators by Power Consumption Level

Area of Science:

• Agricultural Engineering
• Automation and Control Systems
• Sensor Technology

Background:

• Automated irrigation systems rely on Information and Communication Technology (ICT) components like actuators for precision watering.
• Actuator malfunctions in these systems lead to inefficient irrigation, negatively impacting crop health and productivity.
• Early detection of actuator failures is crucial for maintaining optimal irrigation management and crop yields.

Purpose of the Study:

• To develop a signal processing technique for detecting actuator malfunctions in automated orchard irrigation systems.
• To utilize power consumption data and operating status of actuators for fault identification.
• To enhance the reliability and efficiency of automated irrigation systems through improved fault detection.

Main Methods:

• A demonstration orchard setup with apple trees was used, featuring a Python-programmed microcontroller controlling a pump and two solenoid valves.
• Commercial current sensors measured actuator power consumption, with threshold values applied to differentiate normal and abnormal operations.
• Signal processing techniques analyzed power consumption, control commands, and operating states to detect actuator failures.

Main Results:

• The proposed method accurately identified normal and abnormal operations of the pump and solenoid valves.
• For the second solenoid valve, the model achieved high accuracy in detecting both normal (316/333) and abnormal (58/60) instances.
• The system demonstrated a mean average precision of 99.9% for channel 1 actuators and 99.7% for the channel 2 solenoid valve.

Conclusions:

• The developed signal processing approach effectively detects actuator malfunctions in automated irrigation systems.
• This technology has the potential to significantly improve irrigation management and boost crop productivity.
• Future work includes integrating machine learning for enhanced fault detection and assessing system scalability for diverse agricultural applications.