Articles
APPLYING EXPERT SYSTEMS CONCEPTS TO REAL-TIME GREENHOUSE CONTROLS
Actual implementation of ‘desired environmental conditions’ occurs indirectly as a function of the grower’s perceptions of crop response characteristics, control system behavior, and the weather.
Simply stated, when a grower selects (or changes) a setpoint, it represents more of an intuitive than an analytical decision.
In concept, an expert system can serve as a tool for mimicking a grower’s perceptions and provide support for real-time environmental controls.
To investigate this concept, an autonomous dynamic controller that chose setpoints for the frequency and duration of misting events was developed and tested.
The setpoints were chosen by an expert system, MISTING, that was based on the perceived optimal misting strategy of an experienced grower.
System software ran on a general purpose microcomputer (IBM-PC) that was located .5 km from the greenhouse.
The remote microcomputer communicated via dedicated telephone line with a monitor/controller (CR21X) located in the greenhouse.
Sensor data from the greenhouse were provided as facts to MISTING which returned setpoints to the CR21X which accordingly, regulated misting line solenoids.
Knowledge representation of the grower’s perceived control strategy was straight forward because the domain was narrow and very well defined.
Implementation was accomplished using commercial software packages as the primary components in the control software and readily available hardware.
The system ran for 30 days with no software failures and successfully followed the grower’s strategy.