Machine Learning-Based Remote Monitoring and Predictive Analytics System for Apple Harvest Storage: A Statistical Model Based Approach

Investigations were made to model air velocity effect on the cooling rates of apples stored in bulk perforated bins in CAS. Prediction of cooling rates by statistical modeling is an important requirement for automation of post harvest cooling operations of fruits. Air velocity of the four fans installed in the evaporator system was controlled using an electronic Variable Frequency Drive (VFD). Four different air flow rates, viz., 10.8, 8.1, 5.4 and 2.7 m3/s were maintained in four different chambers to find the effect on cooling time. Apples procured from orchards were sorted, graded and filled into 300 kg capacity perforated self stackable plastic bins and stored in controlled atmosphere store rooms of 150 Metric tons (MT) each. Four chambers of 10 m × 9 m × 8 m dimensions (720 cubic metre space) were filled with apple bins for cooling test. Different air flow rates resulted in different pull down time. Bulk cooling from 28 °C to desired temperature of 1 °C was achieved in 109 h at 10.8 m3/s air flow. Lowest air flow rate resulted in pull down time of 140 h. Cooling curves were modeled with three Statistical equations. Out of the three models equations tested, polynomial equations fitted with regression coefficients of 0.96–0.99. The modeling equation can be helpful to automate the post harvest storage processes and design of refrigeration systems for storage of agriculture produce at low temperatures.