Development of a Predictive Model for Wild Blueberry Fruit Losses Using Bio-systems
Authors: Farooque, A.
Description: Development of a Predictive Model for Wild Blueberry Fruit Losses Using Bio-systems Modeling Techniques A. A. Farooque1, Q. U. Zaman2, T. Quang 2, A. W. Schumann3, T. J. Esau1 and W. Jameel 1 1Department of Mechanical Engineering, Dalhousie University, 1360 Barrington St, Halifax, Nova Scotia, B3H 4R2, Canada. 2 Engineering Department, Faculty of Agriculture, Dalhousie University, 39 Cox Road, Truro, Nova Scotia, B2N 5E3, Canada. 3Citrus Research and Education Center, University of Florida, 700 Experiment Station Road, Lake Alfred, FL 33850, USA. Understanding the relationships between the machine operating parameters, fruit losses, topography and crop characteristics can aid in better berry recovery of mechanical blueberry harvester. This study was designed to develop a model for prediction of fruit losses during harvesting and to identify the factors responsible for increased fruit losses. Wild blueberry fields were selected and completely randomized factorial experiments were constructed at each site. Yield plots were made in the path of operating harvester. The harvester was operated at specific levels of ground speed (1.20, 1.6 and 2.0 km hr-1) and head rotational speed (26, 28 and 30 rpm) to collect the total fruit yield, loss through blower, un-harvested berries on the plants and berries on the ground from each plot within selected fields. The readings of slope, plant height and fruit zone were also recorded from each plot. The collected data were normalized and separated into training and validation datasets. The correlation analysis was performed to identify the input variables having significant impact on the output (fruit losses) during harvesting. The feed-forward back-propagation multi-layer artificial neural network (ANN) was used for model development and predictions. Tanh-sigmoid transfer function between the hidden layer and output layer was found to the best for this study. The developed models were validated internally and externally and the best performing model was identified based on mean square error, root mean square error (RMSE), coefficient of efficiency and coefficient of determination. The RMSE were regressed against epoch to determine the epoch size for fruit loss processes. The multiple regression (MR) models were also developed and compared with ANN model to verify the prediction accuracy of the ANN model. The MR models under performed in modeling the fruit loss processes due to the non-linear nature when compared with ANN. Overall this study will help us to predict fruit losses, and to identify the factors responsible for losses during harvesting.
Keywords: Artificial neural network, fruit losses, mechanical harvester, wild blueberry.
Technical field: Bioprocess Systems Engineering
Conference name: CSBE/SCGAB 2015 Annual Conference, Edmonton, AB, 5-8 July 2015.
Session name: Poster
Citation: Farooque, A.. 2015. Development of a Predictive Model for Wild Blueberry Fruit Losses Using Bio-systems. CSBE/SCGAB 2015 Annual Conference, Edmonton, AB, 5-8 July 2015.
Publisher: Canadian Society for Bioengineering
Language 1: en
Rights: Canadian Society for Bioengineering