Prediction of mass‐based process designing parameters of amla fruit using different modeling techniques

Abstract

AbstractThe machine designing parameters such as size, volume, criteria projected area are important for quality evaluation at the time of packaging, grading, and sorting of fruits. All these designing parameters also depend on mass of the fruit sample. So, the correlation between mass and designing parameters could be useful for automation of the process industry. The research was inducted to predict the dimension, projected area, and volume of fruit as a function of fruit mass. Prediction modeling was done using different models, that is, mathematical model (MM), artificial neural network (ANN), and support vector regression (SVR). The mathematical model (linear, quadratic, and power) showed better prediction results than SVR modeling. In addition, the best results were achieved by neural network‐based prediction modeling. It showed the highest correlation coefficient of .884, .986, .992, .999 and the lowest error of 0.570, 0.241, 8.080, 0.059 for height, diameter, projected area, and volume, respectively. It is observed that the physical properties of fruit can be predicted using ANN‐based modeling which could be useful in automation and efficient designing of post‐harvest equipment.Practical ApplicationThe machine designing parameters are the essential and fundamental parameters for designing of grading, sorting, packaging, and handling equipments of various horticultural products including amla. Fruits and vegetables are non‐uniform in size and shape. So, minor changes in shape and size of fruits could decrease the efficiency of various processes and its equipment. The design parameters of fruits and vegetables are dependent on the mass of samples. So, accurate knowledge of mass and design parameters could help in designing the automatic grading system. Different models can be used to predict various designing parameter of amla or similar fruits. By using various models, fruit can be separated more accurately based on minor changes in shape and size rather than its mass. The ANN‐based prediction will be beneficial for the automation of the fruits grading system at an industrial scale.