Machine learning approaches for estimating apricot drying characteristics in various advanced and conventional dryers
Mohammad Kaveh,Necati Çetin,Esmail Khalife,Yousef Abbaspour‐Gilandeh,Maryam Sabouri,Faroogh Sharifian
DOI: https://doi.org/10.1111/jfpe.14475
2023-10-12
Journal of Food Process Engineering
Abstract:Apricots were dried with the use of five different dryers. Moisture ratio, moisture content, and drying rates were determined. Different machine learning algorithms were used to estimate drying characteristics. Random forest and machine learning models could potentially be used for estimation and discrimination. Drying plays a crucial role in preserving the quality of agricultural products. Nevertheless, suboptimal conditions in drying systems have an adverse effect on drying characteristics and energy efficiency. Machine learning approaches are innovative and reliable that have been successfully used to solve such challenges and achieve optimization in drying processes. In this study, five machine learning approaches (multilayer perceptron [MLP], gaussian processes [GP], support vector regression [SVR], k‐nearest neighbors [kN], and random forest [RF]) were used to estimate moisture content and moisture ratio of apricot in five various dryers (convective [CV], microwave [MW], infrared [IR], microwave‐convective [MW‐CV], and infrared‐convective [IR‐CV]). Also, the values of specific energy consumption (SEC) and effective moisture diffusivity (Deff) were calculated in these dryers. Accordingly, the best result of the Deff (3.14 × 10−10 m2/s) and the minimum value of the drying time (130 min) and SEC (18.67 MJ/kg) were obtained using MW‐CV hybrid dryer. While the lowest values of Deff (2.09 × 10−11 m2/s) and highest drying time (18.5 h) and SEC (209.34 MJ/kg) were detected in CV dryer at 50°C. The best correlation coefficients (R) for the estimation of moisture content were gained using RF technique for k‐fold cross validation and train‐test split with the values of 0.9908 and 0.9912, respectively. Moreover, moisture ratio results showed that the MLP achieved the highest R value over 0.9985 for both validation methodologies. In the discrimination of the drying methods, the MLP had the greatest accuracy as 82.00% and 86.00% for k‐fold cross validation and train‐test split, respectively. The results showed that the RF and ML models could potentially be used for estimation and discrimination for drying applications. Practical Applications Recently, there has been an increased interest in healthy food choices such as foodstuffs, snacks, and dried products. This trend has captured the attention of both dietitians and conscious consumers. Apricots are a prime example of a valuable dried product that can be dry in various conditions. Machine learning techniques can be used for rapid and non‐destructive determination of drying characteristics and such techniques yield objective and accurate results. Present findings revealed that texture machine learning models could be used as an effective and reliable discrimination tool for dried products.
food science & technology,engineering, chemical