Comparison of Gaussian process regression, artificial neural network, and response surface methodology modeling approaches for predicting drying time of mosambi (<i>Citrus limetta</i>) peel

Preconditioning of rice is heat curing of salt infused moist parboiled rice prior to microwave puffing process. The sudden expansion of moisture present in the interstices of the starch granules caused by microwave heating of the preconditioned rice results in highly expanded smooth‐surface puffed rice. The expansion ratio (ER) and percentage of puffing (PP) of the microwave puffed rice preconditioned under different conditions were varied between 5.16–7.51 and 84.32–93.12%, respectively. Optimum temperature, air velocity, and salt concentration for fluidized bed preconditioning obtained by using integrated artificial neural network and genetic algorithm approach were estimated to be 70°C, 2.9 m s−1, and 4%, respectively. The predicted values of ER and PP at optimum combination of independent variables were 7.5 and 92.82%, respectively. Scanning electron microscopy revealed that preconditioning transformed the compact polyhedral structure of starch granules to fused mass with interlocking and overlapping structure. Practical applications Preconditioning with fluidized bed drying of salt infused rice would be feasible as a substitute with traditional direct heat curing process with saving of time, labor, and energy. This study would enhance understanding of the effects of preconditioning process on the mechanisms of puffing. Artificial neural network and genetic algorithm integrated optimization can be applied for prediction and control of mechanization of the preconditioning and puffing process. Preconditioned rice can be puffed directly by using a microwave oven that will make the puffing process convenient to consumers.

Section: Introductionmentioning

confidence: 99%

mentioning

confidence: 99%

Optimization of fluidized bed preconditioning for microwave puffed rice using integrated artificial neural network and genetic algorithm approach

Dash

Das

2019

“…Yang et al (2007) also employed Hummeida K ‐value model and modified Oswin EMC model to study the thin layer drying of peanut, which showed good prediction. Furthermore, artificial neural network (ANN) had successfully described the drying kinetics of rice (Alam et al, 2018; Beigi, Torki‐Harchegani, & Tohidi, 2017), mosambi peel (Chaurasia, Younis, Qadri, Srivastava, & Osama, 2018) and quince (Taghinezhad, Kaveh, Jahanbakhshi, & Golpour, 2020). However, Faridi, Golian, Mottaghitalab, López, and France (2013) reported that the predictive ability of ANN models showed poor predictive performance on newly introduced data.…”

Section: Introductionmentioning

confidence: 99%

A moisture content prediction model for deep bed peanut drying using support vector regression

Wang

Jin

et al. 2020

In order to make the moisture content monitoring more convenient and rapid during peanut drying process, the drying characteristics of peanut were investigated and a real‐time SVR moisture content monitoring model was established in this paper. The results showed that hot air temperature, initial moisture content, airflow rate and the layer height were the key factors on peanut drying, and the peanut variety showed little effect on drying. The SVR model exhibited a good performance with R2: 0.91, RMSE: 4.38, and bias: −7.5e‐3. Compared with the results of linear regression models and multilayer perceptron model, SVR model showed a better performance. In addition, the SVR model was validated by the drying data of other three varieties of peanuts. And the relative errors between the predicted values by SVR model and the measured values were within 20%, which suggested that SVR was a promising modeling algorithm for peanut drying. Practical application Peanut drying is essential for peanut production due to its high moisture content at harvest (about 30–50% on the wet basis), which makes it susceptible to mildew, or even produces aflatoxins. Peanut has a special physiological structure. In the drying process, the peanut kernels are gradually shrunk, which makes the air layer volume between the shell and the kernel gradually increases, thus hindering mass and heat transfer and leading to its different drying characteristics from other grain and oil crops. Furthermore, it is necessary to monitor the moisture content changes of peanut during deep bed drying to ensure the drying uniformity and prevent energy waste. Therefore, the drying characteristics of peanuts were studied and a moisture content prediction model for deep bed drying was established to assist the actual drying process.

“…They are effective in modeling complex, nonlinear processes. ANNs have been used in the food industry to model image-based classification (Kılıç, Boyacı, Köksel, & Küsmeno glu, 2007), drying of food products (Chaurasia, Younis, Qadri, Srivastava, & Osama, 2018;Erenturk & Erenturk, 2007;Guiné et al, 2015), modeling rheological properties of foods (Ramzi, Kashaninejad, Salehi, Mahoonak, & Razavi, 2015), and so forth. The Gaussian process regression (GPR) models are a powerful nonparametric Bayesian approach to regression.…”

mentioning

confidence: 99%

“…It uses theoretically infinite number of parameters to understand the inherent function in the training data (Schulz, Speekenbrink, & Krause, 2018). These models are simpler and easier to understand and can help in understanding the underlying function describing the data (Chaurasia et al, 2018).…”

mentioning

confidence: 99%

Optimization of de‐bittering process of mosambi (Citrus limetta) peel: Artificial neural network, Gaussian process regression and support vector machine modeling approach

Younis

Ahmad

Osama

et al. 2019

Self Cite

The process of di‐bittering of mosambi peel powder (MPP) using sodium chloride solution (0, 5, and 10%) was optimized. The optimization was based on the acceptable taste, maximum total polyphenol content, and maximum antioxidant properties of peel powder. Support vector machine (SVM)‐based classification model was developed for sensory evaluation. Artificial neural network (ANN) and Gaussian process regression (GPR) was used for development of regression models to predict total polyphenol and total antioxidant properties. The two combinatorial models of SVM–ANN and SVM–GPR were developed for optimization using genetic algorithm. The optimized powder had total polyphenol content and antioxidant properties of 0.65 ± 0.10% and 57.05 ± 0.08%, respectively, while taste was in acceptable limits and showed good, swelling power, water binding capacity, oil binding capacity, emulsion stability, and emulsion activity. The optimized dibittered powder also showed good results for in vitro glucose retardation index and showed antimicrobial activity against some pathogenic microorganisms. Practical Applications Mosambi peel powder is rich in dietary fiber and other bioactive compounds. However, its utilization is limited due to the presence of some bitter compounds. The present study deals with the optimization of di‐bittering process of mosambi peel. The optimized powder had convincing functional properties and hence opens the doors for its commercial application in food industry. Mosambi peel powder being rich in dietary finder can enhance the fiber content of same fiber deficient foods.