Artificial Neural Network Modeling for Temperature and Moisture Content Prediction in Tomato Slices Undergoing Microwave-Vacuum Drying

Machine learning-based modeling in food processing applications: State of the art

Food processing is a complex, multifaceted problem that requires substantial human interaction to optimize the various process parameters to minimize energy consumption and ensure better-quality products. The development of a machine learning (ML)-based approach to food processing applications is an exciting and innovative idea for optimizing process parameters and process kinetics to reduce energy consumption, processing time, and ensure better-quality products; however, developing such a novel approach requires significant scientific effort. This paper presents and evaluates ML-based approaches to various food processing operations such as drying, frying, baking, canning, extrusion, encapsulation, and fermentation to predict process kinetics. A step-by-step procedure to develop an ML-based model and its practical implementation is presented. The key challenges of neural network training and testing algorithms and their limitations are discussed to assist readers in selecting algorithms for solving problems specific to food processing. In addition, this paper presents the potential and challenges of applying ML-based techniques to hybrid food processing operations. The potential of physics-informed ML modeling techniques for food processing applications and their strategies is also discussed. It is expected that the potential information of this paper will be valuable in advancing the ML-based technology for food processing applications.

Development, validation, and comparison of FE modeling and ANN model for mixed-mode solar drying of potato cylinders

This work aims to develop a finite element (FE) model for predicting temperature and moisture ratio of potato cylinders having diameters of 8, 10, 13 mm and 50 mm length during solar drying using COMSOL Multiphysics software. The developed model computed conduction, convection, and radiation with appropriate governing and boundary conditions by coupling heat transfer in solid, laminar flow, transport of diluted species, and moving mesh modules together. Moving mesh module was employed to embrace the effect of inevitable shrinkage parameter all through solar drying. Experimentations and calculations were done based on the requirement of FE model. The developed model showed the increment of product temperature from 299.51-313.73 K, 299.07-313.03 K, and 298.34-314.57 K in case of 8, 10, and 13 mm diameter samples for an effective drying period of 3 h 15 min, 4 h 15 min, and 5 h, respectively. At the same time, the moisture content reduced from 83.57%, 86.57%, and 82.12% (wb) to 9.08%, 9.99%, 10.44% (wb) for the respective samples. To prove the reliability of the FE model predicted results, an attempt was made through the artificial neural network (ANN) model for describing the drying performance of the potato as well. It was found that the FE model better simulated the drying behavior with higher R² values (R²  = 0.988-0.995). The drying chamber air temperature was also simulated from FE model and validated with experimental data during drying of samples. The prediction capability of FE proposed model based on statistical error analysis showed lower values than ANN model. PRACTICAL APPLICATION: In the present study, the potential of mixed-mode solar drying in food processing industries was established showing detailed investigation of transport processes throughout the solar drying process of potato cylinders. The established finite element (FE) model can be considered as a realistic alternative to experimentation. The food processing industries and dryer engineers can achieve better quality dried products by precisely operating the dryers at the optimum condition by help of the proposed FE model and product shrinkage analysis.

Predictive models of minimum temperatures for the south of Buenos Aires province

Depending on the time of development of a crop temperature below 0 °C can cause damage to the plant, altering its development and subsequent yield. Since frosts are identified from the minimum air temperature, the objective of this research paper is to generate forecast -(predictive) models at 1, 3 and 5 days of the minimum daily temperature (T_min) for Bahía Blanca city. Non-linear numerical models are generated using artificial neural networks and geometric models of finite elements. Six independent variables are used: temperature and dew point temperature at meteorological shelter level, relative humidity, cloudiness observed above the station, wind speed and direction measured at 10 m altitude. Data have been obtained between May and September from 1956 to 2015. Once the available data had been analyzed, this period was reduced to 2007-2015. For the selection of the most suitable model, the correlation coefficient of Pearson (R), the determination coefficient (R²) and the Mean Absolute Error (MAE) are evaluated. The results of the study determine that the geometric model of finite elements with 4 variables, over 9 years (2007-2015) and separated by the season of the year is the one that presents better adjustment in the forecast of T_min with up to 5 days of anticipation.

Moisture Estimation in Cabinet Dryers with Thin-Layer Relationships Using a Genetic Algorithm and Neural Network

Nowadays, industrial dryers are used instead of traditional methods for drying. When designing dryers suitable for controlling the process of drying and reaching a high-quality product, it is necessary to predict the gradual moisture loss during drying. Few studies have been conducted to compare thin-layer models and artificial neural network models on the kinetics of pistachio drying in a cabinet dryer. For this purpose, ten mathematical-experimental models with a neural network model based on the kinetic data of pistachio drying were studied. The data obtained was from a cabinet dryer evaluated at four temperatures of inlet air and different air velocities. The pistachio seeds were placed in a thin layer on an aluminum sheet on a drying tray and weighed by a scale attached to the computer at different times. In the neural network, data was divided into three parts: Educational (60%), validation (20%) and testing (20%). Finally, the best mathematical-experimental model using a genetic algorithm and the best neural network structure for predicting instantaneous moisture were selected based on the least squared error and the highest correlation coefficient.

Artificial Neural Network Modeling for Predicting Wood Moisture Content in High Frequency Vacuum Drying Process

The moisture content (MC) control is vital in the wood drying process. The study was based on BP (Back Propagation) neural network algorithm to predict the change of wood MC during the drying process of a high frequency vacuum. The data of real-time online measurement were used to construct the model, the drying time, position of measuring point, and internal temperature and pressure of wood as inputs of BP neural network model. The model structure was 4-6-1 and the decision coefficient R2 and Mean squared error (Mse) of the training sample were 0.974 and 0.07355, respectively, indicating that the neural network model had superb generalization ability. Compared with the experimental measurements, the predicted values conformed to the variation law and size of experimental values, and the error was about 2% and the MC prediction error of measurement points along thickness direction was within 2%. Hence, the BP neural network model could successfully simulate and predict the change of wood MC during the high frequency drying process.

Comprehensive treatments for hepatocellular carcinoma with portal vein tumor thrombosis

Portal vein tumor thrombosis (PVTT) is one of the most common complications in hepatocellular carcinoma (HCC). HCC with PVTT usually indicates poor prognosis, which has a number of characteristics including a rapidly progressive disease course, worse liver function, complications connected with portal hypertension, and poorer tolerance to treatment. The exact mechanisms of PVTT remain unknown, even though some concerned signal transduction or molecular pathways have been identified. In western countries, sorafenib is the only recommended therapeutic strategy regardless of PVTT types. However, multiple treatment options including transhepatic arterial chemoembolization, hepatectomy, radiotherapy, and sorafenib available in the clinic. In this review, we enumerate and discuss therapeutics against patients with HCC having PVTT available in the clinic and put forward directions for future research.

Modeling of drying kiwi slices and its sensory evaluation

In this study, monolayer drying of kiwi slices was simulated by a laboratory-scale hot-air dryer. The drying process was carried out at three different temperatures of 50, 60, and 70°C. After the end of drying process, initially, the experimental drying data were fitted to the 11 well-known drying models. The results indicated that Two-term model gave better performance compared with other models to monitor the moisture ratio (with average R² value equal .998). Also, this study used artificial neural network (ANN) in order to feasibly predict dried kiwi slices moisture ratio (y), based on the time and temperature drying inputs (x₁, x₂). In order to do this research, two main activation functions called logsig and tanh, widely used in engineering calculations, were applied. The results revealed that, logsig activation function base on 13 neurons in first and second hidden layers were selected as the best configuration to predict the moisture ratio. This network was able to predict moisture ratio with R² value .997. Furthermore, kiwi slice favorite is evaluated by sensory evaluation. In this test, sense qualities as color, aroma, flavor, appearance, and chew ability (tissue brittleness) are considered.

Modeling of the lycopene extraction from tomato pulps

The inputs of this network were the concentration of pectinase and time of incubation, and the outputs were extracted lycopene and the activity of radical scavenging activity. Two different networks were designed for the process under the sonication and without it. For optimal network, networks' transfer functions and different learning algorithms were evaluated and the validity of each one was determined. Consequently, the feedforward neural network with function of logarithmic transfer, Levenberg Marquardt algorithm and 4 neurons in the hidden layer with the correlation coefficient of 0.96 and 0.99 were respectively observed for the treatments under sonication and without it, furthermore, root mean squared error and standard error values were obtained 0.46 and 0.22 respectively for the treatments under sonication and 0.77 and 0.38 without it as respectively optimal networks. The selected networks could determine the chosen responses, individually and in combined effect of both inputs as well (R(2) > 0.98).

Egg volume prediction using machine vision technique based on pappus theorem and artificial neural network

Egg size is one of the important properties of egg that is judged by customers. Accordingly, in egg sorting and grading, the size of eggs must be considered. In this research, a new method of egg volume prediction was proposed without need to measure weight of egg. An accurate and efficient image processing algorithm was designed and implemented for computing major and minor diameters of eggs. Two methods of egg size modeling were developed. In the first method, a mathematical model was proposed based on Pappus theorem. In second method, Artificial Neural Network (ANN) technique was used to estimate egg volume. The determined egg volume by these methods was compared statistically with actual values. For mathematical modeling, the R(2), Mean absolute error and maximum absolute error values were obtained as 0.99, 0.59 cm(3) and 1.69 cm(3), respectively. To determine the best ANN, R(2) test and RMSEtest were used as selection criteria. The best ANN topology was 2-28-1 which had the R(2) test and RMSEtest of 0.992 and 0.66, respectively. After system calibration, the proposed models were evaluated. The results which indicated the mathematical modeling yielded more satisfying results. So this technique was selected for egg size determination.

BP-ANN for fitting the temperature-germination model and its application in predicting sowing time and region for Bermudagrass

Temperature is one of the most significant environmental factors that affects germination of grass seeds. Reliable prediction of the optimal temperature for seed germination is crucial for determining the suitable regions and favorable sowing timing for turf grass cultivation. In this study, a back-propagation-artificial-neural-network-aided dual quintic equation (BP-ANN-QE) model was developed to improve the prediction of the optimal temperature for seed germination. This BP-ANN-QE model was used to determine optimal sowing times and suitable regions for three Cynodon dactylon cultivars (C. dactylon, 'Savannah' and 'Princess VII'). Prediction of the optimal temperature for these seeds was based on comprehensive germination tests using 36 day/night (high/low) temperature regimes (both ranging from 5/5 to 40/40°C with 5°C increments). Seed germination data from these temperature regimes were used to construct temperature-germination correlation models for estimating germination percentage with confidence intervals. Our tests revealed that the optimal high/low temperature regimes required for all the three bermudagrass cultivars are 30/5, 30/10, 35/5, 35/10, 35/15, 35/20, 40/15 and 40/20°C; constant temperatures ranging from 5 to 40°C inhibited the germination of all three cultivars. While comparing different simulating methods, including DQEM, Bisquare ANN-QE, and BP-ANN-QE in establishing temperature based germination percentage rules, we found that the R(2) values of germination prediction function could be significantly improved from about 0.6940-0.8177 (DQEM approach) to 0.9439-0.9813 (BP-ANN-QE). These results indicated that our BP-ANN-QE model has better performance than the rests of the compared models. Furthermore, data of the national temperature grids generated from monthly-average temperature for 25 years were fit into these functions and we were able to map the germination percentage of these C. dactylon cultivars in the national scale of China, and suggested the optimum sowing regions and times for them.

Thin-layer modeling of convective and microwave-convective drying of oyster mushroom (Pleurotus ostreatus)

Oyster mushroom samples were dried under selected convective, microwave-convective drying conditions in a recirculatory hot-air dryer and microwave assisted hot-air dryer (2.45 GHz, 1.5 kW) respectively. Only falling rate period and no constant rate period, was exhibited in both the drying technique. The experimental moisture loss data were fitted to selected semi-theoretical thin-layer drying equations. The mathematical models were compared according to three statistical parameters, i.e. correlation coefficient, reduced chi-square and residual mean sum of squares. Among all the models, Midilli et al. model was found to have the best fit as suggested by 0.99 of square correlation coefficient, 0.000043 of reduced-chi square and 0.0023 of residual sum of square. The highest effective moisture diffusivity varying from 10.16 × 10(-8) to 16.18 × 10(-8) m(2)/s over the temperature range was observed in microwave-convective drying at an air velocity of 1.5 m/s and the activation energy was calculated to be 16.95 kJ/mol. The above findings can aid to select the most suitable operating conditions, so as to design drying equipment accordingly.

Microwave-vacuum drying of sour cherry: comparison of mathematical models and artificial neural networks

Drying characteristics of sour cherries were determined using microwave vacuum drier at various microwave powers (360, 600, 840, 1200 W) and absolute pressures (200, 400, 600, 800 mbars). In addition, using the artificial neural networks (ANN), trained by standard Back-Propagation algorithm, the effects of microwave power, pressure and drying time on moisture ratio (MR) and drying rate (DR) were investigated Based on the evaluation of experimental data fitting with semi-theoretical and empirical models, the Midilli et al. model was selected as the most appropriate one. Furthermore, the ANN model was able to predict the moisture ratio and drying rate quite well with determination coefficients (R(2)) of 0.9996, 0.9961 and 0.9958 for training, validation and testing, respectively. The prediction Mean Square Error of ANN was about 0.0003, 0.0071 and 0.0053 for training, validation and testing, respectively. This parameter signifies the difference between the desired outputs (as measured values) and the simulated values by the model. The good agreement between the experimental data and ANN model leads to the conclusion that the model adequately describes the drying behavior of sour cherries, in the range of operating conditions tested.