Article

Artificial neural network approach to simultaneously predict shelf life of two varieties of packaged rice snacks

Authors:
To read the full-text of this research, you can request a copy directly from the authors.

Abstract

Actual storage shelf life test by storing a packaged product under typical storage conditions is costly and time consuming. A new approach using an artificial neural network (ANN) algorithm for shelf life prediction of two varieties of moisture-sensitive rice snacks packaged in polyethylene and polypropylene bags and stored at various storage conditions was established. The ANN used to predict the shelf life was based on multilayer perceptron with back propagation algorithm. The ANN algorithm employed the data of product characteristics, package properties and storage conditions. The neural network comprised an input, one hidden and one output layers. The network was trained using Bayesian regularisation. The performance of ANN was measured using regression coefficient (R2 = 0.23–0.28) and root mean square error (RMSE = 0.96–0.99). The ANN-predicted shelf lives agreed very well with actual shelf life data. ANN could be used as an alternative method for shelf life prediction of moisture-sensitive food products as well as product/package optimisation.

No full-text available

Request Full-text Paper PDF

To read the full-text of this research,
you can request a copy directly from the authors.

... ANN model is an interconnected group of nodes, parallel to the vast network of neurons in the human brain. The ANN predicted shelf lives, agreed very well with actual shelf life data pertaining to rice snacks, and ANN could be used as an alternative method for shelf life prediction of moisture-sensitive food products [1]. ...
... Linear layer (design) and timedelay methods of intelligent computing expert system were developed by Goyal and Goyal [6] for shelf life prediction of soft mouth melting milk cakes stored at 6 o C. Both the methods were compared with each other and it was observed that linear layer (design) method had superiority in predicting shelf life of soft mouth melting milk cakes stored at 6 o C. ANN predicts soya bean equilibrium moisture content more accurately than mathematical model [7]. The ANN predicted shelf lives, agreed very well with actual shelf life data pertaining to rice snacks, and ANN could be used as an alternative method for shelf life prediction of moisture-sensitive food products [1]. Neuron based artificial intelligent scientific computer engineering models for estimating shelf life of instant coffee sterilized drink were impelemented by Goyal & Goyal [8]. ...
Article
This paper highlights the capability of artificial neural networks for predicting shelf life of milky white dessert jeweled with pista-chio. Linear layer (train) and generalized regression models were developed and compared with each other. Neurons in each hidden layers varied from 1 to 30. Data samples were divided into two sets, i.e., 80% of data samples were used for training and 20% for validating the network. Mean Square Error, Root Mean Square Error, Coefficient of determination and Nash -Sutcliffo Coefficient were applied in order to compare the prediction performance of the developed models. The experimental shelf life is 21 days and the developed intelligent artificial neural network model predicted 20.15 days shelf life for milky white dessert jeweled with pistachio.
... In the final section, the network's output pattern is compared with the target, and special criteria for evaluating the network are calculated (Goncalves et al. 2005). Recently, ANNs have been applied in different fields of food science, such as classification (Jacobsen et al. 2001;Boonmung et al. 2006), prediction (Mittal and Zhang 2001;Alvarez 2009;Siripatrawan and Jantawat 2009;Saiedirad and Mirsalehi 2010) and food-quality evaluation (Goyache et al. 2001;Srisawas and Jindal 2003;Çakmak and Boyacı 2011), simulating processes like drying behavior of different agricultural materials (Kerdpiboon et al. 2006;Martynenko and Yang 2006;Movagharnejad and Nikzad 2007;Youssefi et al. 2009;Boeri et al. 2011), osmotic dehydration (Trelea et al. 1997), cross-flow microfiltration (Dornier et al. 1995) and rheological properties (Bhattacharya and Patel 2007) There are several important parameters from the architecture of specific ANNs that influence the performance of them. Determining the optimum number of neurons to include in each hidden layer is of crucial importance in designing neural network structures. ...
Article
Wheat-flour dough is a viscoelastic material with nonlinear rheological behavior. Extensograph is a useful system for dough rheological measurement. Our purpose in this research was to apply soft computation tools for predicting the extensograph properties of dough from several physicochemical properties of flour. This study used the resulting model to suggest modifications of processing conditions for reducing economic loss and minimizing product quality deterioration. A generalized feed-forward artificial neural network (ANN) with a back-propagation learning algorithm was employed to estimate the extensograph properties of dough. Trial and error and genetic algorithm (GA) were applied in the training phase for developing an ANN with an optimized structure. Developed ANN using GA has excellent potential for predicting the extensograph properties of dough. Sensitivity analyses were conducted to explore the ability of inputs in predicting the extensograph properties of dough. Results showed gluten index was the most sensitive input in dough extensograph characterizations.
... Results showed useful predictions were possible with low mean square error (0.092) and high regression coefficient (0.98) between actual and predicted data. Moreover, ANN have been successfully applied for predicting shelf life of packaged rice snack (Siripatrawan and Jantawat, 2009). Singh et al. (2009) found that the ANN-based models were better than the kinetic models for predicting sensory quality of UHT milk. ...
... However, linear transforms typically extract information from only the second-order correlations in the data (covariance matrix) and ignore higher order correlations in the data. Many researchers have suggested that several chemical and physical measurements of foods are inherently nonsymmetric (Scholkopf, Smola, & Muller, 1998; Siripatrawan & Jantawat, 2008). A number of nonlinear transformation methods for pattern recognition exist. ...
Article
An alternative freshness index method for abalone (Haliotis asinina) muscle packaged under atmospheric air (Air) and modified atmosphere (MA) of 40% CO2: 30% O2: 30% N2 packaging conditions and stored at 2 ± 1 °C was developed. Biochemical indices covering pH, total volatile basic nitrogen (TVB-N), trimethylamine (TMA) and nucleotide degradation products, as well as instrumental texture and color of the packaged abalones, were determined. Sensory characteristics including odor, color and appearance were evaluated and then summarized into overall freshness scores (freshness index). The biochemical and instrumental analyses were then calibrated with the freshness index, using an artificial neural network algorithm. The neural network was shown to be capable of correlating biochemical and instrumental analyses with the freshness index. A useful prediction was possible, as measured by a low mean square error (MSE = 0.092) and a regression coefficient (R2 = 0.98) between true and predicted data.
... Artificial Neural Network (ANN) predicts soya bean equilibrium moisture content more accurately than mathematical model [6]. The ANN predicted shelf lives, agreed very well with actual shelf life data pertaining to rice snacks, and neurocomputing could be used as an alternative method for shelf life prediction of moisture-sensitive food products [7]. Presently, the consumers are extremely conscious about quality of the foods they buy. ...
... Recently, ANNs have been applied in different fields of food science, such as simulating processes like drying behavior of different agricultural materials (Movagharnejad and Nikzad 2007;Boeri et al. 2011), osmotic dehydration (Trelea et al. 1997) and crossflow microfiltration (Dornier et al. 1995). They have also been used in other fields of food science, such as classification (Jacobsen et al. 2001;Boonmung et al. 2006), prediction (Alvarez 2009;Siripatrawan and Jantawat 2009;Saiedirad and Mirsalehi 2010) or food-quality evaluation (Srisawas and Jindal 2003;Çakmak and Boyacı 2011). ...
Article
Alveograph is a useful device for rheological measurement of dough from biaxial extension. Our purpose in the current research was to apply soft computing tools for predicting the alveograph properties (maximum pressure for blowing the bubble [P], abscissa at bubble rupture [L], deformation energy [W]) using physicochemical properties of flour (protein, ash, wet gluten, gluten index, amylase activity, zeleny and particle size). Generalized feed-forward artificial neural networks (ANN) with back-propagation learning algorithms were employed to model the alveograph properties. A genetic algorithm (GA) was applied to optimize the parameters of the ANN's structure and inputs. Sensitivity analyses were conducted to explore the ability of the networks with using physicochemical properties of flour as inputs in predicting alveograph properties. The developed ANNs using GA were shown to have excellent potential in predicting the alveograph properties. Sensitivity analyses showed that zeleny (Sedimentation value) is the most sensitive input in predicting alveograph characteristics.
... Neural networks do not need much of a detailed description or formulation of the underlying process. ANN approach has been successfully applied to process modeling in recent years (Siripatrawan and Jantawat 2009;Yuceer 2010;Cakmak and Boyaci 2011;Khataee et al. 2011;Karadurmus et al. 2012). ...
Article
Effects of microwave power output and sample mass on drying behavior, color parameters, rehydration characteristics and some sensory scores of thyme leaves were investigated. Within the range of the microwave power outputs, 180-900W, and sample amounts, 25-100g, moisture content of the leaves were reduced to 0.1±(0.01) from 4.05kg water/kg dry base value. Drying times of the leaves were found to be varying between 3.5 and 15.5min for constant sample amount, and 6.5 and 20.5min for constant power output. Experimental drying data obtained were successfully modeled using artificial neural networks methodology. Statistical values of the test data were found to be 0.9999, 4.0937 and 0.025 for R-square, MAPE (%) and RMSE, respectively. Some changes were recorded in the quality parameters, and acceptable sensory scores for the dried leaves were observed in all of the experimental conditions (P<0.05). © 2012 Wiley Periodicals, Inc.
... Shelf-life studies provide important information to overcome food spoilage and ensure consumers a high-quality product for a significant period of time (Siripatrawan and Jantawat 2009). The food quality during storage is primarily affected by intrinsic factors like pH, water activity, titratable acidity, free fatty acids, peroxide value, preservatives, and extrinsic factors like storage conditions, relative humidity, and timetemperature profile during processing and storage. ...
Article
Artificial neural network (ANN) is a simplified model of the biological nervous system consisting of nerve cells or neurons. The application of ANN to food process engineering is relatively novel. ANN had been employed in diverse applications like food safety and quality analyses, food image analysis, and modeling of various thermal and non-thermal food-processing operations. ANN has the ability to map nonlinear relationships without any prior knowledge and predicts responses even with incomplete information. Every neural network possesses data in the form of connection weights interconnecting lines between the input to hidden layer neurons and weights of hidden to output layer neurons, which has a significant role in predicting the output data. The applications of ANN in different unit operations in food processing were described that includes theoretical developments using intelligent characteristics for adaptability, automatic learning, classification, and prediction. The parallel architecture of ANN resulted in a fast response and low computational time making it suitable for application in real-time systems of different food process operations. The predicted responses obtained by the ANN model exhibited high accuracy due to lower relative deviation and root mean squared error and higher correlation coefficient. This paper presented the various applications of ANN for modeling nonlinear food engineering problems. The application of ANN in the modeling of the processes such as extraction, extrusion, drying, filtration, canning, fermentation, baking, dairy processing, and quality evaluation was reviewed. • Highlights • This paper discusses application of ANN in different emerging trends in food process. • Application of ANN to develop non-linear multivariate modeling is illustrated. • ANNs have been shown to be useful tool for prediction of outcomes with high accuracy. • ANN resulted in fast response making it suitable for application in real time systems.
... ANNs were used to evaluate capability in predicting process parameters involved in thermal/pressure food processing or to build a model of the degree of proteins hydrolysis (Bucinski, Karamac, & Amarowicz, 2004; Torrecilla et al., 2004 ) and in classification of rapeseed and soybean oils (Wesolowski & Suchacz, 2001). ANNs were also used as an alternative method for shelf life prediction of moisture-sensitive food products or product packaging optimization (Siripatrawan & Jantawat, 2009) or to predict pork drip loss from pH and color measurements (Prevolnik, Candek-Potokar, Novic, & Skorjanc, 2009), as well as determination of volatile characteristics of honey samples and their classification (Tananaki, Thrasyvoulou, Giraudel, & Montury, 2007). However, as jet, ANNs have not been used for interpretation of the results obtained in declarative surveys. ...
... al., 2011;Karadurmus et. al., 2012;Khataee et al., 2011;Siripatrawan and Jantawat, 2009;Yuceer, 2010). ...
Article
Full-text available
Effects of infrared power output and sample mass on drying behaviour, colour parameters, ascorbic acid degradation, rehydration characteristics and some sensory scores of spmach leaves were investigated. Within both of the range of the infrared power outputs, 300-500 W, and sample amounts, 15-60 g, moisture content of the leaves was reduced from 6.0 to 0.1 ± (0.01) kg water/ kg dry base value. It was recorded that drying times of the spinach leaves varied between 3.5-10 min for constant sample amount, and 4-16.5 min for constant power output. Experimental drying data obtained were successfully investigated by using artificial neural network methodology. Some changes were recorded m the quality parameters of the dried leaves, and acceptable sensory scores for the dried leaves were observed m all of the expernnental conditions.
... Artificial neural engineering and regression models were developed to forecast shelf life of instant coffee drink [GOYAL & GOYAL, 2011B]. GOYAL & GOYAL (2011C) & JANTAWAT, 2009). ANN has also been successfully applied for predicting soya bean equilibrium moisture content more accurately than mathematical model (CHAYJAN & ASHARI, 2010). ...
Article
Full-text available
This paper highlights the significance of feedforward artificial neural network models for predicting shelf life of roasted coffee falvoured sterilized drink. Coffee is one of the most important products for trade in international market. Single as well as multilayer models were explored and different backpropagation algorithms were investigated, Root mean square error and coefficient of determination R2 were used to compare the prediction performance of single and multilayer feedforward ANN models. Experimental results suggested that multilayer models take less time and give better results as compared to single layer ANN models for prediction of sensory quality of roasted coffee falvoured sterilized drink..
Article
Migration testing of cyclic organosiloxane oligomer molecules from silicone rubber into food simulants was performed as a function of three factors: temperature, time, and food simulant type (i.e., n-hexane, n-heptane, and ethyl alcohol). A back-propagation artificial neural network (BP-ANN) was proposed to study the migration properties with an average prediction accuracy of 0.997. The analysis of the ANN indicated that high-temperature and n-hexane environments accelerate the migration of cyclic organosiloxane oligomer molecules from silicone rubber. A molecular dynamics simulation was employed to illustrate the migration mechanism at the molecular level. The results indicated that the high-temperature and n-hexane condition led to a high interaction energy, fractional free volume, and diffusion coefficient and a low difference solubility parameter of the system, which accelerate the molecules transfer process and threaten the safety of the materials. These fundamental studies provide a comprehensive understanding of the migration of cyclic organosiloxane oligomer contaminants and guidance for the failure prediction of materials.
Chapter
: A wide array of cereal and snack food products exists on the market. Many of these products have shelf lives as little as 90 days and up to a maximum of one year. As a result of this variability, designing methods to evaluate shelf life and oxidative stability is very much dependent on the product category. This chapter will cover oxidation products of cereals, the effect of oxidation on sensory quality, factors that affect the antioxidant compounds of cereals, and methods used to protect against oxidation. Information about mechanisms and analytical techniques will not be presented in this chapter.
Article
Frozen boiled shrimp and dried shrimp are among the high-value fishery products of Thailand. During the production of these products boiling is one of the most important steps that affects significantly the product physicochemical properties, especially the quantity and quality of proteins, which in turn affect other apparent properties perceived by consumers. The protein changes are, however, difficult to evaluate comparing to other typical physical properties of shrimp. The objective of this study was therefore to develop an artificial neural network (ANN) model to predict the protein changes of shrimp in terms of protein loss and protein denaturation as a function of the boiling conditions, namely, concentration of salt solution and boiling time, as well as a rather easily determined change of shrimp, that is, cooking loss. Other apparent physical changes of shrimp viz. shrinkage and hardness were also predicted. The optimized ANN structure for the prediction of protein changes consists of two hidden layers and twelve neurons per layer, while the optimized ANN structure for the prediction of physical changes consisted of two hidden layers and eight neurons per layer. The predicted protein and physical changes agreed very well with the experimental data with R2 > 0.994 and 0.972, respectively.
Article
Full-text available
Deep-fried Gluten free snacks were prepared using the composition of legumes flours with commercially available starches and evaluated for their proximate composition, nutritional quality and sensory properties. The maximum reduction in peak viscosity (414 BU) of blends was observed compared to refined wheat flour (860 BU). Proximal analysis indicated, protein content was improved from 6% (wheat snack) to 17% (red cowpea based snack) and also increased in mineral content such as iron and calcium. Difference in the iron content with highest in white cowpea based snack (FSW) (9.2 mg%), whereas calcium content was highest in red cowpea based snack (FSR) (65.5 mg%). On the basis of storage studies, there was a significant reduction in the crispiness indicated by instrumental and sensory analysis. The highest PV obtained for FSW and FSR was recorded to be 1.2 and 1.48 respectively at 30 days. Free fatty acid increased in all the three i.e. field bean based snack (FSF), FSW, FSR from 0.13, 0.12 and 0.13 to the highest of 0.35, 0.5 and 0.48 respectively at 0–60 days. Fat content in FSF, FSW and FSR decreased from 10.35 to 8.9%; 13.6–8.9% and 14.5–7.8%. Gluten free confirmation was carried out in the developed snacks by immunochemically. Immunochemical validation showed that, the extracted proteins were not reacted to wheat gliadin antibodies. Hence, the developed legume flours based gluten free snacks can be included in the diet of non-allergic to gluten sensitive population.
Article
Full-text available
Information Technologies is a commanding factor in managing the retail market ever. Advertising is main factor in retail industry or any other type of market. After evolution of Internet the advertising industry has changed completely. Internet is overpowering the business of traditional advertising trends. Online advertising is affecting almost every tradition advertising media. Online advertising is also has an upper edge over traditional methods. Online advertising is far cheaper and effective over all other means of advertising.
Article
Full-text available
Potato chips, popcorn, puffed corn curls, and saltines were equilibrated to various water activities (aw) over saturated salt solutions. Sensory panels determined the crispness and textural hedonic quality of the humidified products by the magnitude estimation technique coupled to a verbal concept scale. Critical water activities (ac), where the products became organoleptically unacceptable, generally fell in the 0.35–0.50 aw range. Instron analyses showed that the force-deformation curve changed distinctly near the ac for saltines and puffed corn curls, while the curve changed more gradually with increasing aw for popcorn. Potato chips did not produce a consistently shaped force-deformation curve. The cohesiveness value of popcorn was found to be a good indicator of its sensory crispness.
Article
Full-text available
An electronic sensor array with 12 nonspecific metal oxide sensors was evaluated for its ability to monitor volatile compounds in super broth alone and in super broth inoculated with Escherichia coli (ATCC 25922) at 37 degrees C for 2 to 12 h. Using discriminant function analysis, it was possible to differentiate super broth alone from that containing E. coli when cell numbers were 10(5) CFU or more. There was a good agreement between the volatile profiles from the electronic sensor array and a gas chromatography-mass spectrometer method. The potential to predict the number of E. coli and the concentration of specific metabolic compounds was investigated using an artificial neural network (ANN). The artificial neural network was composed of an input layer, one hidden layer, and an output layer, with a hyperbolic tangent sigmoidal transfer function in the hidden layer and a linear transfer function in the output layer. Good prediction was found as measured by a regression coefficient (R2 = 0.999) between actual and predicted data.
Article
Full-text available
This study was performed to investigate the ability of using discriminant factor analysis (DFA) and an artificial neural network (ANN) to identify and quantify the number of Escherichia coli (ATCC 25922) in nutrient media from data generated by analysis of E. coli volatile metabolic compounds using solid-phase microextraction (SPME) coupled with gas chromatography (GC) and mass spectrometry (MS). E. coli was grown in super broth and incubated at 37 degrees C for 2 to 12 h. Numbers of E. coli were followed using a colony counting method. An SPME device was used to collect the volatiles from the headspace above the samples, and the volatiles were identified using GC-MS. DFA was used to classify the samples from different incubation times. From DFA, it was possible to differentiate super broth from media containing E. coli when cell numbers were 10(5) CFU or more. The potential to predict the number of E. coli from the SPME-GC-MS data was investigated using a multilayer perceptron (MLP) neural network with back propagation training. The MLP comprised an input layer, one hidden layer, and an output layer, with a hyperbolic tangent sigmoidal transfer function in the hidden layer and a linear transfer function in the output layer. Good prediction was found as measured by a regression coefficient (R2 = 0.996) between actual and predicted data.
Article
Full-text available
A rapid method for the detection of Escherichia coli (ATCC 25922) in packaged alfalfa sprouts was developed. Volatile compounds from the headspace of packaged alfalfa sprouts, inoculated with E. coli and incubated at 10 degrees C for 1, 2, and 3 days, were collected and analyzed. Uninoculated sprouts were used as control samples. An electronic nose with 12 metal oxide electronic sensors was used to monitor changes in the composition of the gas phase of the package headspace with respect to volatile metabolites produced by E. coli. The electronic nose was able to differentiate between samples with and without E. coli. To predict the number of E. coli in packaged alfalfa sprouts, an artificial neural network was used, which included an input layer, a hidden layer, and an output layer, with a hyperbolic tangent sigmoidal transfer function in the hidden layer and a linear transfer function in the output layer. The network was shown to be capable of correlating voltametric responses with the number of E. coli. A good prediction was possible, as measured by a regression coefficient (R2 = 0.903) between the actual and predicted data. In conjunction with the artificial neural network, the electronic nose proved to have the ability to detect E. coli in packaged alfalfa sprouts.
Article
Artificial neural networks (ANN) recently gained attention as a fast and flexible vehicle to microwave modeling and design. Fast neural models trained from measured/simulated microwave data can be used during microwave design to provide instant answers to the task they have learned. We review two important aspects of neural-network-based microwave modeling, namely, model development issues and nonlinear modeling. ii systematic description of key issues in neural modeling approach such as data generation, range and distribution of samples in model input parameter space, data scaling, etc., is presented. Techniques that pave the way for automation of neural model development could be of immense interest to microwave engineers, whose knowledge about ANN is limited. As such, recent techniques that could lead to automatic neural model development, e.g., adaptive controller and adaptive sampling, are discussed. Neural modeling of nonlinear device/circuit characteristics has emerged as an important research area. An overview of nonlinear techniques including small/large signal neural modeling of transistors and dynamic recurrent neural network (RNN) modeling of circuits is presented. Practical microwave examples are used to illustrate the reviewed techniques. (C) 2001 John Wiley & Sons, Inc.
Article
Moisture sorption isotherms of cowpea (Vigna unguiculata L. Walp), powdered cowpea and protein isolate of cowpea at 10, 20 and 30°C were derived. The sorption data were treated according to many well-known sorption isotherm equations. The goodness of fit was evaluated on the basis of criteria such as the mean relative deviation modulus, coefficient of determination and standard errors of models. It was found that the GAB model was the most satisfactory model for representation of the sorption data. Monolayer moisture contents (M0) were determined from the GAB and the BET equations. The temperature dependence of M0 was examined. Heats of sorption data were evaluated using the Clausius–Clapeyron equation. An attempt was made to determine the moisture contribution of the protein component of cowpea to the total moisture content.
Article
Moisture sorption isotherms of Thai Jasmine rice crackers were determined at 30, 45 and 60°C over a water activity range of 0.10 to 0.95 using a static gravimetric technique. Moisture sorption isotherms of rice crackers exhibited the sigmoid (Type II) shape. The moisture content of rice crackers decreased as temperature increased at a given water activity of the storage environment. The Brunauer, Emmett and Teller (BET) and Guggenheim-Anderson-de Boer (GAB) models were applied to fit the experimental data. The isosteric heat of sorption at different moisture levels was also determined using the Clausius–Clapeyron thermodynamic equation. A nonlinear regression analysis method was determined to evaluate the parameters of sorption equations. The criteria used to evaluate the goodness of fit of each model were the mean relative percentage deviation modulus (E) and the percentage root mean square error (RMSE). The more extended range of application of the GAB equation over the BET equation was evident. The GAB model gave the best fit to the experimental sorption data for a wide range of water activity (0.10–0.95) while the BET model gave the best fit for a water activity range of less than 0.60. The GAB model is considered suitable to predict the moisture sorption isotherm of rice crackers since it gave low E and RMSE values. The heat of sorption values of rice crackers were found to be large at low moisture content and decreased with an increase in food moisture content.
Article
A simple mathematical procedure for the estimation of water vapour transmission rate under different conditions is presented and discussed. The ASTM dynamic and gravimetric methods were used for measuring the water vapour transmission rate of four thin plastic films (PET 19 μm, PVC 16 μm, EVA 15 μm and LDPE 23 μm) under 38 different conditions of temperature and relative humidity and the corresponding permeance values were drawn. On the basis of the Clausius Clapeyron's relationship between temperature and water vapour pressure, a simple equation was derived for a rapid transformation of water vapour transmission rate data. The mathematical procedure was applied to the data collected for the four thin films and good agreement was found between observed and predicted values for the PET and LDPE films, while for the PVC and EVA films the equation developed led to an overestimate of water vapour transmission at the lower temperatures.
Article
() ABSTRACT: Artificial neural networks ANN recently gained attention as a fast and flexible vehicle to microwave modeling and design. Fast neural models trained from measured simulated microwave data can be used during microwave design to provide instant answers to the task they have learned. We review two important aspects of neural-network-based microwave modeling, namely, model development issues and nonlin- ear modeling. A systematic description of key issues in neural modeling approach such as data generation, range and distribution of samples in model input parameter space, data scaling, etc., is presented. Techniques that pave the way for automation of neural model development could be of immense interest to microwave engineers, whose knowledge about ANN is limited. As such, recent techniques that could lead to automatic neural model development, e.g., adaptive controller and adaptive sampling, are discussed. Neural model- ing of nonlinear device circuit characteristics has emerged as an important research area. An overview of nonlinear techniques including small large signal neural modeling of () transistors and dynamic recurrent neural network RNN modeling of circuits is presented. Practical microwave examples are used to illustrate the reviewed techniques. 2001 John Wiley & Sons, Inc. Int J RF and Microwave CAE 11: 421, 2001.
Article
Equilibrium moisture sorption isotherms of fried cassava-shrimp chips were investigated by measuring water activity at 20, 30, and 40°C for modeling transient diffusion of moisture. The sorption isotherms followed a type III shape in water activity range of 0.1–0.9 showing negligible influence of temperature, and were best represented by the GAB, Henderson, and modified Henderson equations. Effective moisture diffusivity of chips based on analytical solution of Fick's law ranged from 0.8 × 10 to 2.1 × 10 m/s during exposure to fixed relative humidity (50, 70, and 90%) and temperature (20, 30, and 40°C) environments. An Arrhenius type relationship adequately described the dependence of effective moisture diffusivity both on temperature and relative humidity, and led to good agreement between experimental and estimated transient moisture sorption data.
Article
An artificial neural network (ANN) was developed to predict heat and mass transfer during deep-fat frying of meatballs. Frying time, radius of meatball, fat diffusivity, moisture diffusivity, heat transfer coefficient, fat conductivity, initial moisture content, thermal diffusivity, initial meatball temperature and oil temperature were all input variables. Temperature at meatball geometrical centre (T0), average temperature of meatball (Tave), average fat content of meatball (mf,ave), and average moisture content of meatball (mave) were outputs. The data used to train and verify the ANN were obtained from validated mathematical models. Trained ANN predicted T0, Tave, mf,ave and mave with 0.54, 0.14, 0.03 and 0.10% mean relative errors, respectively.
Article
Wafers are low moisture food products whose crunchiness or crispness is considered a primary textural attribute, highly affected by the product's physical state (glassy or rubbery). The water activity–water content–glass transition relationships for commercial wafers are reported, using Gordon and Taylor's equation to model the water plasticization effect and also Brunnaver-Emmett-Teller (BET) and Guggenheim-Anderson-DeBoer (GAB) sorption models. BET monolayer moisture content was 6.2% and the moisture limit to fit this model was about 11.5%. Critical water activity and critical water content for the glass transition of the product at 20 °C were 0.591 and 0.118 (mass fraction) respectively. Abrupt changes in the mechanical product properties, evaluated from a three-point bend test, could be observed at these limits. Moisture levels between 6 and 11% give rise to a glassy state matrix in the product and so to acceptable product crispness. Below this range, the glassy matrix seems to turn fragile and, above this range, the product becomes rubbery.
Article
The loss or gain of moisture through a package depends on the sorption isotherm of the food material, the conditions of temperature and relative humidity at which it is stored, and the water vapour permeability of the packaging material. The water vapour permeability (WVP) characteristics of paperboard, polypropylene, and polyethylene used for pasta were determined at 30 to 45°C and from 11 to 85% r.h. Both temperature and relative humidity affected the water vapour permeability of all the three films. Paperboard showed a greater WVP at 30°C compared to 45°C due to a higher moisture content which swells the matrix. Thus, it did not follow the typical Arrhenius relationship of a higher permeability at higher temperature. Polyethylene and polypropylene, on the other hand, showed a higher WVP as the temperature was increased. A second degree polynomial was fitted to predict the WVP of paperboard as a function of the ambient conditions. It was found that the creases and end openings on a typical paperboard box increase the overall transmission rate by two times over that of paperboard alone, therefore reducing the shelf life to half under adverse conditions. A dynamic mathematical model to predict mixture transfer for pasta packages under controlled unsteady state conditions of temperature and r.h. was developed and tested; good predictions were obtained.
Article
In this paper, a relevant automated electromagnetic (EM) optimization method and a novel, fast, and accurate artificial neural network are proposed for the efficient CAD modeling of microwave circuits. We lay the groundwork for our investigation of radial wavelet neural networks WNNs trained by BFGS (Broyden-Fletcher-Goldfarb-Shanno) and LBFGS (limited memory BFGS) algorithms and their application to determine the scattering parameters of the circuit under study. Wavelet theory may be exploited in deriving a good initialization for the neural network, and thus improved convergence of the learning algorithm. The optimization method combines a rigorous and accurate global EM analysis of device performed with a finite-element method (FEM) and a fast neural model deduced from its segmented EM analysis. Finally, experimental results, which confirm the validity of the WNN model, and good agreement between theoretical optimization results and experimental ones are reported. ©1999 John Wiley & Sons, Inc. Int J RF and Microwave CAE 9: 297–306, 1999.
Article
The objective of this work was to evaluate three different mathematical models (linear, middle point and logarithm interval) to predict the stability of commercial cornflakes in flexible packages. The limiting factor for cornflakes' shelf-life was gain of moisture from the external environment, which caused loss of crispness and thus was unacceptable to the consumer. It was concluded that both middle point and logarithm interval methods are suitable for estimating the stability of cornflakes with a good approximation. The accuracy of these two methods depends on the precision of the experimental points of the sorption isotherms. The best fit to adjust the moisture isotherm was the GAB model. Knowledge of the mode of deterioration of the product and the sensory critical moisture content were fundamental to understanding and setting up the limits to confirm the models. Copyright © 2005 John Wiley & Sons, Ltd.
Article
In this study, we investigated the quality and shelf-life of a dragee product obtained by coating a confectionery sunflower kernel with sugar syrup. The product was packed in oriented polypropylene/oriented polypropylene, oriented polypropylene/metallised oriented polypropylene, polyester/polyethylene and kept at room temperature in daylight for 5 months. At the beginning of the experiment, the dragee product was in the category of excellent sensory quality (total score of sensory evaluation is between 18.0 and 20.0) in terms of its colour, smell, taste, mastication and structure. During the storage, these properties changed and the product lost its stability. Peroxide value in the dragee product increased from 0.5 to 9.0 mmol O/kg during the 5 months of storage. The free fatty acid content also increased from 1.3% in fresh dragee product to 2.5% after 5 months of storage. The packaging material metallised polyester/polyethylene, labelled metPET/PE, has the lowest oxygen permeability (8.0 mL m−2/dan Δp 1 bar) and because of this it had the strongest influence in the prevention of hydrolytic and oxidative changes in the final product.
Article
Moisture content plays a significant role on the texture of ready-to-eat (RTE) snacks as it directly affects the crispness that is a key factor for their acceptance. The popular low-density RTE snacks such as corn balls has been selected as a model system to assess the detailed textural attributes when the moisture content was varied between 2% and 10%. The compression curve (force offered by the sample at different time of compression) apparently shows three prominent zones. The slope of the curve is high in the first and last zones but marginally increases in between these two zones when most of the fractures occur.Good indices to judge the texture of a snack are firmness, Young’s modulus and number of peaks; the latter has been defined as a drop in force of magnitude ⩾1 N. Critical moisture content for corn ball is 4% above which these three indices is markedly affected and product approaches unacceptable domain. The microstructure of corn balls shows the presence of air cells with thin walls and vacuoles. Deformed/damaged cells, fracture(s) in the cell walls and line of fracture for compressed samples have been observed.
Article
The issue of selecting appropriate model input parameters is addressed using a peak and low flow criterion (PLC). The optimal artificial neural network (ANN) models selected using the PLC significantly outperform those identified with the classical root-mean-square error (RMSE) or the conventional Nash–Sutcliffe coefficient (NSC) statistics. The comparative forecast results indicate that the PLC can help to design an appropriate ANN model to improve extreme hydrologic events (peak and low flow) forecast accuracy. Copyright © 2001 John Wiley & Sons, Ltd.
Article
In this article we discuss artificial neural networks-based fault detection and isolation (FDI) applications for robotic manipulators. The artificial neural networks (ANNs) are used for both residual generation and residual analysis. A multilayer perceptron (MLP) is employed to reproduce the dynamics of the robotic manipulator. Its outputs are compared with actual position and velocity measurements, generating the so-called residual vector. The residuals, when properly analyzed, provides an indication of the status of the robot (normal or faulty operation). Three ANNs architectures are employed in the residual analysis. The first is a radial basis function network (RBFN) which uses the residuals of position and velocity to perform fault identification. The second is again an RBFN, except that it uses only the velocity residuals. The third is an MLP which also performs fault identification utilizing only the velocity residuals. The MLP is trained with the classical back-propagation algorithm and the RBFN is trained with a Kohonen self-organizing map (KSOM). We validate the concepts discussed in a thorough simulation study of a Puma 560 and with experimental results with a 3-joint planar manipulator. © 2001 John Wiley & Sons, Inc.
Article
Three cookies and two corn snacks were analyzed for major components and their moisture adsorption characteristics were evaluated at 25, 35 and 45 °C. The main composition differences were in fat and total carbohydrate content. The isotherms of each product were different (p < 0.05) and significantly affected by temperature. The mathematical description of the adsorption data was obtained applying some of the most common sorption equations. Peleg's model gave the best description of the experimental data, followed by GAB equation. The mean relative deviations for Peleg's equation were higher as the temperature increased and varied from 2.81 to 10.60%. Monolayer moisture content evaluated with BET and GAB models were in general lower for the cookies. The maximum net isosteric heats of adsorption were lower than 11 kJ/mol.
Article
In this work, an artificial neural network (ANN) is used to predict two parameters of interest for high-pressure food processing: the maximum or minimum temperature reached in the sample after pressurization and the time needed for thermal re-equilibration in the high-pressure system. Both variables together represent in a reliable form the temperature evolution during the high-pressure process. The ANN was trained with a data file composed of: applied pressure, pressure increase rate, set point temperature, high-pressure vessel temperature and ambient temperature altogether with the parameters to predict. After a proper training, the ANN was able to make predictions accurately and therefore, it becomes a useful tool to design and optimize high-pressure processes in the food industry where the pressure/temperature evolution is an essential factor to control the microbiological and/or enzymatic activity of the products.
Article
The moisture sorption isotherms of potato were determined using a gravimetric static method at 30, 45 and 60 °C, and over a range of relative humidities. The isotherms exhibited Type II behaviour, with the sorption capacity decreased with increasing temperature. The Guggenheim–Anderson–de Boer (GAB) and Halsey models were found to adequately describe the sorption characteristics. Calculation of the thermodynamic properties (differential enthalpy, integral enthalpy, differential entropy and integral entrophy) was further used to provide an understanding of the properties of water and energy requirements associated with the sorption behaviour. Isosteric heats (differential enthalpies) were calculated through direct use of moisture isotherms by applying the Clausisus–Clapeyron equation. The differential enthalpy and entropy decreased with increasing moisture content and were adequately characterised by a power law model. A plot of differential heat versus entropy satisfied the enthalpy–entropy compensation theory. The spreading pressures (adsorption and desorption) increased with increasing water activity, and decreased with increasing temperature. The net integral enthalpy increased with moisture content to a maximum value (around the monolayer moisture content) and then decreased. In a reverse manner, the net integral entropy decreased with moisture content to a minimum value and then increased.
Article
Concerning the learning problems of recurrent neural networks (RNNs), this paper deals with the problem of approximating a dynamical system (DS) by an RNN as one extension of the problem of approximating trajectories by an RNN. In particular, we systematically investigate how an RNN can produce a DS on the visible state space to approximate a target DS. First, it is proved that RNNs without hidden units uniquely produce a certain class of DSs. Next, a neural dynamical system (NDS) is proposed as such a DS that an RNN with hidden units can produce on the visible state space, and affine neural dynamical systems (A-NDSs) are constructed as concrete examples of NDSs. Moreover, we prove that any DS on a Euclidean space can be finitely approximated by some A-NDS with any precision, and propose adopting an A-NDS as such a DS that an RNN with hidden units produces to approximate a target DS. © 2000 Scripta Technica, Syst Comp Jpn, 31(4): 77–86, 2000
Article
Fluctuation-driven learning rule is proposed for continuous-time recurrent neural networks. In so doing, random fluctuations nj(p, t)(j: neuron number, p: input pattern number, 0 ⩽ t ⩽ Tp, Tp: pattern length) are superimposed on every neuron's threshold. Probability density Nj(nj) of fluctuation amplitude is treated as a time-invariant, and auxiliary function gj(nj): −dNj/dnj = gjNj is introduced. For fluctuations nj(p, t), neuron outputs rj(p, t) and instantaneous error e(p, t) are probabilistic quantities. In so doing, learning rule for synaptic weight wji from i-th neuron is Rji(p, t) = ∫gj ˙ ridτ/τj, &Deltapwji = −μ ∫e ˙ Rjidt/Tp (rj: time constant of membrane potential, μ : learning coefficient). It is shown theoretically that expected mean error (∫edt/Tp) may be minimized by steepest descent. This learning rule does not require any additional functions such as adjoint system or sensitivity system, and can be executed in time-forward direction by simple integrating, which is distinctive of previous algorithms. The features of the proposed method are confirmed through numerical experiments with JK flip-flop, dynamical system's inverse model, and speed control of moving object. © 2001 Scripta Technica, Syst Comp Jpn, 32(3): 14–23, 2001
Article
Beech (Fagus sylvatica L.) seeds indicate intermediate storage behaviour. Properties of water in seed tissues were studied to understand their requirements during storage conditions. Water sorption isotherms showed that at the same relative humidity (RH) the water content is significantly higher in embryo axes than cotyledons. This tendency maintains also after recalculating the water content for zero amount of lipids in tissues. Differential thermal analysis (DTA) indicated water crystallization exotherms in the embryo axes at moisture content (MC) higher than 29% and 16% in the cotyledons. In order to examine the occurrence of glassy state in the cytoplasm of beech embryos as a function of water content, isolated embryo axes were examined using electron spin resonance (ESR) of nitroxide TEMPO probe located inside axes cells. TEMPO molecules undergo fast reorientations with correlation time varied from 2 x 10(-9) s at 180 K to 2 x 10(-11) s at 315 K. Although the TEMPO molecules label mainly the lipid bilayers of cell membranes, they are sensitive to the dynamics and phase transformation of the cytoplasmic cell interior. The label motion is clearly affected by a transition between liquid and glassy state of the cytoplasm. The glass transition temperature (T(g)) raises from 253 to 293 K when water content decreases from 18% to 8%. Far from T(g) the motion is described by Arrhenius equation with very small activation energy E(a) in the liquid state and is relatively small in the glassy state where E(a)=1.5 kJ/mol for 28% H(2)O and E(a)=4.7 kJ/mol for 8% H(2)O or less. The optimal storage conditions of beech seeds are proposed in the range from 255 K for 15% H(2)O to 280 K for 9% H(2)O.
Article
A rapid method for detection of Salmonella typhimurium contamination in packaged alfalfa sprouts using solid phase microextraction/gas chromatography/mass spectrometry (SPME/GC/MS) integrated with chemometrics was investigated. Alfalfa sprouts were inoculated with S. typhimurium, packed into commercial LDPE bags and stored at 10+2 degrees C for 0, 1, 2 and 3 days. Uninoculated sprouts were used as control samples. A SPME device was used to collect the volatiles from the headspace above the samples and the volatiles were identified using GC/MS. Chemometric techniques including linear discriminant analysis (LDA) and artificial neural network (ANN) were used as data processing tools. Numbers of Salmonella were followed using a colony counting method. From LDA, it was able to differentiate control samples from sprouts contaminated with S. typhimurium. The potential to predict the number of contaminated S. typhimurium from the SPME/GC/MS data was investigated using multilayer perceptron (MLP) neural network with back propagation training. The MLP comprised an input layer, one hidden layer, and an output layer, with a hyperbolic tangent sigmoidal transfer function in the hidden layer and a linear transfer function in the output layer. The MLP neural network with a back propagation algorithm could predict number of S. typhimurium in unknown samples using the volatile fingerprints. Good prediction was found as measured by a regression coefficient (R(2)=0.99) between actual and predicted data.
Article
Control of initial moisture content and moisture migration is critical to the quality and safety of multidomain foods. Moisture loss or gain from one region or food component to another region will continuously occur in order to reach thermodynamic equilibrium with the surrounding food components and the environment. Two main factors influencing the amount and rate of moisture migration are water activity equilibrium (thermodynamics) and factors affecting the diffusion rate (dynamics of mass transfer). Adding an edible layer between domains, changing the water activity of the food ingredients, changing the effective diffusivity of the water, and changing the viscosity (molecular mobility) in the entrapped amorphous phases are several means to control the water migration between domains in food systems. ####################################################################################################################
A mathematical model for the prediction of water vapor transmission rate at different temperature and relative humidity
  • Pieglovanni
Solid phase microextraction/gas chromatograph/mass spectrometer coupled with discriminant factor analysis and multilayer perceptron neural network for detection of Escherichia coli
  • Siripatrawan
Fluctuation-driven learning rule for continuous-time recurrent neural networks and its application to dynamical system control
  • Watanabe