Artificial neural networks in food industry
-
Upload
pragati-singham -
Category
Food
-
view
283 -
download
3
Transcript of Artificial neural networks in food industry
ARTIFICIAL NEURAL NETWORK IN
FOOD INDUSTRY
PRAGATI SINGHAMPH.D. (FS & PHT)
10703ICAR-IARI
CONTENTS
1. Introduction 2. Biological inspiration 3. Architecture4. Applications5. Advantages6. Disadvantages7. Conclusion 8. References
INTRODUCTION Food Industry (ibef, 2015)
High growth and high potential
Current value : US$ 39.71 billion
Indian Food Industry• Investment in food processing sector of Rs.100,000 crores (Union
Budget 2015-16) • Contributes about 14% of manufacturing GDP • 1st International Mega Food Park worth Rs.136 crores at Punjab, India
DIFFERENT SECTORS OF FOOD INDUSTRY
Sectors of
Food industry
Dehydration
Baking
Canning
Extrusion
PROBLEMS ASSOCIATED WITH..
Lack of validity of empirical models in simulating wide range of temperatures, air velocity and humidity during drying (Tohidi et al., 2012)
Complexity of mathematical models and large computation time required for modeling of drying process of food (Singh and Pandey, 2011)
Dehydration
Baking
Extrusion
Canning
CONTD.
Lack of non-linear interdependence of viscoelastic properties and gas retention properties on rheological properties of dough (Abbasi, Djomeh and Seyedin, 2011)
Insufficiency in bake level inspection of biscuits (Yeh and Leonard, 1994)
Dehydration
Baking
Canning
Extrusion
CONTD.
Lack of precision in simulation of dynamic temperature during retort processing (Llave, Hagiwara and Sakiyama, 2012)
Dehydration
Baking
Canning
Extrusion
CONTD.
Complexity in modelling of non-linear relationship among variables of extrusion (Popescu et al.,2000)
Dehydration
Baking
Extrusion
Canning
MAJOR PROBLEMS IDENTIFIED
Complexity of biomaterial
Non-linearity of process
Large computational time
Wide range of parameters
Precision
ARTIFICIAL NEURAL NETWORK
HISTORY
ARTIFICIAL NEURAL NETWORK
• It is a dynamic computational modeling tool to solve real-world problems (Chen et al., 2007)
• It is comprised of densely interconnected adaptive simple processing elements that are capable of performing massively parallel computations for data processing.
BIOLOGICAL INSPIRATION
An artificial neuron is an imitation of the human neuron
WORKING
CONTD.
CONTD.
MODELING WITH ANN
R2
Root Mean Square ErrorRMSE
Training Testing
Back Propagation
TRAINING
Supervised Learning
ReinforcementLearning
Unsupervised Learning
ARCHITECTURE
Forward feed network
Radial Basic Function (RBF) Network
Self Organizing Maps
(SOMs)
APPLICATIONS
Prediction
Optimization
Control
Classification
PREDICTION OF HYDRATION CHARACTERISTICS OF PADDY (KALE ET AL., 2013)
Hydration : Important process in parboiling (pre-treatment) to attain complete gelatinization of paddyModel used : Generalized Page Model
: Artificial Neural NetworkANN
Multilayer perceptron Neural Network
RESULTS
Data Points 108Training 60
Testing 21
Validation 27
Modelling of ANN
Model R2 MSE
Generalized Page Model 0.65 0.0018
Multilayer perceptron Network 0.99 0.0013
Comparison between Generalized Page Model and Multilayer Perceptron Network
RESULTS
(a) Generalized Page Model (B) MLP network
MOISTURE RATIO
ADVANTAGES
Exploits non-linearity
High computational speed
Offers wide range
Learning ability
Fault tolerance
DISADVANTAGES
Works as black-box
Large amount of training data
Overfitting of data
CONCLUSION
• ANN can be successfully used for modeling complex food materials
• Prediction of food characteristics in various thermo-physical processes at high computational rate
• Optimization of the supply chain process, parameters, cost and manpower
• Control of the quality of the finished or new product can be quantified
REFERENCESAbbasi, H., Djomeh, E.J. and Seyedin, S.M (2011). Applicatin of Artificial Neural Network and Genetic Algorithm for predicting three important parameters in Bakery Industries, 2, 51-63.
Chen, C.R., Ramaswamy, H.S. and Marcotte, M. (2007). Neural network applications in heat and mass transfer operation in food processing chapter Heat transfer in food processing, © WIT Press, 13, 39-59.
Kale, S.J., Jha, S.K., Jha, G.K., and Samuel, V.K. (2013) Evaluation and modelling of Water absorption characteristics of paddy. J of Agricultural Engg. 50 (3), 29-38.
Llave, Y.A., Hagiwara, T. and Sakiyama, T. (2012). Artificial neural network model for prediction of cold spot temperature in retort sterilization of starch based foods. Journal of food engineering, 109, 553-560.
Singh, N.J. and Pandey, R.K. (2011). Neural Network approches fr prediction of drying kinetics during drying of sweet potato. Agricultural Engineering International, 13, 11-22.
Tohidi, M., Sadeghi, M., Mousavi, S.M. and Mireei, S.A (2012). Artificial neural network modeling of process and product indices in deep bed drying of rough rice. Turk Journal of Agriculture, 36, 738-748.
Yeh, J.C.H. and Haney, L.C.G (1994). Biscuit bake assessment by an Artificial Neural Network, 5, 266-269.
http://www.ibef.org/industry/indian-food-industry.aspx
Thank you