Use of neural networks and neural network inverse in optimizing food processes.

Abstract

Artificial neural networks (ANN) are widely accepted as a tool that offers an alternative way to tackle complex problems in optimizing the food processes. The interest in using ANN in optimization arises from their capacity to model without any assumptions about the nature of underlying mechanisms and their ability to take into account non-linearities and iterations between variables as well as to perform rapid calculations. They have been used in optimization of parameters, resulting in optimal production of the food preparation. ANN inverse (ANNi) is also being applied to optimize the operating conditions or parameters in food processes. ANNi is a new tool that inverts an ANN and uses an optimization method to find the optimum parameter value (or unknown parameter) for given required conditions in the processes. In order to perform ANNi, it is necessary to build the ANN that simulates the output parameters of a food process. In general, this class of ANN model is constituted of a feedforward network with one hidden layer, considering one or more well-known input parameter(s) of the process. Normally, a Levenberg-Marquardt learning algorithm, a hyperbolic tangent sigmoid and a linear transfer-function with several neurons in the hidden layer (because of the process complexity) are considered in the built model. With the required or optimized output and some input parameters in ANNi, it is possible to calculate the unknown input parameters. Consequently, the ANNi is applied to determine the optimal parameters and it already has applications in different process with very short elapsed time (seconds). The great advantage of using ANNi lies in the fact that it makes it faster and easier to predict optimal parameters to control the food preparation.

Hybrid Smart Strategies to Predict Amine Thermal Degradation in Industrial CO2 Capture Processes

“…The ANN is trained by adjusting the weights, attempting to minimize the error between the calculated values/outputs and the expected/target values . A training or learning algorithm is considered a procedure to adjust the coefficients (weights and bias) of ANN outputs for the proposed/given inputs and correct the known outputs (target values) . Different training algorithms have been developed in the past few years.…”

“… 49 A training or learning algorithm is considered a procedure to adjust the coefficients (weights and bias) of ANN outputs for the proposed/given inputs and correct the known outputs (target values). 50 Different training algorithms have been developed in the past few years. However, they might not be suitable/efficient in some specific cases, and they might not lead to optimum solutions.…”

Employing artificial neural networks and fluorescence spectrum for food vegetable oils identification

“…Neurons can be found in each stratum. Each neuron receives information and changes it before sending it to other neurons with whom it is linked (Hernández, 2009;Pouladzadeh et al, 2016). The receiving neurons are used to calculate weights and biases.…”

A neuro‐fuzzy computational approach for multicriteria optimisation of the quality of espresso coffee by pod based on the extraction time, temperature and blend

“…, 2008; Peng et al. , 2008; Hernández, 2009; Liang & Cheng, 2009; Lu et al. , 2010; Çakmak & Boyacı, 2011 to cite just a few).…”