Sorption isotherms of ready‐to‐puff preconditioned brown rice: Development of classical models and artificial neural network approach

Abstract

Moisture sorption isotherm data of ready‐to‐puff preconditioned brown rice shall enable in finding solutions for prolonging the period between preconditioning and puffing. Designed full factorial experiments were conducted with ready‐to‐puff preconditioned brown rice at five different salt concentrations (SC) ranging 0–4%, varying temperature (T), 20–30°C within eight levels of moisture content (MC), and 3–25% to obtain sorption isotherms. Preconditioning was carried out at 75(±3)°C using a fluidized bed dryer to attain a final MC of 11%, optimum for puffing. Water activity (aw) of the sample increased with increasing T and decreased with increasing SC. The individual and interaction effects of T, SC, and MC on aw are highly significant (p < .001). The experimental sorption data at each SC were fitted with the traditional model at a given temperature. All the models were statistically (p < .001) significant. A feedforward artificial neural network model was developed with three input variables T, SC, and MC as the input layer and aw as the output layer. Inter‐relationship among the process variables is explained better by artificial neural network (R2 = 0.991) and second‐order model (R2 = 0.973) as compared with traditional models, which are limited in their ability to express the relationship among the variables. Practical Applications The key to quality puffed rice is the preconditioning process during its manufacture. During preconditioning, salts and moisture penetrate the rice matrix to give a desirable product. The preconditioned rice should be immediately processed, else absorption or desorption of moisture from the atmospheric air takes place. This leads to finished puffed rice with compromised quality. Liberty of using the preconditioned rice for further processing will wrest upon the processor if suitable packaging is provided to this rice. Extensive knowledge of moisture sorption isotherms is required for designing such packaging material and conditions. The present study is carried out to study the sorption isotherm of ready‐to‐puff preconditioned brown rice. The information available in the present study is useful for food industries to develop a suitable packaging material for time‐independent use of preconditioned rice.

Design and development of resistance heating apparatus‐cum‐solar drying system for enhancing fish drying rate

“…Compared with RSM, ANN was able to predict all the responses with a single model rather than separate one. This may be due to the fact that ANN models are able to handle nonlinear responses better than RSM (Mahanti, Chakraborty, & Babu, 2019). However, interaction and individual effects cannot be interpreted using the ANN model compared with RSM.…”

Puffed rice: A materialistic understanding of rice puffing and its associated changes in physicochemical and nutritional characteristics

“…Joshi, Mohapatra, and Joshi (2014) again performed pressure parboiling at 1 atm pressure in an autoclave with hot soaking of paddy followed by milling and sand roasting at 250 C. This method was previously adopted by Kumar and Prasad (2013). Mahanti, Chakraborty, and Babu (2019) performed pressure parboiling with cold soaking and subsequent steaming and drying. In a recent study, Dash and Das (2019a) optimized the parboiled rice salt preconditioning process using a fluidized bed dryer and used microwave puffing.…”

Non-destructive Estimation of Spinach Leaf Area: Image Processing and Artificial Neural Network Based Approach

“…leaf length (L L ) and width (L W ), the number of output neurons was equal to the dependent variable, i.e., leaf area (L A ). There are no fixed, definite rules for determining the number of hidden neurons needed in a hidden layer [26]. The number of neurons in the hidden layer varied between 2 to 5, it was selected based on the trial and error method and the optimum number of neurons were finalized based on the statistical parameters, coefficient of determination (R 2) and root mean square error (RMSE).…”