Original scientific paper

Artificial neural network and mathematical modeling of drying of apples treated with high intensity ultrasound

Sven Karlović orcid.org/0000-0002-6117-2063 ; Prehrambeno – biotehnološki fakultet, Pierottijeva 6, 10000 Zagreb, Hrvatska
Damir Ježek ; Prehrambeno – biotehnološki fakultet, Pierottijeva 6, 10000 Zagreb, Hrvatska
Mladen Brnčić ; Prehrambeno – biotehnološki fakultet, Pierottijeva 6, 10000 Zagreb, Hrvatska
Branko Tripalo ; Prehrambeno – biotehnološki fakultet, Pierottijeva 6, 10000 Zagreb, Hrvatska
Tomislav Bosiljkov ; Prehrambeno – biotehnološki fakultet, Pierottijeva 6, 10000 Zagreb, Hrvatska
Filip Dujmić ; Prehrambeno – biotehnološki fakultet, Pierottijeva 6, 10000 Zagreb, Hrvatska
Ana Skupnjak ; Prehrambeno – biotehnološki fakultet, Pierottijeva 6, 10000 Zagreb, Hrvatska
Ferdinand Troha ; Veleučilište u Požegi, Vukovarska 17, 34000 Požega, Hrvatska

Abstract

The work involves experimentation on infrared drying of apple slices treated with high intensity ultrasound. Variables covered are amplitude of ultrasound (0, 20, 40, 60, 80 and 100 %) and drying temperature (50, 60 and 70 °C). Obtained data was modeled using various empirical mathematical models as well as artificial neural network. Simulated data obtained from models were compared to experimental data and statistically processed to calculate mean square error. Although all models showed good fit to sets of experimental data, Pages model had smallest error thus suggesting best fit to experimental data. Comparison of artificial neural network to tested mathematical models showed that it has significantly lower prediction error and significantly better fitting (R2 = 0,9980) to experimental data through full range of experiments compared to mathematical models.

Keywords

Artificial neural network; high intensity ultrasound; drying

Hrčak ID:

105657

URI

https://hrcak.srce.hr/105657

Publication date:

30.6.2013.

Article data in other languages: croatian

Visits: 1.785 *