Comparison of two artificial intelligence methods (ANNs and ANFIS) for estimating the energy and exergy of drying cantaloupe in a hybrid infrared‐convective dryer.

In this investigation, energy and exergy analysis of a hybrid infrared‐convective (IR‐CV) dryer is presented for drying cantaloupe slices. The experiments were performed at three temperature levels (40, 55, and 70°C), one level of air velocity (0.5 m/s), and three IR powers (250, 500, and 750 W). The relationships between the input process parameters (IR power, input air temperature, and drying time) and the thermodynamic properties of the dried product (moisture ratio, drying rate, energy efficiency, exergy efficiency, and exergy loss) were modeled by implementing the artificial neural network (ANN) and ANFIS. Results indicated that high IR power and air temperature can shorten the drying time, meanwhile increasing the energy efficiency. Based on the obtained results, input air temperature and IR power highly affect the exergy efficiency. The highest exergy efficiency was obtained at the input air temperature of 70°C and IR power of 750 W. The exergy loss was increased by increasing both parameters of the air temperature and IR power. Models developed using ANN and ANFIS indicated that the ANFIS model predicted the moisture ratio, energy efficiency, and exergy loss better than the ANN model, as it estimated these thermodynamic parameters at a higher regression coefficient (>0.9889) than ANN (0.9850), while, the accuracy of ANN model for predicting drying rate and exergy efficiency was better than ANFIS. Novelty impact statement: In this study, the effects of both different infrared powers (250, 500, and 750 W) and air temperature (40, 55, and 70°C) on drying kinetics, energy, and exergy of cantaloupe slices during the drying process were investigated. The prediction model of cantaloupe slice moisture ratio, energy, and exergy in the infrared hot air drying process was established based on artificial neural network and adaptive neuro‐fuzzy inference system. Artificial neural network and adaptive neuro‐fuzzy inference system model predictions agreed well with testing data sets and they could be useful for understanding and controlling the factors affecting drying behaviors. [ABSTRACT FROM AUTHOR]