---

Prediction of spray droplet diameter distribution depends on the various parameters such as physical properties, fluid velocity, and discharge environment and injector geometry. The stage of forming droplets has a great variety in size and therefore will be predictable with a statistical approach. The maximum entropy principle is one of the most popular and best ways to predict the spray droplet size distribution along with the conservation equations. Due to some drawbacks in this model, the predicted results do not match well with the experimental data. It is suggested to improve the available energy source in the MEP model equation by numerical solution of flow inside the injector based on the CFD technique. This will enhance the calculation accuracy of the turbulent kinetic energy of the output spray. In fact, by using this sub-model in the maximum entropy model, the prediction accuracy of the spray characteristics is improved. Also, the requirement of the maximum entropy model to the experimental data as inputs has been reduced. By the present coupled model, the effect of spray upstream on the droplet size distribution can be considered with a good accuracy. The results show a close agreement with the available experimental data.

---

Lipid oxidation is important issues that can lead to the degradation and destruction of foods containing lipids. A number of antioxidants have been used to solve this problem. Stachys lavandulifolia is a medicinal herb with antioxidant properties. Given that the impact of new technologies compared to traditional methods in terms of saving time, energy, and increase the efficiency of extraction have been identified. The aim of this study was modeling the extraction of antioxidant compounds from Stachys lavandulifolia by ultrasound-assisted extraction method. For this purpose, to model the extraction efficiency of neural network antioxidant compounds, artificial neural network hybrids - genetic algorithm and response surface methodology were used. The best model was obtained based on the results of the neural network model with gradient optimization method, with trainbr training and tansig transfer function and the number of hidden layers of this combination with two neurons 8 in the first layer and 4 in the second layer. For this network structure, an error of 0.0128 and a correlation coefficient of 97.30% were determined. By comparing this method with the response level, the model accuracy increased from 92% to 94.68%. The best result for the hybrid model occurred in the trainbr learning algorithm with the tansig transfer function with a hidden layer and 18 neurons. The error rate and correlation coefficient in this method were equal to 0.0693 and 83.27%, respectively. According to the results of the neural network with the logger method, it answered better and the hybrid method of the genetic algorithm with the neural network was not a suitable model for prediction. Finally, it can be said that mountain tea can be considered as a potential source of antioxidant compounds and neural network can be considered as a successful application method to predict the extraction efficiency of antioxidant compounds.