ARTIFICIAL NEURAL NETWORK MODEL FOR THE PREDICTION OF DRYING PARAMETERS OF.pptx

Neural networks for drying parameters determination

1. ARTIFICIAL NEURAL NETWORK MODEL FOR THE PREDICTION OF DRYING PARAMETERS OF SWEET POTATOES.     

2. OUTLINE BACKGROUND OF STUDY STATEMENT OF THE PROBLEM AIM AND OBJECTIVES OF THE STUDY SCOPE AND LIMITATIONS LITERATURE REVIEW MATERIALS AND METHODS RESULTS AND DISCUSSION CONCLUSION RECOMMENDATION

3. BACKGROUND OF STUDY Drying operations can help in reducing the moisture content of feed materials for avoidance of microbial growth and deterioration and for shell life elongation, to minimize packaging and improving storage for easy transportation. Drying is considered one of the methods that are used to preserve some perishable agricultural produces like sweet potato, to ensure their availability all year round, reduce post-harvest and achieve food security.

4. STATEMENT OF RESEARCH PROBLEM sweet potato have high moisture content, this can lead to deterioration easily if not properly processed especially after harvest and since sweet potatoes are seasonal. T he quality of dried products depends on the entire drying conditions, so it is important to understand the drying process parameters and characteristics of sweet potato, since the ultimate aim is to increase the shelf life or preserve the product by reducing its moisture content.

5. AIM AND OBJECTIVES OF THE STUDY AIM: To develop an Artificial neural network (ANN) model that will predict the drying parameter of the agricultural food material which is sweet potato . Objectives are to: Characterizes the sample sweet potato before and after drying Develop the mathematical model for the drying kinetics of the sample Determine the main parameter for moisture removal with the samples and Fit the experimental data to the developed model which is Artificial Neural Network (ANN).

6. SIGNIFICANCE OF STUDY The developed ANN model will help in determining the moisture ratio and drying rate of the product study and also w ith the successful completion of the research, it will solve some food preservation and storage problems by reducing waste of agricultural product like sweet potato.

7. SCOPE AND LIMITATIONS The scope of this study shall be limited to the development of ANN model for the prediction of the drying parameters of sweet potato: moisture content and drying time.

8. LITERATURE REVIEW S/N AUTHORS/YEARS INPUT PARAMETERS OUTPUT PARAMETERS LEARNING ALGORITHM MODEL PERFORMANCE RESULT 1 Elijah et al (2020) 3(drying time, air speed, temperature) 1(Moisture content) Levenberg-Marquardt algorithm ANFIS, ANN and RSM gave R 2 of 0.998, 0.997 and 0.998, RMSE of  0.0282, 0.1178 and 0.0273, and ARE of 1.665, 4.282 and 1.778 respectively The comparative analysis showed that the RSM and ANFIS were better in predicting the moisture content reduction of potato drying. 2 Yaghoubi et al (2013) 3(drying time, drying temperature and different methods) 1(Moisture content) Back propagation algorithm R2 values 0.9972 for ANN and 0.996, for logarimitic and page models The comparison of the obtained results of ANNs and classical modeling indicated that, the neural networks have a higher capability for predicting moisture ratio compared with classical modeling

9. MATERIALS AND METHODS The model was developed using matrix laboratory (MATLAB) software

10. RESULT AND DISCUSSION Figure 4:1 Neural network progress

11. RESULT AND DISCUSSION (Contd.) Figure 4:2 The developed ANN regression performance  

12. RESULT AND DISCUSSION (Contd.) Figure 4.4: Cross plot of experimental and predicted drying rate  

13. RESULT AND DISCUSSION (Contd.)     Samples MSE R Training 70% 221 3.95236e-4 0.99774 Validation 15% 48 5.12589e-4 0.99777 Testing 15% 48 5.43952e-4 0.99562

14. CONCLUSION The R and MSE values obtained from the network training, validation and testing were 0.99774 and 3.9523x10 -4 for training, 0.99777 and 5.1259x10 -4 for validation, and 0.9956 and 5.4395x10 -4 for testing; the developed ANN model prediction of the sweet potatoes drying moisture content resulted in R, R 2 , MSE, RMSE and AARE values of 0.9966, 0.9933, 2.2563, 1.5021 and 1.1673, respectively; the developed ANN model predicted drying rate resulted in R, R 2 , MSE, RMSE and AARE values of 0.9974, 0.9948, 0.0008, 0.0287 and 0.0196, respectively.

15. RECOMMENDATION F uture study should consider increasing the network input parameters for the better performance of the ANN model; and Future work should consider developing hybridize ANN model that optimizes the prediction of the sweet potatoes drying parameters