Title Page
Contents
Abstract 12
Chapter 1. Introduction 14
Chapter 2. Mathematical Modeling of LIB and Model Predictive Control 21
2.1. Diffusion of Li ions 22
2.2. Evolution of Cell Voltage 25
2.3. Effect of Temperature 27
2.4. Capacity Degradation Physics 28
2.4.1. New SEI Growth 28
2.4.2. Particle Cracking 30
2.4.3. Capacity Fade Mechanism 31
Chapter 3. Results and discussion of LIB Modeling and control 35
3.1. Model validation 35
3.1.1. Temperature and Voltage Evolution 35
3.1.2. Stress Analysis 37
3.1.3. Capacity Degradation 40
3.2. Optimized Battery Charging Profile 43
3.2.1. MPC Formulation 43
3.1.2. MPC results 45
Chapter 4. AI based Simulation of Graphene Growth in the CVD system 50
4.1. Chemical Vapor Deposition system 50
4.2. Machine learning for measurement of graphene specificaions 53
4.2.1. R-CNN 54
4.3. Analysis of graphene growth 55
4.3.1. Feed Forward Neural Network (FFNN) 55
4.4. SEM image generation 58
4.4.1 Generative Adversarial Network 58
Chapter 5. AI based Simulation Results of Graphene Growth 60
5.1. Measurement results 60
5.2. CVD system modeling results 61
5.3. SEM prediction results and optimization 65
Chapter 6. Conclusion 70
References 73
Table 1. Design parameters of the enhanced SPM 32
Table 2. Design parameters of chemical/mechanical degradation model. 33
Table 3. RMSE values between predicted and experimental values of voltage and temperature at different discharging and charging rates. 37
Table 4. Operating conditions of four case studies for the evaluation of the capacity fade. 41
Table 5. Final values of charging time, SOC, and capacity fade under different initial temperatures of 15℃, 25℃, and 45℃. 47
Table 6. Final values of charging time, SOC, and capacity fade under different initial SOC of 10%, 20%, and 30%. 48
Table 7. Experimental conditions for graphene growth. 52
Table 8. Training result of Artificial Neural Networks optimized by Bayesian optimization. 62
Table 9. Manually measurement results, ANN and SVM predicted results, and SEM images results measured by R-CNN at experimental conditions used in Fig.12 69
Fig. 1. Schematic illustration of an enhanced SPM with chemical/mechanical degradation mechanisms. 21
Fig. 2. Experimental validation results for voltage and temperature evolution. 35
Fig. 3. Stress changes with respect to the distance over time (3600 s) under a constant charging rate of 1C on (a) the radial and (b) the tangential directions. 38
Fig. 4. Stress changes with respect to the distance over time (3600 s) under a constant discharging rate of 1C on (a) the radial and (b) the tangential directions. 39
Fig. 5. Maximum tensile stress changes with respect to C-rate under different temperatures. 40
Fig. 6. Comparison between predicted (simulation) and experimental capacity fade. 41
Fig. 7. Capacity fade under different charging/discharging rates, DODs and temperatures: (a) Case 1 (2C/2C, DOD=50%, and T=15℃) (b) Case 2 (2C/2C,... 43
Fig. 8. A schematic framework of the proposed MPC. 44
Fig. 9. Current profile, SOC, capacity fade, and voltage results of 1C CC-CV (black), 2C CC-CV (green), and MPC (red) under different temperatures of (a)... 47
Fig. 10. Current profile, SOC, capacity fade, and voltage results of 1C CC-CV (black), 2C CC-CV (green), and MPC (red) under different initial SOC of (a)... 48
Fig. 11. a) Schematic illustration of CVD growth with process variables and representative SEM image of the synthesized graphene. b) Experimental... 50
Fig. 12. a) Detected graphene domains from the SEM image, b) measuring size by transforming various graphene to regular hexagon 53
Fig. 13. Schematic structure of feed-forward neural network. 55
Fig. 14. Structure of GAN with training and generated images. 58
Fig. 15. a comparison manually measured size with measured size by R-CNN. 60
Fig. 16. a) Comparison predicted size, size deviation, coverage, and domain density with the measured values. Predicted tendency of graphene growth b) at... 64
Fig. 17. Graphene growth observation through developed simulation. a) graphene growth at different growth temperatures, b) graphene growth at different growth... 65
Fig. 18. a-h) Comparison actual experimental SEM images and predicted SEM images at not-trained experimental conditions. 66
Fig. 19. a) Simulation results with the recommended experimental condition to synthesize graphene which has higher size and lower domain density. b)... 66