Title Page
Contents
LIST OF ABBREVIATIONS AND ACRONYMS 8
ABSTRACT 15
한글요약 18
Ⅰ. INTRODUCTION 20
A. Optimization Strategies 20
B. Advanced Control 22
C. AI-Driven Control Strategies 23
Ⅱ. Background Study 25
A. Literature Review 25
B. Thesis Objective 28
Ⅲ. Mathematical Modeling of PMSMs 29
A. Electrical Model 30
B. Mechanical Model 31
C. Modulation Scheme 32
1. SVPWM 32
2. Simplified SVPWM 35
Ⅳ. Optimization Strategies 41
A. Cost Function 44
B. Particle Swarm Optimization 45
C. Cuckoo Search Optimization 48
D. JAYA Optimization 51
E. Results and Discussion 55
Ⅴ. Advanced Control 61
A. Maximum Torque per Armature 62
B. Torque Pulsation Analysis 64
C. Reaching Law Sliding Mode Control 65
D. Results and Discussion 72
Ⅵ. AI-Driven Control Strategies 80
A. Feed-Forward Neural Network for Switching Classification 81
1. Mathematical Model of VSI 81
2. Model-Based Predictive Current Control 83
3. Proposed Model-free Predictive Current Control 84
4. Results and Discussion 88
B. Deep Symbolic Regression (DSR) 98
1. Proposed Current Controller 99
2. Test Setup 101
3. Test Results 102
Ⅶ. CONCLUSION 108
A. Future Work 110
A. Appendix 111
1. Particle Swarm Optimization 111
2. Cuckoo Search Optimization 115
3. Jaya Optimization 118
4. Experimental Results of DSR 121
REFERENCES 123
Table 2. Amplitude of output phase and line voltages at each sector 33
Table 3. Simulation test parameters. 36
Table 4. SPMSM Parameter Values. 53
Table 5. Optimization algorithm hyperparameters. 56
Table 6. Algorithm performance comparison. 60
Table 7. Control Performance Comparison 60
Table 8. Test parameters. 71
Table 9. Comparative Performance of conventional and proposed control scheme. 78
Table 10. Inverter switching configuration. 86
Table 11. Performance analysis of trained NN under different number of hidden layers and test cases. 89
Table 12. Motor Parameters. 91
Table 13. Comparative analysis of model-based and model-free control design. 98
Figure 1. Three-phase two-level VSI-fed SPMSM. 29
Figure 2. Voltage vectors produced by a two-level, three-phase voltage source inverter. 33
Figure 3. Calculation of dwell time for n sectors. 35
Figure 4. offset voltage. 36
Figure 5. Phase and Line to Line voltage (a and c) Voltages(b and d) Gating Pulses. 37
Figure 6. Inverter Output (a) SVPWM and Simplified SVPWM output(b) RL load with LC filter. 38
Figure 7. Experimental platform. 38
Figure 8. Experimental Results: Pole voltage (Ta,Tb,Tc) and phase current .[이미지참조] 39
Figure 9. Experimental Results: Inverter Output (a) SVPWM (b)Simplified SVPWM output. 39
Figure 10. Experimental Results: Motor drive response. 40
Figure 11. An overview of Controller tuning methods. 42
Figure 12. Block diagram of speed control loop (a) PI controller(b) IP controller. 43
Figure 13. Structural representation of outer loop speed control of PMSM. 44
Figure 14. Selection of controller optimal gain using PSO. 47
Figure 15. Flowchart of Particle Swarm Optimization. 48
Figure 16. Selection of controller optimal gain using CS. 50
Figure 17. Flowchart of Cuckoo Search Optimization. 51
Figure 18. Selection of controller optimal gain using JAYA. 53
Figure 19. Flowchart of JAYA Optimization. 54
Figure 20. Optimization convergence trajectory. 54
Figure 21. Experimental Platform. 55
Figure 22. Motor response at step speed reference with constant load. 57
Figure 23. Experimental results for speed convergence to speed step reference. 57
Figure 24. Experimental results for speed convergence at variable speed reference. 58
Figure 25. Motor response at variable step speed reference. 59
Figure 26. Motor response at step speed reference with Variable load input. 59
Figure 27. Block diagram of speed control loop. 64
Figure 28. Sliding phase mechanism of the SMC. 66
Figure 29. Structural diagram of proposed compensation scheme. 71
Figure 30. Analysis of SRF for various speed and torque. 73
Figure 31. Speed profile at low speed operating range under load variation. 74
Figure 32. IPMSM current and torque response under nominal parameters with load step variation. 75
Figure 33. Reference output state's variable. 76
Figure 34. IPMSM drive speed response under sinusoidal speed input. 77
Figure 35. Visualization of mpc scheme. 83
Figure 36. Three-layer FNN with measured and reference currents (Iαβ[k,k¯¹]) and switching states (S₁,₃,₅k¯¹) as inputs, whereas the output constitutes optimal voltage vector Vi.[이미지참조] 85
Figure 37. An overview of proposed design. 86
Figure 38. Visualization of Softmax output in a trained network. 88
Figure 39. Model-free current controller of an SPMSM fed by 2L3P VSI. 89
Figure 40. Confusion matrix for multi-class classification. 90
Figure 41. Motor speed response in the wide speed operating range under no load condition. 92
Figure 42. Motor dynamic response under varied load conditions at 800 rad/s. 93
Figure 43. Motor dynamic response under varied load conditions. 94
Figure 44. Dynamic performance at wide speed operation range under load conditions. 95
Figure 45. Expanded result of Figure 44, wide speed operating range under load torque. 96
Figure 46. Performance evaluation of proposed and conventional control approaches in steady-state and transient-state under normal, parameter mismatch (Lₛ=5.5mH) and... 97
Figure 47. Deep symbolic regression core architecture. 100
Figure 48. Rewards for generated numerical expression. 101
Figure 49. Conventional and proposed field-oriented control architecture of SPMSM. 102
Figure 50. Simulation Results − Steadystate Controller Performance: Constant Speed and Torque Responses. 103
Figure 51. Simulation Results − Controller Performance: under Varied Speed and Load Conditions. 104
Figure 52. Comparative analysis of current THD between proposed and traditional control method for Permanent Magnet Synchronous Motor. 104
Figure 53. Simulation Results − Dynamic load response of SPMSM with Constant Speed. 105
Figure 54. Experimental Result: Motor Response at varied step Speed reference employing PI current Controller 106
Figure 55. Experimental Result: Motor Response at varied step Speed reference employing DSR current Controller 107
Figure 56. Experimental platform. 122
Figure 57. Motor drive dq-current tracking response at varied speed. 122
Figure 58. IM performance 123