OPTIMIZATION OF THE FUZZY CONTROLLER FOR AN IMPLANTABLE INSULIN DELIVERY SYSTEM
Download as pptx, pdf1 like616 views
The document summarizes research on developing a fuzzy logic controller to regulate glucose levels for type 1 diabetic patients. It discusses: - Building a system to mimic the function of healthy pancreatic beta cells and tightly regulate glucose levels. - A literature review on previous control methods like PID and model predictive control. - A physiological model of glucose-insulin interaction. - Developing PID, PD, and PI fuzzy logic controllers and testing their robustness using a glucose-insulin model. - Optimization of the controllers and comparison of their performance to show fuzzy controllers more closely regulate glucose levels.
OPTIMIZATION OF THE FUZZY CONTROLLER FOR AN IMPLANTABLE INSULIN DELIVERY SYSTEM
- 2. building a fuzzy system that imitates the natural functionality of 硫-cells in healthy persons to tightly and optimally regulate glucose concentration in Type 1 diabetic patients
- 3. Introduction Literature Review Physiological Model PID-FLC , PD-FLC, and PI-FLC MFB PID-FLC, MFB PD-FLC, and MFB PI-FLC Optimization Robustness Tests Conclusions
- 5. Diabetic patients : Type 1 or Type 2 Type 2 : (90-95)% , Type 1 : (10-5)% In 2001 :120 million by the year 2025 :300 million In Jordan : in 2002:6.4%, in 2004: 7.5% United States : 21 million in 2005, sixth rank among causes of death, 132 billion dollars , 10% of the health care budget Complications : kidney failure, blindness, heart attack, immune system
- 6. Classical control : PID by E. Renard Model Predictive Control (MPC) : L. Magni comparative studies between FLCs and conventional techniques: Ibbini and Masadeh :closed-loop fuzzy logic controller, classical PID-controller. Ibbini :PI- FLC, a classical PI, a classical PID, an optimal LQR and a classical fuzzy logic controller. M. Khalil : P-FLC, PD-FLC, PI-FLC, and PID-FLC, glucose-insulin model by H.Wang .
- 7. dG G in ( t ) f 2 ( G ( t )) f 3 ( G ( t )) f 4 ( I ( t 3 )) f 5 ( I ( t 2 )) dt 5 dI 0 . 05 t 0 t 15 (min) I in ( t ) d i ( I ( t )) 15 dt 45 t G in ( t ) 0 . 05 5 15 t 45 (min) 45 15 0 . 05 45 t 240 (min) 0 . 25 0 t 5 (min ) t 30 0 . 25 (1 ) 5 t 30 (min) 30 25 I in ( t ) t 30 0 . 25 (1 ) 30 t 120 (min) 120 30 0 . 25 120 120 240 (min)
- 8. di>0=0.0076, 2= 15 min, 3= 5 min. An individual has a meal every 4 hours (i.e. 240 min) and the glucose intake duration is 45 min. Lispro insulin
- 9. Externally Plasma injected insulin, Glucose-insulin physiological model Glucose Iin level Glucose intake rate, Gin Setpoint Fixed/ Error Iin Fuzzy Plasma Change in Error Glucose-insulin Time- controller Glucose level physiological varying model/Patient Glucose intake rate, Gin model
- 17. n 1 MAPE n ( G mod el G fuzzy ) / G mod el t0 150 140 130 glucose mg/dl 120 110 100 90 80 0 500 1000 1500 simulation time
- 19. E-NL E-NM E-NS E-Z E-PS E-PM E-PL CE-NL CE-NM CE-NS CE-Z CE-PS CE-PM CE-PL 1 1 Degree of membership 0.8 Degree of membership 0.8 0.6 0.6 0.4 0.4 0.2 0.2 0 0 -1 -0.5 0 0.5 1 CH-ERROR -1 -0.5 0 0.5 1 ERROR NL 1 NM NS Z PS PM PL paramet GE GCE GIin er Degree of membership 0.8 0.6 value 0.0646 0.4036 0.7705 0.4 0.2 0 0 0.2 0.4 0.6 0.8 1 INSULIN
- 20. 了 (PI-FLC) 0.9875 硫 (PD-FLC) 0.0132 725 723.4995 daily infused insulin (mU/kg/day) 3 2.5938 720 2.5 1.3787 715 2 708.8412708.1953 MAPE(%) 710 1.4136 705.25 1.5 705 1 700 0.5 695 reference PD-FLC PI-FLC PID-FLC 0 model PD-FLC PI-FLC PID-FLC
- 21. 150 reference model 140 PI-FLC PD-FLC 130 PID-FLC gluocse mg/dl 120 110 100 90 80 0 500 1000 1500 simulation time
- 22. 了 (MFB PI-FLC) 0.0054 硫 (MFB PD-FLC) 1.0465 5.995 730 725.5572 daily infused insulin (mU/kg/day) 6 725 5 720 715 4 MAPE (%) 710 705.25 703.2602 3 705 700.4594 1.7921 1.7704 700 2 695 1 690 685 0 reference MFB PD- MFB PI-FLC MFB PID- MFB PD- MFB PI-FLC MFB PID- model FLC FLC FLC FLC
- 23. controller Pervious work My work (MPE %) controller MPE (%), (MAPE %) , (MAPE %) (MAPE %) PD-FLC 12.2407 2.5938 MFB PD-FLC 1.7921 PI-FLC 5.73413 1.3787 MFB PI-FLC 5.995 PID-FLC 6.08743 1.4136 MFB PID-FLC 1.7704 controller Pervious work My work (daily controller (daily insulin (daily insulin insulin mU/kg/day) mU/kg/day) mU/kg/day) PD-FLC 1786.72 723.4995 (40.5 MFB PD-FLC 700.4594 %) (39.2 %) PI-FLC 1087.18 708.8412 MFB PI-FLC 725.5571 (65.2 %) (66.7 %) PID-FLC 940.379 708.1953 MFB PID-FLC 703.2602 (75.31 %) (74.8 %)
- 24. Uncertainty in the Clearance Rate Factor d i UP d i * d i d i Down d i * d i Error in Sensor Measurements Severe Initial Condition Changing the Glucose Intake Profile Unexpected Glucose Intake
- 25. 25 21.4467 14.995 15.0234 16 14 11.7188 20 12 MAPE (%) MAPE (%) 10 15 8 6 10 4 4.4458 3.5441 5 2 0 0 PD-FLC PI-FLC PID-FLC MFB PD-FLC MFB PI-FLC MFB PID-FLC 839.388 836.765 daily infused insulin (mU/kg/day) daily infused insulin (mU/kg/day) 840 865 860.4559 860 835 855 830 850 845 825 840 832.345 833.381 820 817.4924 835 830 815 825 810 820 815 805 PD-FLC PI-FLC PID-FLC MFB PD-FLC MFB PI-FLC MFB PID-FLC
- 26. 250 245 250 uncertain reference model 245 230 MFB PI-FLC PI-FLC 230 215 MFB PD-FLC PD-FLC 215 200 PID-FLC MFB PID-FLC 185 200 uncertain reference model gluocse mg/dl gluocse mg/dl 170 185 155 170 140 155 125 140 110 125 95 110 80 65 95 50 80 0 500 1000 1500 0 500 1000 1500 simulation time simulation time
- 27. 17.3114 17.2457 12 10.6656 18 10 16 7.1329 7.1581 14 8 MAPE (%) 12 6 MAPE(%) 10 7.7514 8 4 6 2 4 0 2 MFB PD-FLC MFB PI-FLC MFB PID-FLC 0 PD-FLC PI-FLC PID-FLC 601.7913 602 610 601.2317 602.2983 601.5 daily imfused insulin daily infused insulin 600 601 (mU/kg/day) (mU/kg/day) 600.5 590 600 599.5973 575.1973 576.2602 580 599.5 570 599 598.5 560 PD-FLC PI-FLC PID-FLC MFB PD-FLC MFB PI-FLC MFB PID-FLC
- 28. 200 195 uncertain reference model 200 195 180 MFB PI-FLC PI-FLC 180 165 PD-FLC MFB PD-FLC 165 150 MFB PID-FLC PID-FLC 150 135 uncertain reference moddel gluocse mg/dl 135 gluocse mg/dl 120 120 105 105 90 90 75 75 60 60 45 45 30 30 15 15 0 0 0 500 1000 1500 0 500 1000 1500 simulation time simulation time
- 29. G measured G fuzzy N (x) Gaussian White noise with Variance=0.1 and Mean=0 1.5 1 0.5 mg/dl 0 -0.5 -1 -1.5 0 500 1000 1500 time
- 30. 14.024 9.2793 25 10 20 MAPE (100%) 8 5.6695 MAPE (%) 6 4.4765 15 11.4596 11.5753 4 10 2 5 0 0 PD-FLC PI-FLC PID-FLC MFB PD-FLC MFB PI-FLC MFB PID-FLC 200 reference model 200 185 reference model PI-FLC 185 MFB PI-FLC 170 PD-FLC MFB PD-FLC 170 PID-FLC MFB PID-FLC 155 155 gluocse mg/dl gluocse mg/dl 140 140 125 125 110 110 95 95 80 80 65 65 50 50 0 500 1000 1500 0 500 1000 1500 simulation time simulation time
- 31. 132 130.675 140 122.5 130 120 89.835 90.04 128 time (minute) time (minute) 100 124.6637 126 80 124 60 120.914 122 40 120 20 118 0 116 PD-FLC PI-FLC PID-FLC MFB PD-FLC MFB PI-FLC MFB PID-FLC 200 reference model 200 185 PI-FLC reference model PD-FLC MFB PI-FLC 170 180 PID-FLC MFB PD-FLC 155 MFB PID-FLC gluocse mg/dl 160 gluocse mg/dl 140 125 140 110 120 95 80 100 65 0 120 240 360 480 600 720 840 9601080 1200 1320 1500 1440 80 0 120 240 360 480 600 720 840 9601080 1200 1320 1500 1440 simulation time simulation time
- 32. 0 . 01 t 2 0 t 120 (min) G in l arg e ( t ) 0 . 01 (120 ( t 120 )) 2 120 t 240 (min) 15 10 Gin disturbance 5 0 0 500 1000 1500 t
- 33. 16.7494 19.1079 18 20 16 12.7602 14 11.9346 15 12 MAPE (%) MAPE(%) 10 8 10 6 5.4796 4.8881 4 5 2 0 0 PD-FLC PI-FLC PID-FLC MFB PD-FLC MFB PI-FLC MFB PID-FLC 713.703 893.1328 714 daily infused insulin (mU/kg/day) daily infused insulin (mU/kg/day) 900 712 710.0188 710 850 708 706 800 747.638 746.994 704 701.2719 750 702 700 700 698 696 650 694 PD-FLC PI-FLC PID-FLC MFB PD-FLC MFB PI-FLC MFB PID-FLC
- 34. 200 MFB PI-FLC 185 MFB PD-FLC MFB PID-FLC 170 reference model gluocse mg/dl 155 140 125 110 95 80 0 500 1000 1500 simulation time
- 35. 100 768 t 772 (min) G un exp ected (t ) 0 otherwise dG G in ( t ) G un exp ected ( t ) f 2 ( G ( t )) f 3 ( G ( t )) f 4 ( I ( t 3 )) f 5 ( I ( t 2 )) dt
- 36. 170 MFB PI-FLC MFB PD-FLC 155 MFB PID-FLC reference model gluocse mg/dl 140 125 110 95 700 720 740 760 780 800 820 simulation time
- 37. PID : high participant rate of the controller with the smallest MAPE. PI-FLC PID-FLC . MFB PD-FLC MFB PID- FLC . MFB PD-FLC and MFB PID-FLC better in 1) uncertainty test 2) changing glucose intake profile
- 38. PI-FLC & PID-FLC better in 1 ) corruption in the sensor output and 2) patient experiences an elevated glucose concentration at the beginning of the simulation. All the techniques retain glucose normal when a sudden glucose intake was assumed at a local peak glucose concentration.
- 39. The performance of the MFB PI-FLC was poor in about four of five robustness tests. This controller is unfit for model-flower applications since its response time is slow. The MFB PID-FLC, the MFB PD-FLC, the PI- FLC, and the PID-FLC are good candidates to be used as control systems to drive a miniaturized insulin pump and regulate glucose elevation in Type 1 diabetic patient with an acceptable error.