Optimal Fuzzy Logic Energy Management Strategy of Hybrid Electric Locomotives

Transcription

1 Optimal Fuzzy Logic Energy Management Strategy of Hybrid Electric Locomotives J. Baert*, S. Jemei*, D. Chamagne*, D. Hissel*, D. Hegy** and S. Hibon** * ** University of Franche-Comte, FEMTO-ST (Energy Department), UMR CNRS 6174, Belfort, France. Alstom Transport, 3 Avenue des Trois Chênes, Belfort, France. jerome.baert@univ-fcomte.fr - samuel.hibon@transport.alstom.com 1

2 Summary 1. Introduction 2. Modeling of the Hybrid Electric Locomotive 3. Optimal fuzzy logic Energy Management Strategy 4. Conclusion and outlooks 2

3 Introduc1on Partners of the project Pr Didier Chamagne Pr Daniel Hissel Dr Samir Jemeï Dominique Hegy Samuel Hibon 3

4 Introduc1on Partners of the project Context and problematic 4

5 Introduc1on Adopted solution 60% less particles 40% less NO 15 db less noise 15% less maintenance 5

6 Summary 1. Introduction 2. Modeling of the Hybrid Electric Locomotive 3. Optimal fuzzy logic Energy Management Strategy 4. Conclusion and outlooks 6

7 Modeling of the Hybrid Electric Locomo1ve Energetic Macroscopic Representation approach Advantages: Physical causality Highlight measures and sensors Control structure identification Implementation under Matlab / Simulink 7

8 Modeling of the Hybrid Electric Locomo1ve Global structure [1] (1) Diesel driven generator set (2) Batteries pack (5) (1) (4) (3) Ultra-capacitors pack (4) Rheostat (2) (3) (5) Bus capacity (6) Energy Management Strategy [1] (6) J. Baert, S. Jemei, D. Chamagne, D. Hissel, S. Hibon, and D. Hegy, Practical Control Structure and Simulation of a Hybrid Electric Locomotive IEEE Vehicle Power and Propulsion Conference, VPPC 12. 8

9 Summary 1. Introduction 2. Modeling of the Hybrid Electric Locomotive 3. Optimal fuzzy logic Energy Management Strategy a. Structure of the EMS b. Fuzzy Logic Controller design c. Genetic algorithm 4. Conclusion and outlooks 9

10 Op1mal fuzzy logic Energy Management Strategy Structure of the EMS Goal: To share the power required by the driving cycle performed by the locomotive between the different on-board sources, taking into account their own specifications. 10

11 Op1mal fuzzy logic Energy Management Strategy Structure of the EMS Ultra-capacitors: Limitation of the State Of Charge (SOC) between 50% and 100%, control of the SOC according to the speed of the vehicle, supply the high frequencies of the power mission, α SOC uc ={ min (1, max (0, SOC uc max SOC uc ref = SOC uc max v veh SOC uc est / v max ( SOC uc max / SOC uc max SOC uc high ) ) if uc min ) P uc ref >0@ min (1, max (0, SOC uc est SOC uc min / SOC uc low SOC uc min ) ) if P uc ref <0 11

12 Op1mal fuzzy logic Energy Management Strategy Structure of the EMS Batteries: Limitation of the SOC between 70% and 90%, control of the SOC according to the acceleration of the vehicle, supply the low frequencies of the power mission with the diesel driven generator set. SOC batt ={ min (1, max (0, SOC batt max SOC α *+, -.// 012 = *+, -.// 3.4 1/ /7/ ( 5 51ℎ /) S) ( P batt est / SOC batt max ( OC* +, batt-.// 3.4 high ) ) if *+, -.// 39: ) batt ref >0@ min (1, max (0, SOC batt est SOC batt min / SOC batt low SOC batt min ) ) if P batt ref <0 12

13 Op1mal fuzzy logic Energy Management Strategy Structure of the EMS Diesel driven generator set: Use of a Fuzzy Logic Controller to determine the power delivered by this source, supply the low frequencies of the power mission with the batteries. IF AND THEN is N is P is P 13

14 Summary 1. Introduction 2. Modeling of the Hybrid Electric Locomotive 3. Optimal fuzzy logic Energy Management Strategy a. Structure of the EMS b. Fuzzy Logic Controller design c. Genetic algorithm 4. Conclusion and outlooks 14

15 Op1mal fuzzy logic Energy Management Strategy Fuzzy Logic Controller design Primary Membership Func1on Secondary Membership Func1on Improve uncertain1es modeling Type- 1 Interval Type- 2 15

16 Op1mal fuzzy logic Energy Management Strategy Fuzzy Logic Controller design Fuzzy Logic Controller of the implemented EMS 7 linguistic variables defined by trapezoid, triangular and Interval membership functions: NH, NM, NL, Z, PL, PM, PH N - Negative P - Positive H - High M - Medium Z - Zero L - Low IF AND THEN is NH is PH is PH => 2 inputs 7 linguistic variables + output 7 linguistic variables = 21 membership functions 16

17 Op1mal fuzzy logic Energy Management Strategy Fuzzy Logic Controller design Fuzzy Logic Controller of the implemented EMS How to define the parameters of the 21 membership functions??? 17

18 Op1mal fuzzy logic Energy Management Strategy Fuzzy Logic Controller design Fuzzy Logic Controller of the implemented EMS How to define the parameters of the 21 membership functions or the parameters??? Realization of a survey based on human behavior and knowledge, optimization of the parameters of the controller using a genetic algorithm in order to minimize the fuel consumption of the diesel driven generator set. PhD defended in February 2012 Energy management of a hybrid electric vehicle: an approach based Computer Science and Electronic Engineering on type-2 fuzzy logic department of the University of Essex by Dr Solano Martinez Colchester - United Kingdom From April to July 2012 [2] Javier Solano Martínez, Robert I. John, Daniel Hissel, Marie-Cécile Péra, A survey-based type-2 fuzzy logic system for energy management in hybrid electrical vehicles, Information Sciences, Volume 190, 1 May 2012, Pages , ISSN , /j.ins

19 Summary 1. Introduction 2. Modeling of the Hybrid Electric Locomotive 3. Optimal fuzzy logic Energy Management Strategy a. Structure of the EMS b. Fuzzy Logic Controller design c. Genetic algorithm 4. Conclusion and outlooks 19

20 Op1mal fuzzy logic Energy Management Strategy Genetic algorithm Initialize population Evaluate fitness Satisfy constraints? yes Selection no New individual Crossover Mutation 20

21 Op1mal fuzzy logic Energy Management Strategy Genetic algorithm - IT2 Fuzzy Logic Initialize population Evaluate fitness Satisfy constraints? yes no New individual Selection Crossover Mutation child1 = p parent1 + ( 1 p) parent 2 tmission tmission tmission = if mutation_probability 0.01 fitness + + => max The gene P battvalue ( t) dt of the max considered Puc( t) chromosome dt min is P child 20 = p parent 2 + ( 10 p) parent 1 0 ge ( t) dt The batteries must be fully charged (90%) at the end of the driving cycle randomly modified 21

22 Op1mal fuzzy logic Energy Management Strategy Results Fuzzy maps 22

23 Op1mal fuzzy logic Energy Management Strategy Results Powers distribution 23

24 Op1mal fuzzy logic Energy Management Strategy Results Powers distribution (zoom) High frequencies of the power are dedicated to the UCs Low frequencies of the power are dedicated to the batteries 24

25 Op1mal fuzzy logic Energy Management Strategy Results States Of Charge and speed UCs SOC is controlled using the speed of the locomotive and is limited between 50% and 100% Batteries SOC is controlled using the acceleration of the locomotive and is limited between 70% and 90% 25

26 Op1mal fuzzy logic Energy Management Strategy Results Bus voltage control The implemented fuzzy EMS ensures the stability of the bus voltage 26

27 Summary 1. Introduction 2. Modeling of the Hybrid Electric Locomotive 3. Optimal fuzzy logic Energy Management Strategy 4. Conclusion and outlooks 27

28 Conclusion and outlooks Conclusion Development of the on-board sources dynamical models with their control Development of an intelligent Energy Management Strategy: No prior knowledge of the driving cycle Fuzzy Logic management of the diesel driven generator set Minimization of the use of the internal combustion engine Outlooks Improvement of the Type-2 membership functions definition Improvement of the GA taking into account batteries ageing for instance 28

29 Thanks for your attention 29