on-line hev energy management using a fuzzy logichev with pcube - gafl 35 cycles – 87.6296% hev...
TRANSCRIPT
Yacine Gaoua 1,2,3, Stéphane Caux 1, Pierre Lopez 2,3 and Josep Domingo Salvany 4
1. Institut National Polytechnique de Toulouse, INPT
2. Laboratoire PLAsma et Conversion d'Energie, LAPLACE
3. Laboratoire d'Analyse et d'Architecture des Systemes, LAAS-CNRS
4. Nexter Electronics, NE
On-Line HEV Energy Management Using a
Fuzzy Logic
Outline of the presentation
I. Introduction to HEV energy chain
II. Sources characteristics
III. Modeling
IV. Solving method
V. Off-line optimization
VI. Results and performance
VII. Conclusion
Hybrid Electrical Vehicle
Battery Super-capacitor Fuel cell
HEV energy chain
I. HEV energy chain
Parameter Meaning
𝑰𝒄𝒉 Demand of the powertrain (A)
𝑰𝒂𝒎𝒊𝒏,𝑰𝒂
𝒎𝒂𝒙 Min/Max current exiting the PCube converter (A)
𝑰𝒔𝒄𝒎𝒊𝒏,𝑰𝒔𝒄
𝒎𝒂𝒙 Min/Max current provided by the super-capacitor (A)
𝑼𝒔𝒄𝒎𝒊𝒏,𝑼𝒔𝒄
𝒎𝒂𝒙, 𝑼𝒔𝒄(0) Min/Max/Initial voltage of the super-capacitor (V)
𝑺𝑶𝑪𝒃𝒂𝒕𝒎𝒊𝒏,𝑺𝑶𝑪𝒃𝒂𝒕
𝒎𝒂𝒙, 𝑺𝑶𝑪𝒃𝒂𝒕(0) Min/Max/Initial energy level in the battery pack (%)
𝑪𝒂𝒑𝒃𝒂𝒕 Battery capacity (Ah)
𝜟𝒕 Time stepsize (s)
𝑹𝒔𝒄 Super-capacitor internal resistance (Ω)
𝑪𝒔𝒄 Super-capacitor capacity (F)
𝑬𝑳𝒐𝒔𝒔𝒃𝒂𝒕 Battery energy losses (kW)
E𝑳𝒐𝒔𝒔𝒄𝒗𝒔 Energy losses of the PCube converter (kW)
Input parameters.
Battery efficiency. Convertor efficiency.
II. Sources characteristics
• 𝑰𝒃𝒂𝒕𝑹: Real battery current
• 𝑰𝒃𝒂𝒕: battery current
• 𝑺𝑶𝑪𝒃𝒂𝒕: Battery State of charge
• 𝑼𝒃𝒂𝒕: Battery voltage
• 𝑰𝒔𝒄: Super-capacitor current
• 𝑼𝒔𝒄: Super-capacitor voltage
• 𝑰𝒂: Convertor current
(nlp)
𝐼𝑏𝑎𝑡 + 𝐼𝑎 = 𝐼𝑐ℎ 𝐼𝑐ℎ > 0
𝐼𝑐ℎ ≤ 𝐼𝑏𝑎𝑡 + 𝐼𝑎 ≤ 0 𝐼𝑐ℎ ≤ 0
𝐼𝑎𝑀𝑖𝑛 ≤ 𝐼𝑎≤ 𝐼𝑎
𝑀𝑎𝑥
𝐼𝑠𝑐𝑀𝑖𝑛 ≤ 𝐼𝑠𝑐 ≤ 𝐼𝑠𝑐
𝑀𝑎𝑥
𝑈𝑠𝑐𝑀𝑖𝑛 ≤ 𝑈𝑠𝑐≤ 𝑈𝑠𝑐
𝑀𝑎𝑥
𝑆𝑂𝐶𝑏𝑎𝑡𝑀𝑖𝑛 ≤ 𝑆𝑂𝐶𝑏𝑎𝑡 ≤ 𝑆𝑂𝐶𝑏𝑎𝑡
𝑀𝑎𝑥
𝑃𝑏𝑎𝑡𝑅 = 𝑃𝑏𝑎𝑡
+ 𝐸𝑙𝑜𝑠𝑠𝑏𝑎𝑡(𝑃𝑏𝑎𝑡 )
𝑃𝑠𝑐 = 𝑃𝑎
+ 𝐸𝑙𝑜𝑠𝑠𝑐𝑣𝑠 𝑃𝑎 + 𝑅𝑠𝑐
𝐼𝑠𝑐 2
𝑆𝑂𝐶𝑏𝑎𝑡 = 𝑆𝑂𝐶𝑏𝑎𝑡 0 +100.𝐸𝑏𝑎𝑡
𝐶𝑎𝑝𝑏𝑎𝑡∆𝑡
𝑈𝑠𝑐 = 𝑈𝑠𝑐 0 + 𝐼𝑠𝑐 + 𝑅𝑠𝑐 +∆𝑡
𝐶𝑠𝑐
𝑈𝑏𝑎𝑡 = 𝑓 𝑆𝑂𝐶𝑏𝑎𝑡 0
𝐸𝑏𝑎𝑡 = 𝑔 𝐼𝑏𝑎𝑡𝑅
Decision variables:
Mathematical modeling
Goal: Minimize battery discharge
Under constrains of (system functioning,
sources design, safety limitation),
𝒈: Computation of electrical quantity
𝒇: Computation of battery voltage
III. Modeling
IV. Solving method using fuzzy logic
Powertrain demand. Super-capacitor voltage. Battery current.
Rules engine.
𝒊𝒇 𝑰𝒄𝒉 = . 𝒂𝒏𝒅 𝑼𝒔𝒄 = . 𝒕𝒉𝒆𝒏 𝑰𝒃𝒂𝒕 = . 𝒐𝒓
Rules generation. Decision surface (centroid method).
Parameters setting: Genetic algorithm (off-line - GPS)
Co
ntro
l an
d c
orre
ctio
n a
lgo
rithm
V. Off-line optimization
Mission profile NE (176s).
(nlp)
𝑴𝒊𝒏 𝟏𝟎𝟎 − 𝑺𝑶𝑪𝒃𝒂𝒕 𝑻 = 𝑴𝒂𝒙 𝑺𝑶𝑪𝒃𝒂𝒕 𝑻
𝐼𝑏𝑎𝑡(𝑡) + 𝐼𝑎(𝑡) = 𝐼𝑐ℎ(𝑡) 𝐼𝑐ℎ(𝑡) > 0
𝐼𝑐ℎ ≤ 𝐼𝑏𝑎𝑡 + 𝐼𝑎 ≤ 0 𝐼𝑐ℎ(𝑡) ≤ 0
𝐼𝑎𝑀𝑖𝑛 ≤ 𝐼𝑎 (𝑡) ≤ 𝐼𝑎
𝑀𝑎𝑥
𝐼𝑠𝑐𝑀𝑖𝑛 ≤ 𝐼𝑠𝑐(𝑡) ≤ 𝐼𝑠𝑐
𝑀𝑎𝑥
𝑈𝑠𝑐𝑀𝑖𝑛 ≤ 𝑈𝑠𝑐 𝑡 ≤ 𝑈𝑠𝑐
𝑀𝑎𝑥
𝑆𝑂𝐶𝑏𝑎𝑡𝑀𝑖𝑛 ≤ 𝑆𝑂𝐶𝑏𝑎𝑡(𝑡) ≤ 𝑆𝑂𝐶𝑏𝑎𝑡
𝑀𝑎𝑥
𝑃𝑏𝑎𝑡𝑅 𝑡 = 𝑃𝑏𝑎𝑡
𝑡 + 𝐸𝑙𝑜𝑠𝑠𝑏𝑎𝑡 𝑃𝑏𝑎𝑡 𝑡
𝑃𝑠𝑐 (𝑡) = 𝑃𝑎
(𝑡) + 𝐸𝑙𝑜𝑠𝑠𝑐𝑣𝑠 𝑃𝑎 (𝑡) + 𝑅𝑠𝑐
𝐼𝑠𝑐(𝑡) 2
𝑆𝑂𝐶𝑏𝑎𝑡 𝑡 = 𝑆𝑂𝐶𝑏𝑎𝑡 𝑡 − 1 +100.𝐸𝑏𝑎𝑡 𝑡
𝐶𝑎𝑝𝑏𝑎𝑡∆𝑡
𝑈𝑠𝑐(𝑡) = 𝑈𝑠𝑐 𝑡 − 1 + 𝐼𝑠𝑐(𝑡) + 𝑅𝑠𝑐 +∆𝑡
𝐶𝑠𝑐
𝑈𝑏𝑎𝑡 = 𝑓 𝑆𝑂𝐶𝑏𝑎𝑡 𝑡 − 1
𝐸𝑏𝑎𝑡 = 𝑔 𝐼𝑏𝑎𝑡𝑅(𝑡)
Global optimization
Optimization using Operations
Research methods:
AMPL+ IpOpt algorithm (Interior
Points)
VI. Results and performance
HEV sources/
Method
Number of cycles / Battery discharge
HEV battery alone 30 Cycles – 88.3143%
HEV with PCube - FL 34 Cycles – 88.8872%
HEV with PCube - GAFL 35 Cycles – 87.6296%
HEV with PCube − IpOpt 39 Cycles – 88.7396%
HEV with PCube Battery discharge (1 cycle)
GAFL 2.546%
IpOpt 2.29315%
NE Mission profile 176s. Mission profile 3h 50min.
HEV sources/
Method
Number of cycles / Battery discharge
HEV battery alone 1 Cycle – 52.3566%
HEV with PCube - FL 2 Cycles – 85.7596%
HEV with PCube - GAFL 2 Cycles – 71.9029%
HEV with PCube − IpOpt 3 Cycles – 89.896%
HEV with PCube Battery discharge (1 cycle)
GAFL 36.1712%
IpOpt 30.49%
VII. Conclusions and perspectives
• Genetic algorithm improve the solution by setting FL parameters off-line,
• Good quality of the results (in regard to the global optimization),
• Development of decision support tool in C + + (implementation in a dsp target).
• Validation of results on a real prototype.
Conclusions:
Perspectives:
Thank you for your attention