Optimizing Performance and Fuel Economy of a Dual-Clutch Transmission Powertrain with Model-Based Design

Transcription

1 Optimizing Performance and Fuel Economy of a Dual-Clutch Transmission Powertrain with Model-Based Design Vijayalayan R, Senior Team Lead, Control Design Application Engineering, MathWorks India Pvt Ltd Pete Maloney, Senior Principal Technical Consultant, MathWorks Inc. Wit Nursilo, Senior Application Engineer, MathWorks Inc The MathWorks, Inc. 1

2 Problem Statement Determine Numerically Optimal Transmission Shift-Schedule Calibration and Axle Ratio For Dual-Clutch Powertrain Design Concept With Accurate Engine Fuel Consumption Model 2

3 Challenges for the Powertrain Engineer Building a System Level Simulation Model of Vehicle Designing and Verifying the Controllers along with the Vehicle Model Optimizing the System Parameters Speeding up the Optimization Process 3

4 DCT Powertrain Axle Sweep Results 4

5 Agenda System Level Simulation An Overview Modeling System Level Simulation of Vehicle Estimation of Fuel Economy from Measured Data Shift Schedule Optimization Summary/Q&A 5

6 System Level Simulation What is it? Use simulation for performance and cost optimization at the system (vehicle) level Establish vehicle level requirements and a system architecture for detail design of subsystems and components 6

7 Design Cycle Time Difficult to Predict System Level Simulation Why do it? Traditional vehicle design process System level sim enabled design process Specify Vehicle Design Specify vehicle design Build Test Vehicle Test Components Many Iterations = Many Costly Prototype Builds Modify Test Vehicle Test Vehicle Performance Evaluate Results Reduce prototype builds & dev cost Optimal vehicle design Improve predictability of design cycle time Build System Level Simulation Build Test Vehicle Validate System Level Simulation Optimize Design through Simulation Modify Test Vehicle Opportunity for further optimization Finalize Vehicle Design Finalize Vehicle Design 7

8 Agenda System Level Simulation An Overview Modeling System Level Simulation of Vehicle Estimation of Fuel Economy from Measured Data Shift Schedule Optimization Summary/Q&A 8

9 Model Development Engine Engine Model Engine Calibration Transmission Dual Clutch Transmission Auto-Driver ( Forward Model ) Vehicle Control Engine Control Transmission Shift Schedule PI Control 9

10 Dual Clutch Transmission Model 6 speed Dual-Clutch Transmission Vehicle Dynamics 10

11 Statistical Modeling Engine and Calibrations Statistical methods produce accurate engine model with optimized calibrations Design experiments and collect data Generate engine models using statistical methods Generate optimized calibrations using analytical methods Measured Data Engine Calibrations Simulink Model Statistical Models Design of Experiments Physical Testing High Fidelity Simulation Data Modeling Calibration Generation Results Engine Model Calibrations 11

12 Control Calibration is Included in the Model - Generate Optimal Engine Calibrations From Engine Model With Numerical Optimization - 12

13 Tuning Abstracted Models to Match Simulation Results Model: Control Detailed Abstracted Problem: Simulation results of detailed and abstracted model do not match Solution: Use Simulink Design Optimization to tune abstracted model parameters >> Shift_Gear_Delay >> Ratio_Time_Const 13

14 Agenda System Level Simulation An Overview Modeling System Level Simulation of Vehicle Estimation of Fuel Economy from Measured Data Shift Schedule Optimization Summary/Q&A 14

15 Use Measured Data To Estimate Fuel Economy Model: Control Problem: Use simulation to calculate a realistic estimate of fuel economy Solution: Use Curve Fitting Technique to import fuel economy data and generate a lookup table 15

16 Agenda System Level Simulation An Overview Modeling System Level Simulation of Vehicle Estimation of Fuel Economy from Measured Data Shift Schedule and Drive Axle Ratio Optimization Summary/Q&A 16

17 Optimize Gear Shift Schedule & Drive Axle Ratio To Minimize Fuel Use Up Shift Final Drive Axle Gear Ratio (sweep) Gear Shift Schedules (32 params) Fuel Consumed On FTP75 Drive-Cycle 0-100KPh Acceleration Time Down Shift Global Optimization Toolbox patternsearch Trade Off Acceleration Performance vs. Fuel Consumption Cost By Optimizing Gear Shift Schedules as Final Drive Axle Ratio is Swept 17

18 Global Optimization Solver : Pattern Search Use Pattern Search Optimization for Robustness to Local Minima Search Systematically Steps Through The Search Parameter Space 18

19 Optimization Process Initialize Optimization Parameters Generate 64 Shift Parameter Variations With Pattern Search Run 64 FTP75 Drive-Cycle Simulations No Size Pattern Search Mesh Smaller Than Tolerance? Yes Report Results (~15,400 FTP cycle simulations) 19

20 Speeding Up Optimization Process Use Rapid Accelerator for Stand Alone Executables for Parallel Computing. Use Parallel Computing To Make Execution Speed Scalable and Controllable Multi Core Parameter Set 1 Parameter Set 2 Parameter Set N for i = 1:numSims out{i} = sim(mdl, SimSettings{i}); end parfor i = 1:numSims out{i} = sim(mdl, SimSettings{i}); end Use Distributed Computing Server to execute optimization on a computer cluster 20

21 Use Nonlinear Numerical Optimization Tools On The Model Run Axle Sweep To Find Performance vs. Fuel Economy Tradeoff Best Axle Ratio: 3.0 at Performance Time <10s Pattern Search Optimizer Re-Optimizes Performance and Normal Shift Schedules at Each Axle Ratio Setting (This Process Takes ~33 Hours for 7 Axle Settings on 64 workers). 21

22 Optimal Shift Schedule Surfaces Optimized Baseline Axle Ratio Shift Schedule Base Opt 3.8 Opt 3.0 Axle Ratio MPG % 12.5% Performance Time 0~100KPH[s]

23 Agenda System Level Simulation An Overview Modeling System Level Simulation of Vehicle Estimation of Fuel Economy from Measured Data Shift Schedule Optimization Summary/Q&A 23

24 Challenges for the Powertrain Engineer Building a System Level Simulation Model of Vehicle Designing and Verifying the Controllers along with the Vehicle Model Optimizing the System Parameters Speeding up the Optimization Process 24

25 Key Message System level simulation addresses the challenges involved in the design and optimization Control Algorithm Transmission and Vehicle Calculate Fuel Use Parameter Values Optimization Algorithm 25

26 Thank You 26