0
TECHNICAL PAPERS

Development of a Friction Component Model for Automotive Powertrain System Analysis and Shift Controller Design based on Parallel-Modulated Neural Networks

[+] Author and Article Information
M. Cao1

 United Technologies Research Center 411 Silver Lane, MS 129-73 E. Hartford, CT 06108caom@utrc.utc.com

K. W. Wang2

Department of Mechanical and Nuclear Engineering,  The Pennsylvania State University, University Park, PA 16802kwwang@psu.edu

Y. Fujii, W. E. Tobler

Research and Advanced Engineering,  Ford Motor Company, Dearborn, MI 48121

1

Post Doctoral Fellow, Member of ASME.

2

William E. Diefenderfer Chaired Professor in Mechanical Engineering and Fellow of ASME.

J. Dyn. Sys., Meas., Control 127(3), 382-405 (Aug 17, 2004) (24 pages) doi:10.1115/1.1978909 History: Received January 27, 2003; Revised August 17, 2004

In this study, a new hybrid-neural-network-based friction component model is developed for powertrain (PT) dynamic analysis and controller design. This new model, with significantly improved input-output scalability over conventional neural network configuration, has the capability to serve as a forward as well as an inverse system model. The structural information of the available physical and empirical correlations is utilized to construct a parallel-modulated neural network (PMNN) architecture consisting of small parallel sub-networks reflecting specific mechanisms of the friction component engagement process. The PMNN friction component model isolates the contribution of engagement pressure on engagement torque while identifying the nonlinear characteristics of the pressure-torque correlation. Furthermore, it provides a simple torque formula that is scalable with respect to engagement pressure. The network is successfully trained, tested and analyzed, first using analytical data at the component level and then using experimental data measured in a transmission system. The PMNN friction component model, together with a comprehensive powertrain model, is implemented to simulate the shifting process of an automatic transmission (AT) system under various operating conditions. Simulation results demonstrate that the PMNN model can be effectively applied as a part of powertrain system model to accurately predict transmission shift dynamics. A pressure-profiling scheme using a quadratic polynomial pressure-torque relationship of the PMNN model is developed for transmission shift controller design. The results illustrate that the proposed pressure profiling technique can be applied to a wide range of operating conditions. This study demonstrates the potential of the PMNN architecture as a new dynamic system-modeling concept: It not only outperforms the conventional network modeling techniques in accuracy and numerical efficiency, but also provides a new tool for transmission controller design to improve shift quality.

Copyright © 2005 by American Society of Mechanical Engineers
Your Session has timed out. Please sign back in to continue.

References

Figures

Grahic Jump Location
Figure 1

Black-box network architecture (26). z−1=One step time delay.

Grahic Jump Location
Figure 2

Advanced hybrid neural network (AHNN) architecture (29)

Grahic Jump Location
Figure 3

Friction component model based on the parallel-modulated-ANN-architecture

Grahic Jump Location
Figure 4

Training error comparison between the conventional and parallel-modulated ANN model

Grahic Jump Location
Figure 5

Neural network training: PMNN output vs analytical result. -Analytical data, -PMNN output. Vertical axis-Friction component torque (N.m). Horizontal axis-time (s) pf=2.758×105N∕m2(40psi)

Grahic Jump Location
Figure 6

Neural network testing: PMNN output vs analytical result. —Analytical data—PMNN output. Vertical axis—Friction component torque (N.m). Horizontal axis—Time (s) pf=4.482×105N∕m2(65psi)

Grahic Jump Location
Figure 7

PMNN model: Slip effect Vertical axis—Friction component torque (N.m). Horizontal axis—time (s)—Total friction component torque,— Output of sub-NN1  ⋆α1⋆ slip, ⋯Output of sub-NN2  ⋆α2⋆papp,–Output of sub-NN3  ⋆α3⋆papp2, pf=2.758×105N∕m2(40psi), Toil0=350K

Grahic Jump Location
Figure 8

PMNN model: Temperature effect. Vertical axis—Friction component torque (N.m). Horizontal axis—time (s)—Total friction component torque,—.Output of sub-NN1  ⋆α1⋆ Slip,⋯Output of sub-NN2  ⋆α2⋆papp,–Output of sub-NN3  ⋆α3⋆papp2, pf=3.793×105N∕m2(55psi), ω0=2300rpm

Grahic Jump Location
Figure 9

PMNN model: Pressure effect. Vertical axis—Friction component torque (N.m). Horizontal axis—time (s)—Total friction component torque,—.Output of sub-NN1  ⋆α1⋆ slip,⋯Output of sub-NN2  ⋆α2⋆papp,–Output of sub-NN3  ⋆α3⋆papp2, Toil0=380K, ω0=2100rpm

Grahic Jump Location
Figure 10

Pressure scalability: PMNN vs conventional ANN Vertical axis—friction component torque (N.m). Horizontal axis—time (s)—Analytical data,—.— Output of parallel-modulated ANN,⋯Output of conventional ANN, pf=45psi

Grahic Jump Location
Figure 11

Pressure scalability: PMNN vs conventional ANN. Vertical axis—Friction component torque (N.m). Horizontal axis—time (s)—Analytical data,—.—Output of parallel-modulated ANN,⋯Output of conventional ANN, pf=75psi

Grahic Jump Location
Figure 12

Dynamometer test stand setup (30)

Grahic Jump Location
Figure 13

Parallel-modulated-neural-network-based Model: Neural network training based on the DYNO data — Backtracked friction component torque (N.m), ∙∙∙∙∙∙∙ Friction component torque estimated by parallel-modulated ANN (N.m), —- Slip speed (rad/s), —. Applied pressure (psi)

Grahic Jump Location
Figure 14

PT (Powertrain) simulation with PMNN friction component model

Grahic Jump Location
Figure 15

(a, b) Dynamic PT simulation based on the PMNN friction component model: Low oil temperature and small throttle opening: Toil0=313.6K, θth=12.3deg

Grahic Jump Location
Figure 16

(a, b) Dynamic PT simulation based on the PMNN friction component model: High oil temperature and small throttle opening: Toil0=368.9K, θth=14.6deg

Grahic Jump Location
Figure 17

(a, b) Dynamic PT simulation based on the PMNN friction component model: High oil temperature and large throttle opening: Toil0=370K, θth=26.8deg

Grahic Jump Location
Figure 18

(a, b) Dynamic PT simulation based on the PMNN friction component model: Low oil temperature and large throttle opening: Toil0=314.8K, θth=25.4deg

Grahic Jump Location
Figure 19

Parallel-modulated-neural-network-model-based pressure profiling scheme

Grahic Jump Location
Figure 20

(a, b) Pressure profiling based on the PMNN friction component model: Toil0=313.6K, θth=12.3deg

Grahic Jump Location
Figure 21

Pressure profiling based on the PMNN friction component model-Toil0=370K, θth=26.8deg, ----∙ Network-model-based prediction with unprofiled pressure, —— Network-model-based prediction with profiled pressure

Grahic Jump Location
Figure 22

Pressure profiling based on the PMNN friction component model-Toil0=370K, θth=26.8deg, ----∙ Network-model-based prediction with unprofiled pressure, —— Network-model-based prediction with profiled pressure

Tables

Errata

Discussions

Some tools below are only available to our subscribers or users with an online account.

Related Content

Customize your page view by dragging and repositioning the boxes below.

Topic Collections

Sorry! You do not have access to this content. For assistance or to subscribe, please contact us:

  • TELEPHONE: 1-800-843-2763 (Toll-free in the USA)
  • EMAIL: asmedigitalcollection@asme.org
Sign In