2015 12-02-opti wind-test-setup-validation-estimation-techniques-drivetrains-ku-leuven

Test setup for virtual sensing on mechatronic drivetrains SBO-OptiWind open project meeting 2-12-2015 Bart Forrier bart.forrier@kuleuven.be

Overview

1. Goal 2. Approach 3. Numerical validation 4. Experimental validation – test setup 5. Conclusions

Goal: improve mechatronic drivetrains

Energies 2014, 7(4), 2595-2630; doi:10.3390/en7042595

AC machine nonlinear transmission Load

multi-physical, non-linear

dynamic input

Tload

Approach: model-based virtual sensing PH

YSIC

AL

VIR

TUAL

Sensor data

Output System info

Control input

Disturbance

Mechatronic drivetrain • Multi-physical • Nonlinear • Motor measurements Unscented Kalman Filter • Nonlinear • Unknown load • State/input estimation

Numerical validation

1D/3D

1D

Reference model

UKF model

PHYS

ICAL

VI

RTU

AL

Forrier, B., Naets, F., & Desmet, W. (2015). Virtual sensing on mechatronic drivetrains using multiphysical models. ECCOMAS Thematic Conference on Multibody Dynamics. Barcelona, Catalonia, Spain, 29 June - 2 July.

Numerical validation – results • Noise reduction

• Input estimation

Low SNR

Measured

High SNR

Reference

Estimate

Reference

Estimate

Motor speed

Load torque 1D vs 3D model mismatch

Measured acceleration

Nonlinearity

Forrier, B., Naets, F., & Desmet, W. (2015). Virtual sensing on mechatronic drivetrains using multiphysical models. ECCOMAS Thematic Conference on Multibody Dynamics. Barcelona, Catalonia, Spain, 29 June - 2 July.

v, i (a,b,c) ϑ,ω,α

ϑ,ω,α

T

Experimental setup • AC induction machines

o 5,5 kW – FOC o 18 Nm / 3000 rpm o FOC

• Nonlinear driveline

o Cardan o variable deflection angle o easy replacement

• Stiff frame

Validation of load torque estimation

Model-based state observer

Parameter identification

Validation

RTT

Kalman-based estimation

Conclusion Use operational information to improve mechatronic drivetrains Obtain information by model-based virtual sensing • Multiphysical 1D torsional model • Combined input/state estimation • Multiphysical measurements

Validation • Numerical – 1D/3D reference

o Dynamic load torque estimation • Experimental – test setup

o Design, construction, instrumentation o Driveline dynamics & frame stiffness o Dynamic load torque estimation

Thank you Questions?