IEEE TRANSACTIONS ON CONTROL SYSTEMS TECHNOLOGY, VOL. 20, NO. 5, SEPTEMBER 2012 1311
Brief Papers
Regularized - and DHOBE: An Adaptive Feedforward
for a Solenoid Valve
Jean-Philippe Gauthier and Philippe Micheau
Abstract—To allow a stable and fast acting hydraulic pressure
control on a continuously variable transmission (CVT) for road
vehicles, an adaptive feedforward strategy is used. The Das-
gupta-Huang outer bounding ellipsoid (DHOBE) and recursive
least squares (RLS) with exponential forgetting factor -
adaptation algorithms are compared to the non-adaptive feedfor-
ward. The experiments show a clear advantage for the adaptive
over the non-adaptive version by compensating for the slow drift
of the valve pressure gain during the warm-up period of the
transmission. Because of highly correlated input data, the adap-
tation algorithms offer deceiving performances with oscillating
identified parameters. A regularization procedure is added to both
adaptation algorithms, giving the - and . The
regularized algorithms offer significantly better performances
and stability than their non-regularized counterparts. Because
of its implicit parametric uncertainty calculation while keeping
an equivalent convergence rate, and a lower number of updates,
the rDHOBE algorithm is regarded as the best solution for the
application. By adapting a simple linear model, the rDHOBE
adaptive feedforward succeeds in responding to an abrupt change
of the external pressure setpoint with no added actuation delay
while keeping the pressure error under 0.5 bar.
Index Terms—Adaptive, bounding ellipsoids, control, continu-
ously variable transmission (CVT), hydraulics, least squares, open-
loop, transmission.
I. INTRODUCTION
I
N VEHICLE control systems, the actuation forces are often
created by hydraulic pressure. The hydraulic technology,
for its compactness, low cost, high power density, self lubri-
cation, and quick response time is particularly well suited for
vehicle control applications [1]. Despite the trend to include a
growing number of electric actuators, hydraulic systems are still
needed in numerous applications where higher force and torque
are needed [2]. A way to get the flexibility of electrical control
while keeping the high power density of hydraulics is to use
electro-hydraulic valves.
In the particular case of a steel belt continuously variable
transmission (CVT) transmission, the required extreme thrust
forces make it a good candidate for an electro-hydraulic actua-
tion system. The electro-hydraulic valves used are pressure reg-
Manuscript received December 02, 2010; accepted May 18, 2011. Manuscript
received in final form June 20, 2011. Date of publication July 25, 2011; date of
current version June 28, 2012. Recommended by Associate Editor M. Guay.
The authors are with the Department of Mechanical Engineering, Uni-
versité de Sherbrooke, Sherbrooke, QC J1K 2R1, Canada (e-mail: jean-
philippeg@gmail.com).
Digital Object Identifier 10.1109/TCST.2011.2160641
ulators that convert an electrical input signal into oil pressure.
According to various factors like the oil and solenoid tempera-
tures, the equation describing the valve pressure versus its input
signal changes significantly. To provide good shifting perfor-
mances, the valve has to provide quick pressure changes with
minimal pressure error.
Numerous authors have proposed different techniques to con-
trol electro-hydraulic valves. Aside from the classic closed-loop
proportional-integral derivative (PID) control, most of the more
advanced control techniques have been developed for motion
control hydraulic systems. The adaptive closed-loop control,
which determines the closed-loop controller gain in function of
the valve characteristics, has been proposed [3]. For flow con-
trol valves with a dead zone, adaptive fuzzy compensation has
been investigated and proved effective [4]. Again for flow con-
trol valves and motion controlled piston type actuators, robust
adaptive control has been tested with satisfactory results [5].
For pressure control systems, open-loop and linear closed-
loop control are mostly used. While the open-loop control cre-
ates unacceptable pressure errors, the closed-loop control brings
additional dynamics to the valve itself. To minimize the added
dynamics, high controller gains have to be used, lessoning the
robustness of the system [6]. For example, the experimental re-
sults from Amirante [7] show that a closed-loop controlled valve
brings much more pressure oscillations than the same valve
controlled in open loop. An adaptive open-loop controller is
proposed here as an alternative for the control. The proposed
controller works the same way as a basic open-loop controller
would, but an adaptive inverse model of the steady state valve
characteristics is used. Since the adaptation loop does not gen-
erate the command itself, filtering of the feedback signal has no
effect on the system dynamics. The problem is to design and
implement an efficient online identification scheme to track the
slow variation of the parameters.
The inverse model adaptation is made using one of two adap-
tation algorithms: The recursive least squares with exponential
forgetting - and the Dasgupta Huang Outer Bounding
Ellipsoids (DHOBE). - has been used in the past for model
reference adaptive control (MRAC) of hydraulic systems and
is considered the state of the art for hydraulic adaptive control
[8]. It is included in this study to provide a comparison basis
to evaluate the DHOBE adaptation scheme. DHOBE is a refine-
ment of the - algorithm [9] which brings the advantages
of implicitly calculating the uncertainties of the adapted param-
eters and accepting a bounded signal noise regardless of its fre-
quency spectrum. DHOBE has been used successfully in var-
1063-6536/$26.00 © 2011 IEEE