2020 7th International Conference on Energy Efficiency and Agricultural Engineering (EE&AE)
Model Reference Adaptive Control for Steering
Application of Agriculture Machines
Alexander Mitov
Dept. of Hydroaerodynamics and Hydraulic Machines
Tehcnical University of Sofia
Sofia, Bulgaria
a_mitov@tu-sofia.bg
Tsonyo Slavov
Dept. of Systems and Control
Technical University of Sofia
Sofia, Bulgaria
ts_slavov@tu-sofia.bg
Abstract— Modern agricultural machines happens to work
for prolonged periods of time in a considerably harsh
environments. This puts a higher demand on the automatic
control systems and steering control in particular. Long-term
machine operation leads to variance of the physical parameters
of the working fluid, positive overlap in proportional valve,
working temperature and machine components. This problem
can be addressed by introduction of adaptive controllers. The
present research is focused on the design and application of
model reference adaptive controller (MRAC) for control of
steering angle of a heavy duty agriculture mobile machine
through a hydraulic cylinder. The synthesized controller is
based on simple integrator model of the steering cylinder and
Lyapunov stability theorems. The closed-loop system achieves
good performance and the adaptive gain is stabilized around its
mean value too.
Keywords—model reference, adaptive control,
steering, agricultural
I. INTRODUCTION
The modern agriculture machines are constructed to
travel across and execute specialized mode of operations in a
various terrains. Auto-guided steering function is a general
function which is required of these mobile machines. To
ensure a desired trajectory tracking for agriculture machines,
the advanced control system should be design. The main
component in this system is an embedded controller.
The embedded controller is a digital device which
implements control algorithm. This algorithm uses the
reference signal and feedback signals to evaluate an electrical
control signal which drives the electrohydraulic steering unit
(EHSU). This unit established or tune the status of the
steering system to reach the desired mode of operation. As
it’s known conventional control algorithms operate by
varying the steering unit control signal according to the
reference and output signals. There are two general control
strategies: feedforward and feedback control. The aim of the
both strategies is to design a robust steering controller, which
ensures control system performance. The embedded
controller generates an suitable steering commands to
navigate the mobile machine under a various surfaces of the
road [1].
978-1-7281-0362-4/20/$31.00 ©2020 IEEE
Authorized licensed use limited to: University of the Phillippines Diliman. Downloaded on April 05,2022 at 08:59:37 UTC from IEEE Xplore. Restrictions apply.
Jordan Kralev, Member IEEE
Dept. of Systems and Control
Technical University of Sofia
Sofia, Bulgaria
jkralev@ieee.org
Ilcho Angelov
Dept. of Hydroaerodynamics and Hydraulic Machines
Technical University of Sofia
Sofia, Bulgaria
ilangel@tu-sofia.bg
Some obstructions should be overcome in order to
incorporate operator experience into controller design
procedure. The first obstruction is to ensures an appropriate
design procedure by integrating control dynamics and
operator experience. The second is the need of a sufficiently
good multivariable real-time model that describes the desired
performance for various operational conditions. The last
obstruction is the need of controller adaptation which means
that the controller parameter should be tuned automatically
according to the current conditions.
The main purpose when development an adaptive control
algorithm is to find an effective approach for tuning the
embedded controller parameters according to plant dynamics
variations, which is due to disturbances and/or unmodelled
dynamics and changes in mode of operation [2]. The
effectiveness of the adaptive control scenario can be
MRAC,
estimated by enhancements in performance EHSU. For
example such improvement can be achieved through
reducing the dead-band of the system or enhancing command
modulation [3]. In the EHSU, the dead-band depend of the
positive overlap of proportional spool valve for control of
steering cylinder piston position.
This paper present design and implementation of model
reference adaptive control (MRAC) for electrohydraulic
power steering system that is applicable in modern
agriculture machine. The goal of control algorithm is to
overcome a number of limitations in the control of hydraulic
systems in mobile machines - variance of the physical
parameters of the working fluid, positive overlap in
proportional valve, working temperature and other.
II. ELECTROHYDRAULIC STEERING SYSTEM
The embedded MRAC controller is intended for a
electrohydraulic power steering system which is based on
EHSU type OSPEC200 LSRM. The laboratory test rig is
design and implemented according to technical data sheet
from the mobile machine manufacturers [2]. The hydraulic
diagrams of the test rig system is described particularly in [3].
Fig.1 presents principle hydraulic scheme of the digital
system for control of steering unit drives by digital valve
actuator. The steering operations are done by three main
constructive elements – an electrohydraulic control module
1 12-14 November 2020, Ruse, Bulgaria
 2020 7th International Conference on Energy Efficiency and Agricultural Engineering (EE&AE)

(type PVE), a pilot operated proportional valve and an where ( ) is the desired steering angle and ( ) is the
executive hydraulic cylinder. When control action ( ) > 0, adaptive gain. The closed loop system then is described by the
the left digital switching valve drive the spool of the valve to following expression
the right direction. When ( ) < 0 valves in the right are
switched by PWM signal to drive the spool to the left ( )= ( ) ( )− ( ) . (3)
direction. The proportional valve determinate a flow rate
direction of hydraulic fluid which supplies the both steering
cylinder chambers. Steering cylinder controller ( )
measures steering piston position and calculates the control
signal ( ) in order to achieve desired reference ( ). The
reference signal can be generated by steering wheel, joystick
or GPS based system.

Fig. 2. Closed-loop system structure for MRAC controller design.

The adaptation of the parameter ( ) will be according to

the output reference model
( )=− ( )+ ( ). (4)

This represents an aperiodic model with a time-constant

and unit gain. So after as much as 5 the output
( ) → ( ). The goal is to select the parameter ( ) such
that reference model error ( ) = ( ) − ( ) to be
Fig. 1. Digitally controled power steering system. minimized. In order to investigate the convergence of the
The main components of embedded control system are ( ) we find its first derivative
joystick JS6000, microcontroller MC012-022 [2], EHSU ( )= ( )− ( ). (5)
OSPEC200LSRM with electrohydraulic convertor PVE
module, steering cylinder, linear position sensor (Fig.1). The After substituting the corresponding terms with the
communication between control system components is done expressions define above we get
by CAN network. We use PLUS+1 Guide software
environment to develop software, to load control algorithm ( )= ( ) ( )− ( ) ( )+ ( )−
and to process output data. MC012-022 controller being CE ( ). (6)
compliant is one of the appropriate devices for use in
distributed machine control system. The microcontroller After grouping the term according to their places in the
architecture is based on Digital Signal Processor (DSP) control loop the derivative of the model reference error
TMS320F2812 from Texas Instruments (TI) providing becomes
150MHz processing speed, 128K internal flash, multifunction
input and 32-bit fixed-point arithmetic. Its output ports can
generate PWM signals and input ports can measure voltage, ( )=− ( )+ − ( ) ( )− ( ) .
frequency and resistance. (7)
III. MODEL REFERENCE ADAPTIVE CONTROLLER DESIGN From this form it is evident that we can select the
The design of a MRAC controller is done based on following Lyapunov function
development closed-loop system scheme presented in Figure
2. ( ), ( ) = ( )+ − ( ) . (8)
Let assume that the steering cylinder and the hydraulic
systems are modeled by a simple integrator equation as To achieve stability of the closed-loop according to the
second Lyapunov method the total derivative of the Lyapunov
( )= ( ), (1) function with respect to time must be non-increasing
where ( ) is the steering cylinder position, ( ) is the ( ), ( )
voltage PWM signal applied to the PVE transducer and is < 0. (9)
an unknown system gain, possibly slowly varying with time
Hence by differentiating the expression for we get
due to physical parameter drift. The proposed controller we
select as an adaptive P-based output feedback as
=2 −2 − ( ) ( ). (10)
( )= ( ) ( )− ( ) , (2)

2 12-14 November 2020, Ruse, Bulgaria

Authorized licensed use limited to: University of the Phillippines Diliman. Downloaded on April 05,2022 at 08:59:37 UTC from IEEE Xplore. Restrictions apply.
 2020 7th International Conference on Energy Efficiency and Agricultural Engineering (EE&AE)

Now substituting the expression for the derivative of the

model error

=2 − ( )+ − ( ) ( )−

Cylinder Piston Stroke (mm)

( ) −2 − ( ) ( ). (11)

By grouping the terms the resultant expression becomes

=− ( )+ − ( ) 2 ( )−2 ( ) , (12)
where ( ) = ( ) − ( ) is the feedback error. Now the
sufficient condition for the stability condition ( ) < 0 to be
satisfied is
2 ( )−2 ( )=0 (13)
or
( )= ( ) ( )− ( ) , (14)
where = is a tunable parameter. Fig. 3. Measured cylinder piston stroke.

IV. EXPERIMENTAL RESULTS

The controller is implemented using Simulink® diagram
which is used for automatic code generation with PLC Coder
[4]. The generated ST code is embedded in Danfoss MC012-
Control Input to PVE (V)

022 microcontroller which used for experimental evaluation

of the proposed adaptive algorithm on a specialized test-
bench. Figure 3 present the experimentally measured
transient response of the steering cylinder piston in the
closed-loop system with embedded MRAC algorithm. The
reference signal is a periodic pulse signal with period of 100
seconds. The amplitude of the reference signal corresponds
to 0.1 m movement of the steering cylinder piston in each
direction from middle position presented with 0. The adaptive
algorithm is implemented in discrete time where the
integrator for is implemented as a discrete-time intregrator
with sample time = 0.025 . Also the aperiodic reference
model is approximated with a discrete-time transfer function
( ) with the same sample time. The desired performance
is selected as aperiodic first order system without overshot Fig. 4. Measured control signal.
.
( )= . (15) The Figure 7 shows experimentally measured spool
. positon of the proportional valve during the step response
The output of the cylinder reaches the desired value after reference tracking. This signal is obtained through LVDT
several iterations during which the feedback gain is feedback sensor integrated in EHSU. As can be seen from
adapted to its steady state value. The corresponding control figure there is a signification correlation between from
action is presented on the Figure 3. The control signal stays dynamic behaviour of control action and spool position
within its operational bounds of ±5 . Also the full range of signal. This is an indicator of the correct performance of the
control values is exploited in the control. hydraulic subsystem as a reaction from control signal
generated by the developed MRAC algorithm. The Figure 8
The feedback gain parameter is presented on the Figure 5 shows the experimentally record of the flow rate passed
where the adaptation process can be observed. The tunable through EHSU and supplying the cylinder chambers during
parameter reaches its steady state value for around 400 the reference tracking. The flow rate is delivered from
seconds. The oscillations of the tunable parameter are priority valve to EHSU in steering test rig. It is also an
induced by the reference change but are small enough indicator to energy efficiency analysis [5] of the system
compared to the steady-value of the parameter. The Figure 6 because the flow rate is approximately 12 l/min in compare
represents the model error used in the adaptation algorithm. with the one of the pump - approximately 30 l/min.

3 12-14 November 2020, Ruse, Bulgaria

Authorized licensed use limited to: University of the Phillippines Diliman. Downloaded on April 05,2022 at 08:59:37 UTC from IEEE Xplore. Restrictions apply.
 2020 7th International Conference on Energy Efficiency and Agricultural Engineering (EE&AE)
Gain (mm*10/mV)
Fig. 5. Proportional gain variation.
Error
Fig. 6. Proportional gain variation.
V. CONCLUSION
The article presents the design and implementation of
MRAC algorithm for electrohydraulic power steering system,
which is intended for installation in agricultural mobile
machines. The main contribution of the developed embedded
control is that it overcomes a number of limitations in the
control of hydraulic systems in mobile machines - variance
of the physical parameters of the working fluid, positive
overlap in proportional valve, working temperature and other.
This performance is confirmed with a physical experiments
on the laboratory test rig.
ACKNOWLEDGEMENTS
The authors would like to thank the Research and
Development Sector at the Technical University of Sofia for
the financial support.
4 12-14 November 2020, Ruse, Bulgaria
Authorized licensed use limited to: University of the Phillippines Diliman. Downloaded on April 05,2022 at 08:59:37 UTC from IEEE Xplore. Restrictions apply.
Fig. 7. Measured spool position.
Fig. 8. Measured flow rate.
REFERENCES
[1] F. Más, Q. Zhang etc. “Mechatronics and Intelligent Systems for Off-
road Vehicles” ISBN 78-1-84996-467-8, Springer-Verlag London
Limited, 2010.
[2] Yang Y, C. Zhou etc. “Model reference adaptive robust fuzzy control
for ship steering autopilot with uncertain nonlinear systems” Applied
Soft Computing, Elsevier, 2003.
[3] Danfoss, “OSPE Steering Valve”, Technical Information, 11068682,
November, 2016.
[4] Il. Angelov, Al. Mitov, “Test Bench for Experimental Research and
Identification of Electrohydraulic Steering Units”, International Fluid
Power Conference, 10thIFK`2016, ISBN 978-3-9816480-1-0,
Dresden, Germany, 2016.
[5] P. Petkov, Ts. Slavov, etc. “Design of Embedded Robust Control
Systems using MATLAB®/Simulink®”, IET Control, 2018.
[6] G. Todorov, K. Kamberov, A. Ivanov, “Decreasing power loss through
control improvement of kinetic UPS system”, 44th International
Conference on Applications of Mathematics in Engineering and
Economics, AMEE 2018, Bulgaria, ISBN: 978-073541774.