Automation and Robotics in Construction XII
E.Budny, A.McCrea, K.Szymanski (Editors)
1995 IMBiGS. All Rights reserved.
179
CONCEPT OF A FLEXIBLE MICROPROCESSOR SYSTEM
OF CONTROLLING ELECTRO-HYDRAULIC DRIVES
OF BUILDING MACHINES
Eugeniusz Budnya, Michal Bartysb
a Institute of Mechanization of Building and Rock Mining in Warsaw
b Institute of Industrial Automation, Warsaw University of Technology
Abstract
This paper presents a concept of a microprocessor control system with flexible structure
and parameters, designed for steering and positioning working components of machines
driven by electro-hydraulic systems. An idea for decomposition of a block structure of
the control system was presented, as well as the idea of transposing block structure into a
machine readable form. Also, an example of the concept of a single-circuit control
system for one of the segments of a concrete pump extension arm has been given. Results
of simulation tests and experimental data for a concrete pump and hydraulic excavator
made in Poland are also provided.
1. Introduction
Technological progress in the development of reliable , ever more powerful and relatively
inexpensive microprocessor systems has opened completely new construction
perspectives also in those fields of engineering , which could be regarded as traditionally
conservative, e.g. in building machines.
Remotely controlled building machines designed for operation under hazardous for
direct operators conditions , efficient use of energy or automation and monitoring of
machine performance cannot be resolved by applying mechanical , electromechanical or
relay systems . Symbiosis of hydraulics and electronics has become a fact and the effect in
the future cannot be over-emphasized.
Electro-hydraulic systems use in building machines are exceptionally complicated nonlinear [1] dynamic systems with distributed parameters . So, an attempt to linearize the
dynamic model of a concrete pump extension arm [1] leads after reduction to an 8-th
order system in vicinity of the operating point . Non-stationary status of the object caused
by rheology, change of model descriptions caused by changes in order , strong nonlinearity effects , such as stick-slip , make automation of objects of this type particularly
complicated . The unprecedented recent development of linguistic logic applications for
building of low sensitivity high output controllers without creating an analytical
180
description of the controlled object gives hope for more rapid progress in automation of
complicated objects such as these.
On the automation side, each building machine may be considered as a
multidimensional automation system. Linearization and decoupling enables
decomposition of such a complicated object into systems of low sensitivity or nonsensitive to cross-coupling, i.e. into systems having a certain degree of autonomy. This
paper presents a concept of a reliable, convenient in application control system with
programmed structure and parameters, made in particular for automation of local
constructional units and, especially, of working units of building machines.
The following classification of control systems using microprocessor technique shall be
used in this paper:
• rigid systems with unalterable structure and parameters;
• parametrized systems with rigid structure and variable parameters
• flexible systems with freely selectable structure and parameters.
It seems that most attention should be given to the group of microprocessor systems
allowing free, though understandably restricted, selection of both structure as well as
controlled parameters. This is particularly important when implementing adaptive, selftuning algorithms, when parameters of the controlled object are being changed, etc.
A concept of such a system will be presented below, assuming additionally, that shaping
of control system properties should be performed using only software tools, without any
hardware modifications. This concept of a control system allows for setting its structure
and parameters through local or remote interaction with an operator console, oxternal
higher level control system or directly by the operator.
Such concept is justified because of the following practical reasons:
• the operator receives a tool permitting him to reproduce in a software environment a
parameter control system presented as a block scheme of the type commonly used in
automation;
• the operator or an external hierarchically higher control system can alter the
structure itself or just parameter values of the control systems according to given
requirements with consideration of natural limitations using software tools, only.
To implement this concept, it was necessary to find a method of decomposing the
structure of block control systems and to transpose such a decomposed block type
structure into machine readable form.
2. Decomposition of a control system block structure
A linear automation system with in inputs and n outputs mat be described by a
transmittance matrix G(s):
181
G11(s) G12( s) Glm(s)
G21(s) G22(s) G2m(s)
G(s) =
(1)
LGni(s) Gn2(s) Gnm(s)J
where: G&(s) transmittances are equal to the ratio of Laplace transforms of the i-th
output value to transform of the k-th input value assuming that all other inputs and
outputs as well as boundary conditions are equal zero.
The G(s) matrix may be transformed into the following column hypermatrix:
(2)
G(s) =
LGnI '
where:
G1('m)(s)
M) (S) J
Gil (s) G;2(s) .... Gin( s)].
Hence, each linear system with in inputs and n outputs may be decompose in to n
disjoint components with one output and in inputs.
Such a scheme will be hereafter referred to as a block. Decomposition of a control system
into multi-input blocks with just one output has practical significance, permitting a
relatively rapid and simple translation of the control system component structure into a
for readable for a microprocessor controller. Of course, technical limitations additionally
restrict the permissible number of in - inputs. Each of the above defined blocks may be a
component with:
- many inputs and one output
- many inputs without an output
- without inputs but with an output.
3. Transposition of control system block structure into machine readable form
Such transposition for efficient processing using a microprocessor controlled unit
requires formal modeling of the system ,e.g. into records of data. A set of appropriately
arranged records describing the block structure of a control system will hereinafter be
referred to as a machine readable structure.
Each of the control system structural blocks may be described using a record containing
the following indexed components:
182
- algorithms performed by the block
- algorithm parameters
- sources of input data
- value of output information.
The record consist of a fixed and variable part. The fixed part includes
- pointing to address of algorithm procedure performed by the block;
- pointing to constant values;
- pointing to sources of input information.
The variable part comprises:
- pointing to value of output information
- pointing to value of variable parameters.
4. Principles for formulation of a machine readable structure of a control system
A machine readable structure of the control system comprises a reproduction of the
control system block structure in internal machine language. As processing of the
structure and performance of algorithms pointed to by the appropriate variables
describing the structure usually takes place sequentially and cyclically within the
microprocessor controller, then adherence to appropriate relations and time regimes will
require a transformation of the structure according to arranging principles. Such
principles may be formulated in the following manner:
1. Each block scheme of a control system transformed into a format described by
matrix.(2) corresponds to exactly one machine readable block.
II. Blocks have unique numbers. The sequence of numbers described time sequence of
performing algorithms assigned to the blocks.
III. The last machine readable block (highest numbered) ends the internal structure of
control system.
IV. Processing of information in blocks to which algorithms are assigned is performed
sequentially in cycles with a sampling regime adapted to the numbering of blocks.
The repeat frequency of processing information encoded into rectors describing
structure of the control system cannot be less than the maximum information
processing time within individual structural blocks.
5. Flexible structure of control systems in a building machine
The above described control system block structure decomposition scheme with
transposition of the structure into machine-readable form has been implemented within
flexibly structured microprocessor control systems [3,6] designed for application in
electro-hydraulic drives of building machines.
183
Application of the flexible structure was possible by writing an appropriate, problem
oriented software systems for such controllers . Hence, it was necessary to develop [3 ] :
- a problem oriented interactive language (VALVIL); and
- high level graphical user software (CAVAL).
VALVIL describes a set of principles and methods for information interchange between
the controller and external environment.
Structure of the language permitted to minimize the number of principles and methods
applied, thus, facilitating apprehension of the language fundamentals by the operator, on
one hand, and gave a compact and time-efficient program code for implementation in
the microprocessor controller , on the other hand . The language permits both definition
of the control system block structure , as well as its parametrization , i.e. allows the
definition of an appropriately arranged sequence of records describing the structure and
parameters of the control system . VALVIL as well as its interpreter are characterized by
exceptionally time- efficient translation and performance of its commands . Hence, it
allows quasi- dynamic changes of the control system structure and its parameters in many
practical applications (adaptive systems ). Its features include:
• programmed alteration of structure and parameters of the microprocessor controller
of the system
• programmed control of a network of coupled controllers;
• programmed diagnostics of controller status
• performance of special functions related with the use of the controller and
positioning of electro-hydraulic units.
CAVAL: is the graphical application software system supporting synthesis of control
systems . Its use permits the direct user of microprocessor controlled units may perform
configuring and parametrization of control systems without having in -depth knowledge
of VALVIL language. The system has been designed in such a way which permits
operation by inexperienced operators not having any knowledge of hardware or
programming . However, the operator should have some experience in fundamentals of
automation which are being referred to by CAVAL. The system performs function ;
similar to these of VALVIL and may be considered as its graphical environment.
CAVAL consists of- block scheme editor
- block scheme translator
- direct communication with controller software.
Block scheme editor permits simple and convenient preparation of structure and
parameters of control systems. An example of such a block structure has been given in
fig. 1.
The translator is responsible for converting the defined structure into a set of VALVIL
language commands. It includes a formal correctness verification of the defined
184
structure. The listing of a translation of the structure given in Fig. 1 has been given in
fig. 2.
Communication software permits direct transfer of the structure and parameters of a
designed control system into the microprocessor controller.
t
OB
I
N
M
0
4
Fig. 1. A simple control system designed using CAVAL. OB - controlled object;
TR - measuring transducer (piston rod position); SP - set value definer (shift);
CR - controller; C - output value comparator; MID - two channel modulator of
pulse width generating the set pulse and power step; S - summation node. The
set value (lead)( is generated in real time and entered from an external,
hierarchically superior control system through the SP interface which is the
value defining unit. The output comprises piston rd shift of a servo motor
measured by the programmed transducer TR.
185
Riiii^ii
t
:ST4800
iU
;Li
0
SQ
CE
Trnsl
S
cc
SB1 SP
Now 1s edltln9 ...
Press Esc to exit
SPI 1 83886080
Sol 0
982 T
SP2 1 83096000
i
g..
1u
P-T
Fig. 2. Translation listing for the structure and parameters of the control system
presented in Fig. 1. Abbreviations have the same meaning as given in Fig. 1.
6. Application of a control system in a servo motor drive of a concrete pump
segment.
The central part of a microprocessor controller including the control system with
programmable structure and parameters is typically general. Its application to control of
electro-hydraulic drives used in machine construction requires inclusion of peripherals
permitting direct mating of the controller with a switching or proportional electrohydraulic manipulator.
A solution of this type conveniently integrates controlling and setting functions.
An example of applying a simple control system in a servo motor drive of one of the
three segments of a concrete pump extension arm [5,7] will be presented below.
6.1. Simplified simulation mode of an electro -hydraulic - servo-mechanism
The hydrostatic drive system of each of the three working segments of a concrete pump
or hydraulic excavator is in each case an electro-hydraulic unit coupled with a linear
hydraulic cylinder. Piston rod movement of a hydraulic servo is transformed into rotation
186
of the driven segment. So, the set and output values of each of the servo-mechanisms are
rotation angles of the driven segment and not direct shift of the piston rod.
Dependence between linear shift of the piston rod and segment rotation angle is nonlinear and depends on geometrical relation imposed by kinematic solution of the
extension arm segment [5].
Design of a servo-mechanism angular motion control system using classical solutions,
requires at least the knowledge of an approximate dynamic model of the controlled
object.
Experimental research has led to formulation of a simplified dynamic servo-mechanism
model with emphasized non-linearity such as the dead zone or power restriction in the
system (fig. 3).
u
-a ;
/..
- a
y IT
1
s
Fig. 3. Simplified dynamic model of an electro-hydraulic servo-mechanism
taking into consideration the dead zone and power restrictions. u - input signal
(setting value); 2a - width of dead zone; y - output signal (shift), T - time
constant, A - dead zone block; B - power limitator; 0 - controlled object; S summation node.
Existence of a dead zone is caused mainly by overlapping of mechanical and hydraulic
systems of the electro-hydraulic manipulator..
The total width of the dead zone was experimentally determined for the W19 concrete
pump extension arm at a = 37% (using Mannesmann-Rexroth manipulators) and defined
as approximately symmetrical.
Since this form of non-linearity dominates, no consideration was given to in the
approximate model of hysteresis caused by friction in the mechanical system or still
more negligible elasticity hysteresis of materials and media in the servo-mechanism
system.
Note that no practical solution permits the supply of unlimited power to the working
unit. Therefore, the model assumes power restriction up to a certain maximum value
which is independent of time and working position of the unit. This power supply
restriction to the system affects maximum speed y and acceleration y of piston rod shift
and limits the linear range of servo-mechanism operation when applying a proportional
controller (see [34]).
Linearization of non-linear dead zone characteristics of the object is theoretically
possible by introducing infinitively high gain in a closed control system. But, this is
practically unacceptable. Instead, it is possible to consider correction of the non-linear
static characteristics by introducing an opposite non-linear correction.
187
The use of such solution is desired for two reasons:
• it permits reducing of control deviation in stationary state without increasing gain of
the system;
• it permits lower gain in the system when maintaining the same value of control
deviation in stationary state.
However, in practice the exact width of the dead zone remains unknown and dependent
on the configuration. Moreover, its width may be changed by non-stationary of
construction and operating parameters which affect both size as well as symmetry.
Therefore full compensation of the dead zone is both difficult and unnecessary. It may
also lead to "over-correction", i.e. introduction of a non-linear correction to the linear
part of the characteristics, which results in other types of non-linearity and generally
increases energy losses.
w
S1
W1 Ni
I I N2 LJ N3
D1
C1
U)4
M1
01
LPI=
SIMULINK
W2
02
Fig. 4. Servo-mechanism simulation model for one of the segments of W19 concrete
pump. w - unit stroke generator, Sl - summation node, Wl - proportional controller
with gain k; Ni - unit restricting output signal of controller Wl; C1 - dead zone
correction unit; N2 - unit restricting output signal of correction unit Cl; D1 - unit
with dead zone; N3 - power limiter; 0 - transmittance of linear part of the object;
MI - multiplexer; 01 - y(t) recorder; y - output value vector; W2 - shift
measurement (y) amplifier; S2 - summation node; 02 - e(t) recorder, e - control
deviation.
The simulation model of the concrete pump extension arm electro-hydraulic servomechanisrm, was prepared using tool software for simulation research (SIMULINK) under
Windows 3.1. The simplified scheme presented in Fig. 4. includes both the above
adopted model assumptions as well as certain practical limitations of implementations.
In particular, this refers to the output signal limiters Ni and N2 which are integral
components of all microprocessor controller structural blocks. The proportional
controller was presented for simplicity. It is easily shown that the existing non-linearitics
which cannot be fully compensated make it practically impossible to assure zero
deviation of control in stationary state even though the object has dynamic integra i n'g
properties (see example of simulation testing in Fig. 5).
188
Range of non-linear operation
of the control system
Range of linear operation
of the control system
Fig. 5. Response of the control system with a proportional controller for step signal with
correction for dead zone of the object. The system includes a non-linear dead zone
compensation unit to a= 5%. With the time constant T = 10 s, controller gain k = 10 and
amplifier W2 gain equal 1, the control deviations in stationary state were 0.7%; y - step
shift of servo motor; y - normalized speed of piston rod.
Simulation tests have shown that because of minimization of power loss in the system
entry into linear operating mode should take place only during start-up and braking of
the servo-mechanism. Those phases can be made aperiodical by appropriately defining
the control system algorithm. Simulation and experiments (fig. 6) show that it is possible
to use proportional regulators with dead zone non-linearity correction. It can be shown in
such a case that the approximate value of control deviation in stationary state is given by
[4]:
e=
a
k
where:
a - effective width of dead zone after correction
k - gain coefficient in the control system.
(3)
189
0
0
2
1
4
3
6
5
8
Czas w (s]
60
-& --------- r--------- *--------+.........
50
_-r
40
30
t
0
2
1
4
3
5
7
6
0
8
10
C:zas w (s]
0
5
10
15
20
25
Czar w (s]
Angular shift in [degrees] time in [s]
Fig. 6. Sample results of experimental responses to step (angular) pulses for
servomechanisms of individual segments of W19 extension arm in automated control as
for fig. 7 with upward stroke.
6.2. Implementation of the control system structure in the microprocessor controller
Implementation of a simulated control system in the microprocessor controller reduces
transfer of its structure and parameters using VALVIL o CAVAL and takes under
consideration special internal properties of the microprocessor controller system. The
30
190
task is simplified because reproduction of the simulated system within SIMULINK as
well as reproduction in the microprocessor controller both share a block type structure.
The product of such reproduction has been presented in Fig . 7. The block scheme has
been additionally supplemented by adding a two-channel pulse with modulator block (M)
permitting direct control of a proportional electro-hydraulic manipulator.
Parametrization of microprocessor controller control system block structure requires not
only reproduction of the simulation scheme parameters but also declaration of additional
parameters which expand the attributed block (e.g. zero shift ). This is conveniently
performed using CAVAL graphical software. A procedure of the type shown in Table 1.
r--_____-----` --------------1 I4CUO Ka' xCJ SIF1fTY NAR711EJ
r
r
wap
r
^ roar, -^ r
!1//
I
0002 .,. i
f..
-----------------i
O/ODA
MEZFI
JVJACYJJY
.bs m.li J - SW7 , i 2W P&
Fig. 7. Simplified block scheme of the servo- mechanism control system structure for one
of the W 19 concrete pump extension arm segments . SP - set value definer, T - measuring
transducer (rotation- encoding transducer), PID - controller, DIOD - diode-type nonlinear block; M - two-channel pulse width modulator ; S - summation node; OB controlled object, cc, i3, y, I - signals in the control system.
7. Conclusions
The concept of a microprocessor control system with flexible structure and parameters,
applied in electro-hydraulic unit positioning systems has been applied in practice [6, 71.
191
Table 1. Listing of the UB93 controller configuration program
•i••ar•r n© ri#ir •••aa• r•••ri naa••rr•rrirrrr • rrrrilr•aaarrrr
••
CONTROL
PROGRAM
•• FOR ONE ARM OF W19 CONCRETE PUMP ••
Copyright Michal Bartys •!
•! TU-GH-Duisburg 26.01.1994
:•a
aafa• q# !!!# f
#4
Made for IMB i GS Poland ••
#q! f!! f!flfr ##galria i# # aqr# #gafasaf a!f i •a
P CONTROLLER WITH NONLINEAR CORRECTION f •
:rlri#!!ai!!!!itii ##! f!#!r!!#!!!!lri+##ia!•iii!!#1#f##!f###
"Control phase ••
L
SA 4 ; Sets an address 41!
U
L4
SE
S
CC
;++ Configuration phase !a
SB 1,SP
SB2,T
SB3,S
SB4,P
SB5,DIOD
SB6,DIOD
SB7,S
SB8,M
SC 1,3,1
SC2.3,2
SC3,4.1
SC4,5,1
SC4,6,1
SC5,7,1
SC6,72
SC7,8,1
++ Parametrisation phase
SPI,1 , 16777216
SZ1,0
SP2,1 ,16777216
SZ2,0
SP3, i,1
SP3,2,-1
SP3,3,0
SP3,4,0
SZ3,0
SP4,1 . 167772160
SZ4,0
SP5,1,10905190
SP5,2,1
SZ5,-5S72026
SP6,1 , 10905190
SP6,2,-1
SZ6,-5872026
SP7,1,1
SP7,2,1
SP7,3,0
SP7,4,0
SZ7,0
SP8,1 ,16777216
SZ8,0
;•• Initialization ••
I
C
192
The central unit of this system was built using two eight bit single chip microprocessors
of the popular MCS-51 range from INTEL linked to a common RAM and EPROM. Each
of the processors also has its own program memory. The processors operate in parallel
with one having priority in access to the common memory base. The higher priority
processor is the master while the lower priority one, is the slave. The distribution of tasks
between both processors is as follows:
• the master processor performs controlling algorithms, transmission of information
to and from the controller and emergency procedures;
• the slave processor performs pulse width modulation, operates the D/A converter,
pules and code shift transducers, digital measurement of shift, speed and
acceleration, digital filtration of measuring signals.
Clock speed of both processors is 12 MHz. This solution can be used for implement a
control system of the type shown in Fig. 7 with sampling cycle of about 5 ms.
The concept of a control system which flexible structure and parameters is extremely
useful, particularly in experimental research, where concept of the control system itself
as well as of its parameters has to change according to results the obtained. Practical
usefulness of the microprocessor controller has been confirmed at the instituted during
analysis of the operation of W109 concrete pump extension arm and of the Warynski 711
hydraulic excavator. Results received so far point to the possible use of formal linguistic
logic referring to heuristic knowledge in system controllers.
References
1. Hirsch U., Jacubasch A., Kuntze H., Eberle F., Goller B.: Modeling and Control of a
Hydraulic Large Range Robot, pp. 609-620
2. Benckert H.: Computer Controlled Concrete Distribution, pp. 11-119.
3. Bartys M. Electro-hydraulic valves with digital input, Internal report A409, Institute
of Industrial Automation, Warsaw University of Technology, Warsaw, 1990.
4. Bartys M.: Simulation testing of an electro-hydraulic servo-mechanism of the W19
concrete pump extension arm. Internal Report, IMBiGS, Warsaw, 1994.
5. Budny E.: Hydraulic systems of multi-segment working machines with elements of
working movement automation. Scientific Library of Building Mechanization,
IMBiGS. Warsaw 1994. p. 10 Lecture presented at the Scientific Conference on
heavy Working Machines, January 1994.
6. Budny E., Bartys M.: System for controlling electro-hydraulic drive units of the
W19 concrete pump extension arm. Internal Report, IMBiGS, Warsaw 1993.
7. Budny E., Bock T.: Excavator Rationalization by Hybrid Control System.
Engineering. Construction and Operations in Space Conference, Albuquerque, New
Mexico, February 26 - march, 1994. p. 9