Machine tools abstract
A device for and method of preventing accidents caused by jamming
of a tool when working with hand-operated machine tools driving
a rotating tool, especially hammer drills, is outfitted with an
interrupter device for interrupting the action of the driving motor
on the rotating tool depending on the operating state detected by
a deflection sensor (1). The device includes a computer unit (R)
in which a probability for an expected accident is determined from
the sensor signals supplied by the deflection sensor (1), and in
that the interrupter device (5 6) can be activated when a preset
accident probability is exceeded. In the method rotational movement
values are determined from the measurement signals supplied by the
deflection sensor (1), these rotational movement values are converted
in a computer unit (R) by nonlinear logic operations to form accident
probabilities and activating the interrupter device as soon as the
estimated accident probability exceeds a preadjustable threshold
value.
Machine tools claims
I claim:
1. Device for preventing accidents caused by jamming a rotating
tool (8) in hand operated machine tools (M), said machine tool (M)
comprising an interrupter device (56) for interrupting the operation
of a motor for driving said rotating tool (8) depending on the operating
conditions of the machine tool (M) detected by a deflection sensor
(1) in said machine tool (M), a computer unit (R) within said machine
tool (M) connected to said deflection sensor (1) for converting
measured values received from said deflection sensor into rotational
movement values characterizing the operating conditions of the machine
tool and for processing the rotation movement values according to
presettable criteria by nonlinear signal processing to form accident
probabilities, and a comparison device (14) connected to said interrupter
device (5 6), so that said comparison device generates a signal
for activating said interrupter device (5 6) when the accident
probability determined by said computer unit (R) exceeds an adjustable
threshold value.
2. Device, as set forth in claim 1 wherein said computer unit
(R) comprises a fuzzy controller (3).
3. Device, as set forth in claim 2 wherein said deflection sensor
(1) comprises an angular accelerometer (1).
4. Device, as set forth in claim 3 wherein two band limited integrators
(11 12) are connected in a cascade arrangement, whereby said fuzzy
controller (3) receives the output signal of said angular accelerometer
(1) and the output signal of the integrators (11 12) as input signals.
5. Device, as set forth in claim 4 wherein said angular accelerometer
(1) is one of a piezoelectric, piezoresistive, and micromechanical
accelerometer.
6. Device, as set forth in claim 4 wherein said angular accelerometer
(1) comprises an electrical micromechanical composite chip having
a pendulum forming a middle electrode of a differential capacitor.
7. Device, as set forth in claim 4 wherein said fuzzy controller
(3) comprises a program in a microprocessor.
8. Device, as set forth in claim 4 wherein said fuzzy controller
(3) comprises a lookup table.
9. Device, as set forth in claim 4 wherein said fuzzy controller
(3) comprises a fuzzy processor.
10. Device, as set forth in claim 3 wherein a bandpass filter
(10) and two band limited integrators (11 12) are connected and
located downstream from said angular accelerometer (1), said bandpass
filter and said band-limited integrators being connected in a cascade
arrangement, whereby said fuzzy controller (3) receives the filtered
output signal of said angular accelerometer (1) and the output signal
of said integrators (11 12) as input signals and whereby the limiting
frequencies of the overall useful frequency band processed by said
fuzzy controller (3) and the two band-limited integrators (11 12)
connected in cascade arrangement lie between a lower limiting frequency
.omega..sub..mu. in a range of 0.5 Hz<.omega..sub..mu. <10
Hz and an upper limiting frequency .omega..sub.0 in a range of 100
Hz<.omega..sub.0 <1000 Hz.
11. Device, as set forth in claim 3 wherein said interrupter device
(5 6) comprises a clutch (5) arranged in a drive train between
said drive motor (7) and said rotating tool (8).
12. Method of preventing accidents caused by jamming a rotating
tool (8) driven by a hand operated machine tool (M) and comprising
an interrupter device (5 6) in said machine tool for interrupting
a rotational action on said rotating tool (8) depending on the operating
state of the rotating tool detected by a deflection sensor (1) in
said machine tool comprising the steps of determining rotational
movement values characterizing the operating conditions of the rotating
tool (8) obtained from measured values provided by said deflection
sensor (1), and processing the rotational movement values in a computer
unit (R) based on presettable criteria by nonlinear signal processing
for forming accident probabilities, and activating said interrupter
device (5 6) as soon as the accident probability determined by
the computer unit (R) exceeds an adjustable threshold value.
13. Method, as set forth in claim 12 comprising the further step
of effecting nonlinear processing of the rotational movement values
to form accident probabilities by a fuzzy controller provided in
the computer unit (R).
14. Method, as set forth in claim 12 or 13 wherein said deflection
sensor (1) is an angular accelerometer (1) having measured values
and converting the measured values by double integration into an
angular velocity (phid) and a deflection angle (phi) with a preset
time constant.
15. Method, as set forth in claim 14 comprising the further step
of fuzzifing by a fuzzy controller an angular acceleration (phidd),
the angular velocity (phid) and the deflection angle (phi) and receiving
estimated values for an anticipated accident by inferencing by means
of linguistic rules, and converting the estimated values into a
probability (.mu..sub.k) of an accident by means of defuzzification.
16. Method, as set forth in claim 15 wherein linguistic variables
for fuzzification are defined as follows:
angular acceleration (phidd) small: up to a value of 150 rad/s.sup.2
a truth value .mu.=1 which drops linearly between 150 rad/s.sup.2
and 350 rad/s.sup.2 to .mu.=0 where it remains for higher values;
angular acceleration (phidd) medium: up to a value of 150 rad/s.sup.2
a truth value=0 which increases linearly between 150 rad/s.sup.2
and 350 rad/.sup.2 to .mu.=1 where it remains for higher values
until it drops linearly between 1500 rad/s.sup.2 and 2000 rad/s.sup.2
to .mu.=0 where it remains for higher values;
angular acceleration (phidd) large: up to a value of 1500 rad/s.sup.2
a truth value .mu.=0 which increases between 1500 rad/s.sup.2 and
2000 rad/s.sup.2 to .mu.=1 where it remains for higher value;
angular velocity (phid) small: up to a value of 0.25 rad/s, a truth
value .mu.=1 which drops linearly between 0.25 rad/s and 19.75
rad/s to .mu.=0 where it remains for higher values;
angular velocity (phid) medium: up to a value of 0.25 rad/s, a
truth value .mu.=0 which increases linearly between 0.25 rad/s and
19.75 rad/s to .mu.=1 where it remains until a value of 20 rad/s
and then drops linearly between 20 rad/s and 29 rad/s to .mu.=0
where it remains for higher values;
angular velocity (phid) large: up to a value of 20 rad/s, a truth
value .mu.=0 which increases linearly between 20 rad/s and 29 rad/s
to .mu.=1 where it remains for higher values;
deflection angle (phi) small: up to a value of 0 rad, a truth value
.mu.=1 which drops linearly between 0 rad and 5.5 rad to .mu.=0
where it remains for higher values; and
deflection angle (phi) large: up to a value of 0 rad, a truth value
.mu.=0 which increases between 0 rad and 5.5 rad to .mu.=1 where
it remains for higher values.
17. Method, as set forth in claim 16 wherein fuzzy variables at
the output are defined as follows:
clutch out (.mu.0) is a fuzzy singleton at 0; and
clutch in (.mu.1) is a fuzzy singleton at 1.
18. Method, as set forth in claim 17 wherein the following linguistic
rules apply:
IF angular velocity (phid) small THEN clutch out (.mu.0),
IF angular velocity (phid) medium AND (angular acceleration (phidd)
NOT large) THEN clutch out,
IF angular acceleration (phidd) small THEN clutch OUT (.mu.0),
IF angular velocity (phid) large THEN clutch in (.mu.1),
IF angular acceleration (phidd) large AND deflection angle (phi)
large THEN clutch in (.mu.1),
IF angular acceleration (phidd) medium AND deflection angle (phi)
large AND (angular velocity (phid) NOT small) THEN clutch in (.mu.1).
19. Method, as set forth in claim 18 comprising the step of carrying
out inferencing by means of Min-max inference.
20. Method, as set forth in claim 19 comprising the further step
of effecting defuzzification according to the following equation:
where .mu..sub.0 .di-elect cons.(01) stands for the truth value
for clutch out, .mu.,.di-elect cons.(01) stands for the truth value
for the clutch in, and .mu..sub.k .di-elect cons.(0 1) stands for
the probability of an accident.
Machine tools description
BACKGROUND OF THE INVENTION
The invention is directed to a device for and a method of preventing
accidents due to tool jamming when working with hand-operated machine
tools driving rotating tools, such as hammer drills, outfitted with
an interrupter device for interrupting the driving action of a motor
for the rotating tool depending on the operating conditions detected
by a deflection sensor.
A serious problem in hand-operated machine tools, especially those
with higher output, such as hammer drills, involves so-called rotation
accidents, such as, injuries to the wrist or falls from ladders
or scaffolding, and the like, as a result of the sudden jamming
of the rotating tool and the consequent rapid increase in the reaction
moment at the machine. Out of the large number of known solutions
devised for this problem, reference is had to DE 44 26 305 A1 and
DE 43 44 817 C2.
The first reference discloses a purely mechanical solution for
this safety problem with an overload slip clutch, whose releasing
moment occurs at a presettable adjusting speed depending on the
occurring torque, wherein the clutch does not disengage during normal,
trouble-free operation. If a rapid rise in the required torque should
occur in the event of malfunction, for example, when the tool is
jammed, the slip clutch, which is affected by inertia or damped
by volume displacement, is released and immediately interrupts the
line of transmission of force from the motor of the device to the
rotating tool.
The second reference discloses a mechanical-electrical solution.
In this reference, an anticipated deflection angle is calculated
by means of an acceleration pickup. If this deflection angle is
greater than a maximum presettable deflection angle, an interrupter
device, in particular an electromagnetic clutch, is released before
the maximum deflection angle is reached, possibly in combination
with a shutting off of power. The acceleration pickup is an angular
acceleration sensor whose measurement values are converted to the
expected rotational angle by double integration over the duration
of a time constant.
The first solution presents difficulties in that an untimely releasing
of the safety clutch can also occur as a result of intended manipulation
of the machine, such as when working with a hammer drill in a concrete
mass of inhomogeneous composition. This has to do with the use of
passive signal evaluation, particularly because the movement sensor
and actuator are comprised in one unit and purely mechanical principles
are applied.
Difficulties arise in the second solution because this model-based
solution is founded on a weighted addition of the instantaneous
angle calculated from the measured angular acceleration with the
predicted amounts of angular velocity and the angular acceleration
relative to the angle occurring after the expiration of a forecast
time, wherein an adjustment of the system parameters must be carried
out empirically based on difficult-to-identify parameters of the
process model. In this case, it is hard to modify the process model
or adapt it to different machines. Further, once the process model
has been set, the summation can not be weighted differently in this
procedure. Thus, it may come about, for example, that the interrupter
device is activated when only one of the input variables, for instance,
only acceleration, is very high, although there is no risk of an
accident.
SUMMARY OF INVENTION
The primary object of the present invention is to improve hand-operated
machine tools of the type mentioned above in such a way that a better
criterion is given for discriminating between permissible reaction
moments and dangerous reaction moments which can lead to rotation
accidents so as to activate a clutch such that work safety is increased
appreciably when working with such a machine tool.
A device for preventing accidents caused by jamming of the rotating
tool in hand-operated machine tools such as hammer drills, which
are outfitted with an interrupter device for interrupting the action
of the driving motor on the tool depending on the operating state
detected by a deflection sensor is characterized, according to the
invention, in that a computer unit is provided in which an expected
accident can be estimated by nonlinear signal processing from the
sensor signals supplied by the deflection sensor, and in that the
interrupter device can be activated via a comparison device when
a presettable accident probability is exceeded.
Accordingly, similar to DE 43 44 817 C2 the invention is based
on the idea of a predictive decision about the future behavior of
the machine so that countermeasures can be taken before the machine
tool receives an angular momentum of such extent that an accident
can no longer be prevented. In contrast to DE 43 44 817 C2 which
provides an electronic evaluating unit that calculates an anticipated
deflection angle of the machine tool from the sensor signal supplied
by the deflection sensor over a fixed time period, wherein the interrupter
device can be actuated by a comparison device when a predetermined
maximum deflection angle is exceeded, the invention does not simply
calculate only the anticipated deflection angle of the machine tool.
The solution according to the invention uses a computer unit operating
on the principle of nonlinear signal processing to estimate the
anticipated accident probability so that a valuated judging of the
sensor signals supplied by the deflection sensor or the signals
derived therefrom can be carried out. In this way, the angular acceleration,
the angular velocity, and the deflection angle covered during a
determined time period can be evaluated in relation to one another
as well as independently from one another, whereupon the likelihood
of an accident is determined by the evaluation of judgment rules.
The computer unit preferably comprises a fuzzy controller. The deflection
sensor is primarily conceived as an angular acceleration sensor,
such as a micromechanical angular accelerometer with a very short
response time, whose measurement value signal is converted into
a rotational velocity and a deflection angle by double integration
over the duration of a time constant. The input signals forming
the basis for the decision of the fuzzy controller about the accident
probability are then the measured angular acceleration and rotational
or angular velocity calculated therefrom as well as the deflection
angle determined therefrom.
According to the invention, a limiting of the useful band within
a range of values of 0.5 to 10 Hz for the lower limit frequency
and within a range of approximately 100 to the order of magnitude
of 1000 Hz for the upper limit frequency can be provided, so that
the influence of ever-present low-frequency and high-frequency interference
need not be taken into account exclusively by the fuzzy controller.
Similar to the known solution according to DE 43 44 817 C2 the
interrupter device is chiefly conceived as an electromagnetic clutch,
that is, a friction clutch, multi-disk clutch or claw clutch, which
interrupts the action of the kinetic energy stored in the rotor
of the motor on the jammed rotating tool spindle in a sudden manner.
It has been determined through extensive measurement series, experience
and experiments that there is an available decoupling time of 5
to 20 msec for this interruption, depending, among other things,
on the type of device in question. This decoupling can be, and generally
is, combined with a simultaneous interruption of current to the
motor. For smaller machines with comparatively light rotors, a fast-acting
brake device can also be considered instead of the clutch, possibly
again in combination with an immediate interruption of current when
the sensor signals of the deflection sensor or the signals calculated
therefrom are interpreted by the fuzzy controller to mean that there
is a high probability of accident.
The fuzzy controller can be realized in any known manner ensuring
a timely interruption, that is, for example, by means of a program
in a microprocessor, by a fuzzy processor or by means of a lookup
table.
A process for preventing accidents caused by jamming in hand-operated
machine tools with a rotating tool, especially hammer drills, which
are outfitted with an interrupter device for interrupting the action
of the driving motor on the tool depending on the operating state
detected by a deflection sensor is characterized, according to the
invention, in that an accident probability is determined from the
measurement signals supplied by the deflection sensor by nonlinear
signal processing and the interrupter device is activated as soon
as the estimated accident probability exceeds a presettable threshold.
The nonlinear logic operation between the measurement signals that
are converted into rotational movement quantities is preferably
effected by means of a fuzzy controller.
The deflection sensor is preferably an angular accelerometer whose
measurement values are converted by double integration into an angular
velocity and a deflection angle with a preset time constant, wherein
the fuzzy controller fuzzifies the angular acceleration, the angular
velocity and the deflection angle and receives estimated values
for an anticipated accident by inferencing by means of linguistic
rules, which estimated values are converted into the probability
of an accident by means of defuzzification.
In this way, a complicated filtering of the sensor signals and
difficult determination of parameters of the integrators can be
avoided by using a fuzzy controller, since, by means of fuzzy logic,
inexact linguistic qualitative information can be manipulated heuristically
through the use of inference rules in order to give a clear and
precise finding about accident probability by superimposing a plurality
of fuzzy rules with input values which do not necessarily match
exactly. It is possible to determine the inference rules in a very
simple manner by experimentation, and parameter changes, for example,
as a result of adapting to another machine tool, can be carried
out very quickly and in a very flexible manner.
The invention and advantageous particulars are explained hereinafter
with reference to the drawings in an embodiment form given by way
of example.
BRIEF DESCRIPTION OF THE DRAWINGS
In the drawings, shown schematically;
FIG. 1 is a block diagram illustrating the invention;
FIG. 2 is a hammer drill, as an example of a hand-operated machine
tool, outfitted in accordance with the invention;
FIG. 3 is a block diagram of the drill hammer shown in FIG. 2;
FIG. 4 is a functional block diagram showing a possible signal
pre-processing using an accelerometer as a deflection sensor;
FIG. 5 is the block diagram of a possible fuzzy algorithm;
FIG. 6 is an example of a definition of the linguistic variables
on which the fuzzification of the state variables is based; and;
FIG. 7 is an example of a characteristic surface of the fuzzy algorithm
according to FIGS. 5 and 6 as it is given before the comparator
shown in FIG. 5.
DETAILED DESCRIPTION OF THE INVENTION
A device for preventing accidents due to sudden jamming of the
rotating tool of a hand-operated machine, especially a hammer drill,
and an associated fuzzy controller with an exemplary fuzzy algorithm
are described hereinafter with reference to the drawings.
As can be seen in FIG. 1 the operating state of a hand-operated
machine tool M as shown in FIG. 2 is monitored by at least one
deflection sensor which can be an angular acceleration sensor, especially
a micromechanical accelerometer, a linear acceleration sensor or
a combination of linear acceleration sensors, a speed sensor, a
path-distance sensor (translatory sensor), etc. or a torque sensor.
In FIG. 1 an arrangement of deflection sensors is provided with
reference numbers 1a, 1b . . . , 1n. By way of an input interface
2 for signal shaping, analog-to-digital conversion, etc., the signal
(or signals) of the sensor or sensors reaches (reach) a fuzzy controller
3 which can be realized by means of a microprocessor with a corresponding
program and data storage, by means of a fuzzy processor, or by means
of a lookup table which predicts the accident probability of the
machine at the response of the sensor or sensors 1a to 1n. An example
of a definition of the linguistic variables and of a control basis
and algorithm for the defuzzification and of an accident probability
predicted by the fuzzy controller is explained more fully hereinafter.
When the determined accident probability exceeds a threshold value,
one or more actuators, for example, a clutch 5a which interrupts
the drive train between a driving motor 7 and the tool holder or
tool shaft, a power switch 5b, and/or a brake 5m, are activated
via a comparison device 14. These actuators 5 prevent a predicted
injurious incident, especially a possible rotation accident, from
actually occurring. As soon as the machine M has been changed over
to a safer state for the operator, work may continue as desired
by the operator.
The embodiment example illustrated in FIG. 2 shows a combination
of an inertial angular acceleration sensor 1 followed by amplification,
an anti-aliasing filter AA, note FIG. 2 a programmable microprocessor
as fuzzy controller 3 with a-d converter and digital interface 2
or d-a comparison device 4 a semiconductor current switch 6 and
an electromagnetically operating clutch 5. FIG. 3 shows a block
diagram of the hardware which can correspond to this embodiment
example. For the sake of simplicity, only the connections essential
to an understanding of the invention are shown within the computer
R in the drawing.
During operation of the hand-held device, high reaction torques
can occur as a result of sudden jamming of a rotating tool 8 which
can lead to a dangerous deflection of the hand-held device about
the axis 9. According to the invention, through the use of a fuzzy
control, these reaction moments are limited to a low value compatible
with the safety of the operator.
For this purpose, the angular acceleration around the axis 9 is
sensed by the angular acceleration sensor 1 wherein the inherent
dynamics of the sensor 1 must be sufficiently fast in accordance
with the above statement of problem. The response time is typically
less than 1 ms. Such angular acceleration sensors are known especially
in micromechanical construction and are commercially available.
The measurement signals of the sensor 1 are amplified and fed to
a microprocessor 3 via the a-d interface of the computer R. The
digitized measurement data are converted by single and double integration
to angular velocities and deflection angles. In the microprocessor
3 which comprises a program storage and a data storage, the measured
and converted rotational movement quantities are further processed
by means of a fuzzy algorithm which will be explained more fully
hereinafter. In accordance with the results for the probability
of an accident which are determined by the microprocessor 3 working
as a fuzzy controller, a signal is guided, via a d-a converter 4
serving as an interface output, to the clutch 5 serving in this
case as actuator and to a pulse-width modulator 6 which works as
a voltage breaker and which normally adjusts the motor speed in
order to interrupt the operation of the machine.
FIG. 4 shows the processing of the measurement signals sent from
the sensor arrangement 1 to form rotational movement quantities
characterizing the operating state. The measurement signal coming
from the acceleration sensor 1 is sent via a bandpass filter 10
to an integrator 11 and subsequently to an integrator 12. Finally,
the filtered angular acceleration phidd which is tapped directly
behind the filter 10 the calculated angular velocity phid which
is tapped behind the integrator 11 and the calculated deflection
angle phi which is provided by the output signal of the integrator
12 are transmitted to the fuzzy controller. The input filter 10
which is fashioned as a bandpass filter reduces high-frequency interference
and filters out possible offsets.
The fuzzy algorithm is illustrated in principle by the definition
of the linguistic variables in FIG. 6 and the block wiring diagram
of the fuzzy algorithm in FIG. 5. It serves to predict the accident
probability of the machine M in the case of a corresponding sensor
signal. The rotational movement variables, angular acceleration
phidd, angular velocity phid, and deflection angle phi characterizing
the operating state are provided to the system as linguistic variables.
A definition of the linguistic variables is shown by way of example
in FIG. 6 with the following results:
phidd small:
.mu.=1 for .infin.<phidd.ltoreq.150 rad/s.sup.2
.mu.=1.75-1/1200.multidot.phidd.multidot.s.sup.2 /rad, for 150
rad/s.sup.2 .ltoreq.phidd.ltoreq.350 rad/s.sup.2
.mu.=0 for 350 rad/s.sup.2 .ltoreq.phidd<+.infin.
phidd medium:
.mu.=0 for -.infin.<phidd.ltoreq.150 rad/s.sup.2
.mu.=1/200.multidot.phidd.multidot.s.sup.2 /rad-0.75 for 150 rad/s.sup.2
.ltoreq.phidd.ltoreq.350 rad/s.sup.2
.mu.=1 for 350 rad/s.sup.2 .ltoreq.phidd.ltoreq.1500 rad/s.sup.2
.mu.=4-1/500.multidot.phidd.multidot.s.sup.2 /rad for 1500 rad/s.sup.2
.ltoreq.phidd.ltoreq.2000 rad/s.sup.2
.mu.=0 for 2000 rad/s.sup.2 .ltoreq.phidd<+.infin.
phidd large:
.mu.=0 for -.infin.<phidd.ltoreq.1500 rad/s.sup.2
.mu.=1/500.multidot.phidd.multidot.s.sup.2 /rad-3 for 1500 rad/s.sup.2
.ltoreq.phidd.ltoreq.2000 rad/s.sup.2
.mu.=1 for 2000 rad/s.sup.2 .ltoreq.phidd<+.infin.
phid small:
.mu.=1 for -.infin.<phidd.ltoreq.0.25 rad/s
.mu.=79/78-2/39.multidot.phidd.multidot.s/rad, for 0.25 rad/s.ltoreq.phid.ltoreq.19.75
rad/s
.mu.=0 for 19.75 rad/s.ltoreq.phidd<+.infin.
phid medium:
.mu.=0 for -.infin.<phid.ltoreq.0.25 rad/s
.mu.=2/39.multidot.phid.multidot.s/rad-1/78 for 0.25 rad/s.ltoreq.phid.ltoreq.19.75
rad/s
.mu.=1 for 19.75 rad/s.ltoreq.phid 20 rad/s
.mu.=29/9-1/9 phid.multidot.s/rad, for 20 rad/s.ltoreq.phid.ltoreq.29
rad/s
.mu.=0 for 29 rad/s.ltoreq.phid<+.infin.
phid large:
.mu.=0 for -.infin.<phid.ltoreq.20 rad/s
.mu.=1/9 phid.multidot.s/rad-20/9 for 20 rad/s.ltoreq.phid.ltoreq.29
rad/s
.mu.=1 for 29 rad/s.ltoreq.phid<+.infin.
phi small:
.mu.=1 for -.infin.<phi<0 degrees
.mu.=12/11.multidot.phi.multidot.1/degrees, for 0 degrees.ltoreq.phi.ltoreq.5.5
degrees
.mu.=0 for 5.5 degrees.ltoreq.phi<+.infin.
phi large:
.mu.=0 for -.infin.<phi<0 degrees
.mu.=2/11.multidot.phi.multidot.1/degrees, for 0 degrees.ltoreq.phi.ltoreq.5.5
degrees
.mu.=1 for 5.5 degrees.ltoreq.phi.ltoreq.+.infin.
Due to the fuzzy algorithm, the input variables are fuzzified by
the linguistic variables listed above. Results for an expected accident
are then obtained by means of linguistic rules and are transformed
into the probability of an accident by defuzzification. The linguistic
rules can be determined as follows:
______________________________________ IF phid small THEN clutch
out IF phid medium AND phidd NOT large THEN clutch out IF phidd
small THEN clutch out IF phid large THEN clutch in IF phidd large
AND phi large THEN clutch in IF phidd medium AND phi large AND phid
NOT small THEN clutch in. ______________________________________
The fuzzy variables at the output are defined as follows:
clutch "out" .mu.0 is a fuzzy singleton at 0 and
clutch "in" .mu.1 is a fuzzy singleton at 1.
Inferencing by means of the linguistic rules and fuzzy variables
at the output is processed by the inference machine 13 on the principle
of MAX-MIN inference, known per se. The formation of a maximum or
(1-.mu.) is used to realize the logical operations AND and NOT,
wherein .mu. stands for the truth value of the term to be negated.
If the inference machine 13 generated the estimated values for clutch
"out" and clutch "in" by means of the individual
linguistic rules and outputted their respective maximum values,
the probability for an accident can be calculated by means of a
defuzzification. The defuzzification is effected according to the
following equation:
where .mu..sub.0 .di-elect cons.[01] stands for the truth value
for clutch out, .mu..sub.1 .di-elect cons.[01] stands for the truth
value for clutch in, and .mu..sub.k .di-elect cons.(01) stands
for the probability of an accident.
Thus, the result of the defuzzification is a real number from the
interval [0 1] which relates to a statement concerning the degree
to which the present situation corresponds to the class "jammed
drill". The closer this number is to the value 1 the truer
the statement that the drill is jammed.
However, since the clutch only recognizes the "in" and
"out" states, the results of the defuzzification are fed
to a comparator 14 whose threshold can be varied. The output of
the comparator 14 accordingly controls the active clutch for opening
the drive train.
FIG. 7 shows the characteristic surface of the fuzzy algorithm
as it results prior to the comparator 14 at an angle phi equals
1.5.degree. which is assumed to be constant. In this case, for the
fixed angle phi equals 1.50.degree., the angular velocity of 0 to
40 rad/s is plotted on the x-axis, the angular acceleration of 0
to 2500 rad/s.sup.2 is plotted on the y-axis, and the probability
.mu..sub.k for an accident is plotted on the z-axis. FIG. 7 illustrates
the complex characteristic surface of the fuzzy algorithm resulting
from fuzzification, inferencing and defuzzification. If one of these
three steps is modified, e.g., by means of a divergent definition
of the linguistic variables, different or modified rules for inferencing
or another defuzzification, the fuzzy controller can easily be adapted
to other machine tools or other use requirements. In this case it
would also be possible for these adaptations to be carried out during
operation by the operator of the machine tool, for example, by switching
existing parameter sets. The signal .mu..sub.k representing the
probability of an accident is fed to the comparator 14 whose threshold
is set, e.g., to 0.5.
All signals .mu..sub.k .ltoreq.0.5 are set to a value 0 and all
signals .mu..sub.k >0.5 are set to a value 1 by means of the
comparator 14. The clutch 5 and the pulse-width modulator 6 are
then so controlled by the comparator that the drive train is opened
and the action of the motor is interrupted at a signal "1"
of the comparator.
An accident can be successfully prevented by means of the fuzzy
algorithm. Whereas the judgment in the model-based solution according
to DE 43 44 817 C2 is based purely on a weighted addition of the
instantaneous deflection angle calculated from the measured angular
acceleration with the predicted amounts of angular velocity and
angular acceleration for the deflection angle occurring after the
forecast period has expired, the fuzzy solution is based on a judgment
of the system state by a nonlinear logic operation between a plurality
of state variables by means of linguistic rules. The fuzzy solution
makes it possible for the decision making to include optional relations
between the instantaneous state variables on the one hand and the
state of the system after the expiration of the forecast time on
the other hand. Due to the fuzzy algorithm, it is possible for physical
relationships which could be recognized as crucial for behavior
during malfunction in further improving and adapting the system
to be integrated within the decision making process.
While specific embodiments of the invention have been shown and
described in detail to illustrate the inventive principles, it will
be understood that the invention may be embodied without otherwise
departing from such principles. |