Note : Les descriptions sont présentées dans la langue officielle dans laquelle elles ont été soumises.
CA 02443748 2003-10-07
WO 02/084232 PCT/US01/40474
TITLE OF THE INVENTION
METHOD OF WEIGHT MEASUREMENT USING
MOVING WEIGH CONVEYOR
TECHNICAL FIELD OF THE INVENTION
The present invention relates to dynamic weight measurement, and particularly
to a continuously moving weigh conveyor for weighing individual quantities of
product
to be packaged.
BACKGROUND OF THE INVENTION
Many different kinds of food loaves are produced in a wide variety of shapes
and sizes. Meat loaves consisting of ham, pork, beef, lamb, turkey, fish and
other
meats have been commercialized. Such meat loaves or cheese loaves or other
food
loaves are commonly sliced and collected in groups in accordance with a
particular
weight requirement, the groups being packaged and sold at retail. The number
of
slices in a group may vary depending on the size and consistency of the food
loaf.
For some products, neatly aligned stacked sliced groups are preferred, while
for
other products the groups are shingled so that a purchaser can see a part of
every
slice through transparent packaging. For bacon or other food products of
variable
shape, the slicing and packaging problems are more challenging.
To properly allocate a sufficient number of slices or a sufficient overall
weight
of the group of slices, a weighing operation is undertaken in line with the
slicing
operation. This is particularly advantageous in the application of high-speed
slicers
employed in meat processing plants.
1
CA 02443748 2003-10-07
WO 02/084232 PCT/US01/40474
Some known high-speed food loaf slicing machines are disclosed, for
example, in U.S. Patent Nos. 5,566,600; 5,704,265; and 5,724,874.
It is known to weigh a stack of sliced product transported on a conveyor from
a slicing operation. Such a "check-weighing" operation is disclosed, for
example, in
U.S. Patent Nos. 3,846,958 and 4,065,911. However, in order to make such a
measurement on a dynamic-weigh basis, the prior art weigh scale methods
utilize an
optical or other external triggering device to activate the weigh scale for
choosing an
accurate sample period. The sample period is set on a fixed timing basis from
the
trigger of the triggering device.
The present inventor has recognized the desirability of providing a dynamic-
weigh checker for a conveyed series of products, or groups or stacks of
products,
which does not rely on an external triggering device to ascertain the correct
sample
period of the product, or groups or stacks of products, moving over the
associated
weigh scale.
SUMMARY OF THE INVENTION
The present invention provides a data acquisition and/or control device for a
conveyor weigh scale or "weigh scale control" and a method for operating a
conveyor weigh scale that automatically determines the correct sample period
for
accurately weighing product carried over the weigh scale. The present
invention
provides an algorithm for effective data acquisition and/or control associated
with
such a weighing operation. The weigh scale and control of the present
invention can
advantageously be configured to be combined with a high speed slicing
apparatus
2
CA 02443748 2003-10-07
WO 02/084232 PCT/US01/40474
and can give feedback on product output weight to be used as a control
parameter
for the slicing apparatus.
According to the invention, a conveyor weigh scale senses a dynamic weight
of product as it passes over the weigh scale. This dynamic weight can be
expressed
as a weight waveform of sensed weight over time as the product passes over the
weigh scale. An accurate weight reading for a moving product can be made only
during a brief sample period within the waveform, where the weight readings
are
substantially constant and representative of the static weight of the product.
Prior
known continuously moving product scales have used devices such as a laser
sensor or photosensitive components to detect when a product has entered the
scale and then uses fixed timing numbers to estimate the position of the
sample
period on the weight waveform to make a weight measurement.
The present invention provides a software algorithm for a weigh scale
associated with a continuously moving conveyor which is capable of positioning
the
sample period on each product weight waveform wherein the weight and speed of
the product passing over the scale does not affect the positioning of the
sample
period. The sample period is calculated mathematically using the slope
characteristics of the waveform.
The algorithm first looks for a minimum preselected positive amount of weight
deviation to activate or establish a "trigger". A first inflection point, that
point where
the rate of weight change over time dW/dt (the slope of the waveform), first
decreases; i.e., the waveform changes from a more positive slope to a less
positive
slope, is determined. The slope at the first inflection point is recorded and
defined as
3
CA 02443748 2003-10-07
WO 02/084232 PCT/US01/40474
the "maximum positive slope" dWl/dt. Once the maximum positive slope is found,
the algorithm begins recording weight samples at a sampling rate. The
algorithm
checks the weight waveform of sampled weights for a slope dW2/dt which is a
first
pre-selected percentage of the maximum slope but negative in slope value. The
first
pre-selected percentage is preferably about -50% of the maximum slope dWl/dt.
This point is determined as the "weight-off-scale" point.
When the weight-off-scale point is reached, then the algorithm will look
backward (reverse chronological order) through the saved data of weight
samples to
find another point which has a slope dW3/dt which is a second preselected
negative
percentage of the maximum positive slope dW1/dt. The second pre-selected
negative percentage is preferably about -10% of the maximum positive slope.
This
point is defined as the "end sample position." The end sample position is
experimentally known to be on or close to a flat part of the waveform
representing
substantially constant weight values.
With the end sample position known, a "start sample position" is determined
to fall within, or at the start of, the flat part of the waveform, such that
the weight
values within the sample period between the start and end sample positions are
substantially constant. The start sample position can be calculated as a first
point
having a predetermined slope on the waveform, reviewing the weight samples in
reverse chronological order from the end sample position; or can be
experimentally
determined to be within a preselected number of sample points in front of the
end
sample position. The weight values within the sample period are then averaged
to
determine a static weight value.
4
CA 02443748 2003-10-07
WO 02/084232 PCT/US01/40474
In an apparatus configured according to the invention no extra hardware cost
is required for a separate triggering device, separate from the weigh scale
device.
The apparatus of the invention requires no adjustment for weight changes of
the
product moving over the scale. According to the invention, no synchronization
is
required between a separate triggering device and the scale device. The
apparatus
of the invention achieves an increased operational reliability by eliminating
the need
for a separate triggering device.
Another advantage of the invention is the ability of the software algorithm to
compensate for product which may have a different orientation from stack to
stack,
either intentionally or accidentally. A narrow product, which moves onto the
weigh
scale with different orientations will produce longer and shorter weight
waveforms.
The algorithm positions the sample period from the trailing edge of the
waveform
which eliminates many orientation-based weighing problems experienced by
trigger
and fixed-timing weigh systems.
Other features and advantages of the present invention will become readily
apparent from the following detailed description and the accompanying
drawings.
BRIEF DESCRIPTION OF THE DRAWINGS
FIG. 1 is a perspective view of a high speed slicing machine which
incorporates the weigh scale control of the present invention;
FIG. 2 is a dynamic weight waveform sensed by the weigh scale; and
FIG. 3 is a flow chart for a computer control used in the weigh scale of the
slicing machine shown at FIG. 1.
5
CA 02443748 2007-05-15
DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS
While the present invention is susceptible of embodiment in various forms,
there is shown in the drawings and will hereinafter be described a presently
preferred embodiment, with the understanding that the present disclosure is to
be
considered as an exemplification of the invention, and is not intended to
limit the
invention to the specific embodiment illustrated.
FIG. 1 illustrates a versatile food loaf slicing machine 50 that can be used
to
carry out a preferred embodiment of the present invention. The slicing machine
50 is
of the type described in U.S. Patent 5,561,600; 5,704,265 and 5,724,874.
Slicing
machine 50 comprises a base 51 mounted upon four fixed pedestals or feet 52
and
having a housing or enclosure 53 surmounted by a top 58. Base 51 typically
affords
an enclosure for a computer 54, a low-voltage supply 55, a high voltage supply
56,
and a weigh checker or weigh scale 57.
Slicing machine 50, as seen in FIG. 1, includes a conveyor drive 61 that is
utilized to drive an output conveyor/classifier system 64. The conveyor
classifier
system 64 is described for example in U.S. Patents 5,704,265 or 5,499,719 and
is
responsive to the weigh scale 57 to direct products within a weight tolerance
to an
"accept" conveyor, and to direct out-of-weight tolerance products to a
"reject"
conveyor.
The slicing machine 50 of FIG. 1 further includes a computer display touch
screen 69 in a cabinet 67 that is pivotally mounted on and supported by a
support
68. Cabinet 67 serves as a support for a cycle start switch 71, a cycle stop
switch
6
CA 02443748 2003-10-07
WO 02/084232 PCT/US01/40474
72, and a loaf feed on/off switch 73. Switches 71- 73 and display/touch screen
69
are electrically connected to the computer 54 in the base 51.
The upper right-hand portion of the slicing machine 50, as seen in FIG. 1,
comprises a loaf feed mechanism 75 which, in machine 50, includes a manual
feed
on the far side of the machine and an automatic feed on the near side of the
machine.
As shown in FIG. 1, slicing machine 50 is ready for operation. There is a food
loaf 91 on tray 85, waiting to be loaded into loaf feed mechanism 75 on the
near-
side of machine 50. Machine 50 can produce a series of stacks 92 of food loaf
slices that are fed outwardly of the machine, in the direction of the arrow A
by
conveyor classifier system 64. Machine 50 can produce a series of stacks 93 of
food loaf slices that also move outwardly of the machine on its output
conveyor
system 64 in the direction of arrow A. Stack 92 as shown comprises slices from
a
rectangular loaf, and stack 93 as shown comprises slices from a round loaf.
Both
groups of slices can be overlapping, "shingled" groups of slices rather than
having
the illustrated stack configuration.
The weigh scale 57 is operatively connected to the conveyor 64 such that the
weigh scale 57 continuously senses the weight of the sliced product or product
groups appearing in succession on the scale. The weigh scale 57 in turn
outputs a
continuous succession of weight readings or samples at regular time intervals
to
define corresponding waveforms which are, in effect, dynamic weight measures
of
the product groups sensed by the scale over time. As an example, the weight
readings are sampled at a sample rate of 500 samples per second, filtered to
150
7
CA 02443748 2003-10-07
WO 02/084232 PCT/US01/40474
samples per second, with a conveyor speed of 107 product stacks per minute.
The
sampling rate can be adjusted to vary with the conveyor speed. The filtering
can be
done electronically or by software methods, preferably the latter. Filtering
is used in
part to compensate for inaccurate readings due to impact loads as product
first
appears over the weigh scale. The product stacks typically range from between
about 25 grams to about 1000 grams.
It should be noted that an analog weight signal from the weigh scale can be
sampled and filtered at the weigh scale 57 with the resulting sampled signal
communicated to the computer 54, or the analog signal from the weigh scale can
be
sampled and filtered by the computer 54.
Analyzing the weight readings over time (the waveform) of a single product
group of slices, can be used to ensure sufficient weight portioning for each
group.
However, since the weight and velocity of the product moving over the weigh
scale
affects the dynamic weight reading, a method must be used to ascertain an
appropriate sample period during each waveform which passes over the weigh
scale
to obtain a measurement which accurately represents a static weight of the
product.
FIG. 2 illustrates one such waveform 100. The waveform 100 is a
representation of a filtered and sampled continuous signal from the scale 57.
The
weigh scale control of the invention uses the sampled weight readings
represented
by the waveform to determine a sample period on the waveform which corresponds
to a static weight of the product being weighted. The weigh scale control
includes a
software algorithm. The algorithm calculations, comparisons, and data
recording of
8
CA 02443748 2003-10-07
WO 02/084232 PCT/US01/40474
the weigh scale control can be done in part or in whole by the weigh scale 57
or in
part or in whole by the machine control computer 54.
The steps of the algorithm are set forth in FIG. 3. The algorithm first looks
for
a minimum preselected positive amount of weight deviation to activate or
establish a
"trigger" position 104. The actual trigger position 104 on the waveform is not
critical
to the algorithm as long as the trigger position on the waveform occurs before
a first
inflection point 106. The first inflection point 106 is that point where the
rate of
weight change over time dW/dt (the slope of the waveform), first decreases;
i.e., the
waveform changes from a more positive slope to a less positive slope.
Once the trigger position 104 is located, the algorithm calculates the slope
of
the weight signal until the first inflection point 106 is determined. This
slope is
recorded and defined as the "maximum positive slope" dW1/dt. Once the maximum
positive slope is found, the algorithm begins recording weight samples at a
sampling
rate. The samples can be recorded in a RAM within the control computer 54. The
algorithm checks the weight waveform 100 of sampled weights for a slope dW2/dt
which is a first pre-selected percentage of the maximum slope but negative in
slope
value. The first pre-selected percentage is preferably about -50% of the
maximum
slope dWl/dt. This point is determined as the "weight-off-scale" point 110.
When
the weight-off-scale point is reached, then the algorithm will look backward
(reverse
chronological order) through the saved data of weight samples to find another
point
which has a slope dW3/dt which is a second preselected negative percentage of
the
maximum positive slope dW1/dt. The second pre-selected negative percentage is
preferably about -10% of the maximum positive slope. This point is defined as
the
9
CA 02443748 2003-10-07
WO 02/084232 PCT/US01/40474
"end sample position" 114. With the end sample position known, a "start sample
position" 118 is determined.
A flat region 119 of the waveform extends backward, in reverse chronological
order, from the end sample position 114, wherein the weight samples have a
substantially constant value. The start sample position 118 can be
experimentally
determined by studying a waveform 100 and selecting an appropriate number of
sample points in front of the end sample position 114 such that the start
sample
position falls within, or at the start of, the flat region 119, preferably as
far in front of
the end sample position 114 as possible, to utilize a maximum amount of
samples
within the flat region to calculate an overall more accurate weight reading.
Alternatively, a start sample position 118' can be calculated by reviewing
waveform slopes dW/dt, i.e., the rate of sampled weight change over time at
each
sample point, from the end sample position 114 backward, in reverse
chronological
order. A calculated start sample position 118' is defined when the calculated
slope
dW4/dt first reaches + or - 10% of the maximum positive slope dW1/dt. This
calculated method would ensure that the entire flat region of the waveform was
sampled to determine the static weight reading.
The static or average weight reading for the dynamically weighed product is
calculated as an average from the recorded weight samples between the start
and
end sample positions 118 (or 118'), 114.
The average weight reading can be used to reject out-of-tolerance product or
products, or can be used as a feedback control for the slicer to slice thicker
or
thinner slices or to include more or less slices for each group of slices. A
slicer
CA 02443748 2007-05-15
control is described in U.S. Patents 3,846,958 or 5,109,936.
The invention recognizes that a reason a percentage of the maximum slope is
used to determine the weight-off-scale point is because different weight
products will
cause different amounts of waveform distortion at the leading edge of the
weight
signal. Greater values of weight will create greater values of distortion. If
a fixed
number of samples after the trigger was used to determine the vveight-off-
scale
position, then the distortion caused by larger weights would cause the weight
sample
period to be positioned improperly on the waveform. Heavier weights cause
premature negative slopes occurring before the sample period which are not
really
the weight moving off the weigh scale. The use of any negative slope to
identify the
weight-off-scale position would most likely yield inaccurately calculated
static weight
results.
The percentage calculation used to determine the weight-off-scale position on
the waveform is empirically derived based on an analysis of the product being
weighed. It has been empirically determined that this percentage should be
about -
50 percent of the maximum positive slope. Likewise, the percentage of maximum
positive slope which determines the end sample position is empirically
determined
and has been found to advantageously be about -10 percent.
According to the invention, a dynamic weigh scale algorithm provides a
method of automatically determining a corrected weight measurement from a
weight
waveform without the need for a separate laser or optical triggering device
with a
11
CA 02443748 2003-10-07
WO 02/084232 PCT/US01/40474
timing calculation, in order to extract the accurate weight measurement of the
product or the product group.
Although the invention is described in regards to weighing a group of slices
to
be shingled or stacked, the invention can also be applied to weighing a series
of
single slices. Also, the weigh scale control can be part of a weigh scale
electronics
and software configuration or can be incorporated into the machine computer or
controls without departing from the invention.
Although the present invention has been described in substantial detail with
reference to one or more specific embodiments, those of skill in the art will
recognize
that changes may be made thereto without departing from the scope and spirit
of the
invention as set forth in the appended claims.
12
