JPH0454418A - Standard deviation calculating device and weight sorting machine - Google Patents

Abstract

PURPOSE:To accurately know and record an actual weight distribution by separating a standard deviation which is found at the time of test weighing from the standard deviation of a weighed value of an article. CONSTITUTION:In test mode, one optional article among plural articles to be weighed in actual operation is selected and run several times on a weighing conveyor 2 under the same weighing conditions with the actual operation to find several weighed value standard deviations (t) by a CPU 8; and they are stored in a memory 12 and displayed on a display part 11. Then a CPU 8 finds the standard deviation of the weight of the articles weighed in the actual operation according to respective weighed values of plural articles weighed in the actual operation and the standard deviation found by the test weighing. The standard deviation found at the time of the test weighing differ with the kinds of articles, so respective standard deviations are stored in the memory 12 by the kinds of article while given article numbers and then a desired standard deviation is read out and used as data for calculating the standard deviation value of actual article weight without carrying out test weighing.

Description

Detailed Description of the Invention (Industrial Application Field) The present invention provides a standard deviation calculation device for calculating the standard deviation σt of the weight of an article to be weighed which does not include at least a weighing error of a weighing machine; The present invention relates to a weight sorting machine that sorts weighed articles that are lighter or heavier than a boundary weight value with a predetermined probability so as not to be sorted into the heavy side or the light side due to a weighing error in a weight section.

[Conventional technology]

Conventionally, when calculating the standard deviation in order to find out the dispersion of the weight of objects to be weighed weighed by a weighing machine, etc., the standard deviation calculation device is used to calculate the weight of each weighed object sent from the weighing machine connected to the standard deviation calculation device. A standard deviation is calculated based on a plurality of weighed values of heavy articles, and this is used as the standard deviation of the variation in weight of the heavy articles to be weighed.

In addition, for example, items are filled into containers using a filling machine, and the containers containing the items (hereinafter referred to as filled items) are weighed using a weight sorter, and the appropriate amount of items with a boundary weight value (standard weight #1) or more is determined. There are times when products are sorted into lightweight products that are less than the weight value.

In this case, the filled article may be weighed by the weighing section of the weighing machine, or it may be weighed at a higher weight value than the boundary weight value even though it is actually a light article that is less than the boundary weight value due to a weighing error in the weighing section. There is a need to prevent such products from being sorted into appropriate quantity products. For this reason, the weighing error when weighing the weighing unit or the item is estimated by the operator or by intuition and experience, and a control sorting weight value that is slightly thicker than the boundary weight value is determined. Set on weight sorting machine. Thereby, the weight sorter sorts the filled articles based on this management sorting weight value.

[Invention or problem to be solved]

However, although the conventional standard deviation calculation device described above can calculate the standard deviation of multiple weight values sent from a weighing machine, these weight values include at least the weighing error of the weighing machine. Therefore, the weight value does not correspond to the actual weight of the item, and therefore even if you calculate the standard deviation of the variation in the weight value, it is not possible to accurately measure the variation in the weight value of the weighed item. The problem is that it is impossible to know the standard deviation.

In addition, with conventional weight sorting machines, the operator estimates the degree of variation in weighing errors in the weighing section based on intuition and experience, and then determines the sorting weight value for control based on this estimated range of variation. However, there are cases where the management sorting weight value is set too large than the boundary weight value or too close to the boundary weight value. If the sorting weight value is set too large than the boundary weight value, it is possible to prevent light products that are lighter than the boundary weight value from being sorted into moderate-weight products that are heavier than the boundary weight value, but the weight or boundary weight value is increased by that much. This increases the number of items that are in a suitable quantity between the weight value and the weight value and is classified as lightweight items that are lighter than the boundary weight value, which becomes uneconomical.
There is a problem of increasing the cost of goods for the manufacturer by filling a large amount of goods into a container with a biased weight.If the weight value of sorting gets too close to the boundary weight value, light weight goods will be sorted into moderate quantity goods. There is a problem that the probability of 1o becomes high.

The present invention aims to provide a standard deviation calculating device and a weight sorting machine that solve the above-mentioned conventional problems.

[Means and effects for solving the problem] 5 The first invention is such that, during test weighing before starting actual operation, the weight of at least approximately equal to the average weight of the weighted articles to be weighed is set in the same manner as during actual operation. When the weight is repeatedly weighed using a weighing machine under these conditions, the first calculating section calculates the standard deviation σt of the weighed value based on the weighing error of the weighing machine etc. obtained thereby. The second calculation unit calculates the variation in the weight of each weighed article itself and at least the weight of the weighing machine based on a plurality of weight values obtained when the weighing machine weighs the weighed article during actual operation. Calculate the standard deviation σo of the weight value, which varies due to error. Then, the third calculation section calculates the first and first! Using the standard deviations σt and σo calculated by the s2 calculation unit (σo2 t2 )
1/2 to calculate the standard deviation σt of the weight of the article to be weighed, which is at least -σ and does not include the weighing error of the weighing machine.

The second invention provides that, during test weighing before starting actual operation, if a heavy object that is at least approximately equal to the average weight of the heavy objects to be weighed is repeatedly weighed using a weighing machine under the same setting conditions as during actual operation, The standard deviation σt of the weight values, which varies due to at least the weighing error of the weighing unit, is calculated based on each weight value obtained by the first calculation unit. The second calculation unit multiplies the standard deviation σt by a coefficient k, calculates the algebraic sum of this multiplied value, σt, and the boundary weight value W8 for weight-sorting the weighed articles to calculate a management sorting weight value. 9 The sorting section compares whether or not the weight value of the object to be weighed weighed by the weighing section during actual operation is greater than the management sorting weight value, and at least determines the weight of the weighing section. To sort weighed articles that are lighter or heavier than a boundary weight value due to an error so as not to be distributed to the heavy side or the light side with sufficient probability.

〔Example〕

First embodiment of the standard deviation calculation device of the present invention
This will be explained with reference to FIGS. 4 to 4. Figure tIS4 shows the general configuration of this weight sorting machine, and a standard deviation calculating device is included in a microprocessor (hereinafter referred to as CPU) that controls the weight sorting machine. In Figure 4, 1
is the weight of the article to be weighed on the feeding conveyor (hereinafter referred to as the article)
The weight of the article sent to the weighing conveyor 2 is detected by the load cell 4 and transmitted to the CPU 8 via the amplifier 5, A/D converter 6, and input port door. Here, the standard deviation of the weight of the article is calculated. When you want to know the calculated standard deviation, if you operate the key 9 having operation keys, numeric keys, etc., the display unit 1 connected to the CPU 8 via the output board 10
1 or the standard deviation is displayed according to instructions from the CPU 8. Note that 12 is a memory 12 having ROM, RAM, etc.
Reference numeral 3 denotes a delivery conveyor for delivering the articles conveyed from the weighing conveyor 2.

Now, the standard deviation σt of multiple weight values obtained by repeatedly feeding one article onto the weighing conveyor 2 and weighing it.
is the weighing error of the weighing section (hereinafter referred to as "weighing section") including the load cell 4, the weighing conveyor 2, etc., the setting conditions (weighing conditions) such as the conveyance speed of the weighing conveyor 2, and the shape and properties of the article. Determined by the weighing error corresponding to In addition, if a number of articles having approximately the same shape and properties as the above article are repeatedly weighed under the same weighing conditions as the above, and the standard deviation σt is calculated, this will be calculated even if the weight is slightly different from the above article. The standard deviation σt′ is σ
It is approximately the same as t.

On the other hand, the standard deviation σo of a plurality of weight values obtained by weighing a plurality of articles having substantially the same shape, properties, and weight variations as the above-mentioned article using a weight sorter under the same weighing conditions is It is determined by the standard deviation σt based on the weighing error, etc., and the standard deviation σt of the weight variation of the article itself, which has no correlation with this, σo= σt 2 +σw ′ (1) Therefore, σt; σ0 ′ −σt 2 ...(2).

Therefore, at the time of test weighing before actual operation, σt
is calculated and stored in the memory 12, and this σt and the standard deviation σo obtained from the weighed values of a plurality of articles obtained during actual operation that have approximately the same shape, properties, and vary in weight, are expressed as (
By substituting into equation 2), it is possible to calculate the standard deviation σ1 of the weight variation of the article itself.

To explain this specifically, first, key 9 in FIG.
to enter test mode. Next, select any one item from among the plurality of items to be weighed during the actual operation, and feed this item onto the weighing conveyor 2 n times under the same weighing conditions (same conveyance speed) as during the actual operation. Obtained n weight values W, , W□
2, W+3,... The standard deviation σt of Wln is determined by the following (4). However, W, □ is i=1.2,...
・The weight value, W, of the nth item weighed is W11 to W
This is the average value of the weight values up to in.

The program of the CPU 8 that calculates this equation (4) will be explained with reference to the flowchart shown in FIG. The first calculation unit according to claim 1 executes this flowchart. First, every time the load cell 4 (weighing unit) weighs an article, the weight value W8. are accumulated to calculate the cumulative value WA, and at the same time, each time the weight is calculated, w,,,'' is calculated and the cumulative value W1 is calculated (steps 100 to 106).Then, the number of weighed articles is counted. In step 108), it is determined whether or not the count value C1 is n (n is an arbitrary number of times of weighing set before the test).Step 110) If the result in step 110 is NO, the count value C1 is n. Repeat steps 100 to 110, and when the count value C1 reaches n and step 110 becomes YES, set WB1 as P, W A 1 / Cr as Q, and C, as R, and store them in the memory 12. The standard deviation σt is stored (step 112). W, = ΣLl + wA, = ΣLi, c, = n. =J'T〒-100nia1 (Step 114), σt is displayed on the display section 1.
1 and store σ1 in the memory 12 (steps 116 and 118). At this time, σt2 is also calculated and stored in the memory 12.

Next, based on the weight value Wk of each of the multiple items weighed during actual operation and σt determined by test weighing, one item!
The program of the CPU 8 for calculating the standard deviation σW of the weight of articles weighed during the I operation will be explained with reference to the flowchart shown in FIG. First, when you set it to production mode, the CPU
8 is the weight value Wk (Wk
Wk'' is calculated from the weight value of the Nth item weighed (=1.2,...), and this is accumulated to obtain the cumulative value W.
B3 is calculated, and the cumulative value WA3 of Wk is also calculated sequentially (steps 200 to 206). Then, the number of weighed articles is counted (step 208), and it is determined whether the count value C3 is N (N is an arbitrary number of articles set before actual operation) (step 210). These steps 200 to 210 are the second calculation unit according to claim 1. When step 210 is NO, steps 200 to 210 are repeated until count value C3 becomes N, and step 210 is YES when count value C3 becomes N.
When σo2=(Wa:+-(WA7/C3)
) /c: Calculate l (step 212). However, Wl]3=Esu, WA3=En, wk, C:I=N. Then, this σo′ and σt2 obtained during test weighing
Substitute and into equation (2) to calculate σt (Step 2)
4) Display on the display unit 11 (step 216)
. This step 2]4 is the third calculating section according to claim 1. In this way, it is possible to calculate the standard deviation σ1 of the variation in the actual weight of the articles weighed during actual operation.

Note that the standard deviation σt determined during test weighing differs depending on the type of article, so the standard deviation σt determined by test weighing each type of article sorted by the weight sorter is If you attach a product number to the item and store it in the memory 12,
When weighing articles of the type stored in the memory 12 during actual operation, you can specify the article number corresponding to the type by operating the key 9 without performing test weighing each time before actual operation. By calling the function, it can be used as data for calculating σ1.

Also, n, WA, (
By storing xΣ” Lt) and Wa+(-1m Wli”) in the memory 12 as a pair with the product number, when test weighing the same product type for the second time, any one of the products of the same type can be For example, m(C,
= m) Weighing with the weighing unit and steps 100 to 110
By executing the flowchart, WA2(=7:,W,i), W,2(=,XT,L
i”) and calculate the first and second (n + m)
By using the test data of n times, it is possible to calculate σt1 with higher precision than σt calculated from the test data of n times.

That is, during the second test weighing, the setting conditions for the conveyance speed of the weighing conveyor 2, etc. (weighing conditions) are the same as the first test, and the weight and shape of the articles used in the first test are , one article with almost the same properties is weighed m times, and the weight is only W21, W2□, ..., W2, then the standard deviation σ1 is σ = A / (n + m )
...(5) can be calculated.

Tatami, A=(Wli w,)2+ (Wl2 Wl)2+
...+(Ln-wl)2+(Lt-L)"+(W2
2 w2)2+−−−+(W2.W2)2=(W
li" + stomach, 22+...W, n")+ (W2.2
+ W222+ #@...+tt2-") -((Lt +'W+2 +...Wln)2/
n + (Ll+ W22 +...+W2-)
2/m) ・・・(
6) Note that W is the average value of the weight values W, , ~W, n, and W2 is the weight value W21 to W2. is the average value of

The calculations of equations (5) and (6) are performed by the CPU 8 or by executing a program, and a flowchart thereof is shown in FIG. Step 3 of this flowchart in Figure 3
Steps 00 to 310 are equivalent to steps 100 to 110 in FIG. 1, and their explanation will be omitted. Step 312 includes the first
P, Q, which are stored in the memory 12 by the second test,
In step 302, Cz (=m) is added to R, respectively, and these added values are placed as new P, Q, and R, and in step 314, the P, Q, and R are stored in the memory 2. Steps 316 to 320 in FIG. 3 are intermediate to steps 14 to 11B in FIG. 1, respectively, so their explanation will be omitted. σ1 obtained in this way is used to calculate σ8 when executing each step in FIG. 2 during actual operation.

In addition, in step 314, P= Σ Wl, 2 + Σ W2 Q = (,X:4+ +)"/n + (7":,
w-;)2/.

R=n+m is stored in the memory 12 instead of the first test weight, so the second P, Q, and R are stored in the memory 12 for the third and subsequent test weighings. You can also use it in the same way.

Next, during test weighing, two articles of the same type 41
A case will be described in which σt is calculated using the mouth. Note that the weight, shape, and properties of these two articles are the same as those during the main GJ operation, and the weighing conditions are the same as during the actual operation. First, the CPU 8 is instructed to perform a test using two items by operating the key 9. Then, when weighing two articles in sequence, for example, n times during test weighing, the article to be weighed is determined each time.

Then, by test weighing, the weight values obtained by weighing items A and N times in order are W, 3, W2□, W, 2, W22, etc.
When W, n, and W2n, the standard deviation σt is as follows: σt=DouA, Fu]Dou−・・・(7) Tatashi A n” (L+ w,)2+ *++ +
(Ln-L)2+ (W21- W2)"+ m+*
+ (Ln-L)2=J,"++*+ +W,n"+
W2. ”+##-+W,,.

−((w++”・”+’i/+n)”+
(W 2. + +... Ln) 2) -- is determined. Then, σ1 is calculated using this σt. However, either two articles were test weighed or three or more articles could be used and similarly test weighed.

A weight sorter in a second embodiment will be explained with reference to FIGS. 5 and 6. However, the structure of the weight sorter of this embodiment includes the structure of the weight sorter shown in FIGS. 1, 3, and 4, and a detailed explanation of that portion will be omitted.

The standard weight value of the items weighed by this weight sorter,
The nominal value of the content (boundary weight value) indicated on the product or product.
When sorting into appropriate weight products of WN or above and light weight products of less than WN, the management sorting weight value Ws shown in Fig. 5 is determined by taking into account weighing errors caused by the weighing conditions of the weight sorter. . Note that the standard deviation σt of the weighing value, which varies due to weighing errors caused by the weighing section and weighing conditions, is determined by multiple measurements obtained by weighing the same article multiple times, as explained in the first example. It can be obtained by calculating 5 standard deviations σt based on the weighted values. Therefore, W5 is W5 = Wx + k
・(rt...”(8) However, k is any positive real number, for example], 1.2.0, 3.0, 3.2
,...etc.

It is possible to make a decision. The second calculation unit according to claim 2 is a means for calculating the management sorting weight value Ws according to the equation (8).

Now, for example, set k = 3.0, determine Ws, and then send the CPU 8 to sort the weighed articles based on this W5.
If you set it to
Even if there are only 5, the percentage of items that are more than the actual weight or WN among the items sorted into appropriate quantities (good item rate) is
99.7%, and on the other hand, the percentage of items that are sorted as actual weight, lightweight items of less than W8, or suitable quantities is 0.3%. In this way, if the value of k is determined in advance for each item by taking other economic conditions into account, Ws can be determined from equation (8) using σt obtained by test weighing. . Therefore,
There is no need to estimate the total tuff difference of the weighing machine for the articles to be weighed by the operator or his/her own judgment as in the past. Here, during actual operation, the sorting unit according to claim 2 is a means for comparing the weight value of the weighed articles and W3 to sort the articles.

In addition, in a three-way sorting weight sorter, the upper limit deviation value is WU and the lower limit deviation value is W with respect to the reference weight value WN, and if the actual weight of the article is WX, then (I) Wr > WuuN (= WN +WL+ ), excessive prize (U) (WN WL = )WLL-≦W
When 1≦Wuus (=WN +wL+), an appropriate amount (
Illr) W, <WLLN (=WN-wL>
CI) TokiR The case of determining the management sorting weight values WUUs and WLLs shown in FIG. 6 for sorting bulk products and articles will be explained.

Now, if the cpus of the weight sorter is set to k corresponding to the article, and σ1 has already been calculated by test weighing, the nominal upper limit weight will be changed due to variations in weight values due to weighing errors of the weight sorter. Value (upper boundary weight value) WIJU
In order to prevent items from being sorted into items with a weight greater than o or appropriate quantity items with a predetermined probability, when the upper limit deviation value Wu is set by operating the key 9, the CPU 8 selects W uus ” W s + W U − k・σ1・
...The upper limit sorting weight value Wu for management according to the formula (9)
ljs is calculated and stored in the memory 12.

Then, when the lower limit deviation value WL is set by operating the key 9 in order to prevent items with a weight smaller than the nominal lower limit weight value (lower limit boundary weight value) WLLN from being sorted into items with a predetermined probability of being classified as appropriate quantity items, CPU8 executes WLLS = WN WL + k ・(rt ・
...The lower limit screening weight value W for management according to the formula (10)
LL5 is calculated and stored in the memory 12.

In this way, according to this weight sorting machine, weighed articles are sorted based on the upper and lower boundary weight values WUIJIJ t LLH for management, so that it is possible to select appropriate quantity and light articles with a predetermined probability. You can try to avoid being selected.

(Effect of the invention) According to the first invention, the standard deviation σo of the weighing values of a plurality of articles obtained by weighing the articles with the weighing machine during actual operation is used to determine the weight of the weighing machine, etc. obtained in advance during test weighing. By separating the standard deviation σ1 of the weight value which varies due to the weight error,
The standard deviation of the weight of the weighed item itself (σo' - σt!
) 172, this standard deviation can be used to accurately know and record the actual weight distribution of the articles weighed by the weighing machine. This has the effect of allowing accurate weight management of articles. Another advantage is that it is possible to grasp the accurate weight distribution of articles to be shipped to the market.

According to the second invention, the standard deviation σ1 of a plurality of weighing values that vary due to weighing errors of a weighing machine etc. by test weighing is
Calculated by the calculation unit, σ is added to the management sorting weight value WN+.
t can be determined. As a result, for example, as explained in the conventional example, when sorting the weighed articles into appropriate weight items that are heavier than the boundary weight value WN and lightweight items that are lighter than the boundary weight value WN, the number of items that are sorted into light weight items or appropriate weight items is reduced. Therefore, it is necessary to increase the coefficient k, and whether it is necessary to decrease the coefficient k in order to reduce the number of items to be sorted into appropriate quantities or lightweight products, or whether it is necessary to make the coefficient k small in order to reduce the number of items sorted to the appropriate quantity or light weight side. By determining a satisfactory coefficient k for each article, it is possible to carry out economically extremely effective weight sorting. Also, when sorting the weighed items into appropriate weight items that are heavier than the boundary weight value W8 and appropriate weight items that are lighter, economical control can be achieved in the same manner as above by determining σ1 for the management sorting weight value W, -. This has the effect of making it possible to carry out extremely effective weight sorting.

[Brief explanation of drawings]

FIG. 1 is a flowchart of a first embodiment of the first calculation section of the standard deviation calculation device according to the present invention, and FIG.
3 is a flowchart of another embodiment of the first computing section according to the same invention, FIG. 4 is a diagram showing a schematic configuration of the same embodiment, and FIGS. 5 and 6 are Two-way and three-way weight sorter according to the invention! 1! It is a diagram showing the management sorting weight value of Example 2. 2...Weighing conveyor, 4...Load cell, 8...
...CPU, 11...Display unit, 12...Memory.

Claims (2)

[Claims] (1) During test weighing, based on multiple weight values obtained when a heavy object that is at least approximately equal to the average weight of the heavy objects to be weighed is repeatedly weighed using a weighing machine under the same setting conditions as during actual operation. At least a first calculation unit that calculates the standard deviation σ_t of weight values that vary due to weighing errors of the weighing machine; and a first calculation unit that weighs each of the weighed articles that are sequentially supplied to the weighing machine during actual operation; a second calculation unit that calculates a standard deviation σ_o of weight values that varies due to variations in the weight of each weighted article itself and at least a weighing error of the weighing machine based on the plurality of weight values obtained; Using the standard deviations σ_t and σ_o, (σ_o^2−σ
_t^2)^1^/^2 and calculates at least the standard deviation σ_w of the weight of the article to be weighed that does not include the weighing error of the weighing machine. Standard deviation calculation device. (2) A weighing section that weighs the weight of the heavy articles to be weighed, and a weighing section that weighs a heavy article at least approximately equal to the average weight of the weighed articles at the time of test weighing under the same setting conditions as during actual operation. a first calculation section that calculates the standard deviation σ_t of the weight value which varies due to the weighing error of the weighing section based on each weight value obtained by repeatedly weighing the weight, and a coefficient k for the standard deviation σ_t; a second calculation unit that calculates a sorting weight value for management by multiplying this multiplication value kσ_t and the boundary weight value W_N for weight-sorting the weighed articles, and a second calculation unit that calculates a sorting weight value for management; 1. A weight sorting machine comprising: a sorting section that compares the weight value of the article to be weighed weighed by the above-mentioned management sorting weight value.