JPH0758218B2 - Electronic counting scale - Google Patents

Description

Description: TECHNICAL FIELD The present invention relates to an electronic counting scale that calculates and displays the number of samples by dividing the weight of an unknown number of samples by the unit weight of the sample to be measured. .

<Prior Art> In an electronic counting scale, generally, the estimated value of the unit weight (weight per piece) of a sample has a great influence on the accuracy of the counting result. Therefore, in the conventional electronic counting scale, various ideas have been made for this unit weight estimation method, but basically, it can be classified into the following two methods.

One is a method in which the number of samples such as 10, 40, and 100 is limited in advance, and the load value when the sample is placed on a plate is divided by the known number.

The other one is to first place a small number of known samples such as 5 on the plate and calculate the unit weight in the same manner, then add the samples as appropriate, and calculate each time before adding. Estimate the total number on the plate by dividing the additional weight by the unit weight to find the additional number and adding them sequentially, divide the total weight on the plate by the estimated number,
This is a method of updating the unit weight.

<Problems to be Solved by the Invention> According to the former conventional method, when the sampling number for unit weight estimation becomes large, counting the number is troublesome,
Further, there is a defect that counting errors may occur.

The latter method has a problem in that it is not guaranteed that all the estimations for each addition of samples are correct, and eventually that the estimation knowledge of the finally obtained unit weight is also correct.

Here, in estimating the unit weight, it is needless to say that a quantitative and clear treatment cannot be performed unless the coefficient of variation of the weight of the sample to be counted is used. In the method, a method has been proposed in which an estimated value of the unit weight is guaranteed by obtaining an allowable additional number that should limit the additional number based on the range of 3σ with the sample weight as a normal distribution (JP-A-60- No. 31023).

By the way, when the coefficient of variation of the sample is, for example, 2 to 3% or more, the allowable additional number based on the range of 3σ of the normal distribution is
There is a problem that the efficiency of sampling work for estimating the unit weight is reduced because the number cannot be set to be too large.

The present invention has been made in view of this point, and in the latter method described above, based on the stochastic treatment of the counting error, it is possible to give a quantitatively clear counting error guarantee, and The conventional 3σ in giving a degree of assurance
It is an object of the present invention to provide an electronic counting scale capable of increasing the allowable additional number as compared with the method based on the range.

<Means for Solving Problems> A structure for achieving the above object will be described with reference to a basic conceptual diagram shown in FIG. 1. The present invention detects a load placed on a plate. Load detector a; 2 of the sample to be counted
First and second unit weight memories b and c for storing the unit weight estimated values μ _A and μ _B ; additional weights based on the values detected by the load detector a
W _i is the content of the first and second unit weight memories b and c μ _A
And the number of samples added on the dish divided by μ _B k _A and
Additional number calculation means d for estimating k _B ; Variation coefficient setting means e for setting the variation coefficient CV of the sample; Setting content CV and calculation at the time of the previous ((i-1) th) addition by the additional number calculation means d Using the result N _i-1 , a preset probability P
As described above, the allowable additional number calculating means f for calculating the maximum allowable number K by which the next additional number can be estimated based on the counting error within the predetermined number; the calculated allowable number K and this time by the additional number calculating means ( The second comparing means h for comparing the (i-th) estimation results k _A and k _B ; the estimation results up to the previous (i-1) -th time by the additional number calculating means d and the additional weights W _i corresponding to the respective estimation results Sampling data memory q for storing data, and contents of first and second single weight memories b and c by estimating two types of unit weights μ _A and μ _B based on different data in the sampling data memory q Unit weight calculation means r;

Then, based on the comparison result of the first and second comparing means g and h, the estimation result k _A or k _{B of} this time (i-th time)
Is equal to or less than the allowable number K and k _A and k _B are equal to each other, the estimated result k _A (or k _B ) and the additional weight W _i are added to the contents of the sampling data memory q (k _A →
N _i ), i + 1, the unit weight calculation means r estimates and updates the unit weights μ _A and μ _B so as to be added.

<Principle and Action of Invention> In general, the weight w when n samples are taken out from a population N (μ, σ ² ) having a normal distribution of mean weight μ and standard deviation σ is Becomes Where ξ ₁ is a random variable and its density function is Is.

On the other hand, when the unit weight of the sample is estimated from the weight w, from Equation (1), Becomes

Next, if k samples are taken out and their weight is W, Becomes Here, ξ ₂ is a random variable independent of ξ ₁ .

Estimating the number using Becomes Therefore, the counting error ξ at this time is Becomes Since ξ is expressed as a linear combination of ξ ₁ and ξ ₂ , from the known theorem, Becomes That is, ξ is a normal random variable, the average value of ξ is 0, and the standard deviation of ξ is Is.

Therefore, when the number is estimated as an integer by rounding to four, the probability P that is correctly obtained, that is, the probability P that k can be estimated with a count error of 0 is Becomes here, And put It can be expressed as.

Also, the probability P has a counting error of +1 or -1
₁ is And the probability P ₂ with ± 2 or more counting errors is Is.

Here, the allowable additional number based on the range of 3σ of the conventional normal distribution corresponds to the value obtained by calculating k in equation (12) with β = 3. In equation (12), if the fluctuation coefficient is CV, μ
/ Σ = 1 / CV.

By the way, assuming that ± 1 counting error is allowed, 3β is the main parameter in the probabilistic discussion instead of β. Speaking to the extreme, the place where β ≧ 3 in the past is set to 3β ≧ 3, that is, β ≧ 1, so that the allowable additional number can be further increased. However, if this is adopted as it is, it cannot be considered as a guarantee that the counting error of ± 1 is always included, and therefore it is possible to prevent the counting error of ± 1 from being generated by some other measure. Will be needed.

Therefore, the present invention has a
Two types of unit weights μ _A and μ _B are estimated from the sampling data up to the first time, that is, the additional weight W _i and the estimated value N _i thereof.

That is, by estimating the n-th additional number with both the unit weight estimation values μ _A and μ _B , and adding the condition that the results k _A and k _B match, the above ± 1 counting error is added. This is a preventive measure. This will be described below.

Now, when the allowable additional number K which is recognized up to ± 1 counting error is obtained, the probability P that the number is estimated to be 0 with counting error when adding below the number is Is.

Add counted by mu _A error -1; where, p _0; mu probability additional number by counting error 0 by _A is estimated p _1; mu probability additional number by counting error +1 by _A is estimated p _-1 Probability that the number is estimated q ₀ ; Addition with counting error 0 by μ _B Probability that the number is estimated q ₁ ; Addition error with counting error +1 by μ _B q ₋₁ ; Count error by μ _B − It is the probability that the additional number is estimated by 1.

Given that the results k _A and k _B estimated by μ _A and μ _B are equal to each other, Becomes p ₁ , p _-1 , q ₁ , q _-1 is sufficiently smaller than p ₀ , q ₀ ,
Therefore, if the allowable number K that allows a counting error of ± 1 is found and k _A and k _B are equal when adding a sample of less than that number, it can be considered that the counting error is substantially zero. .

By the way, equation (17) is Therefore, P is a monotonic increase with respect to p ₀ and q ₀ , and if μ _A and μ _B are calculated so that p ₀ ≦ q ₀ , Becomes

When p ₀ and β corresponding to 3σ (β = 3) of the normal distribution are obtained, Therefore, p ₀ ≧ 0.9315 …… (21), and β ≧ 1.83 …… (22) is enough for that. On the contrary, since 3β ≧ 5.49, ±
The probability of occurrence of two or more counting errors is virtually 0 (it cannot be larger than 10 ^{-5 according} to the normal distribution table). That is, k that satisfies equation (22) is obtained from equation (12), and the allowable additional number K is calculated by rounding down the fractions, etc., and the additional number estimation result by μ _A and μ _B is ± 2 There is no probability that the above counting error is included, and if the additional number estimated values k _A and k _B by μ _A and μ _B are equal to each other when adding the sample below the allowable additional number K, ± 1 counting error will also occur. Can be guaranteed not to.

<Example> An example of the present invention will be described below with reference to the drawings.

FIG. 2 is a block diagram showing the configuration of the embodiment of the present invention.

The load detection unit 1 includes a load sensor and AD which are engaged with the dish 1a.
It has a built-in converter etc. and can output digital data corresponding to the load on the plate 1a.

The data from the load detector 1 is CPU21, ROM22 and RAM2.
It is incorporated into the microcomputer 2 equipped with 3 etc. moment by moment. The microcomputer 2 is provided with a keyboard 3 mainly including a numeric keypad for inputting the coefficient of variation CV and the like.
Further, a display 4 for digitally displaying the number of samples, and an indicator lamp 5 for informing whether or not an additional number k _{B, which} will be described later, is within a predetermined range of an allowable number K or less are connected.

In addition to the work area, RAM23 has two types of unit weight μ _A ,
Areas as unit weight memories A and B for storing μ _B respectively, area for total number memory for sequentially adding and storing calculated values of additional number, area for storing all sample weights on the plate, and last addition An area or the like is set as a sampling data storage memory for storing the weight and the estimated value of the additional number.

FIG. 3 is a flow chart showing the contents of the unit weight estimation routine of the program written in the ROM 22, and the operation will be described below with reference to this figure.

Prior to the measurement, the coefficient of variation CV of the individual weight of the sample to be measured is entered from the keyboard 3 (ST1). This coefficient of variation CV is obtained by extracting an appropriate m number of samples, and using the standard deviation σ and the average weight of the individual weights w _i , It can be calculated by and the result is keyboard 3
You can enter from.

Next, K ₀ pre-set samples, for example 5,
Place only on dish 1a (ST2). From the load data at this time, the sample weight W on the dish 1a is determined by a known method and stored in the RAM 23 (ST3). Then this weight W
Is divided by K ₀ , the first unit weight μ _B is calculated (ST4) and stored in the unit weight memory B in the RAM 23, and at the same time, K ₀ is stored in the total number memory (ST5). . In addition,
The contents of the total number memory are displayed on the display 4.

Next, using the content T of the total number memory and the coefficient of variation CV previously input, the maximum number of additions at the first time at which a counting error does not occur with a probability P ₀ sufficiently close to 1 set in advance, that is, allowable number K ₁ is calculated, further predetermined number than the K _1, for example, five as few K _'1 is determined, they are stored in the RAM 23 (ST6). Regarding the allowable number K ₁ at the time of this first addition, since there is only one unit weight estimated value, for example, 3σ of normal distribution similar to the conventional one
Based on the range of, it is desirable that β in the equation (12) is set to 3 and n in the equation is set as T (= K ₀ ), and the fraction is rounded down to obtain K ₁ .

Next, when the operator adds a sample to the dish 1a (ST7), the total sample weight on the dish 1a is determined and stored from the load data at that time, and the difference W ₁ from the previously stored weight, that is, The first additional weight W ₁ is calculated (ST8). By dividing the additional weight W ₁ by the content μ _B of the unit weight memory B and rounding, the first additional number k _B is calculated (ST9). Then, it is determined whether or not this k _B is K ′ ₁ or more and K ₁ or less obtained in ST6 (ST10). If k _B is less than K ′ ₁ , that is determined, and if it exceeds K ₁ , To that effect, each is notified by the indicator lamp 5 (ST11, ST12). When an operator adds or removes an appropriate number of samples based on this notification (ST13, S
After returning to T14) and ST8, the additional number is recalculated.

When k _B falls within the range of K ′ _{1 to} K ₁ , this estimated number k
_B is added and stored in the total number memory (ST15),
Stored in sampling data memory with additional weight W ₁ (ST16). Then, two types of unit weights μ _A and μ _B are calculated and stored in unit weight memories A and B, respectively (ST17).

Here, μ _A is the content N _s and W _s of the sampling memory, and μ _A = W _s / N _s = W ₁ / k _B (24), and μ _B is the content of the total number memory and the time From the total sample weight W on the dish 1a, μ _B = W / T = W / (K ₀ + k _B ) ... (25).

Next, the second allowable number K ₂ is calculated using the variation coefficient CV and the content N _s of the sampling data memory, that is, the immediately preceding estimated number k _B (ST18, ST19). Allowable number K after this second time
_i (i ≧ 2) is the maximum additional number that allows up to ± 1 counting error. As described above, β in equation (12) is set to 1.83, and n in the equation is set to N _s. Calculated by At this time, predetermined five than K _i similarly small K _i is calculated and stored in the RAM 23.

When an operator adds a sample in this state (ST20), the total sample weight W on the dish 1a is similarly determined and stored, and the difference W _{i from the} previous weight is calculated (ST21). And this additional weight W _i
Are divided by the two types of unit weight estimation values μ _A and μ _B , respectively, and rounded to obtain the additional numbers k _A and k _{B of} the two types (S
T22).

Next, one of the additional number estimation values, for example, k _B is calculated as K ′ _i.
It is determined whether or not it is within the range of to K _i (ST23),
If it does not fit, the fact is displayed by the indicator lamp 5 to instruct addition or removal (ST24, ST25). When an operator adds or removes an appropriate number of samples based on this display (ST26, ST27), the process returns to ST21 and below, and k _A and k _B are recalculated.

If k _B is within the range K ′ _{i to} K _i , then k _A and k _B
It is determined whether or not are equal (ST28). If k _A and k _B are equal, it means that there is no counting error in the additional number estimation value k _B (= k _A ), as described above, and assuming that normal sampling was performed, that value k _B is the total number. In addition to storing in memory (ST29), k _B , W in sampling data memory
Store _i (ST30).

Then, the two unit weights μ _A and μ _B are updated (ST
31). mu _A, mu _B is, when the n-th sampling is finished, the additional estimated number of 1~n th _{k B1, k B2, ...... k} Bi
…… When k _Bn , Can be generally expressed as

The above operation is repeated until sampling for unit weight estimation is completed (ST32, ST33).

In the above embodiment, K ′ _i , which is smaller than K _i by a predetermined number, is used.
Was calculated, and an example in which the additional number estimated value k _B does not proceed to the next operation until it falls within the range of K ′ _{i to} K _i is shown.
_{It is also possible} to calculate _i without calculating _i as long as k _B is K _i or less. However, the allowable number at the time of the next addition increases as k _B increases. In view of this point, the addition or removal of the sample may be notified so that the estimated additional number value becomes equal to K _i . That is, instead of the indicator lamp 5, a display device for digitally displaying the excess or deficiency number of k _B with respect to K _i is provided so that the next operation is not performed until k _B = K _i . By doing this, the maximum allowable number of next additions can be taken.

Further, the method of estimating the two types of unit weights μ _A and μ _B is not limited to the above-mentioned example, and for example, the data sampled up to that time may be {k ₀ , W ₀ }, {k _B1 , W ₁ }, {k. _{B2, W 2} ... {k} Bn, as W _n}, the J {0, 1, one of mu _a when a subset of the ...... n}, You can also ask. In this case, the sampling data memory may store not only the immediately preceding data but also sampling data of a plurality of past times.

Further, in the above example, the allowable number is calculated with β = 1.83, but this value 1.83 corresponds to the probability P = 0.9973 based on the range of 3σ of the normal distribution in the related art, and the present invention is not limited to this. Of course, if the probability P for guarantee is changed, it is needless to say that β corresponding to the value of P should be selected.

Furthermore, the present invention selectively employs the sample weight and the estimated number thereof as sampling data in the number estimation routine after finishing the sampling operation for estimating the unit weight as described above. It can be configured to update _{A 1} , μ _B or an allowable number.

<Effects of the Invention> As described above, according to the present invention, the allowable additional number is, for example, up to ± 1 counting error instead of the number that guarantees a counting error of 0 with a predetermined probability or more as in the conventional case. Calculate the number of admitted pieces, and estimate the number of additional pieces of two kinds using the unit weights μ _A and μ _B of two kinds when adding a sample below the number, and only when the estimated values of both agree. By configuring to proceed to the operation, it is configured to guarantee that the counting error of ± 1 etc. allowed by the allowable additional number does not occur, so increase the allowable additional number compared to the conventional method. The work efficiency can be improved. Moreover, the occurrence of the counting error can be guaranteed with the same probability as the conventional method. For example, if the initial number K _{0 is set} to 5 and the number of samples equal to the allowable additional number K _i is added, β = 1.83 in the present invention, and the estimated number of each time is guaranteed with a probability of 0.9973 or more. _{when, {K 1, K 2,} K 3, K 4} is the coefficient of variation 0.01 ... {34,143,262,330} also 0.02 in.. {16,47,73,85} also 0.03 in .... It becomes {10,24,34,38}. By the way, in the conventional way of thinking that guaranteeing a counting error of 0 within the range of 3σ, even if β is set to a value of 2.8, which is smaller than 3, the coefficient of variation is 0.02 ... {17,31,42,50} Similarly, at 0.03 ... It becomes {11,17,22,25}. It is clear that the allowable additional number after K ₂ is large according to the concept of the present invention.

When k _A and k _B do not match, it is necessary to remove an appropriate number to reduce the additional number of samples in the present invention, but the probability that k _A and k _B do not match is 1 / A work rate of about 11 does not significantly reduce work efficiency.

[Brief description of drawings]

FIG. 1 is a basic conceptual diagram showing the configuration of the present invention, FIG. 2 is a block diagram showing the configuration of an embodiment of the present invention, and FIG. 3 is a flowchart showing a program written in the ROM 22. 1 …… Load detection part 1a …… Plate 2 …… Microcomputer 21 …… CPU 22 …… ROM 3 …… Keyboard 4 …… Display unit 5 …… Indicator lamp

Claims (3)

[Claims] 1. A load detector for detecting a load placed on a plate, first and second unit weight memories for storing two kinds of estimated unit weight values of a sample to be counted, and the load detector. And an additional number calculating means for estimating the number of samples k _A and k _B added on the dish by dividing the additional weight based on the detected value of by the contents μ _A and μ _B of the first and second unit weight memories. Using the coefficient of variation setting means for setting the coefficient of variation of the sample and the setting contents and the calculation result at the time of the previous addition by the additional number calculation means, the counting error within a predetermined number with a probability higher than a preset probability And an allowable additional number calculating means for estimating the next maximum additional number, and a first comparing means for comparing the calculated allowable number with the current estimation result k _A or k _B by the additional number calculating means. When, the present estimated result k _a and k _B comparison Based on different data in the sampling data memory for storing the respective estimation results up to the previous time by the second comparing means and the additional number calculating means and the data relating to the additional weight corresponding thereto _A unit weight calculating means for estimating two kinds of unit weights μ _A and μ _B and updating the contents of the first and second unit weight memories is provided, and based on the comparison result by the first and second comparing means. Then, the above estimation result k _A or
k _B is below the allowable number, and only if the k _A and k _B are equal to each other, the additional weight that a current estimation result k _A (or k _B) in addition to the contents of the sampling data memory the Estimation of unit weight μ _A and μ _B by unit weight calculation means
An electronic counting scale configured to perform the update. 2. The unit weights μ _A and μ _B are the total sample weight on the dish divided by the total estimated sample number, and the additional weight at the time of the previous addition divided by the estimated number. The electronic counting scale according to claim 1, characterized in that. 3. The allowable additional number calculating means is configured to calculate the maximum allowable number with which the next additional number can be estimated when the counting error is ± 1 or less with the probability or more. The electronic counting scale according to claim 1 or 2, which is characterized.