JP2012101910A - Basic stock setting system - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a proper safety stock calculation system by determining the stationarity performance of demand fluctuation and the volume of a demand.SOLUTION: The basic stock setting system is composed of a memory 101 storing demand data for different products and different time points, feeding lead time data for different products, various kinds of statistics data, and basic stock quantity data for different products and different time points, an input interface 102, an output interface 103, a demand stationarity determination section 104 determining whether a time series change of the demand data is stationary or non-stationary to output various kinds of statistical data, a demand distribution determination section 105 determining a probability distribution used for the calculation of a basic stock based on the determination result of the demand stationarity determination section 104, statistical data, and feeding lead time data for the different products to output various kinds of statistical data characterized by the probability distribution including an average and dispersion, and a basic stock calculation sections 106 calculating the basic stock for different products and different time points using the probability distribution and statistical data selected from the determination result of the demand distribution determination section 105.

Description

  The present invention is a technique for calculating an appropriate safety stock quantity according to the characteristics of demand quantity data when it is necessary to have a safety stock in preparation for fluctuations in demand quantity.

  A method for calculating a safety stock amount in preparation for fluctuations in demand has been established for a long time as a safety margin calculation method for parts and products. On the other hand, due to recent diversification of customer needs, shortening of product life cycle and intensifying changes in market environment, the change in demand has become severe, and the need to properly review safety inventory according to the change in demand due to the point in time Is growing. If demand trends (average demand and magnitude of fluctuations in demand) are changing, but the safety stock is not reviewed, there will be excess inventory or a large number of missing items. This will increase the possibility of losing management.

  If the demand is stable and the demand fluctuations from time to time are steady, the distribution of the demand is expressed as a constant probability distribution (normal distribution is common), and the probability that a shortage will occur is a predetermined value. A general method is to set the safety stock so as to keep it below. However, at the present time when the change in demand is intense as described above, it is no longer appropriate to consider the probability that a shortage will occur on the premise of a constant probability distribution.

  Thus, for example, Patent Literature 1 is disclosed as a technique capable of calculating an appropriate safety stock amount even when the demand fluctuation at each time point is non-stationary. In Patent Document 1, even when the fluctuation amount of the demand amount at each time point changes depending on the time point, in order to calculate the necessary safety margin at each time point according to the change, the error rate (error error of demand prediction independent of the time point) A method to calculate the safety stock quantity using (quantity / predicted value). However, in this method, the probability distribution of demand fluctuation is expressed as an error distribution, and a normal distribution is used. Therefore, when the absolute value of demand is small, the demand will be negative in reality. Since the calculation includes the probability, there is a problem that the accuracy of the shortage prevention probability is deteriorated. Especially in industries where customer needs are diversifying, demand per product type is decreasing, and there is a need for a method that can calculate the safety stock appropriately even when the absolute value of demand is small. .

  Assuming a normal distribution, there is a problem that the accuracy of the shortage prevention probability deteriorates when the demand amount is small as described above, so a technology that provides a method to calculate the safety stock quantity without assuming the normal distribution There is also. For example, Patent Document 2 discloses a method that does not cause an unreasonable case in which a demand amount becomes negative by estimating a probability distribution from a sample of actual demand data and calculating a safety stock amount. However, in this method, it is necessary to identify the probability distribution by statistical estimation, so there is a problem that a large amount of samples of actual data are required, and the past demand trend is regarded as the same regardless of the time point. Therefore, there is a problem that it is not possible to cope with unsteady demand fluctuations.

  As described above, there has not been provided a technique for calculating an appropriate safety stock amount that covers both the response to unsteady demand fluctuation and the response to the case where the absolute value of demand is small.

JP 2004-359415 A JP 2009-104359 A

K. Funaki, et.al., Piecewise Pseudo Estimation for NHPP Using Data in The Adjacent Intervals, J. Operations Research Society of Japan, Vol. 42, No. 4, 1999

The problems to be solved by the present invention are the following two.
(1) Appropriate safety stock quantity can be calculated even when the fluctuation in demand changes depending on the time (unsteady).
(2) Appropriate safety stock can be calculated even when the absolute value of demand is small.

  In order to solve the above problems, a reference stock quantity setting system according to the present invention is provided with the following configuration.

  That is, the demand data for each time point by product and the supply lead time data for each product are stored as input data, and various statistical data and the standard inventory data for each time point by product are stored as output data. A storage unit, an input interface unit that allows the input data to be input from the outside, an output interface unit that allows the output data to be output to the outside, and a time-series change in the demand amount data by time for each product, A demand continuity determination unit that determines whether the demand of the product is steady or non-stationary, calculates various statistical data such as average demand and unbiased variance, and stores the data in the storage unit; and the demand continuity determination unit The probability distribution to be used for the calculation of the reference inventory quantity is determined based on the determination result, the statistical data, and the supply lead time data for each product, and the probability distribution is characterized. A demand distribution determination unit that calculates various statistical data such as distribution and variance and stores the data in the storage unit, a probability distribution selected based on a determination result of the demand distribution determination unit, and a time point for each product using the statistical data A standard inventory quantity calculation unit for calculating another standard inventory quantity and outputting it to the storage unit.

  In particular, in order to determine the continuity of the demand amount data in the demand continuity determination unit, a continuous test based on the median value of the demand amount data is executed.

Further, in order to determine the demand amount probability distribution in the demand distribution determination unit, when the demand amount is stationary, a normal distribution or a Poisson distribution is selected using the average and unbiased variance. In particular, when the demand is steady, the average and unbiased variance are used, and if the value of (average −3 × √unbiased variance) is zero or more, the normal distribution is selected, and if it is negative, the Poisson distribution is selected. .
On the other hand, in the determination of the demand probability distribution in the demand distribution determination unit, if the demand amount is non-stationary, the demand amount data is treated as a non-stationary Poisson process, and the probability distribution according to the demand amount at each time point is used as the probability distribution. A Poisson distribution having an average of λ (t) calculated based on the demand data for each time point is selected. Alternatively, in the determination of the demand amount probability distribution in the demand distribution determination unit, when the demand amount is non-stationary, the demand amount data is treated as a non-stationary Poisson process, and the probability distribution that the demand amount at each time point follows is Based on the demand amount data at each time point, a Poisson distribution having an average demand amount during the continuous period including the time point and the time points before and after that time point as λ (t) is assumed.

  The reference inventory quantity calculation unit can calculate the necessary inventory quantity for each probability that the demand quantity can be covered.

  The output interface unit outputs the probability distribution selected by the demand distribution determination unit, the statistic data characterizing the probability distribution, and the reference stock amount calculated by the reference stock amount calculation unit, and displays them on the screen. Or print it on a form.

With the reference inventory quantity setting system according to the present invention, it is possible to accurately grasp the change in demand quantity at each time point and to select the probability distribution for calculating the shortage prevention probability. As a result, opportunity loss due to shortage can be prevented, and cash flow deterioration and loss due to excess inventory can be suppressed.

It is a figure explaining an example of a functional structure of the reference | standard stock quantity setting system by this invention. It is a figure explaining one Example of the demand amount data input into the reference | standard inventory amount setting system by this invention. It is a figure explaining one Example of the supply lead time data input into the reference | standard stock quantity setting system by this invention. It is a figure explaining one Example of the processing flow which calculates the safety stock quantity of each model by the reference | standard stock quantity setting system by this invention. It is a figure explaining one Example of the processing flow which determines the steady and unsteady of a demand fluctuation in the demand steadiness determination part of the reference | standard stock quantity setting system by this invention. It is a figure explaining the example which identified the demand data in the upper / lower side of the median value of the demand data of FIG. It is a figure which shows the result of having determined whether the demand amount data of each model of FIG. 2 is steady or non-steady. It is a figure which shows the some determination processing flow of the demand amount by the demand distribution determination part of the reference | standard inventory amount setting system by this invention. The average demand amount, unbiased variance, and judgment value of the models A and B calculated by the processing flow of FIG. 8 are shown. It is a figure which shows the result of having calculated average demand amount (lambda) (t) of each month of the model C of FIG. Safety calculated with a normal distribution, expected value μ, variance σ ² , β = 0.95 (95%) to 0.99 (99%) in 1% increments for each month of model A in FIG. It is a figure which shows a stock quantity. 2 shows the Poisson distribution, expected value λ, β = 0.95 (95%) to 0.99 (99%), and the safety stock amount calculated in 1% increments for each month of model B in FIG. FIG. When λ (t) is calculated using Method 1 for each month of model C in FIG. 2, unsteady Poisson distribution, expected value λ (t), β = 0.95 (95%) to 0.99 It is a figure which shows the safety stock quantity calculated in 1% increments until (99%). When λ (t) is calculated for each month of model C in FIG. 2 using method 2, unsteady Poisson distribution, expected value λ (t), β = 0.95 (95%) to 0.99 It is a figure which shows the safety stock quantity calculated in 1% increments until (99%). It is a figure explaining the example of a screen for data input of the standard inventory quantity setting system by this invention. The figure which the reference | standard stock amount of every month is displayed for every value of the demand corresponding | compatible probability (beta) for every model is represented. It is a figure explaining an example of the hardware software configuration of the standard stock quantity setting system by this invention.

  Embodiments of a reference inventory quantity setting system according to the present invention will be described below with reference to the drawings.

  The target scene in the present embodiment is a case where a safety stock is prepared in preparation for a demand fluctuation of a product, and a business for calculating a necessary safety stock quantity based on a fluctuation of demand data for each model is targeted. It does not depend on the structure of the target product or the way of dealing.

  FIG. 1 shows an example of the functional configuration of this system. The reference inventory quantity setting system 10 stores demand quantity data for each time point by product and supply lead time data for each product as input data, and calculates various statistical data and results for each time point for each product as output data. A storage unit 101 that stores reference inventory data, an input interface unit 102 that allows the input data to be input from the outside, an output interface unit 103 that allows the output data to be output to the outside, and the demand by time for each product A demand continuity determination unit 104 that determines whether it is steady or non-stationary by looking at a time series change of the amount data, outputs various statistical data such as an average of demand and unbiased variance, and stores the data in the storage unit 101; Probability distribution to be used for calculation of reference inventory based on the determination result of the demand continuity determination unit 104, the statistical data, and the supply lead time data for each product A demand distribution determination unit 105 that determines and outputs various statistical data such as an average and variance that characterize the probability distribution and stores the data in the storage unit 101; and a probability selected based on a determination result of the demand distribution determination unit 105 A reference inventory quantity calculation unit 106 that calculates a reference inventory quantity for each time point of each product using the distribution and the statistical data and outputs it to the storage unit.

  The demand amount data and the supply lead time data registered in the storage unit 101 are held in a format as shown in FIGS. 2 and 3, for example. FIG. 2 shows demand data, and the monthly demand (forecast) is registered for each model for identifying a product. In the present embodiment, an example is shown in which the demand amount is registered for each month. However, a time unit for grasping the demand amount, such as a week or a day, may be arbitrarily handled depending on the application. FIG. 3 shows supply lead time data, in which the time length necessary to replenish the product inventory is registered for each model. In this example, the supply lead time is registered as 2 months for all models.

  The reference inventory quantity setting system 10 calculates the safety inventory quantity of each model by the processing flow shown in FIG. 4 using the demand quantity data and supply lead time data registered in the storage unit 101. The process flow shown in FIG. 4 is executed for each model registered in the demand data shown in FIG. 2, for example.

  First, in step S1, the demand continuity determination unit 104 determines whether the demand fluctuation of the model is steady or non-stationary using the demand amount data and supply lead time data registered in the storage unit. The stationarity determination in the demand stationarity determination unit 104 is performed by continuous testing, and is executed according to the processing flow shown in FIG.

  First, in step S11, a median is obtained from demand data of the model, and is set as M. For example, according to the demand data in FIG. 2, the median of model A is 3, the median of model B is 3, and the median of model C is 9.

Next, in step S12, the number of samples below the median value M and the number of samples above the median value M are counted, and are set as p and q, respectively. In the present embodiment, a sample having the same amount of demand as the median value M is treated as the upper side, but the effectiveness of the technique of the present invention remains the same regardless of whether it is on the upper side or the lower side. FIG. 6 shows the amount of demand data on the upper side and the lower side of the median M with respect to the amount of demand data shown in FIG. In FIG. 6, the value of the cell painted in light color is the upper side of the median value, and the value of the cell painted in dark color is the lower side data. In such a case, a data group in which the upper data continues or the lower data continues is called a series. (In this embodiment, even if only one upper or lower data continues and is switched, it is counted as one ream.)
In step S13, the number of runs extracted in this way is counted and set to r0. In the example of FIG. 6, the number of stations is 7 for model A, 7 for model B, and 3 for model C.

Next, in step S14, the probability function {P _{p, q} (x) of the series: the random variable x represents the number of series. } To find the confidence interval for the number of runs of each model. The ream probability function represents the probability that the ream number will occur according to the combination of the values of p and q, and is given by, for example, Equation 2.

信頼区間を求める際には信頼確率αが必要であるが、αの値は外部から与えても良いし、内部で一意に決めた値を保持、使用しても良い。本実施の形態では、α＝０．９５として予め設定されているものとし、信頼区間の計算時は両側確率を均等にする。連の確率関数Ｐ_p,q(ｘ) を用いて連の数の信頼区間[r1＋１，r2−１]を求める。但し、r1、r2は次式により求める。

When obtaining the confidence interval, the reliability probability α is required, but the value of α may be given from the outside, or a value uniquely determined internally may be held and used. In this embodiment, it is assumed that α = 0.95 is set in advance, and the two-sided probability is made equal when calculating the confidence interval. The confidence interval [r1 + 1, r2-1] of the number of runs is obtained using the run probability function P _{p, q} (x). However, r1 and r2 are obtained by the following equations.

図６の例では、信頼区間は機種Ａでは［４，８］、機種Ｂでは［５，１１］、機種Ｃでは［４，８］とそれぞれ計算される。最後にステップＳ１５により、先にカウントした連の数ｒ０がそれぞれの信頼区間に含まれるか否かで定常か非定常かを判定する。以上の需要定常性判定部１０４の処理により得られる判定結果を図７に示す。図７において、９５％判定の欄にあるＨは定常を、Ｎは非定常を意味している。本実施の形態では、機種Ａ、機種Ｂの需要量データは定常、機種Ｃの需要量データは非定常であると判定されている。

In the example of FIG. 6, the confidence intervals are calculated as [4,8] for model A, [5,11] for model B, and [4,8] for model C, respectively. Finally, in step S15, it is determined whether it is steady or non-stationary depending on whether or not the number r0 of the series counted previously is included in each confidence interval. FIG. 7 shows a determination result obtained by the above-described processing of the demand continuity determination unit 104. In FIG. 7, H in the 95% determination column means steady and N means non-stationary. In the present embodiment, it is determined that the demand amount data of the model A and the model B is steady and the demand amount data of the model C is non-stationary.

  Next, the reference inventory quantity setting system 10 specifies the demand distribution to be used by the demand distribution determination unit 105 based on the determination result of FIG. The demand distribution determination unit 105 performs some determination of the demand amount in step S2 for the model determined to be stationary as a result of step S1 in FIG. The determination of the amount of demand by the demand distribution determination unit 105 is executed according to the processing flow shown in FIG. First, in step S21, average demand and unbiased variance are calculated from demand data of the model. Next, in step S22, if the average demand amount−3 × the unbiased variance value is 0 or more, the demand amount of the model is large, and if the average demand amount−3 × the unbiased variance value is less than 0, the model is concerned. It is determined that there is little demand. FIG. 9 shows the average demand amount, unbiased variance, and judgment value calculated by the processing flow of FIG. In the present embodiment, among the models A and B determined to be steady by the demand steadiness determination unit 104, it is determined that the demand amount of the model A is large and the demand amount of the model B is small.

Based on these determination results, the demand distribution determination unit 105 selects the normal distribution using the average demand amount and unbiased variance as the demand amount distribution when the demand amount is steady and large, and when the demand amount is steady and small. Selects a Poisson distribution having a parameter λ (= expected value) calculated using the average demand. Specifically, when the demand amount is steady and large, a normal distribution having parameters of expected value μ = average demand amount × supply lead time and variance σ ² = unbiased variance × supply lead time is selected (step S3). . When the demand amount is steady and small, a Poisson distribution having the expected value λ = average demand amount × supply lead time as a parameter is selected (step S4).

  On the other hand, when the demand steadiness determination unit 104 determines that the demand amount is non-stationary in step S1 of FIG. 4, the demand distribution determination unit 105 has a parameter λ (t) (= expected value) for each time point. A non-stationary Poisson distribution is selected (step S5). Here, several methods of calculating the value of λ (t) are conceivable. In the present embodiment, examples of the following two methods are shown. Even if the value of λ (t) is calculated by a method other than the method described below, the effectiveness of the reference stock quantity setting system according to the present invention is not impaired.

  First, as Method 1, each month's value of the demand data of the model is treated as the average demand of the month, and the value obtained by adding up the average demand from the month to the month before the supply lead time is λ This is a method of handling as (t). This method is based on the assumption that the value of the demand amount for each month is a representative value that represents the actual demand for that month, and is sufficient information to represent the unsteady Poisson distribution for that month. For example, since the supply lead time of model C is two months from FIG. 3, the value of λ (t) in January 2010 is the demand amount 15 in January 2010 and 10 years from the demand data in FIG. It is calculated as 25, which is the total amount of demand 10 in February.

  Next, as method 2, when calculating the value of λ (t) for each month of the model, the average demand for each month is also calculated using the demand data for the months before and after the month. . This method is based on the premise that demand in the months before and after the month should follow a similar distribution, and the extent to which the preceding and following months are considered is determined statistically. is there. The logic and theoretical basis for realizing this method are disclosed in Non-Patent Document 1, for example.

  In the present embodiment, the average demand for each month of the model C is calculated as shown in FIG. 10 by the algorithm disclosed in Non-Patent Document 1. In FIG. 10, for example, the average demand in January 2010 is calculated as a value 12.5 obtained by averaging the demand 15 in January 2010 and the demand 10 in February 2010 in the demand data in FIG. The average demand amount in February 2010 is calculated as a value 11.67 obtained by averaging the demand amount 15 in January 2010, the demand amount 10 in February 2010, and the demand amount 10 in March 2010. In this example, the average demand for the month is calculated by averaging the demand for the three months of the previous and next months across the month (the first month of the target period). (January 2010) and last month (March 2011) are the average of the demand for the next month and the previous month, respectively). Using the average demand for each month calculated as described above, a value obtained by adding up the average demand from the month to the month preceding the supply lead time is calculated as λ (t). For example, since the supply lead time of model C is 2 months, the value of λ (t) in January 2010 is, for example, the average demand in January 2010 when the value of average demand in FIG. 10 is used. It is calculated as 24.17, which is 12.50 and the average demand volume of 11.67 in February 2010.

  Finally, the reference inventory quantity setting system 10 uses the demand distribution selected for each model in steps S3, S4, and S5 in FIG. 4, and the reference inventory quantity calculation unit 106 calculates the safety inventory quantity in step S6. The amount is registered in the storage unit 101. The safety stock quantity is calculated by the following formula.

ここでｐ（ｚ）は当該機種に対して選択された分布の確率関数である。また、βは当該供給リードタイム中の需要変動に対応できる確率を表すパラメータであり、０から１．０の間の任意の値である。βの値は予め一つの値を与えても良いし、値の範囲を与えてその範囲に含まれるいくつかの値ごとに上記数式で安全在庫量を計算しても良い。期待値は、使用する確率分布によって異なり、正規分布の場合はμ、ポアソン分布の場合はλ、非定常ポアソン分布の場合はλ（ｔ）の値を用いる。すなわち、安全在庫量はβ以上の確率で供給リードタイム内の需要変動に対応できるようにバッファとしてもつ量として計算される。

Here, p (z) is a probability function of the distribution selected for the model. Β is a parameter representing the probability of being able to cope with demand fluctuations during the supply lead time, and is an arbitrary value between 0 and 1.0. One value may be given in advance as the value of β, or a range of values may be given, and the safety stock quantity may be calculated by the above formula for each of several values included in the range. The expected value varies depending on the probability distribution to be used, and the value of μ is used for the normal distribution, λ is used for the Poisson distribution, and λ (t) is used for the non-stationary Poisson distribution. That is, the safety stock quantity is calculated as an amount held as a buffer so that it can cope with demand fluctuation within the supply lead time with a probability of β or more.

FIG. 11 to FIG. 14 show the results finally calculated by the reference stock quantity calculation unit 106 by the above processing. FIG. 11 shows the results calculated for the model A for which the demand data is determined to be steady and large. Usage distribution for each month (Normal = normal distribution), expected value μ, variance σ ² , β = The safety stock amount calculated in increments of 1% between 0.95 (95%) and 0.99 (99%) is shown. FIG. 12 shows a result calculated for the model B in which the demand data is determined to be steady and small. Usage distribution for each month (Poisson = Poisson distribution), expected value λ, β = 0.95 ( 95%) to 0.99 (99%), the safety stock amount calculated in 1% increments. FIG. 13 shows the result when λ (t) is calculated using Method 1 for the model C for which the demand data is determined to be unsteady, and the monthly usage distribution (NHPP = non- (Stationary Poisson distribution), expected value λ (t), β = 0.95 (95%) to 0.99 (99%), and the safety stock amount calculated in 1% increments. FIG. 14 shows the result when λ (t) is calculated using Method 2 for the model C, and the monthly usage distribution (NHPP = unsteady Poisson distribution) and expected value λ (t). , Β = 0.95 (95%) to 0.99 (99%), the safety stock amount calculated in increments of 1%.

  Demand amount data and supply lead time data for performing the above processing are input via the input interface unit 102 of the reference inventory amount setting system 10 and registered in the storage unit 101. The entity of the input interface unit 102 may be realized by a data fetching program or the like when importing data from an external system, and may be realized by a data input screen when the user needs to input data. FIG. 15 is an example of a data input screen. In FIG. 15, a table for inputting the demand amount for each month for each model is prepared, and a registration button for registering data after input is provided. A graph may be displayed as needed to visually display changes in demand.

  Further, the reference inventory data obtained as a result of the above processing and various statistical data necessary for the calculation are output to the outside through the output interface unit 103 of the reference inventory setting system 10. The entity of the output interface unit 103 may be realized by a data transfer program or the like when exporting data to an external system, and may be realized by a screen for data output when displaying so as to be confirmed by the user. FIG. 16 is an example of a screen for outputting the reference inventory quantity calculation result. In FIG. 16, the reference stock amount for each month is displayed for each model for each value of the demand corresponding probability β (in this example, in units of 1% from 95% to 99%). For each model, the demand distribution used to calculate the standard inventory and the parameter value of its probability distribution (expected value and variance for normal distribution, λ for Poisson distribution, λ (t for non-stationary Poisson distribution) )) Is also displayed. If necessary, a graph may be displayed to visually display changes in the reference inventory. In the example of FIG. 16, a graph of the standard inventory amount of model C is displayed.

  Finally, FIG. 17 shows an example of a hardware / software configuration for realizing the reference stock quantity setting system 10 according to the present invention.

  A standard inventory quantity setting computer 20 that implements the standard inventory quantity setting system 10 includes a CPU (Central Processing Unit) 201 that performs various operations and instructions necessary for system operation, an OS (Operating System), and a standard inventory quantity setting. Communication that controls connection and communication between an application program such as a reference inventory setting program that describes the operation contents of the system and a memory 202 that stores data necessary for the program, and externally via a network as necessary. And a control unit 203. Further, by connecting the auxiliary storage device 204 to the reference inventory quantity setting computer 20, the OS, program and data stored in the memory 202 can be stored in the auxiliary storage device 204. A user interface for the user to operate the reference inventory quantity setting system and to input / output data can be provided inside the reference inventory quantity setting computer 20. In order to make the user accessible from a location physically separated from the user 20, the user interface terminal 205 is placed outside the reference stock quantity setting computer 20 as shown in FIG. The communication control unit 203 and the network may be connected so as to be able to communicate with the reference inventory quantity setting computer 20.

  In the hardware / software configuration described above, each part of the reference stock quantity setting system 10 shown in FIG. 1 corresponds to the components shown in FIG. 17 as follows.

  When other means executes processing, the storage unit 101 realizes its function mainly by the memory 202. On the other hand, the storage unit 101 stores auxiliary data when storing a large amount of data or storing it in a fixed manner. The function is realized by 204.

  The functions of the input interface unit 102, the output interface unit 103, the demand steadiness determination unit 104, the demand distribution determination unit 105, and the reference inventory quantity calculation unit 106 are the OS and the reference inventory quantity setting program stored in the memory 202, and these These are realized by the interaction of the CPU 201 that controls the data, and various data stored in the memory 202 or the auxiliary storage device 204 are referred to or updated at that time. In addition, when the input interface unit 102, the output interface unit 103, the demand steadiness determination unit 104, the demand distribution determination unit 105, and the reference inventory amount calculation unit 106 require communication with the user interface terminal 205, a communication control unit The function is realized by 203.

  The reference inventory quantity setting system according to the present invention is capable of calculating a necessary safety inventory quantity and setting it as a reference inventory quantity when the safety inventory is prepared for demand fluctuation. In particular, it is possible to cope with a case where the demand amount is small due to unsteady demand fluctuations in which an appropriate safety stock amount cannot be calculated by the conventional technology.

DESCRIPTION OF SYMBOLS 10 Reference | standard inventory quantity setting system 101 Memory | storage part 102 Input interface part 103 Output interface part 104 Demand steadiness determination part 105 Demand distribution determination part 106 Reference | standard inventory quantity calculation part 20 Reference | standard inventory quantity setting computer 201 CPU (Central Processing Unit)
202 Memory 203 Communication control unit 204 Auxiliary storage device 205 User interface terminal S1 Demand steadiness determination processing step S2 Demand amount somewhat determination processing step S3 Normal distribution selection processing step S4 Poisson distribution selection processing step S5 Unsteady Poisson distribution selection processing Step S11 One step S12 in demand continuity determination processing One step S13 in demand continuity determination processing One step S14 in demand continuity determination processing One step S15 in demand continuity determination processing One step S21 in demand continuity determination processing Demand distribution determination 1 step in the process S22 1 step in the acceptance distribution determination process

Claims (7)

A system that calculates an appropriate base stock amount that should be held as a safety stock even when fluctuations in the demand amount of a product change from time to time, or even if the absolute value of the demand amount is small,
Demand data for each unit period by product and supply lead time data for each product are stored as input data, and various statistical data as a result of calculation and basic inventory data for each unit period are stored as output data A storage unit to
An input interface unit that allows the input data to be input from the outside;
An output interface unit capable of outputting the output data to the outside;
CPU and
The CPU executes a reference inventory amount setting program stored in the storage unit,
A time series change of the demand amount data for each product by unit period is a data group in which the number of samples p, q on the lower side / upper side of the median value of the demand amount data for each unit period, and the data on the upper side or the lower side continues. Counts the number r0 of a certain series, obtains a confidence interval of the number of series of each model using a probability function of the number of series generated by the combination of p and q values, and whether r0 is included in the confidence interval A demand steadiness determination unit that determines whether the demand amount data is steady or non-stationary depending on whether or not,
For the demand data determined to be steady by the demand steadiness determination unit, the average demand amount, unbiased variance is calculated, and the demand data is large according to the value of (average demand amount-3 × unbiased variance), Or, it is determined that the demand amount data is (1) steady and large, (2) steady and small, and (3) non-steady, and each of the predetermined demand distributions is different. A demand distribution determination unit for selecting
The probability distribution function p (z) of the demand distribution selected based on the determination result of the demand distribution determination unit, where z is a random variable representing the demand amount data of the corresponding product per unit period, and the safety stock amount is calculated according to Equation 1. And

Here, β is a parameter representing the probability of being able to cope with demand fluctuation during the supply lead time, the expected value is a value calculated according to the probability function p (z) of the selected demand distribution,
A function of a reference stock quantity calculation unit that stores the calculated safety stock quantity in the storage unit as reference stock quantity data for each product-specific unit period;
The reference interface quantity setting system, wherein the output interface unit outputs a standard inventory quantity for each set value of the parameter β for each unit period for each product. In the demand distribution determination unit, an average demand amount, unbiased variance is calculated for the demand amount data determined to be steady by the demand steadiness determination unit, and a value of (average demand amount-3 × unbiased variance) is calculated. In the process of determining that the demand data is large or small,
(Average demand-3 x unbiased variance) ≥ 0, it is determined that the demand data is large,
2. The reference inventory quantity setting system according to claim 1, wherein if it is <0, the demand quantity data is determined to be small. In the demand distribution determination unit, the demand data is a predetermined demand that is different in any of the cases of (1) steady and large, (2) steady and small, and (3) non-steady. In the process of selecting a distribution,
When the demand data is (1) steady and large, select a normal distribution with parameters of expected value μ = average demand × supply lead time, variance σ ² = biased variance × supply lead time,
When the demand data is (2) steady and small, select a Poisson distribution having an expected value λ = average demand x supply lead time as a parameter,
2. The standard inventory quantity according to claim 1, wherein when the demand quantity data is (3) non-stationary, an unsteady Poisson distribution having an expected value λ (t) for each time as a parameter is selected. Configuration system.   In the demand distribution determination unit, when the demand amount data is (3) non-stationary, in the process of selecting an unsteady Poisson distribution having an expected value λ (t) for each time as a parameter, an expected value λ ( t) is a value obtained by treating the value of each unit period of the demand amount data of the model as the average demand amount of the unit period and summing the average demand amount from the unit period to the unit period ahead of the supply lead time. The reference inventory quantity setting system according to claim 3, wherein:   In the demand distribution determination unit, when the demand amount data is (3) non-stationary, in the process of selecting an unsteady Poisson distribution having an expected value λ (t) for each time as a parameter, 4. When calculating the expected value λ (t) of a unit period, the average demand amount of each unit period is calculated using the demand amount data of the unit periods before and after the unit period. Standard inventory quantity setting system as described in.   In the determination of the demand probability distribution in the demand distribution determination unit, when the demand amount is non-stationary, the demand amount data is treated as a non-stationary Poisson process, and the probability distribution that the demand amount at each time point follows is the time point of the product 2. The reference inventory quantity setting according to claim 1, wherein a Poisson distribution having an average demand quantity for a continuous period including the current time point and a time point before and after the current time point as λ (t) is based on the different demand amount data. system.   The output interface unit outputs the probability distribution selected by the demand distribution determination unit, the statistic data characterizing the probability distribution, and the reference stock amount calculated by the reference stock amount calculation unit, and displays them on the screen. The standard inventory quantity setting system according to claim 1, wherein the system is printed on a form.

Images