Demand predicting device, demand predicting method and program

Abstract

To provide a demand predicting device, a demand predicting method and a program, capable of reducing time and cost when a demand is predicted.SOLUTION: A demand predicting device 100 includes a decomposing unit 11 and a predicting unit 12. The decomposing unit 11 decomposes order acceptance actual record data on parts into a regular component and an irregular component using ensemble empirical mode decomposition. The predicting unit 12 predicts transitions of each of the regular component and the irregular component using dynamic mode decomposition, and predicts a demand for the parts based on the result of the transition prediction.SELECTED DRAWING: Figure 1

Description

[Technical field]

[0001] The present disclosure relates to a demand forecasting device, a demand forecasting method, and a program. [Background technology]

[0002] In the logistics of service parts, production preparation and inventory adjustments are carried out based on the results of demand forecasts in order to reduce stockouts and excess inventory, and a high level of forecast accuracy is desirable. However, since the demand characteristics of service parts are a mixture of regular and irregular elements, it is not easy to ensure good forecast accuracy.

[0003] Regarding demand forecasting methods that take into account regular factors, Patent Document 1 discloses a method of forecasting demand using order records and associated information (manufacturing information, usage information, repair information, etc.) for a group of parts that have many common points (for example, a collection of multiple parts obtained by clustering). In addition, when it comes to demand forecasting methods that take irregular factors into account, statistical methods such as average forecasting are used. In addition, Non-Patent Documents 1 and 2 are used in the description of the embodiments, and therefore will be described below. [Prior art documents] [Patent documents]

[0004] [Patent Document 1] JP 2020-098388 A [Non-patent literature]

[0005] [Non-Patent Document 1] NE Huang, Z. Shen, SR Long et al., "The empirical mode decomposition and the Hilbert spectrum for nonlinear and nonstationary time series analysis", Proceedings of the Royal Society of London Series A, vol. 454, pp. 903-995, 1998. [Non-Patent Document 2] Schmid, Peter J. 2010. “Dynamic Mode Decomposition of Numerical and Experimental Data.”, Journal of Fluid Mechanics 656. Cambridge University Press:5-28. Summary of the Invention [Problem to be solved by the invention]

[0006] In the technology disclosed in Patent Document 1, in order to forecast demand for parts, it is necessary to build a database that stores a large amount of information such as order records for parts that have many things in common and their associated information. However, building a database is not easy, and there is an issue that it takes time and costs to forecast demand.

[0007] In consideration of such problems, the present disclosure aims to provide a demand forecasting device, a demand forecasting method, and a program that can reduce the time and cost required for demand forecasting. [Means for solving the problem]

[0008] The demand forecasting device of the present disclosure is a decomposition unit that decomposes the order record data of the parts into regular components and irregular components using ensemble empirical mode decomposition; A prediction unit that performs a transition prediction for each of the regular components and the irregular components by using dynamic mode decomposition, and performs a demand prediction for the part based on the transition prediction result.

[0009] The demand forecasting method of the present disclosure includes: The computer The order history data of the parts is decomposed into regular and irregular components using ensemble empirical mode decomposition. A transition prediction is performed for each of the regular components and the irregular components using dynamic mode decomposition, and a demand prediction for the part is performed based on the transition prediction results.

[0010] The program of the present disclosure is The order history data of the parts is decomposed into regular and irregular components using ensemble empirical mode decomposition. The computer is caused to execute a process of predicting trends for each of the regular components and the irregular components using dynamic mode decomposition, and predicting demand for the part based on the results of the trend prediction. Effect of the Invention

[0011] The present disclosure makes it possible to provide a demand forecasting device, a demand forecasting method, and a program that can reduce the time and cost required for demand forecasting. [Brief description of the drawings]

[0012] [Figure 1] 1 is a block diagram showing an example of a configuration of a demand prediction device according to a first embodiment. [Diagram 2] FIG. 2 is a diagram showing an example of decomposition of actual order data of parts into regular components and irregular components using EEMD by the demand forecasting device according to the first embodiment. [Diagram 3] FIG. 2 is a diagram illustrating an example of an operation of the demand prediction device according to the first embodiment. [Figure 4] FIG. 2 is a diagram showing an example of an algorithm of EEMD of the demand prediction device according to the first embodiment. [Diagram 5]FIG. 11 is a diagram showing a comparison between a demand forecast result for service parts by a demand forecasting method according to a comparative example and a demand forecast result for service parts by a demand forecasting method according to the demand forecasting device according to the first embodiment. [Figure 6] FIG. 11 is a diagram showing a comparison between a demand forecast result for service parts by a demand forecasting method according to a comparative example and a demand forecast result for service parts by a demand forecasting method according to the demand forecasting device according to the first embodiment. [Figure 7] FIG. 11 is a diagram showing a comparison between a demand forecast result for service parts by a demand forecasting method according to a comparative example and a demand forecast result for service parts by a demand forecasting method according to the demand forecasting device according to the first embodiment. [Figure 8] FIG. 11 is a diagram showing a comparison between a demand forecast result for service parts by a demand forecasting method according to a comparative example and a demand forecast result for service parts by a demand forecasting method according to the demand forecasting device according to the first embodiment. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0013] (First embodiment) First, a configuration of a demand prediction device 100 according to the first embodiment will be described with reference to Fig. 1. Fig. 1 is a block diagram showing an example of the configuration of the demand prediction device 100 according to the first embodiment.

[0014] 1, the demand forecasting device 100 is a device capable of forecasting future demand by inputting actual order data for a part such as a service part with a single product number. The demand forecasting device 100 includes a decomposition unit 11 and a prediction unit 12.

[0015] The decomposition unit 11 receives the order history data of each part. The order history data is time-series signal data with the number of orders of parts on the vertical axis and time on the horizontal axis. The decomposition unit 11 uses Ensemble Empirical Mode Decomposition (EEMD) to decompose the part order history data into regular components and irregular components as shown in FIG. 2.

[0016] The prediction unit 12 performs trend prediction for each of the regular components and the irregular components by using dynamic mode decomposition (DMD), and performs demand prediction for parts based on the trend prediction results.

[0017] Next, the operation of the demand prediction device 100 according to the first embodiment will be described with reference to FIGS.

[0018] Fig. 3 is a diagram showing an example of the operation of the demand prediction device 100 according to the first embodiment. As shown in Fig. 3, the demand prediction device 100 performs the following processes of Phase 1 and Phase 2. In the following example, the parts for which demand is predicted are service parts. <Phase 1> 3, in phase 1, the decomposition unit 11 of the demand prediction device 100 decomposes the order record data of each service part into a plurality of intrinsic mode functions (IMFs) using EEMD. In this embodiment, the decomposition unit 11 uses the order record data as an input signal, performs mode decomposition using the following EEMD algorithm, and outputs IMF1 to IMF4.

[0019] TIFF2025070652000002.tif69166

[0020] In the EEMD of this embodiment, the maximum order of the IMF was set to 4. This is because monthly order result data over several years is used in the demand forecast of service parts. In addition, as the order number increases, the IMF changes from short-term fluctuations (irregularity) to long-term fluctuations (regularity). Therefore, it can be seen that the waveform of the order result data can be appropriately separated into irregular and regular components.

[0021] <Phase 2> Returning to the explanation of Fig. 3, next, in phase 2, the prediction unit 12 of the demand prediction device 100 constructs a regression model based on the output of dynamic mode decomposition with each IMF as an input. The prediction unit 12 calculates a predicted value corresponding to each IMF using the constructed regression model, and calculates a demand forecast value by summing up the predicted values ​​corresponding to each IMF. Specifically, the prediction unit 12 performs the processes from step S1 to step S7.

[0022] TIFF2025070652000003.tif24166

[0023]

number

[0024] TIFF2025070652000005.tif20166

[0025]

number

[0026] TIFF2025070652000007.tif34166

[0027]

number

[0028] TIFF2025070652000009.tif95166

[0029]

number

[0030] Furthermore, when equation (4) is expanded using eigenmode information of the DMD analysis, it can be expressed as the following equation (5).

[0031]

number

[0032] TIFF2025070652000012.tif16166

[0033]

number

[0034] TIFF2025070652000014.tif79166

[0035]

number

[0036] TIFF2025070652000016.tif9166

[0037]

number

[0038] Next, the effectiveness of the demand forecast by the demand forecasting device 100 according to the first embodiment will be described with reference to FIGS.

[0039] 5 to 8 are diagrams illustrating a comparison between the demand forecasting results for service parts according to the demand forecasting method of the comparative example and the demand forecasting results for service parts according to the demand forecasting method of the demand forecasting device 100 according to the first embodiment.

[0040] Figures 5A, 6A, 7A, and 8A show order record data (records) for each representative part of service parts over about five years and a demand forecast result (MT1) for the most recent year using a demand forecasting method according to a comparative example based on this order record data. Meanwhile, Figures 5B, 6B, 7B, and 8B show order record data (records) for each representative part of service parts over about five years and a demand forecast result (MT5) for the most recent year using a demand forecasting method according to demand forecasting device 100 based on this order record data.

[0041] 5A and 5B, a demand forecast is made for service parts whose demand characteristics in the order history data have seasonality (annual cycle). In FIGS. 6A and 6B, a demand forecast is made for service parts whose demand characteristics in the order history data have a two-year cycle. In FIGS. 7A and 7B, a demand forecast is made for service parts whose demand characteristics in the order history data have a decreasing trend. In FIGS. 8A and 8B, a demand forecast is made for service parts whose demand characteristics in the order history data have irregular elements.

[0042] In the demand forecasting method of the comparative example described above, similar features between parts are clustered and feature quantities are extracted using an order history database of approximately 9,000 parts, and then demand is forecast. In the demand forecasting method of the demand forecasting device 100 according to the first embodiment, prediction is performed using only order history data (single time-series data) for each part.

[0043] 5 to 8, the demand prediction method related to the demand prediction device 100 shows improved demand prediction accuracy for service parts compared to the demand prediction method related to the comparative example. This shows that the demand prediction method related to the demand prediction device 100 is effective. The reasons for this are considered to be, for example, as follows. In the demand forecasting method according to the comparative example, there is a possibility that highly irregular data is mixed in among the actual data classified as having high regularity. On the other hand, in the demand forecasting method according to the demand forecasting device 100, the accuracy of the demand forecast can be improved by classifying the actual data into data having high regularity and data having high irregularity by EEMD. In the demand forecasting method according to the comparative example, an interpolation prediction is performed to predict unknown values ​​within the range of actual order data. On the other hand, in the demand forecasting method according to the demand forecasting device 100, an extrapolation prediction is performed to predict unknown values ​​outside the range of actual order data. As a result, the accuracy of the demand forecast for service parts, especially service parts whose demand characteristics have a decreasing trend, has been improved.

[0044] Furthermore, in the demand forecasting method according to the demand forecasting device 100, compared to the demand forecasting method according to the comparative example, a database of order records of approximately 9,000 parts is not required to forecast demand for a specific service part. Therefore, the demand forecasting device 100 can reduce the time and cost required for demand forecasting.

[0045] In addition, there are cases where the desired data information or data volume cannot be obtained due to the circumstances of the base, region, or country, or information system constraints. Therefore, there is also the problem that the application of demand forecasting is highly dependent on the database state and is limited by the deterioration of prediction accuracy due to database constraints. Furthermore, even in a secured database, much of the data contains irregular elements (or noise), and there is concern that the influence of noise will reduce prediction accuracy. However, the demand forecasting device 100 does not require a large-scale database, so it is possible to prevent such problems from occurring.

[0046] The demand prediction device 100 according to the above embodiment is configured by hardware or software, or both. The demand prediction device 100 may be configured by one piece of hardware or software, or may be configured by multiple pieces of hardware or software. Specifically, the demand prediction device 100 is configured by a computer including a processor and a memory. The processor may be, for example, a microprocessor, an MPU (Micro Processing Unit), or a CPU (Central Processing Unit). The processor may include multiple processors. The memory is configured by a combination of a volatile memory and a non-volatile memory. The memory may include a storage located away from the processor. In this case, the processor may access the memory via an I / O interface not shown. The processor executes one or more programs including a set of instructions for causing a computer to perform an algorithm (method) described using the drawings. The program includes a set of instructions (or software code) for causing a computer to perform one or more functions described in the embodiment when the program is loaded into the computer. The program may be stored in a non-transitory computer-readable medium or a tangible storage medium. By way of example and not limitation, computer readable media or tangible storage media include random-access memory (RAM), read-only memory (ROM), flash memory, solid-state drive (SSD) or other memory technology, CD-ROM, digital versatile disc (DVD), Blu-ray® disk or other optical disk storage, magnetic cassette, magnetic tape, magnetic disk storage or other magnetic storage device. The program may be transmitted on a transitory computer readable medium or communication medium. By way of example and not limitation, transitory computer readable medium or communication medium includes electrical, optical, acoustic or other form of propagated signal.

[0047] The present invention is not limited to the above-described embodiment, and can be modified as appropriate without departing from the spirit and scope of the present invention. [Explanation of symbols]

[0048] 100 Demand forecasting device, 11 Decomposition unit, 12 Forecasting unit

Claims

1. a decomposition unit that decomposes the order record data of the parts into regular components and irregular components using ensemble empirical mode decomposition; a prediction unit that performs a transition prediction for each of the regular components and the irregular components by using dynamic mode decomposition, and performs a demand prediction for the part based on the transition prediction result. Demand forecasting device.

2. The decomposition unit includes: Decomposing the order record data of the parts into a plurality of intrinsic mode functions representing the regular components and the irregular components using ensemble empirical mode decomposition; The prediction unit is A regression model is constructed based on an output result of dynamic mode decomposition using each of the intrinsic mode functions as an input, a predicted value corresponding to each of the intrinsic mode functions is calculated using the regression model, and a demand forecast for the part is performed by summing up the predicted values ​​corresponding to each of the intrinsic mode functions. The demand forecasting device according to claim 1 .

3. The computer The order history data of the parts is decomposed into regular and irregular components using ensemble empirical mode decomposition. A transition prediction is performed for each of the regular components and the irregular components using dynamic mode decomposition, and a demand prediction for the part is performed based on the transition prediction result. Demand forecasting methods.

4. The order history data of the parts is decomposed into regular and irregular components using ensemble empirical mode decomposition. A process of predicting a transition of each of the regular components and the irregular components by using dynamic mode decomposition and predicting a demand for the part based on the transition prediction result is performed by a computer. program.

Images