Maintenance work assistance device, maintenance work assistance method, and maintenance work assistance program

Abstract

A processor (11) of a maintenance work assistance device (10) acquires operation information of a banknote conveyance device to be subjected to maintenance work, calculates a feature amount on the basis of the operation information, uses a machine learning model to predict, on the basis of the feature amount, the number of days until a failure occurs in the banknote conveyance device, and provides notification of said number of days. For example, the processor (11) determines the presence or absence of notification on the basis of the degree of contribution of the feature amount to the prediction of the number of days until a failure occurs in the banknote conveyance device. Furthermore, for example, the processor (11) calculates a SHAP value as the degree of contribution.

Description

Maintenance work support device, maintenance work support method, and maintenance work support program 【0008】 【0001】 The present disclosure relates to a maintenance work support device, a maintenance work support method, and a maintenance work support program. 【0002】 When a failure occurs in the banknote transport device during operation, first, an operator of the banknote transport device (hereinafter sometimes referred to as "device operator") performs maintenance work. The device operator identifies the cause of the failure and takes countermeasures based on, for example, an error code displayed by the banknote transport device. 【0003】 In addition, when it is difficult for the device operator to handle the failure, a request for dispatch is sent to a worker (hereinafter sometimes referred to as "maintenance worker") who specializes in performing maintenance work on the banknote transport device. The maintenance worker who receives the dispatch request uses specialized knowledge in addition to the error code to identify the cause of the failure and take countermeasures. 【0004】 International Publication No. 2020 / 110208, Japanese Unexamined Patent Application Publication No. 2022-126285, Japanese Unexamined Patent Application Publication No. 2020-173496 【0005】 However, since the error codes and the causes of failures are diverse, even maintenance workers with specialized knowledge may have variations in the accuracy of cause identification and the man-hours required for failure countermeasures, and it may take a long time to recover the banknote transport device. In addition, since the request for dispatch to the maintenance worker occurs suddenly, if it is difficult to adjust the schedule of the maintenance worker, it may take a long time to recover the banknote transport device. Thus, if it takes a long time from the occurrence of a failure in the banknote transport device to its recovery, the operation efficiency of the banknote transport device decreases. 【0006】 Therefore, the present disclosure proposes a technology that can improve the operation efficiency of the banknote transport device. 【0007】 The maintenance work support device of the present disclosure has a processor, and the processor acquires operation information of a banknote transport device that is a target of maintenance work, calculates a feature amount based on the operation information, predicts the number of days until a failure occurs in the banknote transport device using a machine learning model based on the feature amount, and notifies the number of days. 【0008】According to this disclosure, the operational efficiency of the banknote transport device can be improved. 【0009】 Figures showing an example configuration of the maintenance system of this disclosure. Figures showing an example configuration of the maintenance work assistance device of this disclosure. Figures showing an example configuration of the banknote transport device of this disclosure. Figures showing an example of the determination table of this disclosure. Figures showing the first calculation example of the model contribution of this disclosure. Figures showing the second calculation example of the model contribution of this disclosure. Figures showing the third calculation example of the model contribution of this disclosure. Figures showing the fourth calculation example of the model contribution of this disclosure. 【0010】 The embodiments of this disclosure will be described below with reference to the drawings. In the following embodiments, the same parts or processes will be denoted by the same reference numerals, and redundant explanations may be omitted. 【0011】 <Configuration of the Maintenance System> Figure 1 shows an example configuration of the maintenance system of this disclosure. In Figure 1, the maintenance system 1 includes a maintenance work support device 10, an automated teller machine 20, an operations company system 30, and a maintenance company system 40. The automated teller machine 20 has a banknote transport device 21. The maintenance work support device 10 is connected to the automated teller machine 20, the operations company system 30, and the maintenance company system 40 via a network 50. 【0012】 The operating company system 30 is an information system used by the company that operates the automated teller machine 20 (hereinafter sometimes referred to as the "operating company"), and it notifies the administrator of the operating company of the operating status of the automated teller machine 20. An example of an operating company is a retail store. 【0013】 The maintenance company system 40 is an information system used by the company responsible for the maintenance of the ATM 20 (hereinafter sometimes referred to as the "maintenance company"), and it notifies maintenance workers of the operational status of the ATM 20. In other words, the banknote transport device 21 installed in the ATM 20 is one of the items subject to maintenance work by the maintenance company. 【0014】 The maintenance support device 10 predicts the occurrence of a malfunction in the banknote transport device 21 and notifies the operation company system 30 and the maintenance company system 40 of the prediction results. 【0015】<Configuration of the Maintenance Assistance Device> Figure 2 shows an example configuration of the maintenance assistance device of this disclosure. In Figure 2, the maintenance assistance device 10 includes a processor 11, a storage unit 12, and a communication module 13. The communication module 13 is connected to a network 50, and the processor 11 can communicate with each of the ATM 20, the operating company system 30, and the maintenance company system 40 via the network 50 using the communication module 13. An example of the processor 11 is a CPU (Central Processing Unit). An example of the storage unit 12 is storage or memory. 【0016】 The memory unit 12 stores feature information 12A calculated from information indicating the operating status of the banknote transport device 21 (hereinafter sometimes referred to as "device operation information"). The memory unit 12 also stores a machine learning model 12B and a judgment table 12C. 【0017】 <Configuration of the banknote transport device> Figure 3 shows an example of the configuration of the banknote transport device of this disclosure. In Figure 3, the banknote transport device 21 includes a top module 201, a recycling stacker module 202, and a cash box module 203. 【0018】 The top module 201 has a banknote deposit / discharge slot 211 as a mechanism for exchanging banknotes with users of the ATM 20 (hereinafter sometimes referred to as "transaction machine users"). The banknote deposit / discharge slot 211 receives banknotes from transaction machine users and sends the deposited banknotes into the top module 201. The banknote deposit / discharge slot 211 also discharges the banknotes that have been sent to the top module 201 within the banknote transport device 21 so that transaction machine users can receive them. 【0019】 The recycling stacker module 202 includes a recycling stacker 212. The recycling stacker 212 temporarily stores banknotes and ejects the stored banknotes when they are withdrawn. For example, the recycling stacker 212 can store banknotes by sequentially inserting them between films and winding them into a roll. 【0020】The cash box module 203 includes a cash box 213 and an optical sensor 214. The cash box module 203 also includes a pusher (not shown) for pushing banknotes into the cash box 213 and a motor (sometimes referred to as the "pusher motor" below) (not shown) for driving the pusher. An example of the optical sensor 214 is a transmissive optical sensor. The cash box 213 collects banknotes when settling banknotes stored in the recycling stacker 212 and also collects damaged banknotes. The optical sensor 214 is used to count the number of banknotes (sometimes referred to as the "number of collected banknotes") collected into the cash box 213. 【0021】 In Figure 3, banknotes are transported using the transport path 210. For example, banknotes inserted into the banknote deposit / discharge slot 211 are transported via the transport path 210 to the recycling stacker 212 or the cash box 213. The transport path 210 is also used when banknotes are sent from the recycling stacker 212 to the banknote deposit / discharge slot 211 or the cash box 213. 【0022】 <Operation of the Maintenance Assistance Device> The maintenance assistance device 10 has two operation phases: a "learning phase" in which a machine learning model 12B is generated by machine learning, and a "prediction phase" in which the occurrence of a malfunction in the banknote transport device 21 is predicted using the machine learning model 12B. Below, an example of the operation of the maintenance assistance device 10 will be explained, divided into the learning phase and the prediction phase. 【0023】 Furthermore, the following explanation will take the case where the optical sensor 214 is a transmissive optical sensor as an example. Also, the following explanation will take the failure of the optical sensor 214 as an example of a failure occurring in the banknote transport device 21. When the ATM 20 detects a failure in the optical sensor 214, it outputs the error code "E001". 【0024】<Operation in the Learning Phase> The processor 11 obtains daily device operation information from the ATM 20 via the network 50 using the communication module 13. For example, if the daily operating hours of the ATM 20 are from 8:00 to 20:00, the processor 11 obtains device operation information from the ATM 20 at 20:30 every day. 【0025】 Here, the collected banknotes are collected in the cash box 213 via a transport path 210 where an optical sensor 214 is located. The number of collected banknotes is counted by the ATM 20 when the optical sensor 214 detects the passage of the banknotes. For example, when a banknote is transported along the transport path 210 with its longitudinal direction as the direction of travel, the optical sensor 214 detects that the transmission of light is blocked by the banknote for a time corresponding to the length of the banknote (hereinafter sometimes referred to as "longitudinal time"). Therefore, the ATM 20 increments the number of collected banknotes by one each time the time the optical sensor 214 blocks the transmission of light (hereinafter sometimes referred to as "transmission blocking time") reaches the longitudinal time. The ATM 20 then stores the cumulative number of collected banknotes. 【0026】 On the other hand, it has been found that an abnormal transmission blockage time of less than a long duration (hereinafter sometimes referred to as "abnormal blockage time") may appear as a precursor to a malfunction in the optical sensor 214. For example, the abnormal blockage time appears as a very short, pulse-like transmission blockage time. Therefore, the ATM 20 acquires the start time of the blockage of light transmission at the optical sensor 214 (hereinafter sometimes referred to as "blockage start time") and the end time of the blockage of light transmission at the optical sensor 214 (hereinafter sometimes referred to as "blockage end time"). The abnormal blockage time may appear, for example, when the optical sensor 214 is contaminated by paper dust or the like generated from banknotes being transported along the transport path 210. 【0027】 Furthermore, the ATM 20 counts the operating time of the pusher motor for each day and stores the cumulative operating time of the pusher motor. 【0028】The device operation information acquired by the processor 11 from the ATM 20 includes the date, the cumulative number of banknotes collected, the start time of the blocking process, the end time of the blocking process, and the cumulative operating time of the pusher motor. The device operation information also includes an error code "E001" associated with the date on which the malfunction of the optical sensor 214 was detected by the ATM 20. In other words, in the device operation information, a date on which the error code "E001" exists indicates the date on which the malfunction of the optical sensor 214 was detected, and a date on which the error code does not exist indicates a date on which no malfunction was detected in the banknote transport device 21. The processor 11 acquires the device operation information from the ATM 20 on a daily basis. 【0029】 Furthermore, the processor 11 calculates the transparent shutdown time from the shutdown start time and shutdown end time, and counts the number of times a transparent shutdown time less than the longest time (i.e., abnormal shutdown time) occurred (hereinafter sometimes referred to as the "number of abnormal shutdown occurrences"). 【0030】 The processor 11 then calculates daily feature quantities from the device operation information acquired daily. The following three types of feature quantities are used as daily feature quantities calculated by the processor 11: the first feature quantity F1, the second feature quantity F2, and the third feature quantity F3. Hereinafter, the first feature quantity F1, the second feature quantity F2, and the third feature quantity F3 may be collectively referred to as "feature quantity F123". ・First feature quantity F1 = Number of abnormal shutdowns per day ÷ Daily increment of pusher motor operating time ・Second feature quantity F2 = Number of abnormal shutdowns per day ÷ Daily increment of the number of collected banknotes ・Third feature quantity F3 = Daily increment of the number of abnormal shutdowns 【0031】 The processor 11 sequentially stores feature information 12A in the storage unit 12, which is information that associates the date, feature quantity F123, and error code with each other. In the feature information 12A, the date on which the error code "E001" exists indicates the date on which a malfunction of the optical sensor 214 was detected, and the date on which the error code does not exist indicates the date on which no malfunction was detected in the banknote transport device 21. In other words, the presence or absence of the error code "E001" corresponds to the label of the training data used in machine learning. 【0032】The processor 11 then generates a machine learning model 12B by performing machine learning using feature information 12A for a predetermined period as training data. This generates a machine learning model 12B that, when a series of time-series feature quantities F123 for an arbitrary period is input, can predict the number of days until an error with error code "E001" (hereinafter sometimes referred to as "E001 error") occurs, starting from the input date. For example, when the machine learning model 12B is input with the feature quantity F123 for today's date and the feature quantities F123 for a series of past dates for an arbitrary period starting from yesterday's date, it predicts how many days from today the E001 error will occur and outputs the number of days from the input date until the E001 error occurs (hereinafter sometimes referred to as "error grace period") as the prediction result by the machine learning model 12B (hereinafter sometimes referred to as "model prediction result"). Machine learning by the processor 11 is performed periodically, and the machine learning model 12B is updated periodically by the processor 11. 【0033】 <Operation in the prediction phase> The processor 11 predicts the number of days of grace period before a failure occurs using the machine learning model 12B. 【0034】 Furthermore, the processor 11 uses the determination table 12C to determine whether or not to notify the operational company system 30 and the maintenance company system 40 of the number of days allowed before failures occur. Figure 4 shows an example of the determination table of this disclosure. 【0035】In the judgment table 12C shown in Figure 4, the error code, module name, feature type, threshold for the SHAP value (SHapley Additive exPlanation value) (hereinafter sometimes referred to as the "SHAP threshold"), priority, recommended action for the failure, and replacement part information are set in association with each other. For example, in the judgment table 12C shown in Figure 4, for failure E001, the "cache box module," which is the location of the failure, is associated with the first feature F1 and the second feature F2. Also, for example, in the judgment table 12C shown in Figure 4, the first feature F1 is associated with a SHAP threshold of "3.5," a priority of "1," the recommended action being "replacement of the cache box module," and the replacement part number being "CBM01." For example, in the judgment table 12C shown in Figure 4, the second feature F2 is associated with a SHAP threshold of "1.5", a priority of "2", and the recommended action being "cleaning the cache box module". 【0036】 The processor 11 predicts the number of days of failure grace period using the machine learning model 12B, and then calculates a SHAP value using the machine learning model 12B and the feature F123 input to the machine learning model 12B when predicting the number of days of failure grace period. The SHAP value calculated by the processor 11 includes the SHAP value of the first feature F1 (hereinafter sometimes referred to as "first SHAP value S1"), the SHAP value of the second feature F2 (hereinafter sometimes referred to as "second SHAP value S2"), and the SHAP value of the third feature F3 (hereinafter sometimes referred to as "third SHAP value S3"). The first SHAP value S1 indicates the contribution of the first SHAP value S1 to the model prediction result, the second SHAP value S2 indicates the contribution of the second SHAP value S2 to the model prediction result, and the third SHAP value S3 indicates the contribution of the third SHAP value S3 to the model prediction result. In this way, the processor 11 calculates a SHAP value as, for example, the contribution of feature quantity F123 to the model prediction result (hereinafter sometimes referred to as the "model contribution"). Hereinafter, the first SHAP value S1, the second SHAP value S2, and the third SHAP value S3 may be collectively referred to as the "SHAP value S123". 【0037】 In other words, in the judgment table 12C shown in Figure 4, the SHAP threshold of "3.5" corresponding to the first feature F1 represents the threshold for the first SHAP value S1 (hereinafter sometimes referred to as "first threshold TH1"), and the SHAP threshold of "1.5" corresponding to the second feature F2 represents the threshold for the second SHAP value S2 (hereinafter sometimes referred to as "second threshold TH2"). 【0038】 The calculation of model contribution will be explained below with reference to the first, second, third, and fourth calculation examples. Figure 5 shows the first calculation example of model contribution in this disclosure. Figure 6 shows the second calculation example of model contribution in this disclosure. Figure 7 shows the third calculation example of model contribution in this disclosure. Figure 8 shows the fourth calculation example of model contribution in this disclosure. In Figures 5, 6, 7, and 8, the vertical axis represents the type of feature, and the horizontal axis represents the SHAP value. In Figures 5, 6, 7, and 8, the larger the value in the negative direction of the SHAP value, the greater the contribution of feature F123, which causes the E001 failure, to the model prediction result. 【0039】 In the first, second, third, and fourth calculation examples, the case where the reference value RV is "5" will be used as an example for explanation. The reference value RV is calculated by the processor 11. For example, the processor 11 uses the failure grace period output from the machine learning model 12B when the average values ​​of the first feature F1, second feature F2, and third feature F3 used during machine learning of the machine learning model 12B are input to the machine learning model 12B as the reference value RV. The failure grace period predicted by the machine learning model 12B is the sum of the reference value RV and the SHAP value S123. 【0040】Furthermore, in the first, second, third, and fourth calculation examples, the processor 11 refers to the determination table 12C based on the first SHAP value S1 and the second SHAP value S2. When the processor 11 refers to the determination table 12C based on the first SHAP value S1 and the second SHAP value S2, it compares the value obtained by reversing the sign of the first SHAP value S1 (hereinafter sometimes referred to as the "first comparison value C1") with the first threshold TH1, and compares the value obtained by reversing the sign of the second SHAP value S2 (hereinafter sometimes referred to as the "second comparison value C2") with the second threshold TH2. 【0041】 <First Calculation Example (Figure 5)> As shown in Figure 5, in the first calculation example, the processor 11 calculates the first SHAP value S1 as "-3", the second SHAP value S2 as "-1", and the third SHAP value S3 as "+1". When the SHAP values ​​S123 are calculated as shown in Figure 5, the number of days allowed for failure occurrence is "5 - 3 - 1 + 1", which is "2 days". 【0042】 The processor 11 refers to the determination table 12C based on the first SHAP value S1 and the second SHAP value S2, and determines that the first comparison value C1 is less than the first threshold TH1 and the second comparison value C2 is less than the second threshold TH2. Therefore, if the first SHAP value S1 is "-3" and the second SHAP value S2 is "-1", the processor 11 does not notify the operations company system 30 and the maintenance company system 40, including the grace period for failure occurrence. 【0043】 <Second Calculation Example (Figure 6)> As shown in Figure 6, in the second calculation example, the processor 11 calculates the first SHAP value S1 as "-4", the second SHAP value S2 as "-1", and the third SHAP value S3 as "+2". When the SHAP values ​​S123 are calculated as shown in Figure 6, the number of days allowed before failure occurs is "5 - 4 - 1 + 2", which is "2 days". 【0044】The processor 11 refers to the determination table 12C based on the first SHAP value S1 and the second SHAP value S2, and determines that the first comparison value C1 is greater than or equal to the first threshold value TH1 and the second comparison value C2 is less than the second threshold value TH2. Therefore, when the first SHAP value S1 is "-4" and the second SHAP value S2 is "-1", the processor 11 notifies the operation responsible enterprise system 30 and the maintenance responsible enterprise system 40 that the number of days until the occurrence of a failure is "2 days", the recommended measure is "replacement of the cash box module", and the replacement part number is "CBM01". 【0045】 <Third calculation example (Fig. 7)> As shown in Fig. 7, in the third calculation example, the processor 11 calculates, for example, the first SHAP value S1 as "-2", the second SHAP value S2 as "-4", and the third SHAP value S3 as "+3". When the SHAP value S123 is calculated as shown in Fig. 7, the number of days until the occurrence of a failure is "2 days" of "5 - 2 - 4 + 3". 【0046】 The processor 11 refers to the determination table 12C based on the first SHAP value S1 and the second SHAP value S2, and determines that the first comparison value C1 is less than the first threshold value TH1 and the second comparison value C2 is greater than or equal to the second threshold value TH2. Therefore, when the first SHAP value S1 is "-2" and the second SHAP value S2 is "-4", the processor 11 notifies the operation responsible enterprise system 30 and the maintenance responsible enterprise system 40 that the number of days until the occurrence of a failure is "2 days" and the recommended measure is "cleaning of the cash box module". 【0047】 <Fourth calculation example (Fig. 8)> As shown in Fig. 8, in the fourth calculation example, the processor 11 calculates, for example, the first SHAP value S1 as "-4", the second SHAP value S2 as "-2", and the third SHAP value S3 as "+3". When the SHAP value S123 is calculated as shown in Fig. 8, the number of days until the occurrence of a failure is "2 days" of "5 - 4 - 2 + 3". 【0048】The processor 11 refers to the determination table 12C based on the first SHAP value S1 and the second SHAP value S2, and determines that the first comparison value C1 is equal to or greater than the first threshold TH1 and the second comparison value C2 is equal to or greater than the second threshold TH2. Also, since the first comparison value C1 is equal to or greater than the first threshold TH1 and the second comparison value C2 is equal to or greater than the second threshold TH2, the processor 11 refers to the priority and compares the priority of the first feature quantity F1 (hereinafter sometimes referred to as "first priority P1") with the priority of the second feature quantity F2 (hereinafter sometimes referred to as "second priority P2"). Then, the processor 11 determines that the first priority P1 is "1" and has a higher priority than "2" of the second priority P2. Therefore, the processor 11 selects "replacement", which is the recommended countermeasure corresponding to the first feature quantity F1, from among a plurality of recommended countermeasures such as "replacement" and "cleaning". Thus, when the first SHAP value S1 is "-4" and the second SHAP value S2 is "-2", the processor 11 notifies the operation responsible enterprise system 30 and the maintenance responsible enterprise system 40 that the number of days until the occurrence of a failure is "2 days", the recommended countermeasure is "replacement of the cash box module", and the replacement part number is "CBM01". 【0049】 In addition, the processor 11 may notify only one of the operation responsible enterprise system 30 or the maintenance responsible enterprise system 40 of the number of days until the occurrence of a failure and the recommended countermeasure. When the notification of the number of days until the occurrence of a failure and the recommended countermeasure is made to the operation responsible enterprise system 30 but not to the maintenance responsible enterprise system 40, a request for the dispatch of maintenance workers may be made from the operating enterprise to the maintenance responsible enterprise as necessary. 【0050】 The above is an explanation of an example of calculating the model contribution degree. 【0051】Here, all or part of the processes described above in the maintenance assistance device 10 may be implemented by having the processor 11 execute a program corresponding to each process. For example, the program corresponding to each process described above may be stored in the storage unit 12, and the program may be read from the storage unit 12 and executed by the processor 11. Alternatively, the program may be stored in a program server connected to the maintenance assistance device 10 via the network 50 and downloaded from the program server to the maintenance assistance device 10 for execution, or it may be stored in a recording medium readable by the maintenance assistance device 10 and read from that recording medium for execution. Recording media readable by the maintenance assistance device 10 include, for example, portable storage media such as memory cards, USB memory, SD cards, flexible disks, magneto-optical disks, CD-ROMs, and DVDs. 【0052】 The examples described above have been explained. 【0053】 As described above, the maintenance work assistance device of this disclosure (maintenance work assistance device 10 in the embodiment) has a processor (processor 11 in the embodiment). The processor acquires operational information of the banknote transport device (banknote transport device 21 in the embodiment) that is the target of maintenance work, calculates feature quantities based on the operational information, and predicts and notifies the number of days until a failure occurs in the banknote transport device using a machine learning model (machine learning model 12B in the embodiment) based on the feature quantities. 【0054】 This allows operators and maintenance personnel to know how many days it will take for a banknote transport system to malfunction. Therefore, operators and maintenance personnel can take preventative measures before a malfunction occurs, thereby improving the operational efficiency of the banknote transport system. 【0055】 1 Maintenance system 10 Maintenance work support device 11 Processor 12 Memory unit 12A Feature information 12B Machine learning model 12C Decision table 13 Communication module 20 ATM 21 Banknote transport device 30 System of the operating company 40 System of the maintenance company

Claims

1. A maintenance assistance device comprising a processor, the processor acquiring operational information of a banknote transport device subject to maintenance work, calculating feature quantities based on the operational information, predicting the number of days until a failure occurs in the banknote transport device using a machine learning model based on the feature quantities, and notifying the user of the number of days.

2. The maintenance work assistance device according to claim 1, wherein the processor calculates the contribution of the feature quantity to the prediction of the number of days, and determines whether or not to notify the number of days based on the contribution.

3. The maintenance work assistance device according to claim 2, wherein the processor calculates the SHAP value as the contribution.

4. The maintenance work assistance device according to claim 2, wherein the processor calculates a feature including a first feature and a second feature based on the operational information, predicts the number of days using the machine learning model based on the first feature and the second feature, calculates a first contribution which is the contribution of the first feature to the prediction of the number of days, and a second contribution which is the contribution of the second feature to the prediction of the number of days, and determines whether or not to notify the number of days based on a first comparison result which is the result of comparing the first contribution with a first threshold, and a second comparison result which is the result of comparing the second contribution with a second threshold.

5. The maintenance work assistance device according to claim 4, wherein the processor notifies the user of the first comparison result and a recommended action for the fault based on the first comparison result.

6. The maintenance work assistance device according to claim 5, wherein the processor selects the recommended action to be notified from among a plurality of recommended actions based on the comparison result between the first priority, which is the priority of the first feature, and the second priority, which is the priority of the second feature.

7. A maintenance work assistance device that acquires operational information of a banknote transport device that is the subject of maintenance work, calculates feature quantities based on the operational information, predicts the number of days until a failure occurs in the banknote transport device using a machine learning model based on the feature quantities, and notifies the user of the number of days.

8. A maintenance support program that causes a processor to perform the following processes: acquire operational information of a banknote transport device subject to maintenance work; calculate feature quantities based on the operational information; predict the number of days until a failure occurs in the banknote transport device using a machine learning model based on the feature quantities; and notify the processor of the number of days.

Images