CN115186906B

CN115186906B - Intelligent prediction method and platform for container business lease returns

Info

Publication number: CN115186906B
Application number: CN202210830100.1A
Authority: CN
Inventors: 李涵; 顾寅俊; 陈玺文
Original assignee: Florence China Co ltd; Cosco Shipping Technology Co Ltd
Current assignee: Florence China Co ltd; Cosco Shipping Technology Co Ltd
Priority date: 2022-07-15
Filing date: 2022-07-15
Publication date: 2024-01-19
Anticipated expiration: 2042-07-15
Also published as: CN115186906A

Abstract

The invention relates to an intelligent prediction method and platform for container business lease boxes, which are characterized in that based on contract history data and the planned box returning month number, the method obtains the actual box returning data of a client, determines the predicted box returning month number and the box returning proportion of a certain port, and performs analysis and calculation respectively by adopting a specific analysis method and a specific calculation mode according to three conditions that the contract period is not expired, the contract period is expired and no actual box returning data is historic, the contract period is expired and the actual box returning data exists in the month before the month where the current time is, and predicts the box returning amount of a certain port of each month according to a preset chart type. The invention can accurately reveal the box returning trend, assist enterprises in predicting future box returning conditions, help enterprises to adjust and manage operation strategies in time, and display the predicting trend in a preset chart type so that the data presentation mode is more visual and clear.

Description

Intelligent prediction method and platform for container business lease returns

Technical Field

The invention relates to the technical field of container management and informatization construction, in particular to an intelligent prediction method and platform for container business lease returns.

Background

In the face of complex and changeable market environments, container renting and box returning are always an important link in business, and the box returning quantity can directly influence the stock quantity of container asset management enterprises in each port, and further influence the sales strategies of the enterprises and the like.

The existing container asset management enterprises face the following problems in the container renting and returning process, namely 1. Uncertainty of the renting and returning places is caused, and whether customers still cannot control the places specified by contracts or not is caused by returning orders of the containers; 2. uncertainty of box returning time of renting is avoided, and a customer is inevitably delayed to return to the box when returning to the box, so that actual box returning time is difficult to reasonably estimate; 3. the renting and box returning behavior is difficult to estimate, and although there are ports with boxes that do not return for a few months in the past, it is not possible to estimate whether the customer will return to the box at this location in the future.

The method is limited by uncertainty of box returning and renting behaviors, upper limit of box returning in each country and port and other factors, and a scientific and effective container box returning and renting prediction model does not exist in the industry at present. Meanwhile, due to the intervention of more internal and external environmental factors such as the operation cost, the way factors, the expected benefits and the like, a container asset management enterprise currently does not have a set of available renting and returning box prediction model for effectively guiding and developing container renting and returning box work, and the operation management efficiency and the asset utilization rate of the enterprise are improved in an auxiliary manner.

Disclosure of Invention

In order to solve the problems faced by the container asset management enterprises in the container renting and returning process, the invention provides an intelligent prediction method for the container operational renting and returning, based on historical data, the returning quantity is intelligently predicted by adopting a specific calculation mode while considering factors such as the upper limit of a regional returning box, the upper limit of a port box type month returning box and the like, the returning trend can be automatically and accurately revealed, the container asset management enterprises are assisted in predicting the future returning situation, the enterprises are helped to timely adjust and manage the operation strategy, in addition, the predicting trend is displayed in the form of a histogram and a line diagram in a region, a box type and a port, and the change trend and the change amplitude of the returning data are displayed, so that the data presentation mode is more visual and clear. The invention also relates to an intelligent prediction platform for the container operational lease also.

The technical scheme of the invention is as follows:

the intelligent prediction method for the container operational lease also is characterized by comprising the following steps:

s1: automatically acquiring actual box returning data of a client according to contract history data, automatically determining the expected box returning month number according to the box returning month number of a client plan and/or the contract history data, intelligently analyzing and determining the box returning proportion of a certain box type of a certain port according to the actual box returning data of the history, automatically judging the contract period, executing the step S2 when the contract period is not expired, and executing the step S3 or the step S4 when the contract period is expired;

S2: when the contract term is not expired, intelligently predicting the box returning amount of a certain box type of a certain port in each month by adopting a Prophet algorithm according to the predicted box returning month number and the box returning proportion of the certain box type of the certain port and combining with the upper limit of the monthly box returning amount of the certain box type of the certain port and/or the upper limit of the box returning amount of the certain box type of the certain region and the historical actual box returning data, and entering into the step S5 after the prediction is finished;

s3: when the contract term expires and no historical actual box returning data exists, but a customer applies for a box returning plan, the first month takes the pre-box returning amount in the customer applies for the box returning plan as a predicted box returning amount, intelligent prediction is carried out on the second month and the post-month box returning amount according to the prediction mode of the step S2, and the step S5 is carried out after the prediction is finished;

s4: when the contract term expires and historical actual box returning data exists in the month before the month of the current time, obtaining the box returning delay time of each box according to the box returning time data in the historical actual box returning data of the month, and calculating the average box returning delay time of each box; calculating the quantity of boxes which can be returned in the month after the month of the current time in all unreturned boxes according to the average box returning delay time of each box, and intelligently predicting the box returning quantity of the month after the month of the current time by adopting a Prophet algorithm according to the quantity of boxes returned in the month after the month of the current time, the average box returning delay time and the upper limit of the monthly box returning quantity of a certain port;

Calculating weekly average box returning amount and weekly average box returning increment according to the actual box returning amount of the month before the month at the current time and the predicted box returning amount of the month after the month at the current time, intelligently predicting the box returning amount of a week in the future according to the weekly average box returning amount and the weekly average box returning increment, and further intelligently predicting the box returning amount of a month in the future;

s5: and displaying the intelligently predicted box returning quantity according to a preset chart type.

Preferably, in the step S4, after predicting the amount of the return box for a month in the future, the method further includes the following steps:

a first step of: comparing the predicted box returning amount of a certain box type of a certain port in the future with the monthly box returning amount upper limit of a certain box type of the certain port, and taking the predicted box returning amount of the certain box type of the certain port in the month as the actual box returning amount of the certain port if the predicted box returning amount of the certain box type of the certain port in the future in the month is smaller than the monthly box returning amount upper limit of the certain box type of the certain port;

if the predicted box returning amount of a certain box of a certain port in the future month is more than or equal to the monthly box returning amount upper limit of the certain box of the certain port, taking the month box returning amount upper limit of the box of the certain port as the actual box returning amount of the month;

and a second step of: comparing the total number of all predicted box returning amounts and the predicted box returning amounts of a box in a certain region in the future and before the month with the upper box returning amount limit of the box in the certain region, and taking all predicted box returning amounts of the box in the region before the month and the month as the actual box returning amount of the region if the total number is smaller than the upper box returning amount limit of the box in the region;

If the total number is greater than or equal to the upper limit of the box returning amount of the box type in the region, taking the upper limit of the box returning amount of the box type in the region as the actual box returning amount of the region, dividing the box amount exceeding the upper limit of the box returning amount into all ports under the region, and subtracting the divided box amount exceeding the upper limit from the predicted box returning amount of the last week of the month of each port.

Preferably, after judging the actual box returning amount of a certain month in the future, if the total predicted box returning amount in the next month or the last months of the month is equal to zero, ending the box returning prediction process; if the total predicted box returning amount in the next month or the last months of the month is larger than zero, automatically judging whether the number of the months and the previous months is smaller than the maximum predicted box returning number of the months, if the number of the months and the previous months is smaller than the maximum predicted box returning number of the months, continuously predicting the predicted box returning amount of the next month of the month, and if the number of the months and the previous months is larger than or equal to the predicted box returning number of the months, ending the box returning prediction process.

Preferably, in the step S5, the chart type includes at least one of a summary histogram, a regional graph, a box-type line graph, and a port histogram.

Preferably, in the step S1, a topic table is also created according to the historical actual box returning data, where the topic table includes a customer ID, a contract number, a contract type, a box returning area, a box returning port, and a box returning amount.

The intelligent prediction platform for the operational lease of the container is characterized by comprising a first module, a second module, a third module, a fourth module and a fifth module, wherein the first module is respectively connected with the second module, the third module and the fourth module, the second module, the third module and the fourth module are respectively connected with the fifth module,

the first module automatically acquires actual box returning data of the customer history according to the contract history data, automatically determines the estimated box returning month number according to the box returning month number of the customer plan and/or the contract history data, intelligently analyzes and determines the box returning proportion of a certain box type of a certain port according to the actual box returning data of the history, then automatically judges the contract period, executes the second module when the contract period is not expired, and executes the third module or the fourth module when the contract period is expired;

the second module intelligently predicts the box returning amount of a certain box type of a certain port in each month by adopting a Prophet algorithm according to the predicted box returning month number and the box returning proportion of the certain box type of the certain port and by combining the upper limit of the monthly box returning amount of the certain box type of the certain port and/or the upper limit of the box returning amount of the certain box type of the certain region and the historical actual box returning data, and enters the fifth module after the prediction is finished;

The third module is used for intelligently predicting the returning quantity of the second month and the following months according to the prediction mode of the second module when the contract period is expired and no historical actual returning data exists, but when a customer applies for a returning plan, the first month is used as the predicted returning quantity according to the pre-returning quantity in the customer applies for the returning plan, and the second month and the following months enter the fifth module after the prediction is finished;

a fourth module for obtaining the time delay of the box according to the time data of the box in the historical actual box-returning data of the month when the contract period is expired and the historical actual box-returning data exists in the month before the current time, and calculating the average time delay of the box; calculating the quantity of boxes which can be returned in the month after the month of the current time in all unreturned boxes according to the average box returning delay time of each box, and intelligently predicting the box returning quantity of the month after the month of the current time by adopting a Prophet algorithm according to the quantity of boxes returned in the month after the month of the current time, the average box returning delay time and the upper limit of the monthly box returning quantity of a certain port;

And a fifth module for displaying the intelligently predicted box returning amount according to a preset chart type.

Preferably, in the fourth module, after predicting the return amount of a certain box in a certain month in the future, the predicted return amount of the certain box in the certain month in the future is compared with the monthly return amount upper limit of the certain box in the certain port, and if the predicted return amount of the certain box in the certain month in the certain port in the future is smaller than the monthly return amount upper limit of the certain box in the certain port, the predicted return amount of the box in the certain port in the certain month is taken as the actual return amount of the certain month;

comparing the total number of all the predicted box returning amounts and the predicted box returning amounts of a box in a certain region in the future and before the month with the upper box returning amount limit of the box in the certain region, and if the total number is smaller than the upper box returning amount limit of the box in the region, taking all the predicted box returning amounts of the box in the region and before the month as the actual box returning amount of the region;

Preferably, in the fourth module, after determining the actual box returning amount for a certain month in the future, if the total predicted box returning amount in the month or months after the month is equal to zero, the box returning prediction process is ended; if the total predicted box returning amount in the following month or the following months of the month is larger than zero, judging whether the number of the months and the previous months is smaller than the maximum predicted box returning number of the months, if the number of the months and the previous months is smaller than the maximum predicted box returning number of the months, continuously predicting the predicted box returning amount of the next month of the month, and if the number of the months and the previous months is larger than or equal to the predicted box returning number of the months, ending the box returning prediction process.

Preferably, the graph type includes at least one of a summary bar graph, a regional graph, a box line graph, and a port bar graph.

Preferably, in the first module, a topic table is also created according to historical actual box returning data, and the topic table includes a client ID, a contract number, a contract type, a box returning area, a box returning port and a box returning amount.

The invention has the following technical effects:

the invention provides an intelligent prediction method for container business lease returns, which is characterized in that actual customer return data is obtained based on contract history data, the expected return month is determined according to the return month number of a customer plan and/or the contract history data, the return proportion of a certain port box type is analyzed and determined according to the actual return month data, and when the contract period is not expired and is expired, the factors such as the upper limit of regional returns, the port box type and the upper limit of monthly returns are considered, and meanwhile, the return quantity under various conditions is intelligently predicted by adopting a specific calculation mode, so that the calculation result is more scientific and reasonable, and the return trend can be accurately revealed by means of a large number of historical data and calculation means on the basis of applying objective rules. According to the invention, according to the lease contract, the historical data is consulted to predict the future case returning situation, the risk of long-term idle of the case can be avoided in advance, the container asset management enterprise is assisted to predict the future case returning situation, the enterprise is helped to adjust the management operation strategy in time, in addition, the prediction trend is displayed in the form of a histogram, a graph and a line diagram in the form of a regional, a case-dividing and a port-dividing, the change trend and the change amplitude of the case returning data are displayed, and the data presentation mode is more visual and clear.

The invention also provides an intelligent prediction platform for the container business lease also, which corresponds to the intelligent prediction method for the container business lease also, and can be understood as a platform for realizing the intelligent prediction method for the container business lease also, the intelligent prediction platform for the container business lease is a background or a server, the platform cooperates with each other through five modules based on historical data, and the special calculation mode is adopted to predict the still box quantity while considering factors such as the upper limit of the regional still box, the port box type, the upper limit of the month still box and the like, so that the multi-channel intelligent prediction still box quantity under different conditions can be accurately disclosed, the future still box situation can be predicted by a container asset management enterprise, and the enterprise can be helped to adjust management operation strategies in time. The invention starts from the actual service scene, integrates the management requirements of the container asset management enterprises, the related internal and external service data and the related historical data, designs and realizes a set of intelligent prediction platform meeting the requirements of the container asset management enterprises on the container renting and returning, helps management staff to know the future returning situation in advance, and further analyzes based on the returning situation.

Drawings

FIG. 1 is a flow chart of the intelligent prediction method for the container business lease also in the present invention.

FIG. 2 is a preferred flow chart of the intelligent prediction method for container business rental boxes of the present invention.

Figures 3-9 are further preferred flow diagrams of the intelligent prediction method for container business rental boxes of the present invention.

FIG. 10 is a graph showing the effect of the month prediction box data of the present invention.

FIG. 11 is a graph showing another effect of the weekly prediction of the present invention on the data of the still box.

FIG. 12 is a representation of predicted detail data of the present invention.

Detailed Description

The present invention will be described below with reference to the accompanying drawings.

The invention relates to an intelligent prediction method for a container returning operation lease case, which is shown in a flow chart of fig. 1, and sequentially comprises the following steps:

s1: acquiring actual box returning data of a client according to contract history data, automatically determining the expected box returning month number according to the box returning month number of a client plan and/or the contract history data, intelligently analyzing and determining the box returning proportion of a certain box type of a certain port according to the actual box returning data of the history, automatically judging the contract term, executing the step S2 when the contract term is not expired, and executing the step S3 or the step S4 when the contract term is expired;

Specifically, as shown in the preferred flow chart of fig. 2, the estimated number of still months MF is first determined according to the number of still months selected by the customer, and when the user does not select the number of still months, that is, the estimated number of still months MF is empty, the estimated number of months according to the combined estimated down from & estimated down to is estimated, and when both are empty, the estimated number of months is estimated by default of 12 months, and when both are not empty, the estimated number of months is estimated according to MF.

And then manually inputting a reference contract number, automatically acquiring historical actual box returning data of the clients in the reference contract within two years according to contract historical data, acquiring the historical actual box returning data of the clients in the corresponding type of the contracts within two years if the reference contract number is not manually input, creating a theme table through the historical actual box returning data, and intelligently analyzing and determining the box returning proportion of a certain box type of a certain port through the historical actual box returning data. The topic table structure is as follows: customer ID, contract number, contract type, box Region, port, and volume.

S2: when the contract term is not expired, intelligently predicting the box returning amount of a certain box in a certain port in each month by adopting a Prophet algorithm (a time sequence prediction algorithm) according to the predicted box returning month number and the box returning proportion of the certain box in a certain port, combining the upper limit of the box returning amount of the certain box in a certain port (namely Monthly Caps of the contract) and/or the upper limit of the box returning amount of the certain box in a certain Region (namely Region Caps of the contract) and the historical actual box returning data, and displaying the intelligently predicted box returning amount according to a preset chart type after intelligently predicting the box returning amount of the certain box in the certain port in each month;

Specifically, when there is no Region Caps & no montaly Caps, a certain box per month is returned to the certain port = at the rented box/the estimated number of returned months (a certain port certain box history returned to the box/a total box history returned to all ports);

for example: customer a, contract currently rents 1000, expects to refund for mf=6 months, historical refund as follows:

AM LAX D20 20 QTY

AM NYC D20 20 QTY

AP SHA D20 20 QTY

AP NGB D20 10 QTY

EU FRA D20 15 QTY

EU LON D20 25 QTY

note that: AM refers to the american region, AP refers to the asian region, EU refers to the european region; LAX, NYC, SHA, NGB, FRA, LON are all port code abbreviations; d20 is box; QTY refers to a unit of measure. The fourth column 20, 10, 15, 25 represents the historical bin load.

Assume that: when no Region Caps and no Monthly Caps exist, the prediction of each box type of each port is further performed by a small box amount: i.e., the round function returns a floating point number five-house six-in value.

AM LAX D20 round(1000/6,0)*(20/(20+20+20+10+15+25)＝31

AM NYC D20 round(1000/6,0)*(20/(20+20+20+10+15+25)＝31

AP SHA D20 round(1000/6,0)*(20/(20+20+20+10+15+25)＝31

AP NGB D20 round(1000/6,0)*(10/(20+20+20+10+15+25)＝16

EU FRA D20 round(1000/6,0)*(15/(20+20+20+10+15+25)＝23

EU LON D20 round(1000/6,0)-(31+31+31+16+23)＝34

The last month is predicted to be the case: (1000-round (1000/6, 0) ×5) =170

AM LAX D20 170*(20/(20+20+20+10+15+25)＝31

AM NYC D20 170*(20/(20+20+20+10+15+25)＝31

AP SHA D20 170*(20/(20+20+20+10+15+25)＝31

AP NGB D20 170*(10/(20+20+20+10+15+25)＝16

EU FRA D20 170*(15/(20+20+20+10+15+25)＝24

EU LON D20 170-(31+31+31+16+24)＝37

When there are no regions Caps & Monthly Caps, monthly box return amount of a certain port = box rental amount/estimated box return month number (historical box return amount of a certain port/total historical box return amount of certain boxes of all ports); if the Monthly return amount of a certain box at a certain port is greater than the montaly Caps, the Monthly return amount of a certain box at a certain port=montaly Caps; if the Monthly box returning amount of a certain box of a certain port is smaller than the Monthly Caps, taking the predicted Monthly box returning amount of the certain box of the certain port as the actual Monthly box returning amount;

Assume that: when there is no Region Caps & there is a montaly Caps, the LAX and NYC have montaly Caps, montaly caps=30, there is a port priority calculation for montaly Caps:

AM LAX D20 round(1000/6,0)*(20/(20+20+20+10+15+25)＝31then 30

AM NYC D20 round(1000/6,0)*(20/(20+20+20+10+15+25)＝31then 30

AP SHA D20 round(1000/6,0)*(20/(20+20+20+10+15+25)＝31

AP NGB D20 round(1000/6,0)*(10/(20+20+20+10+15+25)＝16

EU FRA D20 round(1000/6,0)*(15/(20+20+20+10+15+25)＝23

EU LON D20 round(1000/6,0)-(30+30+31+16+23)＝36

the last month is predicted to be the case: (1000-round (1000/6, 0) ×5) =170

AM LAX D20 170*(20/(20+20+20+10+15+25)＝31then 30

AM NYC D20 170*(20/(20+20+20+10+15+25)＝31then 30

AP SHA D20 170*(20/(20+20+20+10+15+25)＝31

AP NGB D20 170*(10/(20+20+20+10+15+25)＝16

EU FRA D20 170*(15/(20+20+20+10+15+25)＝24

EU LON D20 170-(30+30+31+16+24)＝39

When there is a Region Caps and no montaly Caps, if the sub Region Caps (a small upper limit range below the Region Caps) is greater than or equal to the box renting amount of a box of the sub Region, then the Monthly box renting amount of a box of a port = the box renting amount/the estimated month number of a box of the sub Region of the port (historical box retuning amount of a box of a port/total historical box retuning amount of a box of a sub Region of the port);

if the sub-Region Caps is smaller than the box renting amount of the sub-Region box, the monthly box returning amount of the port is equal to the month number of the sub-Region Caps/the estimated box returning amount (the historical box returning amount of the port/the total historical box returning amount of the sub-Region box);

assume that: with Region Caps & no montaly Caps, am=300, with preference for Region Caps, region Caps need to be calculated by subtracting the number of bins:

AM LAX D20 round(300/6,0)*(20/(20+20)＝25

AM NYC D20 round(300/6,0)*(20/(20+20)＝25

AP SHA D20 round(1000-300)/6,0)*(20/(20+10+15+25)＝34

AP NGB D20 round(1000-300)/6,0)*(10/(20+10+15+25)＝17

EU FRA D20 round(1000-300)/6,0)*(15/(20+10+15+25)＝25

EU LON D20 round(1000-300)/6,0)-(34+17+25)＝41

the last month is predicted to be the case: AM 300-300/6*5 =50 (700-round (700/6, 0) ×5) =115

AM LAX D20 50*(20/(20+20)＝25

AM NYC D20 50*(20/(20+20)＝25

AP SHA D20 115*(20/(20+10+15+25)＝33

AP NGB D20 115*(10/(20+10+15+25)＝17

EU FRA D20 115*(15/(20+10+15+25)＝25

EU LON D20 115-(33+17+25)＝40

1) When there is a Region Caps & a montaly Caps, if the sub Region Caps are equal to or greater than the sub Region box in-box renting amount, then the port box per month box returning amount=the sub Region box in-box renting amount/the estimated number of box returning months (a port box history box returning amount/a sub Region box total history box returning amount);

If the port box per month is greater than the montaly Caps, the per month box per month=montaly Caps

If the Monthly box amount of port boxes is smaller than the montaly Caps, the Monthly box amount=the Monthly box amount of port boxes

2) If the sub-Region Caps is smaller than the sub-Region box renting amount, the port box monthly box returning amount=the sub-Region Caps/the estimated number of box returning months (a port box historical returning amount/a sub-Region box total historical returning amount);

if the port box per month is greater than the montaly Caps, the per month box per month = montaly Caps

If the Monthly box amount of port boxes is smaller than the montaly Caps, monthly box amount = Monthly box amount of port boxes

Assume that: there are regions Caps & there are montaly Caps, AM (upper limit of america) =300, LAX and NYC montaly caps=24, SHA and NGB montaly caps=32, there are priorities for regions Caps, there are second priorities for montaly Caps, region Caps need to be calculated by subtracting the number of bins already in place:

AM LAX D20 round (300/6, 0) × (20/(20+20))=25then 24 (upper regional & upper lunar limit)

AM NYC D20 round (300/6, 0) × (20/(20+20))=25then 24 (upper regional & upper lunar limit)

AP SHA D20 round ((1000-300)/6, 0) × (20/(20+10+15+25))=34then32 (upper month limit)

AP NGB D20 round((1000-300)/6,0)*(10/(20+10+15+25))＝17

EU FRA D20 round((1000-300)/6,0)*(15/(20+10+15+25))＝25

EU LON D20 round((1000-300)/6,0)-(32+17+25)＝43

The last month is predicted to be the case:

AM area: total bin amount (48×5=240) =60 for the first 5 months in 300-AM region

AP & EU area: total bin amount (117×5=585) =115 for the first 5 months of the 700-AP & EU region

AM LAX D20 60 (20/(20+20) =30then24 (upper limit of area & upper limit of month)

AM NYC D20 60 x (20/(20+20) =30then24 (upper region & month limit)

AP SHA D20 115 (20/(20+10+15+25) =33then32 (upper month limit)

AP NGB D20 115*(10/(20+10+15+25)＝17

EU FRA D20 115*(15/(20+10+15+25)＝25

EU LON D20 115-(32+17+25)＝41

S3: when the contract term expires and no actual box returning data exists, but a customer applies for a box returning plan, the first month is taken as a predicted box returning amount according to the preset box returning amount in the customer applies for the box returning plan, the second month and the following months (namely, future MF-1 month) are intelligently predicted according to the prediction mode when the contract term does not expire, after intelligently predicting the box returning amount of a certain box type of a certain port in each month, the step S5 is carried out, and the intelligently predicted box returning amount is displayed according to a preset chart type.

S4: when the contract term is expired and historical actual box returning data exists in the month before the month of the current time, obtaining the box returning delay time of each box according to the box returning time pre date data in the historical actual box returning data of the month, and then averaging to obtain the average box returning delay time GI LAG of each box; calculating the quantity of boxes which can be returned in the month after the month of the current time in all unreturned boxes according to the average box returning delay time of each box, and intelligently predicting the box returning quantity of the month after the month of the current time by adopting a Prophet algorithm according to the quantity of boxes returned in the month after the month of the current time, the average box returning delay time and the upper limit of the monthly box returning quantity of a certain port;

Specifically, as shown in fig. 3-4, pre-clear data is obtained according to contract numbers, namely actual box returning data of each box type of each port in the month (1 st to 4 th weeks) before the month in which the current time is located is obtained; obtaining open pre-clear data according to contract numbers, namely how many boxes are not yet applied for, and pre-clear date of each open pre-clear, namely the box returning time of all boxes returned by the contract, obtaining the box returning delay time of each box according to the box returning time pre-clear date of all boxes returned by the contract, and then averaging to obtain average box returning delay time GI LAG; then calculating the quantity of boxes which can be returned in the next month (namely 5-8 weeks and the 1 st month in the future) of the month in which the current time is located in all unreturned boxes according to the average box returning delay time; and finally, intelligently predicting the final box returning amount of the month (namely 5-8 weeks) after the month of the current time by adopting a Prophet algorithm according to the quantity of boxes returned in the month (namely 5-8 weeks) after the month of the current time, the average box returning delay time GI LAG and the upper limit of the monthly box returning amount of a certain box type of a certain port.

And then calculating weekly average box returning amount and weekly average box returning increment according to the actual box returning amount of the month before the current time and the predicted box returning amount of the month after the current time, intelligently predicting the box returning amount of a week in the future according to the weekly average box returning amount and the weekly average box returning increment, and further intelligently predicting the box returning amount of a month in the future.

Specifically, as shown in fig. 5, firstly, obtaining the case renting amount according to contract numbers and all pre-clear case not yet, and calculating total case returning amount which is not yet applied and needs to be predicted in 9 to (4 x mf+4) weeks;

then, as shown in fig. 6, the amount of reduction in the future month i is predicted: calculating an average per week box returning quantity Q and an average box returning increment QD by using the box returning quantity Q and the average box returning increment QD of 8 weeks according to the actual box returning quantity of the month before the current time (namely 1-4 weeks) and the predicted box returning quantity of the month after the current time (namely 5-8 weeks), calculating the box returning quantity of the K th week in the future by Q+K, taking K=1, 2,3,4 … n and corresponding to the 9,10,11,12 … n weeks in the future and the 2,3, … n months (4 weeks of the i th month are 4i+1-4i+4);

taking the 4-week bin amount as an example, taking 5, 10 and 10 from the first week to the fourth week bin amount in turn, the average bin amount q= (5+5+10+10)/4=7.5≡8, and the average bin increment qd= [ (5-5) + (10-5) + (10-10) ]/3=1.666≡2.

After predicting the return amount of the next week K, as shown in FIG. 7, comparing the predicted return amount of a certain box type of a certain port in the future month (i is more than or equal to 2) (namely 4i+1-4i to +4 weeks) with the upper limit Monthly cap of the box type of the certain port in each month, judging the return amount of the future month according to the Monthly cap, and taking the predicted return amount of the box type of the certain port in the month as the actual return amount if the predicted return amount of the box type of the certain port in the i month is smaller than the upper limit of the box type of the certain port in each month, namely the predicted return amount of the box type of the port in the month < Monthly cap;

If the predicted box returning amount of a certain box in a certain port in the ith month is more than or equal to the upper limit of the Monthly box returning amount of the certain box in the certain port, namely the predicted box returning amount of the box in the certain port in the month is more than or equal to Monthly cap, taking the upper limit of the Monthly box returning amount of the box in the certain port as the actual box returning amount;

for example: if a month, the box-type box-returning amount of a port is 210,230,250,270, and the box-type Monthly cap of the port is 800, 210+230+250+270=960 >800, 160 boxes cannot be returned, and the month 4 can only be returned to 110, namely the month box-returning amount is 210,230,250,110.

As shown in fig. 8, comparing the total number of all predicted and returned boxes of a certain box in a certain region in the future (i is more than or equal to 2) (i.e. 4i+1 to 4i to +4 weeks) and before the month with the upper limit region cap of the box of the certain region, judging the returned boxes of the future (i) month according to the region cap, and taking all predicted and returned boxes of the box in the region before the month and the month as actual returned boxes if the total number is smaller than the upper limit of the box of the region, i.e. all predicted and returned boxes of the box in the region before the month and the month, +returned boxes < region cap;

if the total number is greater than or equal to the upper limit of the box-shaped box-returning amount of the region, namely, the total predicted box-returning amount+the returned box-shaped amount before the month and the month of the region is greater than or equal to the region cap, taking the upper limit of the box-shaped box-returning amount of the region as the actual box-returning amount, dividing the box-shaped amount exceeding the upper limit of the box-shaped amount into all ports under the region, and subtracting the box-shaped amount exceeding the upper limit from the predicted box-returning amount of the last week of the month of each port.

For example: if the total amount of the box in the AP area is 3900, the region cap of the box in the AP area is 3700, 200 boxes in the box in the AP area cannot be returned in the month, the 200 boxes are divided into two ports under the area on average, the box in the month D20 and the box in the Port 1 are 230,250,270,190, and the box in the AP area is 210,230,250,170,Port 2, and the box in the AP area is 210,230,250,270, the box in the month and the box in the Port 1 are 1900 and 230,250,270,290.

After the i-th month's recycle amount is judged, as shown in fig. 9, the total recycle amount of 9 to (4×mf+4) weeks-the predicted recycle amount of 2 to i-th months (MF month is predicted if the predicted recycle amount MF is filled, MF 12 months is predicted if the MF is not filled, 2 nd month is the 9 th to 12 th weeks, … …, mf+1 to 4×mf+4 weeks are the future MF month), whether the remaining recycle amount is greater than zero is judged, and if no recycle amount is left, the process is ended; if the remaining box returning amount is greater than zero, judging whether i is smaller than MF, if i is smaller than MF, circulating i=i+1, and continuously predicting the box returning amount of the next month; otherwise, if i is greater than or equal to MF, ending.

S5: and displaying the intelligently predicted box returning quantity according to a preset chart type. Preferably, the chart type includes at least one of a summary bar chart, a regional graph, a box-type line chart, and a port bar chart, wherein the summary bar chart shows a total trend of the box-type predictions; the regional graph shows a return box prediction trend according to the region; the box-type line graph displays a box-returning prediction trend according to the box type; the port histogram shows the trend of the box forecast according to the port.

The step S5 is to graphically display the month-to-tank amount or the week-to-tank amount intelligently predicted in the steps S2, S3, S4, etc., wherein fig. 10 is a graph showing the effect of the embodiment of the predicted month-to-tank data, showing the month prediction trend according to the region, the box type and the port, respectively, wherein the curves showing the america region, the asia region and the europe region show the region prediction trend according to the month, the a curve shows the prediction trend of the D4H box type according to the month, the B bar graph shows the port with the month-to-tank amount being greater than or equal to the month upper limit, and the other bar graphs show the port with the month-to-tank amount being less than the month upper limit. Fig. 11 is an example showing effect graphs of predicted peri-bin data showing predicted trends by region, bin and port, respectively, wherein curves representing american, asia and european regions show predicted trends by region, C curves show predicted trends of D4H bin by region, D bar graphs show ports with peri-bin numbers equal to or greater than the upper month limit, and other bar graphs show ports with peri-bin numbers less than the upper month limit. The map can also be displayed, the map can be provided with a prediction detail of a box type of a port, the port is marked with a box, the predicted box quantity of each box type can be displayed in the future for several months by clicking the port, and the box type can be displayed with the box type prediction detail data of the box type of the port. As shown in fig. 12, the detailed prediction result may be downloaded in a list, and the box-in prediction detail data may be downloaded by clicking a download button, including: time, region, port, box, and also box quantity QTY, TEU.

The invention also relates to a container operation lease case intelligent prediction platform which corresponds to the container operation lease case intelligent prediction method, which can be understood as a platform for realizing the container operation lease case intelligent prediction method, the platform is a background or a server, can be inherited as a container operation lease case intelligent prediction APP, and comprises a first module, a second module, a third module, a fourth module and a fifth module, wherein the first module is respectively connected with the second module, the third module and the fourth module, the second module, the third module and the fourth module are respectively connected with the fifth module,

then calculating weekly average box returning amount and weekly average box returning increment according to the actual box returning amount of the month before the month of the current time and the predicted box returning amount of the month after the month of the current time, intelligently predicting the box returning amount of a certain week in the future according to the weekly average box returning amount and the weekly average box returning increment, and further intelligently predicting the box returning amount of a certain month in the future;

A fifth module: and displaying the intelligently predicted box returning quantity of the second module, the third module or the fourth module according to a preset chart type.

Preferably, in the fourth module, after predicting the return amount of a certain box type of a certain port in the future in a certain month, the predicted return amount of the certain box type of the certain port in the future in a certain month is compared with the upper limit of the return amount of the certain box type of the certain port in the month, and if the predicted return amount of the certain box type of the certain port in the future in a certain month is smaller than the upper limit of the return amount of the certain box type of the certain port in the month, the predicted return amount of the box type of the certain port in the month is taken as the actual return amount of the certain port in the month;

comparing the total number of all predicted box returning amounts and the predicted box returning amounts of a box in a certain region in the future and before the month with the upper box returning amount limit of the box in the certain region, and taking all predicted box returning amounts of the box in the region before the month and the month as the actual box returning amount of the region if the total number is smaller than the upper box returning amount limit of the box in the region;

Preferably, the chart type includes at least one of a summary bar chart, a regional graph, a box line chart, and a port bar chart.

Preferably, in the first module, a topic table is also created according to the historical actual box returning data, wherein the topic table comprises a client ID, a contract number, a contract type, a box returning area, a box returning port and a box returning amount.

The PC end framework of the platform is based on the packaging of a light-weight J2EE framework of Spring boot+Hibernate JPA, and the front end interface layer adopts a Vue technology and uses an Ant design UI framework; the back end adopts a micro-service architecture, adopts RPC communication, takes a Zookeeper as a registration center, uses Redis cache service, takes a Nacos configuration center and introduces a Docker container technology, adopts a Tomcat as a method application server, and is deployed on Linux. The front-end infrastructure adopts an MVVM design mode and communicates with the network using http. The MVVM mode is an event-driven programming mode that simplifies the user interface.

The platform PC end frame uses a currently popular B/S structure, and is divided into three layers:

front end interface layer: according to the WEB front end of the data interaction standard required by the ant design or the front end of the development framework based on the ant design UI, a user can operate only by a common webpage browser.

Application server layer: the method adopts the industry mature technology, uses Tomcat7 as a method application server, and is deployed on Linux to provide encapsulated application service support.

Database access layer: in the aspect of tuning, the Hibernate JPA has a Session mechanism and a secondary cache, and can optimally design SQL. The method can perfectly support large database methods such as Oracle, mySQL, SQL Server and the like.

Spring is a lightweight control inversion (IoC) and tangential plane (AOP) facing container frame. It has the following characteristics:

lightweight-Spring is lightweight in both size and overhead.

Control inversion-Spring facilitates loose coupling by a technique called control inversion (IoC).

Section-oriented Spring provides rich support for section-oriented programming, allowing cohesive development by separating application business logic from method-level services (e.g., transaction (transaction) management, etc.).

The Spring Boot is a brand new framework provided by the Pivotal team, and the design purpose of the Spring Boot is to simplify the initial construction and development process of a new Spring application. The framework is configured in a particular manner so that developers no longer need to define a templated configuration.

Dock is an open-source application container engine that allows developers to package their applications and rely on packages into a portable container and then release them to any popular Linux machine, and also allows virtualization. The container is completely using a sandbox mechanism without any borrowing of each other.

RPC (Remote Procedure Call) -remote invocation procedure, which is a protocol that requests services from a remote computer program over a network without requiring knowledge of the underlying network technology. The RPC protocol assumes the existence of certain transport protocols, such as TCP or UDP, to carry information data between communication programs. In the OSI network communication model, the RPC spans a transport layer and an application layer. RPC makes it easier to develop applications including network distributed multiprogramming.

The RPC adopts a client/server mode. The requesting program is a client and the service provider is a server. First, the client calling process sends a call message with process parameters to the service process and then waits for a response message. At the server side, the process remains dormant until the call information arrives. When one call information arrives, the server obtains the process parameters, calculates the result, sends the reply information, then waits for the next call information, finally, the client calls the process to receive the reply information, obtains the process result, and then calls the execution to continue.

Nacos is a service infrastructure that builds modern application architectures (e.g., micro-service paradigm, yun Yuansheng paradigm) centered on "services". Support DNS-based and RPC-based service discovery (which may be a registry of a springgroup), dynamic configuration services (which may be a configuration center), dynamic DNS services. In an effort to facilitate discovery, configuration, and management of micro-services. A set of simple and easy-to-use feature sets is provided to facilitate dynamic service discovery, service configuration management, service and traffic management. The micro-service platform is more agile and easy to build, deliver and manage.

Data Transfer Object (DTO) (Data Transfer Object) is a software application method for transferring data between design modes. The data transfer objective is often for the data access object to retrieve data from a database. The difference between a data transfer object and a data interaction object or data access object is a data (access and accessor) that does not have any behavior other than storage and retrieval.

Hibernate is an object relational mapping solution in Java. Object relational mapping or ORM framework is a technique that maps application data model objects to relational database tables. Hibernate focuses not only on the mapping of Java classes to database tables, but also on the mapping of Java data types to SQL data types.

Web2.0 is a generic term for a new class of Internet applications relative to Web1.0. Web2.0 is more focused on the interaction of users, and is a new generation mode of the internet realized by new theories and technologies such as xml, ajax and the like. Better interaction modes and user experience can be provided for users.

The reverse proxy server can be used for uniformly forwarding the request to a plurality of application servers or directly returning the cached data to the client, so that the access speed can be improved to a certain extent by the acceleration mode, and the purpose of load balancing is achieved. The reverse proxy can combine load balancing and proxy server caching techniques to provide beneficial performance and stability, and is an effective guarantee to provide 7 x 24 services.

The invention provides an objective and scientific intelligent prediction method and platform for container business leasing, which are based on historical data, predict the box returning quantity by adopting a specific calculation mode while considering factors such as the upper limit of a regional box returning, the upper limit of a harbor box returning and the like, accurately reveal the box returning trend, assist a container asset management enterprise to predict the future box returning condition, help the enterprise to timely adjust the management operation strategy, display the prediction trend in a form of a histogram and a line diagram according to regions, boxes and harbors, and display the change trend and the change amplitude of the box returning data, so that the data presentation mode is more visual and clear.

It should be noted that the above-described embodiments will enable those skilled in the art to more fully understand the invention, but do not limit it in any way. Therefore, although the present invention has been described in detail with reference to the drawings and examples, it will be understood by those skilled in the art that the present invention may be modified or equivalent, and in all cases, all technical solutions and modifications which do not depart from the spirit and scope of the present invention are intended to be included in the scope of the present invention.

Claims

1. The intelligent prediction method for the container operational lease also is characterized by comprising the following steps:

S4: when the contract term expires and historical actual box returning data exists in the month before the month of the current time, obtaining the box returning delay time of each box according to the box returning time data in the historical actual box returning data of the month, and calculating the average box returning delay time of each box; calculating the quantity of boxes which can be returned in the month after the month of the current time in all unreturned boxes according to the average box returning delay time of each box, and intelligently predicting the box returning quantity of the month after the month of the current time by adopting a Prophet algorithm according to the quantity of boxes which can be returned in the month after the month of the current time in all unreturned boxes, the average box returning delay time and the upper limit of the box returning quantity per month of a certain port;

2. The intelligent prediction method for container business rental boxes according to claim 1, wherein in the step S4, after predicting the box returning amount for a month in the future, the method further comprises the steps of:

if the predicted box returning amount of a certain box of a certain port in the future month is more than or equal to the monthly box returning amount upper limit of the certain box of the certain port, taking the monthly box returning amount upper limit of the box of the certain port as the actual box returning amount of the month;

If the total number is greater than or equal to the upper limit of the box returning amount of the box type in the area, the upper limit of the box returning amount of the box type in the area is taken as the actual box returning amount of the area, the number of boxes exceeding the upper limit of the box returning amount of the box type in the area is evenly distributed to all ports under the area, and the number of boxes exceeding the upper limit is subtracted from the forecast box returning amount of the last week of the month of each port.

3. The intelligent prediction method for container business lease boxes according to claim 2, characterized in that after judging the actual boxes for a certain month in the future, if the total predicted boxes for the month or several months after the month is equal to zero, the boxes are predicted; if the total predicted box returning amount in the last month or the last months of the month is larger than zero, automatically judging whether all the months before the month are smaller than the maximum predicted box returning month number, if the months before the month are smaller than the maximum predicted box returning month number, continuously predicting the predicted box returning amount of the next month of the month, and if the months before the month are larger than or equal to the maximum predicted box returning month number, ending the box returning prediction process.

4. The intelligent prediction method for container operations rental boxes according to any one of claims 1 to 3, wherein in the step S5, the graph type includes at least one of a summary bar graph, a regional graph, a box line graph, and a port bar graph.

5. The intelligent prediction method for container operations rental boxes according to any one of claims 1 to 3, wherein in step S1, a topic table is created based on historical actual box return data, and the topic table includes a customer ID, a contract number, a contract type, a box return area, a box return port, and a box return amount.

6. The intelligent prediction platform for the operational lease of the container is characterized by comprising a first module, a second module, a third module, a fourth module and a fifth module, wherein the first module is respectively connected with the second module, the third module and the fourth module, the second module, the third module and the fourth module are respectively connected with the fifth module,

a fourth module for obtaining the time delay of the box according to the time data of the box in the historical actual box-returning data of the month when the contract period is expired and the historical actual box-returning data exists in the month before the current time, and calculating the average time delay of the box; calculating the quantity of boxes which can be returned in the month after the month of the current time in all unreturned boxes according to the average box returning delay time of each box, and intelligently predicting the box returning quantity of the month after the month of the current time by adopting a Prophet algorithm according to the quantity of boxes which can be returned in the month after the month of the current time in all unreturned boxes, the average box returning delay time and the upper limit of the box returning quantity per month of a certain port;

7. The intelligent predicting platform for container operation leasing and returning of claim 6, wherein in the fourth module, after predicting the returning amount of a certain month in the future, the predicting returning amount of a certain case type of a certain port in the future is compared with the upper limit of the returning amount of a certain case type of a certain port in the month, and if the predicting returning amount of a certain case type of a certain port in the future in a certain month is smaller than the upper limit of the returning amount of a certain case type of a certain port in a certain month, the predicting returning amount of the certain case type of the port in the month is taken as the actual returning amount of the certain month;

8. The intelligent prediction platform for container operations rental boxes according to claim 7, wherein in the fourth module, after judging the actual box returning amount for a certain month in the future, if the total predicted box returning amount in the month or months after the month is equal to zero, the box returning prediction process is ended; if the total predicted box returning amount in the following month or the following months of the month is larger than zero, judging whether the month and all the months before the month are smaller than the maximum predicted box returning month number, if the month and all the months before the month are smaller than the maximum predicted box returning month number, continuously predicting the predicted box returning amount of the next month of the month, and if the month and all the months before the month are larger than or equal to the maximum predicted box returning month number, ending the box returning prediction process.

9. The container business rental further box intelligent prediction platform of any one of claims 6-8, wherein the chart types include at least one of a summary bar chart, a regional graph, a box line chart, and a port bar chart.

10. The intelligent prediction platform for container operations rental further comprises a first module for creating a topic table based on historical actual container data, wherein the topic table comprises a customer ID, a contract number, a contract type, a box type, a container area, a container port, and a container amount.