KR102623622B1 - Improved delivery time prediction system through AI algorithm - Google Patents

Abstract

Pertaining to a method of grouping delivery destinations, according to the present invention, a plurality of first center coordinates are set in a certain range of areas according to a specific standard, and a certain number of first center coordinates are set based on each of the plurality of first center coordinates. A plurality of delivery destinations included within a distance may be grouped into a plurality of first groups through a first grouping procedure.

Description

Improved delivery time prediction system through AI algorithm}

The present invention is a dispatch system for product delivery, which includes an algorithm for grouping each delivery location for goods, a delivery route algorithm for setting the minimum route according to the delivery location, and an algorithm for predicting delivery time based on the delivery information of the deliverer. It is about the dispatch system through.

In cases where the volume or volume is too large to use courier or quick service, logistics services at cargo terminals are often used. Conventionally, work is ordered using a radio such as TRS (Trunked Radio System), and the borrower also receives the order through the radio. A response is in progress.

In the conventional cargo transportation process, the customer who wants to transport cargo makes a transport request to a call center via phone or radio. In other words, the customer makes a reservation at the call center by verbally explaining the loading location, vehicle type, and time at which the cargo will be loaded.

Afterwards, the call center informs the borrower members through radio broadcast that there is a reserved transportation request, and the borrower members who received the radio broadcast also reply via radio and request that the vehicle be dispatched to them. [0006] At this time, it is common to accept requests for dispatch from borrower members by pressing a key on the radio, and in a call center, dispatch is made on a first-come, first-served basis, that is, to the member who first presses the key on the radio.

However, in the prior art dispatch system, when a maximum of 2 to 4 products are bundled and delivered, optimal grouping according to the distance between each product is not possible because the delivery address is selected directly by the deliverer, and more than 4 delivered products cannot be assigned at once. can not do it.

In addition, since it only provides information on the minimum route according to the distance between each item during the delivery stage, it cannot provide the optimal route according to the overall delivery location. Since it only provides information on the delivery time for each item, it does not provide information on how many items the deliverer is delivering and the number of deliveries by the deliverer. There is a problem that delivery time cannot be predicted.

The present invention was created to solve the above-mentioned problems, and provides an optimal dispatch system to deliverers using a grouping algorithm and a route setting algorithm.

Additionally, the present invention provides a method for optimal grouping of delivery destinations based on region and distance.

Additionally, the present invention provides a method for setting an optimal delivery route based on a plurality of delivery destinations according to the number of delivery items of the deliverer.

Additionally, the present invention provides a delivery time prediction algorithm for predicting delivery time according to the number of deliveries of a deliverer.

The technical problems to be achieved in this specification are not limited to the technical problems mentioned above, and other technical problems not mentioned will be clearly understood by those skilled in the art from the description below. You will be able to.

The present invention provides an integrated dispatch system to solve the above-mentioned problems.

Specifically, the grouping method according to an embodiment of the present invention includes setting a plurality of first center coordinates in a certain range of areas according to a specific standard, where the specific standard is the number of delivery destinations according to each of the plurality of center coordinates, It is at least one of the distance between center coordinates, population number, traffic volume, and population density; Grouping a plurality of delivery destinations included within a certain distance from the plurality of center coordinates into a plurality of first groups based on each of the plurality of first center coordinates through a first grouping procedure; Determining whether the number of the plurality of delivery destinations included in each of the plurality of first groups is greater than or equal to a specific value, at least one where the number of the plurality of delivery destinations among the plurality of first groups is greater than or equal to the specific value Grouping the first group into at least one second group through a second grouping procedure such that the number of the plurality of delivery destinations is less than or equal to the specific value, wherein the second grouping procedure includes the at least one second group included in the second group. This is repeatedly performed until the number of at least one delivery address becomes less than or equal to the specific value.

The present invention was created to solve the above-mentioned problems, and can provide an optimal dispatch system to deliverers using a grouping algorithm and a route setting algorithm.

In addition, the present invention has the effect of enabling efficient delivery because it sets the optimal delivery route based on a plurality of delivery destinations according to the number of delivery items of the deliverer.

In addition, the present invention has the effect of calculating an accurate delivery time because it predicts the delivery time based on the number of deliveries by the deliverer.

The effects that can be obtained from the present invention are not limited to the effects mentioned above, and other effects not mentioned can be clearly understood by those skilled in the art from the description below. will be.

The accompanying drawings, which are included as part of the detailed description to aid understanding of the present invention, provide examples of the present invention and explain technical features of the present invention along with the detailed description.
Figure 1 briefly shows the configuration of the service dispatch system proposed in this specification.
Figure 2 is a flowchart showing the dispatch sequence of the service dispatch system proposed in this specification.
Figure 3 is a flowchart showing an example of a method for grouping delivery destinations proposed in this specification.
4 to 9 show an example of a method for grouping delivery destinations proposed in this specification.
Figure 10 is a flowchart showing an example of a method for setting a route to a delivery destination proposed in this specification.
11 to 17 show an example of the route setting method proposed in this specification.
Figure 18 is a flowchart showing an example of a method for predicting delivery time proposed in this specification.
Figure 19 shows an example of linear regression data for predicting delivery time proposed in this specification.
Figure 20 shows an example of a sigmoid function for predicting delivery time proposed in this specification.

The above-described objects, features and advantages of the present invention will become more apparent through the following detailed description in conjunction with the accompanying drawings. However, since the present invention can make various changes and have various embodiments, specific embodiments will be illustrated in the drawings and described in detail below. Like reference numerals throughout the specification in principle refer to the same elements. Additionally, if it is determined that a detailed description of a known function or configuration related to the present invention may unnecessarily obscure the gist of the present invention, the detailed description will be omitted.

Figure 1 briefly shows the configuration of the service dispatch system proposed in this specification.

Referring to FIG. 1, the service dispatch system (or dispatch system, 100) proposed in this specification may be composed of a grouping algorithm 200, a route setting algorithm 300, and a delivery time prediction algorithm 400.

Specifically, the service dispatch system 100 groups delivery destinations in a certain area into a plurality of groups using the grouping algorithm 200, and establishes a delivery route between at least one delivery destination delivered by the deliverer among the grouped delivery destinations. The optimal path can be set through the path setting algorithm 300.

Additionally, the delivery time calculated based on the delivery information of the deliverers can be provided to recipients who receive the delivered goods through the delivery time prediction algorithm 400.

The grouping algorithm 200 may group a plurality of delivery destinations distributed in a certain area according to a specific criterion so that each of the plurality of groups includes a certain number of delivery destinations or less. At this time, grouping can be performed by including delivery locations located at a certain distance in a group using a certain location or a specific delivery location as a reference position. Hereinafter, the grouping algorithm 200 will be described in detail below.

The route setting algorithm 300 allows a deliverer to set the shortest distance to a certain number of delivery destinations among a plurality of delivery destinations as a delivery route. That is, at least one delivery location among delivery locations grouped or located within a certain distance based on the current location of the deliverer may be selected using a specific algorithm (eg, an improved TRS algorithm, etc.). Thereafter, the delivery order of the selected at least one delivery address may be changed to set the shortest delivery route that can pass through the selected at least one delivery location once. Hereinafter, the route setting algorithm 300 will be described in detail below.

The delivery time prediction algorithm 400 may be provided by a service dispatch system, a server providing a service dispatch system, a deliverer's terminal, or a recipient's terminal. Specifically, the delivery time prediction algorithm 400 may calculate and provide a delivery time for the recipient to receive the delivered goods based on delivery information of the goods delivered by the deliverer. At this time, the delivery information may include the number of goods delivered by the deliverer and the delivery order of the goods.

At this time, the delivery time is the delivery time of each of the first section of at least one product delivered before delivery of the specific product received by the recipient, and the second section of delivery of the specific product after the last product among the at least one product. It may be the combined delivery time. Each of the first section delivery time and the second section delivery time may be a sum of a distance-based delivery time using a first algorithm based on distance and a weighted time using a second algorithm based on surrounding environment information. At this time, the first algorithm may be modeled based on linear regression analysis, and the second algorithm may be modeled based on a sigmoid function based on the difference in delivery time according to environmental information of the delivery location of each delivery item. . Hereinafter, the delivery time prediction algorithm 400 will be described in detail below.

In this way, by providing a dispatch service through a service dispatch system, the optimal delivery service can be provided to the deliverer and recipient.

Figure 2 is a flowchart showing the dispatch sequence of the service dispatch system proposed in this specification.

Referring to FIG. 2, the service dispatch system can provide a delivery service by grouping delivery destinations through a grouping algorithm and providing the deliverer with an optimal route for the grouped groups.

Specifically, 1. Deliverers can receive deliveries from business partners through the Partners app. At this time, delivery items can be received through Excel or API on the homepage. 2. Afterwards, a collection engineer can collect the delivered product from the customer, 3. deliver it to the logistics center, and have it stocked.

4. Afterwards, the received delivered goods are grouped into a plurality of groups through a grouping algorithm, and 5. The grouped groups can be exposed to deliverers (or riders delivering goods, etc.) through the delivery group list.

Afterwards, 6. The deliverer can directly select the delivery group to be delivered and apply for dispatch. 7. After receiving a QR code, the deliverer can move to the logistics center. 8. Afterwards, the deliverer can recognize the QR code at the logistics center, confirm the dispatch group, and receive approval for dispatch.

9. Afterwards, deliverers can receive the delivered goods that have been dispatched, and 10. The delivered goods can be displayed on the list to be delivered (or the in-delivery list that displays goods being delivered, etc.).

11. Afterwards, the optimal delivery route for delivering the items can be provided through the delivery route algorithm, and delivery can be carried out based on the provided route.

12. After delivery is completed, deliverers can complete the delivery through the app.

Figure 3 is a flowchart showing an example of a method for grouping delivery destinations proposed in this specification.

Referring to FIG. 3, when there are a plurality of delivery destinations, the delivery destinations can be grouped into a plurality of groups according to specific criteria in order to set an optimal delivery route.

Specifically, a plurality of first center coordinates may be set in a certain range of areas according to specific criteria. At this time, the specific criterion may be at least one of the number of delivery destinations according to each of the plurality of central coordinates, the distance between the central coordinates, the number of population, traffic volume, and population density.

Based on each of the plurality of first center coordinates, a plurality of delivery destinations included within a certain distance from the plurality of center coordinates may be grouped into a plurality of first groups through a first grouping procedure, each of the plurality of first groups It may be determined whether the number of the plurality of delivery destinations included in is greater than or equal to a certain value.

Among the plurality of first groups, at least one first group in which the number of the plurality of delivery addresses is greater than or equal to the specific value is divided into at least one second group through a second grouping procedure such that the number of the plurality of delivery addresses is less than or equal to the specific value. may be grouped, and the second grouping procedure may be repeatedly performed until the number of at least one delivery address included in the at least one second group is less than or equal to the specific value.

The specific value is determined between a minimum value of 2 or more and a maximum value of 10 or less, and may be flexibly determined depending on the region in which each of the plurality of center coordinates is set.

The minimum and maximum values are the minimum and maximum number of delivery destinations at which the user can complete delivery within a certain delivery time, and the constant distance is based on the density, population density, traffic congestion, and type of delivery destination of the delivery destination based on the plurality of center coordinates. It can be decided depending on

The delivery destination type is one of residential type, office type, corporate type, and commercial type, and the certain distance may change depending on the morning, afternoon, and night based on the delivery destination type. For example, if the delivery address type is a residential type, there may be a lot of deliveries at night, so the certain distance may be reduced, and may be increased in the morning and afternoon. On the other hand, if the delivery destination type is an office type, there may be more deliveries in the afternoon than at night, so the certain distance may become smaller and increase in the morning and night.

The second grouping procedure may include the following procedures.

- Select one first group from at least one first group

- Select a specific delivery address located at the greatest distance from the center coordinates among the plurality of delivery locations included in the selected first group.

- Grouping at least one delivery address located within a certain distance from the selected specific delivery address among a plurality of delivery addresses into a second group

At this time, the number of at least one delivery address in the second grouping procedure may be less than half of the number of the plurality of delivery locations included in the first group.

In addition, the distance between at least one delivery address and a specific delivery address may be smaller than the distance between the at least one delivery address and the center coordinate, and the second grouping procedure may be performed so that the number of the at least one delivery address included in the second group is the specific value. This can be done repeatedly until it becomes smaller.

Through this method, delivery destinations can be grouped using a grouping algorithm.

For example, as shown in FIG. 3, when there are n delivery locations, the latitude and longitude of k surrounding cities, counties, and districts can be initialized as the center coordinates of the cluster (or group) (S3010). Afterwards, grouping can be performed by executing k-means clustering (S3020).

If the number of groups is 0, group information can be output (S3030).

However, if the number of groups is not 0, the first group among the k groups is checked and the number of delivery destinations included in the first group (center coordinate C, first group) is checked (S3040). At this time, the value of K is decreased by 1.

If the number of delivery addresses in the first group is less than a certain number (for example, g, which is the lub of the group size), all delivery addresses included in the first group are grouped into one group (first group) (S3050) .

However, if the number of delivery destinations in the first group is more than a certain number (for example, g, which is the lub of the group size), the delivery destination D furthest from the center position C is selected, and a certain number of delivery destinations are selected in order of proximity to the selected delivery address D. Select (S3060). At this time, the number of delivery destinations selected in order of proximity to delivery destination D may be smaller than g divided by 2.

Thereafter, for g/2 delivery destinations that are next to the selected delivery destination and closest to the delivery destination D, for each of the selected delivery destinations, the selected delivery destination has a shorter distance to the delivery destination D than the center coordinate C, and the distance to the delivery destination D is shorter than the center coordinate C, and the distance to the delivery destination D is shorter than the center coordinate C, Determine whether the number is greater than g divided by 2.

If the selected delivery location is not shorter in distance to delivery location D than the center coordinate C, or the number of unselected delivery locations based on center coordinate C is not greater than g divided by 2, the delivery location does not change the group. Leave it as is (S3070).

However, if the selected delivery address has a shorter distance to the delivery address D than the center coordinate C, and the number of unselected delivery locations based on the center coordinate C is greater than g divided by 2, the corresponding delivery location is selected (S3080), and the selected delivery address is Delivery destinations are separated from a group based on the center coordinate C and grouped into a group based on delivery destination D (S3090).

4 to 9 show an example of a method for grouping delivery destinations proposed in this specification.

First, as shown in FIG. 4, city, county, and district coordinates can be set as the initial center coordinates. Afterwards, as shown in Figure 5, 2. K-Means Clustering can be performed based on the initial coordinates. K-Means Clustering refers to a clustering method that sets K central coordinates and then assigns data to the nearest central coordinate cluster.

Afterwards, as shown in Figure 6, the cluster with the most delivery destinations (Seocho-gu in Figure 6) is selected, and 1 is subtracted from the value of K.

Thereafter, as shown in Figure 7, the 4. Seocho-gu cluster is subject to division because it exceeds the maximum size of the group (9 in this example). Therefore, first, we find the coordinates furthest from the cluster center coordinates (delivery destination shown in green).

Thereafter, as shown in FIG. 8, in order to have a minimum group size of 5 (for example, the number of delivery destinations is 5), the four coordinates close to the corresponding coordinates may be further grouped and separated into other groups. Afterwards, it can be determined whether to place it in a new group or an existing group by comparing the distances to the next closest 9 - 5 = 4 coordinates.

Thereafter, the above processes from Figures 6 to 8 can be repeatedly performed until the value of K becomes 0 (e.g., until all clusters are confirmed), and the final clusters are grouped in the form shown in Figure 9. You can.

Figure 10 is a flowchart showing an example of a method for setting a route to a delivery destination proposed in this specification.

Referring to FIG. 10, when there are a plurality of delivery destinations (or grouped groups), the optimal delivery route for delivering the plurality of delivery destinations (or grouped groups) can be set through a specific algorithm. Hereinafter, in the present invention, the delivery address may refer to a grouped group.

Specifically, at least one delivery location among a plurality of delivery locations may be determined using a specific algorithm, and the shortest distance passing through each of the determined at least one delivery location may be determined.

At this time, determining at least one delivery location can be performed through the following steps.

- If the number of the plurality of delivery destinations is more than a certain number, select the specific number of delivery destinations in order of shortest distance based on the current location.

- Select one delivery location among the selected specific number of delivery locations using the specific algorithm

At this time, the specific algorithm is used to select the specific number of at least one delivery destination among the plurality of delivery destinations, and the specific number is the shortest route from the current location to the at least one delivery destination using the specific algorithm. This may be the maximum number of delivery destinations for setting a route.

The procedure of selecting one delivery address using a specific algorithm can be performed through the following steps.

- Using the specific algorithm, set the optimal route for the specific number of delivery destinations selected in order of shortness based on the current location.

- Select the first delivery location based on the current location in the optimal route set above

- Select the specific number of delivery destinations in order of closest distance based on the first delivery address.

- Using the specific algorithm, set the optimal route for the specific number of delivery destinations selected in order of short distance based on the location of the first delivery destination.

- Select the first second delivery location based on the location of the first delivery location in the optimal route set above.

At this time, the optimal route may be determined based on the starting location, arrival location, set of the specific number of delivery destinations, and distance information.

For example, when the starting location is start, the arrival location is last, the set of the specific number of delivery destinations is V, and the distance information is D, the optimal path find_path is calculated by Equation 1 below: You can.

[Equation 1]

The shortest route is the shortest distance from the current location that passes through all of the selected delivery destinations once, and the shortest distance can be set through the following procedure.

- Determine a first route passing through the at least one delivery destination in the selected order

- Determine the second route by partially changing the delivery order of the at least one delivery address.

- Compare the first path and the second path

- Select the shorter path between the first path and the second path

At this time, if the first path is shorter than the second path, the shortest distance can be set by further performing the additional procedure below.

-Determine a third route by changing the delivery order of the at least one delivery destination to be different from the second route.

- Compare the first path and the third path

- Select the shorter path between the first path and the third path.

However, if the second path is shorter than the first path, the shortest distance can be set by further performing the additional procedure below.

-Determine a third route by changing the delivery order of the at least one delivery destination to be different from the second route.

- Compare the second path and the third path

- Select the shorter path between the second path and the third path.

At this time, the procedure for selecting the short route may be repeatedly performed depending on the number of the at least one delivery destination.

For example, as shown in Figure 10, it is determined whether n delivery destinations are less than the maximum value N for applying the customized TSP.

At this time, the Traveling Salesman Problem (TSP) is an algorithm that finds a cycle. When there are several cities and the cost of moving from one city to another is given, all cities are visited only once and the original This refers to an algorithm that finds the shortest path back to the starting point.

However, customized TSP is an algorithm that calculates a path by modifying TSP to suit the delivery system.

Specifically, since it has changed from finding circulation through TSP to finding a route through customized TSP, the starting point and destination can be set. In particular, the destination can be set to a place desired by the rider, such as a logistics center or the delivery person's residence, so it is rider-friendly. path can be provided.

Customized TSP can be implemented with dynamic programming. For example, when start is the starting point, last is the destination point, V is the delivery address set, and D is the distance information matrix, this can be implemented through Equation 1.

According to Equation 1, this can be done by calling a recursive function that visits all cities that have not yet been visited for a specific start, last, and V, and selecting the path that shows the minimum value. For example, find_path( 1, 5, {1, 2, 3, 5} ) is equivalent to find_path( 1, 2, {1, 2, 3} ) + D[5][2],

It can be determined as the minimum value among find_path( 1, 3, {1, 2, 3} ) + D[5][3], and whether or not to visit can be expressed using a bitmasking method. For example, if there are 8 delivery destinations and the 1st, 2nd, and 5th delivery destinations are visited, it can be expressed as 00010011 => 19.

If the value of n is greater than N, find the k closest to the current location, perform TSP (S10010), find the place to visit first, and add the delivery address to the delivery route (S10020).

Afterwards, find the k closest ones based on the delivery address and perform TSP. Afterwards, repeat the above process and select n delivery destinations with a value smaller than N.

If the value of n is smaller than N, n delivery destinations are selected using the customized TSP (S10030).

Afterwards, for the selected n delivery destinations, each delivery destination is replaced with the first, second, ..., last delivery destination, and when the moving distance becomes shorter, the order is changed to calculate the shortest distance (S10040). Afterwards, if any of the delivery routes intersect, the twisted part is calculated and reset (S10050).

Through this method, the optimal delivery route for multiple delivery destinations can be established.

11 to 17 show an example of the route setting method proposed in this specification.

Specifically, as shown in FIG. 11, when 30 delivery destinations are given, since the given delivery destinations are more than a certain number (15), 7 delivery destinations close to the current location (B) are selected as shown in FIG. 12.

Afterwards, as shown in Figure 13, a delivery route is set by applying the customized TSP to the seven selected delivery destinations, and the 16th delivery destination, which is the first delivery destination, is selected from the set delivery route.

Afterwards, as shown in Figure 14, 7 nearby delivery locations are selected based on the 16th delivery location, a customized TSP is applied to the selected 7 delivery locations to set a delivery route, and the 7th delivery location, which is the first delivery location in the set delivery route, is selected. Select .

Afterwards, the method described in FIGS. 12 to 14 is repeatedly performed until k delivery destinations remain, and as shown in FIG. 15, condensation can be set for the k delivery destinations using a customized TSP.

Afterwards, as shown in FIG. 16, the delivery order of the selected delivery destinations is changed and the order with the shortest route is checked.

For example, if the delivery addresses are sorted in the order [1, 3, 2, 4, 5], then [3, 1, 2, 4, 5], [3, 2, 1, 4, 5], [3, 2 , 4, 1, 5], [3, 2, 4, 5, 1] … Check whether the path is shortened by inserting each element into every order as shown. In the case of Figure 16, the delivery order of numbers 21, 12, and 17 has been changed.

Afterwards, as shown in Figure 17, if there is a part where the paths intersect, it is released. In the case of Figure 17, the paths 24 - 23, 7 - 12 have been changed to 24 - 7, 23 - 12.

Figure 18 is a flowchart showing an example of a method for predicting delivery time proposed in this specification.

Referring to FIG. 18, the delivery time of a specific product can be accurately calculated by obtaining delivery information about the deliverer's delivered items and delivery order and calculating the delivery time for each delivered item.

Specifically, delivery information related to a specific product may be obtained. At this time, the delivery information may include the number of a plurality of delivery items delivered by the deliverer, and a delivery order of the plurality of delivery items.

The plurality of delivery items may be sorted according to the number of the plurality of delivery items included in the delivery information and the delivery order of the plurality of delivery items, and among the plurality of delivery items sorted according to the delivery order, Delivery time for a specific item can be calculated.

At this time, the delivery time of the specific product is the delivery time of the first section of each of the at least one product delivered before delivery of the specific product and the second section of the delivery time of the specific product after the last product among the at least one product. It may be the sum of delivery times.

Each of the first section delivery time and the second section delivery time is the sum of the distance-based delivery time using the first algorithm based on the distance and the weighted time using the second algorithm according to the surrounding environment information, and the first algorithm is the current delivery time It may be modeled based on a linear regression analysis based on the distance between the first delivery location of the product and the second delivery location of the previous delivery product delivered before the current delivery product.

Additionally, the second algorithm may be modeled based on a sigmoid function based on the difference in delivery time according to environmental information of the delivery location of each delivery item, and the environmental information is at least one of rainfall, snowfall, wind volume, and traffic volume. may include.

The linear regression analysis is modeled based on the delivery distance and the time required for each time slot according to the delivery distance. Based on the linear regression analysis, if the delivery distance is The values of a and b of 2 can be determined.

[Equation 2]

If the value based on the surrounding environment information is x, the weight time can be calculated by Equation 3 below based on a and b determined by Equation 2.

[Equation 3]

At this time, the delivery information may be obtained through the deliverer's terminal or a server for the dispatch service system.

At this time, before performing the delivery time prediction algorithm, the following two algorithm models must be obtained.

- Delivery time linear regression analysis model

1. Classify delivery data by delivery completion time and region.

2. Perform linear regression analysis on delivery time using straight line distance.

- Reflection of weather conditions (sigmoid model)

1. Find how much the delivery time varies depending on rainfall and snowfall.

2. Model with a sigmoid function.

3. Find the weights of the model through regression analysis.

For example, as shown in FIG. 18, the algorithm for predicting delivery time can obtain delivery information (or rider information) of the deliverer including the product for which delivery time is to be predicted (S18010). At this time, as discussed above, the delivery information may be in the form of the number of the plurality of delivery items and the delivery order of the plurality of delivery items.

Thereafter, the items can be aligned according to the optimal path set through the path setting method described in FIGS. 10 to 17 (S18020).

Afterwards, the straight line distance between the current location of the deliverer and the first product to be delivered (i=1) is obtained (S18030), and the first delivery time (total_time) according to the distance is calculated using the linear regression model, which is the first algorithm ( S18040). If the i-th product to be delivered is not the target specific product, the value of i is increased by 1, and the straight-line distance to the delivery destination of the i-th product with the increment of 1 is calculated. Afterwards, the delivery time based on the linear regression model is calculated based on the distance from the delivery location of the previous product to the delivery location of the ith product (S18050). This process is performed repeatedly until the ith product is the specific desired product.

If the item i is a specific target product, the weighted time is calculated using the sigmoid algorithm, which is a second algorithm, according to the environmental impact (e.g., rainfall, snowfall, etc.) on the first delivery time (total taime). Add (S18060).

In this case, the second algorithm can be applied individually to the distance between each delivery destination of the delivered goods, and can be applied separately and combined to the first delivery time according to the distance between each delivery destination of the delivered goods.

Thereafter, the sum of each first delivery time and weight time calculated up to the ith time may be calculated and output as the delivery time (S18070).

Through this method, the delivery time for a specific delivery item can be accurately calculated.

Figure 19 shows an example of linear regression data for predicting delivery time proposed in this specification.

Referring to FIG. 19, the linear regression data can be divided by deliverer, city, county, and district, and time zone, and linear regression can be performed on the travel distance (x) for the time required (y).

At this time, the linear equation is obtained as y = ax + b, and the coefficients a and b can be stored separately. If there is not enough lidar data, linear regression values for all deliveries in the region and time zone can be used.

Figure 20 shows an example of a sigmoid function for predicting delivery time proposed in this specification.

Referring to FIG. 20, the sigmoid function may be introduced to reflect environmental factors that may affect delivery time, such as precipitation and/or snowfall.

The sigmoid function may have a reformed form as shown in FIG. 20, and an equation such as Equation 3 below may be used.

[Equation 3]

In Equation 3, the coefficients a and b can be obtained through the linear regression analysis described in FIG. 19.

If environmental factors such as rain, snow, and/or typhoons apply to the delivery time, the regression analysis estimate obtained above can be multiplied by the sigmoid function value.

For example, the distance from the constituency of Rider Kim Cheol-soo's delivery item and the previous delivery location is as follows.

1. Gangnam-gu, 4km (distance from current location for first item)

2. Gangnam-gu, 2km

3. Gangnam-gu, 3km

4. Seocho-gu, 0.5km

5. Seocho-gu, 1km

The values of linear regression with existing data are stored as follows:

…

Kim Cheol-soo Gangnam-gu 3-4 p.m .: a = 5, b = 6

Kim Cheol-soo Gangnam-gu 4-5 p.m .: a = 7, b = 5

…

Kim Cheol-soo Seocho-gu 4-5 p.m .: a = 4, b = 7

…

When Rider Kim Cheol-soo starts delivery at 3:30, the estimated time is determined as follows.

1. The current time is 3:30, and since it is a Gangnam-gu product, Kim Cheol-soo uses data from 3 to 4 o'clock in Gangnam-gu . Since it takes 4km, the estimated time is 5*4 + 6 = 26 minutes.

Estimated completion at 3:56.

2. The estimated delivery start time is 3:56, and since it is a Gangnam-gu product, Kim Cheol-soo used data from 3 to 4 o'clock in Gangnam-gu. Since it takes 2km, the estimated time is 5*2 + 6 = 16 minutes.

Estimated completion at 4:12.

3. The estimated delivery start time is 4:12, and since it is a Gangnam-gu product, Kim Cheol-soo used data from 4 to 5 o'clock in Gangnam-gu. It takes 3km, so 7*3 + 5 = 26 minutes.

Estimated completion at 4:38.

4. The estimated delivery start time is 4:38, and since it is a Seocho-gu item, Kim Cheol-soo used data from 4 to 5 o'clock in Seocho-gu. It takes 0.5km, so 4*0.5 + 7 = 9 minutes.

Estimated completion at 4:47.

5. The estimated delivery start time is 4:47, and since it is a Seocho-gu item, Kim Cheol-soo used data from 4 to 5 o'clock in Seocho-gu. It takes 1km, so 4*1 + 7 = 11 minutes.

Estimated completion at 4:58.

At this time, it is raining at 22 mm per hour, and if the coefficients of the sigmoid function for rain are a = 0.5 and b = 20, at each stage = Multiplied by a weight of 1.44 (In this case, the second delivery is expected to start after 4 o'clock, so when calculating the expected time, the 4-5 o'clock data is used, not the 3-4 o'clock data.)

The description of the present invention described above is for illustrative purposes, and those skilled in the art will understand that the present invention can be easily modified into other specific forms without changing the technical idea or essential features of the present invention. will be. Therefore, the embodiments described above should be understood in all respects as illustrative and not restrictive. For example, each component described as unitary may be implemented in a distributed manner, and similarly, components described as distributed may also be implemented in a combined form.

The scope of the present invention is indicated by the claims described below rather than the detailed description above, and all changes or modified forms derived from the meaning and scope of the claims and their equivalent concepts should be construed as being included in the scope of the present invention. do.

Claims (8)

In a method for predicting delivery time by a server or terminal, the method includes:
Obtaining delivery information related to a specific product;
The delivery information includes the number of a plurality of delivery items delivered by the deliverer, and a delivery order of the plurality of delivery items;
Sorting the plurality of delivery items according to the number of the plurality of delivery items included in the delivery information and the delivery order of the plurality of delivery items; and
Comprising the step of calculating a delivery time of the specific product among the plurality of delivery items arranged according to the delivery order,
The delivery time of the specific product includes a first section delivery time of each of the at least one product delivered before delivery of the specific product and a second section delivery time of the specific product delivered after the last product among the at least one product. It is the sum of time,
Each of the first section delivery time and the second section delivery time is a sum of a distance-based delivery time using a first algorithm based on distance and a weighted time using a second algorithm based on surrounding environment information,
The first algorithm is modeled based on linear regression analysis based on the distance between the first delivery location of the current delivery item and the second delivery location of the previous delivery item delivered before the current delivery item,
The second algorithm is modeled based on a sigmoid function based on the difference in delivery time according to the environmental information of the delivery location of each delivery item, and
The linear regression analysis is modeled based on the delivery distance and the time required for each time period according to the delivery distance.
Based on the linear regression analysis, if the delivery distance is
Equation 1

When the value based on the surrounding environment information is x, the weight time uses equation 2 below based on a and b,
Equation 2

calculating a delivery time of the specific product among the plurality of delivery items arranged according to the delivery order; Is
When a plurality of delivery destinations exist, grouping the delivery destinations into a plurality according to specific criteria to set an optimal delivery route; Including,
When the plurality of delivery destinations exist, grouping the delivery destinations into a plurality according to specific criteria to set an optimal delivery route; Is
A plurality of first center coordinates are set according to specific criteria in a certain range of areas, and the specific criteria include the number of delivery destinations for each of the plurality of center coordinates, the distance between center coordinates, the number of people, traffic volume, and population density. contains at least one,
Based on each of the plurality of first center coordinates, a plurality of delivery locations included within a certain distance from the plurality of center coordinates are grouped into a plurality of first groups through a first grouping procedure, and are included in each of the plurality of first groups. Determine whether the number of the plurality of delivery destinations is greater than or equal to a certain value,
Among the plurality of first groups, at least one first group in which the number of the plurality of delivery addresses is greater than or equal to the specific value is divided into at least one second group through a second grouping procedure such that the number of the plurality of delivery addresses is less than or equal to the specific value. grouping, and the second grouping procedure is repeatedly performed until the number of at least one delivery address included in the at least one second group becomes less than or equal to the specific value.







delete delete delete According to claim 1,
A method wherein the environmental information includes at least one of rainfall, snowfall, wind volume, and traffic volume.




delete delete delete