Transportation time limit prediction method and device, computer device, and storage medium

Abstract

The application provides a method and device for predicting the time limit for transportation, a computer device and a storage medium. The target sending time, target sending network point and target delivery network point of a target waybill are obtained, and then the target logistics link is determined according to the target sending network point and the target delivery network point. The time consumption of the target logistics link is predicted based on a preset waybill time limit prediction model, and the estimated transportation time consumption on the target logistics link is obtained. The estimated delivery time of the target waybill is determined according to the target sending time and the estimated transportation time consumption. The transportation time consumption of the logistics link corresponding to the sending network point and the delivery network point is obtained, and the transportation time consumption of the logistics link and the sending time are combined to predict the delivery time limit of the waybill, thereby improving the prediction accuracy of the transportation time limit.

Description

Technical Field

[0001] This application relates to the field of artificial intelligence technology, specifically to a method, apparatus, computer equipment, and storage medium for predicting transportation timeliness. Background Technology

[0002] With the development of e-commerce, users are shopping online more and more frequently. The non-virtual goods purchased need to be delivered to customers by merchants through logistics networks, and logistics transportation often takes one to several days. Therefore, when making online shopping decisions, users often pay attention to when the goods will arrive. The delivery time of goods is of great significance to users' online shopping behavior and will affect their choice of products. However, existing logistics companies or e-commerce applications do not provide estimated delivery times for express shipments, or only provide delivery times in days, resulting in poor accuracy in predicting delivery times. Summary of the Invention

[0003] Therefore, it is necessary to provide a method, apparatus, computer equipment, and storage medium for predicting transportation timeliness in order to improve the accuracy of transportation timeliness prediction, in order to address the aforementioned technical problems.

[0004] In a first aspect, this application provides a method for predicting transportation timeliness, the method comprising:

[0005] Obtain the target shipping time, target shipping location, and target delivery location for the target waybill;

[0006] Determine the target logistics link based on the target parcel delivery point and the target parcel collection point;

[0007] Based on the preset waybill timeliness prediction model, the time consumption of the target logistics link is predicted to obtain the estimated transportation time on the target logistics link;

[0008] Determine the estimated delivery time of the target shipment based on the target shipment time and the estimated transportation time.

[0009] In some embodiments of this application, determining the target logistics link based on the target sending point and the target delivery point includes:

[0010] Obtain the shipping area corresponding to the target shipping point and the delivery area corresponding to the target delivery point;

[0011] The logistics link from the sending area to the delivery area is defined as the target logistics link.

[0012] In some embodiments of this application, the target logistics link includes multiple transportation stages. Based on a preset waybill timeliness prediction model, the time consumption of the target logistics link is predicted to obtain the estimated transportation time on the target logistics link, including:

[0013] The time-consuming prediction data for each transportation stage on the target logistics link is obtained by determining the quantile values ​​of different transportation stages on the target logistics link and the transportation time values ​​at the quantile values ​​through the waybill time-consuming prediction model.

[0014] Based on the time-prediction data of each transportation stage in the target logistics link, the estimated transportation time in the target logistics link is determined.

[0015] In some embodiments of this application, the waybill timeliness prediction model includes a Markov chain model, which determines the quantiles of different transportation stages on the target logistics link, including:

[0016] Acquire training data, which includes training samples corresponding to different logistics links and quantile time of each transportation stage in different logistics links at different quantiles. The training samples include sample shipment time and sample actual transportation time.

[0017] Each training sample is input into a pre-built Markov chain model. Based on the preset state values ​​of each Markov state in the Markov chain model, the sample time prediction data of each transportation stage in the sample logistics chain corresponding to each training sample is obtained from the quantile time corresponding to different quantiles of each transportation stage in different logistics links. Each Markov state in the Markov chain model corresponds to a different transportation stage.

[0018] Based on the sample time prediction data of each transportation stage in the logistics chain of each sample, obtain the sample estimated transportation time of each logistics chain.

[0019] The estimated delivery time of the training samples is determined based on the sample shipping time of each training sample and the estimated transportation time of the sample logistics link corresponding to each training sample.

[0020] Based on the actual delivery time and the estimated delivery time of each training sample, the preset state values ​​of each Markov state in the Markov chain model are adjusted to obtain the target state value.

[0021] The target state value is determined as the quantile of the value at different transportation stages.

[0022] In some embodiments of this application, the preset state values ​​of each Markov state in the Markov chain model are adjusted according to the actual delivery time and the estimated delivery time of each training sample to obtain the target state value, including:

[0023] Based on the actual delivery time and the estimated delivery time of each training sample, obtain the sample time consumption error and the time consumption achievement rate.

[0024] Based on the time achievement rate being greater than or equal to the preset achievement rate, the preset state value of the Markov state is adjusted according to the time error until the sample time error is minimized, thus obtaining the target state value.

[0025] In some embodiments of this application, the estimated delivery time of the training samples is determined based on the sample dispatch time of each training sample and the estimated transportation time of the sample logistics link corresponding to each training sample, including:

[0026] Based on the sample shipping time of each training sample, obtain the sample transportation start time of each training sample;

[0027] The estimated delivery time of the training samples is determined based on the sample transportation start time of each training sample and the estimated transportation time of the corresponding sample logistics link.

[0028] In some embodiments of this application, obtaining training data includes:

[0029] Obtain multiple historical waybills and classify them into groups corresponding to different logistics links based on the sending and receiving points of the historical waybills.

[0030] Based on the historical transportation time of each historical waybill under each historical waybill group at each transportation stage, obtain the quantile time corresponding to different quantiles of each transportation stage on the logistics link corresponding to each historical waybill group.

[0031] Training samples for each logistics link are generated based on the dispatch time, actual transportation time, dispatch point, and delivery point of each historical waybill group.

[0032] In some embodiments of this application, historical waybills are divided into groups corresponding to different logistics links based on the sending and receiving points of the historical waybills, including:

[0033] Based on the preset regional division granularity, logistics outlets are divided into different logistics regions;

[0034] Obtain the first logistics region to which the sending point of the historical waybill belongs, and the second logistics region to which the sending point of the historical waybill belongs;

[0035] Based on the first and second logistics regions of historical waybills, historical waybills are divided into historical waybill groups corresponding to different logistics links.

[0036] In some embodiments of this application, the estimated delivery time of the target waybill is determined based on the target shipping time and the estimated transportation time, including:

[0037] Based on the target shipment time, determine the estimated start time of transportation for the target waybill;

[0038] The estimated delivery time of the target waybill is obtained based on the estimated start time of transportation and the estimated transportation time.

[0039] In some embodiments of this application, determining the estimated start time of transportation for a target waybill based on the target shipping time includes:

[0040] Obtain the shipping time period corresponding to the target shipping time;

[0041] The estimated departure time of the target waybill is obtained based on the shipping time period, and the estimated departure time is determined as the estimated start time of transportation.

[0042] Secondly, this application provides a transportation timeliness prediction device, the device comprising:

[0043] The waybill acquisition module is used to obtain the target dispatch time, target dispatch point, and target delivery point for a target waybill.

[0044] The logistics link acquisition module is used to determine the target logistics link based on the target sending point and the target delivery point;

[0045] The transportation time estimation module is used to predict the time of the target logistics link based on the preset waybill timeliness prediction model, and obtain the estimated transportation time on the target logistics link.

[0046] The delivery time estimation module is used to determine the estimated delivery time of the target waybill based on the target shipment time and the estimated transportation time.

[0047] Thirdly, this application also provides a server, the server comprising:

[0048] One or more processors;

[0049] Memory; and

[0050] One or more applications, wherein the one or more applications are stored in the memory and configured to be executed by the processor to implement the method for predicting transportation timeliness.

[0051] Fourthly, this application also provides a computer-readable storage medium having a computer program stored thereon, the computer program being loaded by a processor to perform the steps in the method for predicting transportation timeliness.

[0052] Fifthly, embodiments of this application provide a computer program product or computer program that includes computer instructions stored in a computer-readable storage medium. A processor of a computer device reads the computer instructions from the computer-readable storage medium and executes the computer instructions, causing the computer device to perform the method provided in the first aspect described above.

[0053] The aforementioned method, apparatus, computer equipment, and storage medium for predicting transportation timeliness acquire the target dispatch time, target dispatch point, and target delivery point of the target waybill. Then, based on the target dispatch point and target delivery point, the target logistics link is determined. A preset waybill timeliness prediction model is used to predict the time consumption of the target logistics link, obtaining the estimated transportation time on the target logistics link. Based on the target dispatch time and the estimated transportation time, the estimated delivery time of the target waybill is determined. By acquiring the transportation time of the logistics link corresponding to the dispatch point and delivery point, and combining the transportation time of the logistics link with the dispatch time, the delivery time of the waybill is predicted, improving the accuracy of transportation timeliness prediction. Attached Figure Description

[0054] To more clearly illustrate the technical solutions in the embodiments of this application, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0055] Figure 1 This is a schematic diagram of a scenario illustrating the method for predicting transportation timeliness in an embodiment of this application;

[0056] Figure 2 This is a flowchart illustrating the method for predicting transportation timeliness in the embodiments of this application;

[0057] Figure 3 This is a schematic diagram of different logistics stages in the embodiments of this application;

[0058] Figure 4 This is a flowchart illustrating the steps for obtaining the estimated transportation time in an embodiment of this application.

[0059] Figure 5 This is a flowchart illustrating the steps for obtaining the quantile values ​​in an embodiment of this application;

[0060] Figure 6 This is a schematic diagram illustrating the relationship between the actual arrival time and the estimated arrival time in the embodiments of this application;

[0061] Figure 7This is a schematic diagram of the model for obtaining the estimated transportation time in an embodiment of this application;

[0062] Figure 8 This is a schematic diagram of the logistics area in an embodiment of this application;

[0063] Figure 9 This is a flowchart illustrating another method for predicting transportation timeliness in an embodiment of this application;

[0064] Figure 10 This is a schematic diagram of the structure of the transportation timeliness prediction device in the embodiments of this application;

[0065] Figure 11 This is a schematic diagram of the structure of the computer device in the embodiments of this application. Detailed Implementation

[0066] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0067] In the description of this application, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of indicated technical features. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of the stated features. In the description of this application, "multiple" means two or more, unless otherwise explicitly specified.

[0068] In the description of this application, the word "for example" is used to mean "used as an example, illustration, or description." Any embodiment described as "for example" in this application is not necessarily to be construed as being more preferred or advantageous than other embodiments. The following description is provided to enable any person skilled in the art to make and use the invention. Details are set forth in the following description for purposes of explanation. It should be understood that those skilled in the art will recognize that the invention can be made without using these specific details. In other instances, well-known structures and processes will not be described in detail to avoid obscuring the description of the invention with unnecessary detail. Therefore, the invention is not intended to be limited to the embodiments shown, but is consistent with the broadest scope of the principles and features disclosed in this application.

[0069] It should be noted that the transportation time prediction method provided in this application is executed in computer equipment. The processing objects of each computer equipment exist in the form of data or information, such as time, which is essentially time information. It can be understood that if logistics outlets, logistics links, etc. are mentioned in subsequent embodiments, they are all corresponding data that exist so that the computer equipment can process them. The specifics will not be elaborated here.

[0070] In the embodiments of this application, it should also be noted that the transportation time prediction method provided in the embodiments of this application can be applied to, for example, Figure 1 The transportation timeliness prediction system shown includes a terminal 100 and a server 200. The terminal 100 can be a device that includes both receiving and transmitting hardware, i.e., a device with receiving and transmitting hardware capable of performing bidirectional communication over a bidirectional communication link. Such a device can include cellular or other communication devices with single-line displays, multi-line displays, or no multi-line displays. Specifically, the terminal 100 can be a desktop terminal or a mobile terminal, and can also be a barcode scanner, tablet computer, laptop computer, etc. The server 200 can be a standalone server or a server network or server cluster, including but not limited to computers, network hosts, single network servers, multiple network server sets, or cloud servers composed of multiple servers. The cloud server consists of a large number of computers or network servers based on cloud computing.

[0071] Those skilled in the art will understand that Figure 1 The application environment shown is merely one application scenario of the solution in this application and does not constitute a limitation on the application scenario of the solution in this application. Other application environments may include those that are more specific to this application. Figure 1 The number of computer devices shown is more or less, for example Figure 1 Only one server (200) is shown in the diagram. It is understood that this transportation timeliness prediction system may include one or more other servers; specific details are not specified here. Additionally, as... Figure 1 As shown, the transportation timeliness prediction system may also include a memory for storing data, such as information related to waybills.

[0072] It should also be noted that, Figure 1The schematic diagram of the transportation timeliness prediction system shown is merely an example. The transportation timeliness prediction system and scenario described in the embodiments of the present invention are for the purpose of more clearly illustrating the technical solutions of the embodiments of the present invention, and do not constitute a limitation on the technical solutions provided by the embodiments of the present invention. As those skilled in the art will know, with the evolution of transportation timeliness prediction systems and the emergence of new business scenarios, the technical solutions provided by the embodiments of the present invention are also applicable to similar technical problems.

[0073] See Figure 2 This application provides a method for predicting transportation timeliness, mainly applied to the above-mentioned... Figure 1 Taking server 200 as an example, the method includes steps S210 to S240, as follows:

[0074] S210, obtain the target shipment time, target shipment location, and target delivery location for the target waybill.

[0075] Among them, the target waybill refers to the waybill whose transportation time needs to be predicted; the target dispatch time can refer to the generation time of the target waybill or the courier's collection time; the target dispatch point refers to the logistics point that generated the target waybill, which can correspond to the logistics point in the area where the target waybill's transportation start location is located; the target delivery point refers to the logistics point in the area where the target waybill's receipt address or transportation destination location is located.

[0076] Specifically, such as Figure 1 In the transportation timeliness prediction system shown, during the express delivery process, when the courier obtains the item to be sent from the user and determines the sender and recipient addresses, the corresponding operation terminal 100 generates a waybill. The terminal 100 sends the information corresponding to the waybill to the server 200. The server 200 obtains the sending time, sending point, and delivery point corresponding to the waybill, and can estimate the transportation timeliness of the waybill, obtain the estimated delivery time of the waybill, and return the estimated delivery time to the terminal.

[0077] S220, determine the target logistics link based on the target sending point and the target delivery point.

[0078] The logistics link is used to identify the entire transportation chain from the sending point to the delivery point. The logistics link begins when the item is picked up by the courier and ends when it is delivered to the customer. Therefore, the logistics link can be divided into multiple transportation stages according to different transportation phases. In one embodiment, such as... Figure 3As shown, the logistics chain can be divided into the receiving stage, the receiving branch stage, the transit stage, the delivery branch stage, and the delivery stage. The receiving stage is the process from when the courier collects the item to when the item arrives at the receiving point. The receiving branch stage is the process from when the item is transported from the receiving point to the starting transit point. The transit stage is the process from when the item is transported from the starting transit point to the ending transit point. The delivery branch stage is the process from when the item is transported from the ending transit point to the delivery point. The delivery stage is the process from when the item is delivered to the customer.

[0079] Specifically, the server can determine the logistics link between the target sending point and the target delivery point as the target logistics link corresponding to the target waybill. For example, if the target sending point of the target waybill is point A and the target delivery point is point B, then the target logistics link corresponding to the target waybill is the logistics link from A to B.

[0080] Furthermore, the server can also determine the logistics link from the area where the target sending point is located to the area where the target delivery point is located as the target logistics link corresponding to the target waybill; specifically, in one embodiment, step S220 includes: obtaining the sending area corresponding to the target sending point and the delivery area corresponding to the target delivery point; and determining the logistics link from the sending area to the delivery area as the target logistics link.

[0081] In this context, the sending area refers to the region where the target sending point is located, and the delivery area refers to the region where the target delivery point is located. Specifically, logistics points can be divided into different regions corresponding to different granularities based on different regional divisions. For example, logistics points can be divided into administrative regions corresponding to different administrative division levels. For instance, if point A is a point under Street A of Nanshan District, Shenzhen City, Guangdong Province, then point A will be divided into the Guangdong region, Shenzhen region, Nanshan region, and Street A region according to different administrative division levels. It should be noted that the sending area corresponding to the target sending point and the delivery area corresponding to the target delivery point are obtained within the same regional division granularity.

[0082] Specifically, the server can obtain the sending area corresponding to the target sending point at different regional granularities, and the delivery area corresponding to the target delivery point at different regional granularities. It can then obtain the logistics link between the sending and delivery areas at the same regional granularity and determine it as the target logistics link. For example, if point A is a point under Street A in Nanshan District, Shenzhen, Guangdong Province, and point B is a point under Street B in Jianghan District, Wuhan, Hubei Province, using prefecture-level administrative regions as the regional granularity, the sending area corresponding to point A is Shenzhen, and the delivery area corresponding to point B is Wuhan. Therefore, the logistics link from Shenzhen to Wuhan is determined as the target logistics link for this waybill. Similarly, if the regional granularity is township-level administrative regions, the sending area corresponding to point A is Street A, and the delivery area corresponding to point B is Street B. Therefore, the logistics link from Street A to Street B is determined as the target logistics link for this waybill. By obtaining logistics links at different regional granularities, timeliness prediction can be achieved at different granularities, enabling refined timeliness prediction for waybills.

[0083] S230, based on the preset waybill timeliness prediction model, predicts the time consumption of the target logistics link and obtains the estimated transportation time on the target logistics link.

[0084] The waybill timeliness prediction model is used to predict the timeliness of different logistics links to obtain the estimated transportation time on the target logistics link. Specifically, the waybill timeliness prediction model can be a trained Markov chain model. To predict the timeliness of the target logistics link and obtain the estimated transportation time, the waybill timeliness prediction model needs to be trained in advance to enable it to predict the transportation time of the target logistics link, thus obtaining the trained waybill timeliness prediction model. Then, after determining the target logistics link corresponding to the target waybill, the transportation time of the target logistics link is obtained from the waybill timeliness prediction model, i.e., the estimated transportation time. The steps involved in this embodiment, which predict the time of the target logistics link based on the preset waybill timeliness prediction model to obtain the estimated transportation time on the target logistics link, will be described in detail below.

[0085] In one embodiment, such as Figure 4 As shown, step S230 includes: S410, determining the quantiles of different transportation stages on the target logistics link and the transportation time values ​​at the quantiles through the waybill timeliness prediction model, and obtaining the time prediction data of each transportation stage on the target logistics link; S420, determining the estimated transportation time on the target logistics link based on the time prediction data of each transportation stage on the target logistics link.

[0086] The quantile refers to the quantile value taken from the observation data of the target historical waybill when making this prediction of transportation timeliness. For example, through the waybill timeliness prediction model, the quantile value of a certain transportation stage is obtained as the 75th quantile. That is, after sorting the multiple historical transportation times corresponding to each target historical waybill in the transportation stage from smallest to largest, the 75th percentile of historical transportation time is obtained.

[0087] Specifically, the server determines the quantile values ​​for different transportation stages on the target logistics link using a waybill timeliness prediction model. Then, based on the historical transportation time of the target historical waybills at each transportation stage, it obtains the transportation time value at the quantile value for each transportation stage, thus obtaining the time prediction data for each transportation stage on the target logistics link. Finally, based on the time prediction data for each transportation stage on the target logistics link, it determines the estimated transportation time on the target logistics link. The target historical waybills include multiple historical waybills for which the logistics link is the target logistics link. The server can obtain historical waybills within a preset time period that are for which the logistics link is the target logistics link, as the target historical waybills. For example, if the target logistics link is from point A to point B, the server can obtain all historical waybills from point A to point B from the past month or the past three weeks as the target historical waybills.

[0088] Furthermore, the server obtains multiple target historical waybills with the same logistics link and target logistics link, as well as the historical transportation time of these target historical waybills at each transportation stage; for any transportation stage, the historical transportation time corresponding to each target historical waybill is sorted from smallest to largest, and then the specific time value of the historical transportation time at the quantile is determined, and this time value is determined as the time prediction data of the target waybill at that transportation stage. For example, taking the receiving branch line stage as an example, the quantile of the value for the receiving branch line stage in the logistics chain is first obtained through the transportation timeliness prediction model. If the quantile of the receiving branch line stage is the median (i.e., the 50th quantile), after obtaining the target historical waybill, the historical transportation time of all target historical waybills in the receiving branch line stage is obtained and sorted from smallest to largest. Then, the historical transportation time value at the 50th quantile is determined. For example, if the historical transportation time at the 50th quantile is 12 hours, then the historical transportation time at the 50th quantile (12 hours) will be used as the timeliness prediction data for the target waybill in the receiving branch line stage. Similarly, the method of obtaining timeliness prediction data for other transportation stages is similar, only the data processing objects are different, which will not be elaborated here.

[0089] After obtaining the predicted time consumption data for each transportation stage, the sum of the predicted time consumption data for each transportation stage can be obtained to get the estimated transportation time of the target waybill on the target logistics link. For any target waybill, its transportation time can be predicted by the transportation time of historical waybills. However, the amount of historical waybill data is large and the transportation time values ​​vary. In this embodiment, based on the waybill timeliness prediction model, the quantile of the value in each transportation stage of the logistics link is determined. Then, based on the quantile of the value in each transportation stage, the transportation time of the target waybill is determined from the transportation time of historical waybills. This realizes the addition of the time consumption data of all historical waybills to the transportation time prediction of the current target waybill, effectively improving the accuracy of transportation time.

[0090] S240, determine the estimated delivery time of the target waybill based on the target shipment time and the estimated transportation time.

[0091] The estimated delivery time refers to the time it takes for the courier corresponding to the waybill to reach the customer. Specifically, after obtaining the estimated transportation time of the target waybill on the target logistics chain, the server can perform time conversion based on the estimated transportation time and the target shipment time to obtain the delivery time of the courier corresponding to the target waybill to reach the customer, thus obtaining the estimated delivery time.

[0092] Furthermore, after receiving the items to be sent from the customer, the courier packs and loads them into the vehicle before couriers begin delivery. However, packing and loading the items does not immediately send them to the next branch; instead, they are only transported to the next branch after a certain volume of packages has been accumulated. Therefore, there is a certain time interval between receiving the package and loading it into the vehicle to begin transportation. Thus, in one embodiment, determining the estimated delivery time of the target waybill based on the target shipping time and the estimated transportation time includes: determining the estimated start time of transportation for the target waybill based on the target shipping time; and obtaining the estimated delivery time of the target waybill based on the estimated start time of transportation and the estimated transportation time.

[0093] The server determines the estimated start time of transportation based on the target shipment time. After obtaining the estimated transportation time of the target waybill on the target logistics chain, it performs time conversion based on the estimated transportation time and the estimated start time to obtain the arrival time of the corresponding express delivery to the customer, thus obtaining the estimated delivery time. Further, the server determines the estimated start time of transportation based on the target shipment time, and obtains the time from receipt of the express delivery to loading and commencement of transportation based on the estimated start time and the target shipment time. The sum of this time and the estimated transportation time is then determined as the estimated delivery time of the target waybill. Determining the estimated start time of transportation for the target waybill through the target shipment time further improves the accuracy of the estimated delivery time.

[0094] In one embodiment, determining the estimated start time of transportation for a target waybill based on the target dispatch time includes: obtaining the dispatch time period corresponding to the target dispatch time; obtaining the estimated departure time of the target waybill based on the dispatch time period; and determining the estimated departure time as the estimated start time of transportation.

[0095] Specifically, the server can divide the entire day's shipping time into several shipping time periods, and obtain the correspondence between shipping time and pickup / departure schedules in advance from historical shipping times and pickup / departure schedules in historical waybills. After obtaining the target shipping time for the target waybill, the server can determine the corresponding shipping time period based on the target shipping time, and estimate the pickup / departure schedule for the target waybill based on the pre-learned correspondence between shipping time and pickup / departure schedules. Thus, the estimated departure time of the pickup / departure schedule is determined as the estimated start time of transportation for the target waybill.

[0096] The aforementioned method for predicting delivery time involves obtaining the target shipping time, target shipping point, and target delivery point for the target waybill. Based on these locations, the target logistics link is determined. A pre-defined waybill delivery time prediction model is used to predict the time consumption of this logistics link, obtaining the estimated transportation time. Finally, based on the target shipping time and the estimated transportation time, the estimated delivery time of the target waybill is determined. By obtaining the transportation time of the logistics link corresponding to the shipping and delivery points, and combining this with the transportation time of the logistics link and the shipping time, the delivery time of the waybill is predicted, improving the accuracy of delivery time prediction. Furthermore, users receive more accurate delivery times, helping them make informed product selections and increasing their trust in the service. Simultaneously, the pre-predicted estimated delivery time provides information related to the promised delivery time, assisting couriers in adjusting their delivery schedule, improving delivery efficiency, and enhancing overall logistics effectiveness.

[0097] In one embodiment, such as Figure 5 As shown, the quantiles of different transportation stages on the target logistics chain are determined using a waybill timeliness prediction model, including:

[0098] S510, acquire training data. The training data includes training samples corresponding to different logistics links and quantile time consumption of each transportation stage on different logistics links at different quantiles. The training samples include sample shipment time and sample actual transportation time.

[0099] The training samples include training samples from different logistics links, that is, training samples from logistics links between any two logistics network points. Specifically, the server can collect historical waybills from the entire network for a preset historical time period, generate a training sample for each historical waybill, and obtain the transportation time at different quantiles for each transportation stage on different logistics links based on the transportation time of all historical waybills corresponding to different logistics links at each transportation stage.

[0100] In one embodiment, acquiring training data includes: acquiring multiple historical waybills; dividing the historical waybills into historical waybill groups corresponding to different logistics links based on the sending and delivery points of the historical waybills; acquiring the quantile time corresponding to different quantiles of each transportation stage on the logistics link corresponding to each historical waybill group based on the historical transportation time of each historical waybill in each transportation stage; and generating training samples corresponding to each logistics link based on the sending time, actual transportation time, sending point, and delivery point of the historical waybills in each historical waybill group.

[0101] Specifically, the server can collect historical waybills from the entire network within a preset historical time period, and determine the logistics link of each historical waybill based on the sending and receiving points of each historical waybill. Historical orders with the same logistics link are grouped together to obtain historical waybill groups corresponding to different logistics links. After obtaining the historical waybill groups, for each historical waybill group, the historical transportation time of the historical waybills in each transportation stage is obtained. For each transportation stage, all historical transportation times (each historical transportation time corresponds to a historical waybill) in the transportation stage are sorted from smallest to largest, and then the historical transportation time value at each quantile is determined. This yields the transportation time at each quantile in each transportation stage, and finally, the quantile time corresponding to different quantiles of each transportation stage on different logistics links is obtained.

[0102] For example, the server can collect historical waybills from the entire network over the past month and group them according to the sending and receiving points of each historical waybill. Among these, there are 100 historical waybills with a logistics link from point A to point B. 100 training samples are generated for these 100 historical waybills. Simultaneously, the transportation time for each stage of the logistics for these 100 historical waybills is obtained and sorted from smallest to largest. This determines the historical transportation time value at each quantile, thus obtaining the transportation time of the logistics link from point A to point B at different quantiles for each logistics stage. Taking the transit stage as an example, after obtaining the transportation time for these 100 waybills in the transit stage, the transportation time value at each quantile is obtained, thus obtaining the quantile time of the transit stage of the logistics link from point A to point B at different quantiles. The process of obtaining the quantile time at different quantiles is the same for other transportation stages of the logistics link from point A to point B. It is understandable that the process of obtaining the quantile time corresponding to different quantiles in various transportation stages of other logistics links is similar, the only difference being the data processing object, which will not be elaborated here.

[0103] Furthermore, after obtaining historical waybills, the server can first perform data cleaning, data feature filtering, and other data processing on the waybill data of the historical waybills, and then generate training data based on the processed waybill data corresponding to the historical waybills. Specifically, in the data processing, firstly, abnormal data in the waybill data corresponding to historical waybills is removed, such as transportation abnormalities caused by vehicle abnormalities, misdelivery, forwarding, and termination of express delivery due to random external factors. Then, the cleaned waybill data is segmented according to each transportation segment to obtain waybill data for each transportation stage. The waybill data for each transportation stage is then processed and transformed, including time dimension information (such as key operation time, time spent in different transportation stages, etc.), location information transformation (such as network type, cities, administrative regions, and network points involved in the transportation of the express shipment corresponding to the waybill), and product information and capacity type related transformation. Finally, training samples corresponding to historical waybills are generated based on the processed waybill data, and based on the processed waybill data, the quantile time corresponding to different quantiles of each transportation stage in the logistics link is obtained.

[0104] S520: Input each training sample into the pre-built Markov chain model. Based on the preset state values ​​of each Markov state in the Markov chain model, obtain the sample time prediction data of each transportation stage in the sample logistics chain corresponding to each training sample from the quantile time corresponding to different quantiles of each transportation stage in different logistics links. Each Markov state in the Markov chain model corresponds to a different transportation stage.

[0105] Since logistics links can be divided into different transportation stages, the state of each stage is often only affected by the state of the previous stage. Taking the transit stage in a logistics link as an example, the time it takes for goods to travel from the initial transit point to the final transit point (i.e., the transit time) is often related to the time it takes for the goods to arrive at the initial transit point in the receiving branch stage (this arrival time is taken as the start time of the transit stage). For example, if the goods arrive at the initial transit point in the afternoon, they will often be transported to the final transit point the next day, resulting in a longer transit time. Therefore, a waybill timeliness prediction model can be constructed using a Markov chain model. The Markov chain model includes multiple Markov states corresponding to each transportation stage, and each Markov state represents the transit time in the corresponding transportation stage. Since in logistics scenarios, the various transportation stages in the logistics link are sent sequentially, with only directional state transitions, using a Markov chain model to construct a waybill timeliness prediction model can simplify the model and reduce its size.

[0106] After acquiring the training data, the server inputs each training sample into a pre-built Markov chain model. The Markov chain model is used to obtain the estimated delivery time of each training sample. The specific parameters of each Markov state in the Markov chain model are set as quantiles. In other words, the transportation time value corresponding to the transportation stage is the value of the historical transportation time data for that transportation stage at that quantile. Specifically, after inputting each training sample into the pre-built Markov chain model, the preset state values ​​of each Markov state in the Markov chain model are first obtained. These preset state values ​​represent the initial quantile values ​​in the corresponding transportation stage. Then, for any training sample, the sample logistics link of the training sample is determined, and the quantile time corresponding to the different quantiles in each transportation stage of the sample logistics link is determined from the quantile time corresponding to the different quantiles in each transportation stage of the different logistics links. Then, based on the preset state values ​​of different Markov states, the quantile time at the quantile corresponding to the preset state value in the corresponding transportation stage is determined. Finally, the obtained quantile time is determined as the sample time prediction data of the training sample in each transportation stage.

[0107] For example, if the preset state value of each Markov state is the 50th quantile, and the sample transportation link of a certain training sample is the transportation link from point A to point B, then first determine the quantile time corresponding to the different quantiles of each transportation stage from point A to point B from the quantile time corresponding to the different quantiles of each transportation stage on different logistics links. Then, obtain the quantile time corresponding to the 50th quantile in each transportation stage, that is, obtain the sample time prediction data of the sample logistics link corresponding to the training sample in each transportation stage.

[0108] S530: Based on the sample time prediction data of each transportation stage in the sample logistics link, obtain the sample estimated transportation time of each sample logistics link.

[0109] S540, based on the sample dispatch time of each training sample and the estimated transportation time of the sample logistics link corresponding to each training sample, determine the estimated delivery time of the training sample.

[0110] Specifically, after obtaining the sample time prediction data for each transportation stage, the server uses this data to obtain the estimated transportation time for each sample's logistics link. Finally, based on the sample dispatch time of each training sample and the estimated transportation time of the corresponding sample logistics link, the estimated delivery time of the training samples is determined.

[0111] Further, in one embodiment, determining the estimated delivery time of a training sample based on the sample dispatch time of each training sample and the estimated transportation time of the sample logistics link corresponding to each training sample includes: obtaining the sample transportation start time of each training sample based on the sample dispatch time of each training sample; and determining the estimated delivery time of the training sample based on the sample transportation start time of each training sample and the estimated transportation time of the sample logistics link corresponding to each training sample.

[0112] S550, based on the actual delivery time and estimated delivery time of each training sample, adjusts the preset state value of each Markov state in the Markov chain model to obtain the target state value.

[0113] S560 determines the target state value as the quantile of values ​​at different transportation stages.

[0114] After obtaining the estimated delivery time of the training samples, the preset state values ​​of each Markov state in the Markov chain model are adjusted based on the actual and estimated delivery times of each training sample. The final parameter value (target state value) corresponding to each Markov state in the Markov model is the quantile of the corresponding transportation stage. The conditions for stopping parameter iteration adjustment can include: 1. The error between the actual and estimated delivery times of the training samples is less than a pre-set small value; 2. The change in state values ​​between two iterations is very small; a threshold can be set, and training stops when the value is less than this threshold; 3. A maximum number of iterations is set, and training stops when the maximum number of iterations is exceeded.

[0115] Furthermore, in one embodiment, the preset state values ​​of each Markov state in the Markov chain model are adjusted according to the actual delivery time and the estimated delivery time of each training sample to obtain the target state value. This includes: obtaining the sample time consumption error and the time consumption achievement rate according to the actual delivery time and the estimated delivery time of each training sample; and adjusting the preset state values ​​of the Markov states according to the time consumption error based on the time consumption error, provided that the time consumption achievement rate is greater than or equal to the preset achievement rate, until the sample time consumption error is minimized, thereby obtaining the target state value.

[0116] The time-consuming achievement rate refers to the percentage of training samples whose actual delivery time is less than their estimated delivery time, out of the total training sample size. The preset achievement rate is a threshold set based on actual conditions. See also Figure 6 In the logistics field, the actual arrival time of a package corresponding to a waybill often falls into three categories: 1. If the estimated delivery time is B, the actual delivery time is earlier than the estimated time; 2. If the estimated delivery time is A, the actual delivery time is later than the estimated time; 3. The actual delivery time is the same as the estimated time. Considering the actual logistics process and transportation capacity, the estimated delivery time given by the model often needs to fall into either category 1 or 3. Therefore, when adjusting the preset state value, the smaller the error between the actual delivery time and the estimated delivery time, the better. At the same time, the more training samples with a positive error between the actual delivery time and the estimated delivery time, the better. Specifically, after obtaining the actual delivery time and estimated delivery time of the sample data, the server obtains the sample time consumption error and time consumption achievement rate between the actual delivery time and the estimated delivery time. Based on the time consumption achievement rate being greater than or equal to the preset achievement rate, the server iteratively adjusts the state values ​​of the Markov states corresponding to each logistics link stage until the sample time consumption error reaches the minimum value. Then, the state values ​​of each Markov state when the sample time consumption error reaches the minimum value are determined as the target state values.

[0117] Specifically, the training objective for a Markov chain model can be expressed as follows:

[0118]

[0119]

[0120] Among them, t i Let t′ be the actual delivery time of the i-th training sample. i The estimated delivery time for the i-th training sample; Ar rate E represents the time taken to achieve the goal. mae This represents the sample time consumption error.

[0121] The preset achievement rate can be set according to the actual situation, allowing for dynamic adjustment of the tightness of the transportation time forecast. For example, when the preset achievement rate is set to greater than 50%, the estimated delivery time is later than when it is set to 50%, meaning that considering the current logistics capacity, the estimated delivery time is a transportation time that is more easily achieved. When the preset achievement rate is set to less than 50%, the estimated delivery time is longer than when it is set to 50%, with more flexibility, meaning that considering the current logistics capacity, the estimated delivery time is shorter than when it is set to 50%.

[0122] Although this embodiment does not distinguish whether the server 200 is a training server that only performs model training tasks, a recognition server that only performs object recognition tasks, or a comprehensive server that can perform both tasks, the server functions can be set according to business needs in actual applications, and this application does not limit the specifics.

[0123] like Figure 7 As shown, the following example further illustrates the estimated delivery time of each training sample. The logistics chain can be divided into the receiving stage, the receiving branch line stage, the transit stage, the delivery branch line stage, and the delivery stage. The sample transportation chain for a certain training sample is the transportation chain from point A to point B. First, the entire day's shipping time can be divided into several shipping time periods. The correspondence between shipping time and receiving / departure times is pre-obtained from historical shipping times and receiving / departure schedules in historical waybills. After obtaining the sample shipping time of the training sample, the corresponding shipping time period can be determined based on the sample shipping time. Then, based on the pre-learned correspondence between shipping time and receiving / departure times, the departure times of the training sample are estimated. The departure time of this estimated departure time is then determined as the sample transportation start time of the training sample.

[0124] Then, based on the initial state value of the Markov state corresponding to the dispatch branch stage (e.g., the initial state value of the Markov state corresponding to the dispatch branch stage is the 70th percentile), the quantile time corresponding to the quantiles at different quantiles in the dispatch branch stage is determined as the time for the dispatch branch stage. Then, based on the time for the dispatch branch stage and the sample transportation start time, the arrival time of historical waybills at the dispatch transit hub in the dispatch branch stage is obtained. Furthermore, the corresponding arrival time of historical waybills at the dispatch transit hub can be determined. The arrival time of historical waybills at the dispatch transit hub is used as the start time of the transit stage.

[0125] Based on the initial state value of the Markov state corresponding to the transit stage (e.g., the initial state value of the Markov state corresponding to the transit stage is the 60th percentile), the quantile time to the 60th percentile is determined from the quantile times corresponding to different quantiles in the transit stage for the transportation link from point A to point B, and this quantile time is used as the transit stage duration. Based on the start time and duration of the transit stage, the arrival time of historical waybills at the delivery transit center during the transit stage is obtained. The arrival time of historical waybills at the delivery transit center is used as the start time of the delivery branch stage.

[0126] Based on the initial state value of the Markov state corresponding to the delivery branch stage, for example, if the initial state value of the Markov state corresponding to the delivery branch stage is the 55th percentile, the quantile time corresponding to different quantiles in the transportation link from point A to point B in the delivery branch stage is determined as the quantile time to the 55th percentile and used as the time for the delivery branch stage. Based on the start time of the delivery branch stage and the time for the delivery branch stage, the arrival time of historical waybills at the delivery point in the delivery branch stage is obtained, and the arrival time of historical waybills at the delivery point is used as the start time of the delivery stage.

[0127] Based on the initial state value of the Markov state corresponding to the delivery stage (e.g., the initial state value of the Markov state corresponding to the delivery stage is the 50th percentile), the quantile time to the 50th percentile is determined from the quantile times corresponding to different quantiles in the delivery link from point A to point B during the delivery stage, and this is taken as the delivery stage time. Based on the start time and delivery stage time, the time when historical waybills arrived at the customer's hands during the delivery stage is obtained, resulting in the sample estimated delivery time.

[0128] Furthermore, the training of the Markov chain model, or the prediction of waybill delivery time, relies on historical waybill data. For logistics links not covered by historical waybills, or logistics links with a small number of historical waybills, the accuracy of the predicted delivery time is greatly reduced. Therefore, in one embodiment, historical waybills are divided into historical waybill groups corresponding to different logistics links based on the sending and receiving points of historical waybills. This includes: dividing logistics points into different logistics regions based on a preset regional division granularity; obtaining the first logistics region to which the logistics point corresponding to the sending point of the historical waybill belongs, and the second logistics region to which the logistics point corresponding to the sending point of the historical waybill belongs; and dividing the historical waybill into historical waybill groups corresponding to different logistics links based on the first and second logistics regions of the historical waybill.

[0129] Logistics outlets can be divided into different logistics regions based on varying degrees of granularity. For example, they can be assigned to different administrative regions based on their administrative division level. Figure 8 As shown in the diagram on the right, each logistics point can be assigned to logistics region A in the diagram on the left. The logistics regions, divided at different granularities, ensure that all network points are covered and there is no overlap. The server can obtain the first logistics region to which the sending point of a historical waybill belongs, and the second logistics region to which the sending point of a historical waybill belongs. Based on the first and second logistics regions of the historical waybill, the logistics link of the historical waybill is determined. Finally, historical waybills with the same logistics link are grouped into the same historical waybill group, resulting in historical waybill groups with different logistics links.

[0130] Taking different administrative division levels as examples, let's assume that point A is a point under Street A in Nanshan District, Shenzhen, Guangdong Province. If we use the township-level administrative division as the regional division granularity, the logistics area corresponding to point A is Street A. If we use the prefecture-level administrative division as the regional division granularity, the logistics area corresponding to point A is Shenzhen.

[0131] By dividing logistics networks into different granularities based on regional divisions, logistics outlets are assigned to different logistics regions corresponding to varying granularity. For logistics links not covered by historical waybills or with a low volume of historical waybills, historical waybills from which the sending and receiving outlets are located in the same logistics region can be used as training data. Through end-to-end model training, the accuracy of delivery time prediction for logistics links not covered by historical waybills or with a low volume of historical waybills can be improved. Furthermore, delivery time prediction at different granularities is implemented, achieving refined delivery time prediction.

[0132] Understandably, when predicting delivery time later, the server can obtain the logistics area corresponding to the target sending point at a certain regional granularity to obtain the sending area, and obtain the logistics area corresponding to the target delivery point at the same regional granularity to obtain the delivery area; then the logistics link between the sending area and the delivery area is determined as the target logistics link.

[0133] In one embodiment, the estimated transportation time on the target logistics link is predicted based on a preset waybill timeliness prediction model, including: determining a timeliness configuration table according to the waybill timeliness prediction model; and looking up the corresponding estimated transportation time in the timeliness configuration table according to the target logistics link.

[0134] The timeliness configuration table includes the estimated transportation time for different logistics links. Specifically, the server determines the quantile values ​​for different transportation stages using a waybill timeliness prediction model. Then, based on the historical transportation times of all historical waybills across the entire network at each transportation stage, it obtains the transportation time values ​​at the quantile values ​​for each transportation stage of the logistics link corresponding to the historical waybill, thus obtaining the timeliness prediction data for each transportation stage on the logistics link corresponding to the historical waybill. Furthermore, based on the timeliness prediction data for each transportation stage on each logistics link, it obtains the estimated transportation time for different logistics links and generates the timeliness configuration table. After obtaining the target logistics link corresponding to the target waybill, the server can directly look up the corresponding estimated transportation time from the timeliness configuration table. By pre-storing the estimated transportation times for different logistics links in the timeliness configuration table, the estimated transportation time for the target waybill can be directly obtained through a table lookup operation, effectively improving the prediction efficiency of transportation timeliness.

[0135] The following combination Figure 9 This embodiment will be further described. See [link to documentation]. Figure 7 The terminal sends a transportation timeliness query request to the server. After receiving the request, the server retrieves relevant information about the corresponding waybill, such as the target dispatch time, target dispatch point, and target delivery point. The server converts the received information to obtain the query key and retrieves the corresponding timeliness value from the timeliness configuration table stored in the cache (Redis). If the retrieval fails, the server retrieves the logistics area corresponding to the target dispatch point at a certain regional granularity to obtain the dispatch area, and retrieves the logistics area corresponding to the target delivery point at the same regional granularity to obtain the delivery area. The logistics link between the dispatch area and the delivery area is then determined as the target logistics link. This adjusts the query key and retrieves the corresponding timeliness value from the timeliness configuration table stored in the cache (Redis). If the retrieval fails, an empty value is returned. If the timeliness value is successfully retrieved, it is output to the terminal as the predicted transportation timeliness. Furthermore, after obtaining the timeliness value, the server can input the timeliness value into the correction process. After correction through static planning and relevant business scenarios, the final timeliness is output. At the same time, the final timeliness is stored in the cache (Redis) for subsequent server use.

[0136] Furthermore, the prediction of transportation timeliness can be incorporated into an evaluation system. The main dimensions of this system include online response performance (e.g., daily request volume, peak QPS, timeout rate, cache hit rate), overall solution coverage (e.g., network / network AOI model coverage, new route model coverage), overall solution effectiveness (e.g., online achievement rate of different timeliness levels, error rate), and individual model effectiveness (e.g., average error, achievement rate, percentage of requests within a certain error range). The server uses this evaluation system to validate the online data related to predicted transportation timeliness. Once validation is successful, the data is used for model training.

[0137] To better implement the transportation timeliness prediction method provided in the embodiments of this application, based on the transportation timeliness prediction method proposed in the embodiments of this application, the embodiments of this application also provide a transportation timeliness prediction device, such as... Figure 10 As shown, the transportation timeliness prediction device 1000 includes:

[0138] The waybill acquisition module 1010 is used to acquire the target dispatch time, target dispatch point and target delivery point of the target waybill.

[0139] The logistics link acquisition module 1020 is used to determine the target logistics link based on the target sending point and the target delivery point.

[0140] The transportation time estimation module 1030 is used to predict the time consumption of the target logistics link based on the preset waybill timeliness prediction model, and obtain the estimated transportation time on the target logistics link.

[0141] The delivery time estimation module 1040 is used to determine the estimated delivery time of the target waybill based on the target shipment time and the estimated transportation time.

[0142] In one embodiment, the logistics link acquisition module 1020 is specifically used to acquire the sending area corresponding to the target sending point and the delivery area corresponding to the target delivery point; and to determine the logistics link from the sending area to the delivery area as the target logistics link.

[0143] In one embodiment, the transportation time estimation module 1030 is specifically used to determine the quantile values ​​of different transportation stages on the target logistics link and the transportation time values ​​at the quantile values ​​through the waybill timeliness prediction model, so as to obtain the time prediction data of each transportation stage on the target logistics link; and to determine the estimated transportation time on the target logistics link based on the time prediction data of each transportation stage on the target logistics link.

[0144] In one embodiment, the transportation time estimation module 1030 is further configured to acquire training data, including training samples corresponding to different logistics links and quantile times corresponding to different transportation stages at different quantiles on different logistics links. The training samples include sample shipping times and sample actual transportation times. Each training sample is input into a pre-constructed Markov chain model. Based on the preset state values ​​of each Markov state in the Markov chain model, the module obtains sample time prediction data for each transportation stage on the sample logistics link corresponding to each training sample from the quantile times corresponding to different quantiles for each transportation stage on different logistics links. In the Markov chain model, each Markov state corresponds to a different transportation stage. Based on the sample time prediction data of each transportation stage in each sample logistics link, the sample estimated transportation time of each sample logistics link is obtained. Based on the sample dispatch time of each training sample and the sample estimated transportation time of the corresponding sample logistics link, the sample estimated delivery time of the training sample is determined. Based on the sample actual delivery time and sample estimated delivery time corresponding to each training sample, the preset state value of each Markov state in the Markov chain model is adjusted to obtain the target state value. The target state value is determined as the quantile of the value in different transportation stages.

[0145] In one embodiment, the transportation time estimation module 1030 is further configured to obtain the sample time error and the time achievement rate based on the actual delivery time and the estimated delivery time of each training sample; and, based on the time achievement rate being greater than or equal to the preset achievement rate, adjust the preset state value of the Markov state according to the time error until the sample time error is minimized, thereby obtaining the target state value.

[0146] In one embodiment, the transportation time estimation module 1030 is further configured to obtain the sample transportation start time of each training sample based on the sample dispatch time of each training sample; and determine the estimated delivery time of the training sample based on the sample transportation start time of each training sample and the estimated transportation time of the sample logistics link corresponding to each training sample.

[0147] In one embodiment, the transportation time estimation module 1030 is further configured to acquire multiple historical waybills, divide the historical waybills into historical waybill groups corresponding to different logistics links based on the sending and receiving points of the historical waybills; acquire the quantile time corresponding to different quantiles of each transportation stage on the logistics link corresponding to each historical waybill group based on the historical transportation time of each historical waybill in each transportation stage; and generate training samples corresponding to each logistics link based on the sending time, actual transportation time, sending point, and receiving point of the historical waybill in each historical waybill group.

[0148] In one embodiment, the transportation time estimation module 1030 is further configured to divide logistics outlets into different logistics regions based on a preset regional division granularity; obtain the first logistics region to which the logistics outlet corresponding to the sender of the historical waybill belongs, and the second logistics region to which the logistics outlet corresponding to the sender of the historical waybill belongs; and divide the historical waybill into historical waybill groups corresponding to different logistics links according to the first logistics region and the second logistics region of the historical waybill.

[0149] In one embodiment, the delivery time estimation module 1040 is specifically used to determine the estimated start time of transportation of the target waybill based on the target dispatch time; and to obtain the estimated delivery time of the target waybill based on the estimated start time of transportation and the estimated transportation time.

[0150] In one embodiment, the delivery time estimation module 1040 is further specifically used to obtain the shipping time period corresponding to the target shipping time; obtain the estimated departure time of the target waybill based on the shipping time period, and determine the estimated departure time as the estimated start time of transportation.

[0151] In the above embodiments, the target dispatch time, target dispatch point, and target delivery point of the target waybill are obtained. Then, the target logistics link is determined based on the target dispatch point and the target delivery point. The time consumption of the target logistics link is predicted based on a preset waybill timeliness prediction model to obtain the estimated transportation time on the target logistics link. Based on the target dispatch time and the estimated transportation time, the estimated delivery time of the target waybill is determined. By obtaining the transportation time of the logistics link corresponding to the dispatch point and the delivery point, and combining the transportation time of the logistics link with the dispatch time, the delivery time of the waybill can be predicted, improving the accuracy of transportation timeliness prediction.

[0152] In some embodiments of this application, the transportation timeliness prediction device 1000 can be implemented as a computer program, which can be implemented in, for example... Figure 11 The computer device shown operates on this device. The computer device's memory can store various program modules that make up the prediction device 1000 for transport timeliness, for example, Figure 10 The diagram shows a waybill acquisition module 1010, a logistics link acquisition module 1020, a transportation time estimation module 1030, and a delivery time estimation module 1040. The computer program comprised of these modules causes a processor to execute the steps in the transportation timeliness prediction methods of the various embodiments of this application described in this specification.

[0153] For example, Figure 11 The computer equipment shown can be used as follows Figure 10The waybill acquisition module 1010 in the transportation timeliness prediction device 1000 shown executes step S210. The computer device can execute step S220 via the logistics link acquisition module 1020. The computer device can execute step S230 via the transportation time estimation module 1030. The computer device can execute step S240 via the delivery time estimation module 1040. The computer device includes a processor, memory, and a network interface connected via a system bus. The processor of the computer device provides computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and internal memory. The non-volatile storage medium stores the operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage medium. The network interface of the computer device is used to communicate with external computer devices via a network connection. When the computer program is executed by the processor, it implements a method for predicting transportation timeliness.

[0154] Those skilled in the art will understand that Figure 11 The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computer device to which the present application is applied. Specific computer devices may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0155] In some embodiments of this application, a computer device is provided, including one or more processors; a memory; and one or more application programs, wherein the one or more application programs are stored in the memory and configured to be executed by the processors as steps of the above-described method for predicting transportation timeliness. The steps of the method for predicting transportation timeliness here may be steps from the methods for predicting transportation timeliness in the various embodiments described above.

[0156] In some embodiments of this application, a computer-readable storage medium is provided, storing a computer program that is loaded by a processor, causing the processor to execute the steps of the above-described method for predicting transportation timeliness. The steps of the method for predicting transportation timeliness here may be steps from the methods for predicting transportation timeliness in the various embodiments described above.

[0157] Those skilled in the art will understand that all or part of the processes in the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium, and when executed, it can include the processes of the embodiments of the above methods. Any references to memory, storage, databases, or other media used in the embodiments provided in this application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, or optical storage, etc. Volatile memory can include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM), etc.

[0158] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0159] The foregoing has provided a detailed description of a method, apparatus, computer device, and storage medium for predicting transportation timeliness provided in the embodiments of this application. Specific examples have been used to illustrate the principles and implementation methods of the present invention. The descriptions of the above embodiments are only for the purpose of helping to understand the method and core ideas of the present invention. At the same time, for those skilled in the art, there will be changes in the specific implementation methods and application scope based on the ideas of the present invention. Therefore, the content of this specification should not be construed as a limitation of the present invention.

Claims

1. A method for predicting transportation timeliness, characterized in that, include: Obtain the target shipping time, target shipping location, and target delivery location for the target waybill; The target logistics link is determined based on the target parcel delivery point and the target parcel collection point; Based on a pre-defined waybill timeliness prediction model, the time consumption of the target logistics link is predicted to obtain the estimated transportation time on the target logistics link. This includes: acquiring training data, where the target logistics link includes multiple transportation stages; the training data includes training samples corresponding to different logistics links and quantile times corresponding to different quantiles for each transportation stage on different logistics links; the training samples include sample shipping times and sample actual transportation times; the waybill timeliness prediction model includes a Markov chain model; each training sample is input into a pre-constructed Markov chain model; based on the pre-defined state values ​​of each Markov state in the Markov chain model, sample time consumption prediction data for each transportation stage on the sample logistics link corresponding to each training sample is obtained from the quantile times corresponding to different quantiles for each transportation stage on different logistics links; wherein, each Markov state in the Markov chain model... Each state corresponds to a different transportation stage; based on the sample time prediction data for each transportation stage on each sample logistics link, the estimated transportation time for each sample logistics link is obtained; based on the sample shipping time of each training sample and the estimated transportation time of the sample logistics link corresponding to each training sample, the estimated delivery time of the training sample is determined; based on the actual delivery time and estimated delivery time of the sample corresponding to each training sample, the preset state values ​​of each Markov state in the Markov chain model are adjusted to obtain the target state value; the target state value is determined as the quantile of values ​​in different transportation stages; the transportation time value at the quantile is determined to obtain the time prediction data for each transportation stage on the target logistics link; based on the time prediction data for each transportation stage on the target logistics link, the estimated transportation time on the target logistics link is determined. The estimated delivery time of the target waybill is determined based on the target shipping time and the estimated transportation time.

2. The method according to claim 1, characterized in that, Determining the target logistics link based on the target sending point and the target delivery point includes: Obtain the parcel sending area corresponding to the target parcel sending point and the parcel delivery area corresponding to the target parcel delivery point; The logistics link from the sending area to the delivery area is determined as the target logistics link.

3. The method according to claim 1, characterized in that, The step of adjusting the preset state values ​​of each Markov state in the Markov chain model based on the actual delivery time and the estimated delivery time of each training sample to obtain the target state value includes: Based on the actual delivery time and the estimated delivery time of each training sample, obtain the sample time consumption error and the time consumption achievement rate. Based on the time achievement rate being greater than or equal to the preset achievement rate, the preset state value of the Markov state is adjusted according to the time error until the sample time error is minimized, thereby obtaining the target state value.

4. The method according to claim 1, characterized in that, The step of determining the estimated delivery time of the training samples based on the sample shipping time of each training sample and the estimated transportation time of the sample logistics link corresponding to each training sample includes: Based on the sample shipping time of each training sample, obtain the sample transportation start time of each training sample; The estimated delivery time of the training samples is determined based on the sample transportation start time of each training sample and the estimated sample transportation time of the corresponding sample logistics link.

5. The method according to claim 1, characterized in that, The acquisition of training data includes: Obtain multiple historical waybills, and classify the historical waybills into historical waybill groups corresponding to different logistics links based on the sending and receiving outlets of the historical waybills; Based on the historical transportation time of each historical waybill under each historical waybill group at each transportation stage, obtain the quantile time corresponding to different quantiles of each transportation stage on the logistics link corresponding to each historical waybill group. Training samples are generated for each logistics link based on the dispatch time, actual transportation time, dispatch point, and delivery point of the historical waybill under each historical waybill group.

6. The method according to claim 5, characterized in that, The step of dividing the historical waybills into historical waybill groups corresponding to different logistics links based on the sending and receiving outlets of the historical waybills includes: Based on the preset regional division granularity, logistics outlets are divided into different logistics regions; Obtain the first logistics region to which the sending point of the historical waybill belongs, and the second logistics region to which the delivery point of the historical waybill belongs; Based on the first logistics region and the second logistics region of the historical waybill, the historical waybill is divided into historical waybill groups corresponding to different logistics links.

7. The method according to claim 1, characterized in that, Determining the estimated delivery time of the target waybill based on the target shipping time and the estimated transportation time includes: Based on the target shipment time, determine the estimated start time of transportation for the target waybill; The estimated delivery time of the target waybill is obtained based on the estimated start time of transportation and the estimated transportation time.

8. The method according to claim 7, characterized in that, The step of determining the estimated start time of transportation for the target waybill based on the target shipment time includes: Obtain the shipping time period corresponding to the target shipping time; Based on the shipping time period, the estimated departure time of the target waybill is obtained, and the estimated departure time is determined as the estimated start time of transportation.

9. A device for obtaining transportation timeliness, characterized in that, The device includes: The waybill acquisition module is used to obtain the target dispatch time, target dispatch point, and target delivery point for a target waybill. The logistics link acquisition module is used to determine the target logistics link based on the target sending point and the target delivery point. The transportation time estimation module is used to predict the transportation time of the target logistics link based on a preset waybill timeliness prediction model, and obtain the estimated transportation time of the target logistics link. This includes: acquiring training data, wherein the target logistics link includes multiple transportation stages, the training data includes training samples corresponding to different logistics links and quantile times corresponding to different quantiles for each transportation stage on different logistics links, the training samples including sample dispatch time and sample actual transportation time; the waybill timeliness prediction model includes a Markov chain model, each training sample is input into a pre-constructed Markov chain model, and based on the preset state values ​​of each Markov state in the Markov chain model, sample time prediction data for each transportation stage on the sample logistics link corresponding to each training sample is obtained from the quantile times corresponding to different quantiles for each transportation stage on different logistics links. The Markov states correspond to different transportation stages. Based on the sample time prediction data for each transportation stage in each sample logistics link, the estimated transportation time for each sample logistics link is obtained. Based on the sample shipping time of each training sample and the estimated transportation time of the corresponding sample logistics link, the estimated delivery time of the training sample is determined. Based on the actual delivery time and estimated delivery time of each training sample, the preset state values ​​of each Markov state in the Markov chain model are adjusted to obtain target state values. The target state values ​​are determined as quantiles for different transportation stages. The transportation time values ​​at these quantiles are determined to obtain the time prediction data for each transportation stage in the target logistics link. Based on the time prediction data for each transportation stage in the target logistics link, the estimated transportation time for the target logistics link is determined. The delivery time estimation module is used to determine the estimated delivery time of the target waybill based on the target shipment time and the estimated transportation time.

10. A computer device, characterized in that, The computer device includes: One or more processors; Memory; and One or more applications, wherein the one or more applications are stored in the memory and configured to be executed by the processor to implement the method for predicting transport timeliness as claimed in any one of claims 1 to 8.

11. A computer-readable storage medium, characterized in that, It stores a computer program, which is loaded by a processor to perform the steps in the method for predicting transport timeliness as described in any one of claims 1 to 8.

Images