CN117993545A - Method, device, computing equipment and storage medium for predicting arrival time of package - Google Patents
Method, device, computing equipment and storage medium for predicting arrival time of package Download PDFInfo
- Publication number
- CN117993545A CN117993545A CN202410033829.5A CN202410033829A CN117993545A CN 117993545 A CN117993545 A CN 117993545A CN 202410033829 A CN202410033829 A CN 202410033829A CN 117993545 A CN117993545 A CN 117993545A
- Authority
- CN
- China
- Prior art keywords
- target
- time
- logistics
- package
- node
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 105
- 238000003860 storage Methods 0.000 title claims abstract description 13
- 230000032683 aging Effects 0.000 claims abstract description 77
- 238000012549 training Methods 0.000 claims description 112
- 230000008569 process Effects 0.000 description 28
- 238000012384 transportation and delivery Methods 0.000 description 27
- 238000009826 distribution Methods 0.000 description 24
- 238000010586 diagram Methods 0.000 description 11
- 238000004590 computer program Methods 0.000 description 10
- 238000005070 sampling Methods 0.000 description 10
- 230000006870 function Effects 0.000 description 8
- 238000004891 communication Methods 0.000 description 6
- 238000005457 optimization Methods 0.000 description 6
- 238000012545 processing Methods 0.000 description 6
- 230000003247 decreasing effect Effects 0.000 description 5
- 238000012544 monitoring process Methods 0.000 description 5
- 230000018109 developmental process Effects 0.000 description 4
- 238000003058 natural language processing Methods 0.000 description 4
- 238000007781 pre-processing Methods 0.000 description 4
- 230000001550 time effect Effects 0.000 description 4
- 238000012546 transfer Methods 0.000 description 4
- 230000009471 action Effects 0.000 description 3
- 238000004422 calculation algorithm Methods 0.000 description 3
- 238000000546 chi-square test Methods 0.000 description 3
- 230000007246 mechanism Effects 0.000 description 3
- 230000002787 reinforcement Effects 0.000 description 3
- 238000012216 screening Methods 0.000 description 3
- 238000004458 analytical method Methods 0.000 description 2
- 238000013473 artificial intelligence Methods 0.000 description 2
- 230000001934 delay Effects 0.000 description 2
- 238000011161 development Methods 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 238000010801 machine learning Methods 0.000 description 2
- 238000004519 manufacturing process Methods 0.000 description 2
- 238000007637 random forest analysis Methods 0.000 description 2
- 238000013468 resource allocation Methods 0.000 description 2
- 230000004044 response Effects 0.000 description 2
- 206010012186 Delayed delivery Diseases 0.000 description 1
- 230000005856 abnormality Effects 0.000 description 1
- 238000013528 artificial neural network Methods 0.000 description 1
- 238000013475 authorization Methods 0.000 description 1
- 230000006399 behavior Effects 0.000 description 1
- 230000033228 biological regulation Effects 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 230000001413 cellular effect Effects 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 238000006243 chemical reaction Methods 0.000 description 1
- 125000004122 cyclic group Chemical group 0.000 description 1
- 238000003066 decision tree Methods 0.000 description 1
- 230000007423 decrease Effects 0.000 description 1
- 238000013136 deep learning model Methods 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 230000008451 emotion Effects 0.000 description 1
- 230000004927 fusion Effects 0.000 description 1
- 230000002452 interceptive effect Effects 0.000 description 1
- 238000011835 investigation Methods 0.000 description 1
- 238000012417 linear regression Methods 0.000 description 1
- 238000007726 management method Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 238000013439 planning Methods 0.000 description 1
- 230000000246 remedial effect Effects 0.000 description 1
- 230000035945 sensitivity Effects 0.000 description 1
- 238000010206 sensitivity analysis Methods 0.000 description 1
- 239000004984 smart glass Substances 0.000 description 1
- 238000000528 statistical test Methods 0.000 description 1
- 230000001360 synchronised effect Effects 0.000 description 1
- 230000002123 temporal effect Effects 0.000 description 1
- 230000036962 time dependent Effects 0.000 description 1
- 238000013519 translation Methods 0.000 description 1
- 238000012795 verification Methods 0.000 description 1
- 230000000007 visual effect Effects 0.000 description 1
- 238000005406 washing Methods 0.000 description 1
Landscapes
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The embodiment of the specification provides a package arrival time prediction method, a package arrival time prediction device, a computing device and a storage medium, wherein under the condition that a target package arrives at a current logistics node, first logistics information of the target package and first actual time of the target package arriving at the current logistics node are determined; inputting first logistics information and first actual time into a first time efficiency prediction model corresponding to a first logistics stage to obtain first prediction time for a target package to reach a target object flow node; under the condition that the target package reaches the next logistics node, determining second physical flow information of the target package and second actual time of the target package reaching the next logistics node; inputting the second physical flow information and the second actual time into a second aging prediction model corresponding to the second physical flow stage to obtain a second predicted time for the target package to reach the target physical flow node; and updating the first prediction time according to the second prediction time to obtain the target prediction time for the target package to reach the target object flow node.
Description
Technical Field
The embodiment of the specification relates to the technical field of computers, in particular to a parcel arrival time prediction method.
Background
With the development of electronic commerce, logistics technology is also mature continuously. When a user purchases a commodity through the e-commerce platform, the commodity can be generally dispatched to the user address through express delivery, the express delivery is required to be accurately delivered to the user address, and the time for the express delivery to reach the user address is required to be ensured to be as early as possible so as to ensure the shopping experience of the user on the e-commerce platform. In general, in order to enable a user to know the specific time of delivering the express package at any time, the predicted arrival time of the express package is displayed to the user. However, due to the influence of various factors, the estimated arrival time displayed to the user is inaccurate, and further the user cannot effectively arrange the taking matters, so that the user taking experience is affected. Therefore, an effective solution is needed to solve the above problems.
Disclosure of Invention
In view of this, the present description embodiments provide two methods of parcel arrival time prediction. One or more embodiments of the present specification relate to two types of package arrival time prediction apparatuses, a computing device, a computer-readable storage medium, and a computer program, which solve the technical drawbacks of the prior art.
According to a first aspect of embodiments of the present disclosure, there is provided a parcel arrival time prediction method, including:
Under the condition that the target package reaches the current logistics node, determining first logistics information of the target package and first actual time of the target package reaching the current logistics node;
Inputting the first logistics information and the first actual time into a first time-efficient prediction model corresponding to a first logistics stage to obtain a first prediction time for the target package to reach a target logistics node, wherein the first logistics stage is a logistics stage from the current logistics node to the target logistics node;
Under the condition that the target package reaches a next logistics node, determining second physical flow information of the target package and second actual time of the target package reaching the next logistics node;
Inputting the second physical flow information and the second actual time into a second aging prediction model corresponding to a second physical flow stage to obtain a second predicted time for the target package to reach a target physical flow node, wherein the second physical flow stage is a physical flow stage from the next physical flow node to the target physical flow node;
and updating the first predicted time according to the second predicted time to obtain the target predicted time for the target package to reach the target logistics node.
According to a second aspect of embodiments of the present specification, there is provided a parcel arrival time prediction apparatus, comprising:
a first determining module configured to determine first logistics information of a target package and a first actual time of arrival of the target package at a current logistics node, if it is determined that the target package arrives at the current logistics node;
The first input module is configured to input the first logistics information and the first actual time into a first time efficiency prediction model corresponding to a first logistics stage, and obtain a first predicted time for the target package to reach a target logistics node, wherein the first logistics stage is a logistics stage from the current logistics node to the target logistics node;
A second determining module configured to determine second flow information of the target package and a second actual time of arrival of the target package at a next logistics node, if it is determined that the target package arrives at the next logistics node;
The second input module is configured to input the second stream information and the second actual time into a second aging prediction model corresponding to a second stream stage, so as to obtain a second predicted time for the target package to reach a target stream node, wherein the second stream stage is a stream stage from the next stream node to the target stream node;
And the updating module is configured to update the first predicted time according to the second predicted time to obtain a target predicted time for the target package to reach the target logistics node.
According to a third aspect of embodiments of the present disclosure, there is provided a parcel arrival time prediction method, applied to cloud side equipment, including:
Responding to a logistics inquiry request of a user, and under the condition that a target package arrives at a current logistics node, determining first logistics information of the target package and first actual time of the target package arriving at the current logistics node;
Inputting the first logistics information and the first actual time into a first time-efficient prediction model corresponding to a first logistics stage to obtain a first prediction time for the target package to reach a target logistics node, wherein the first logistics stage is a logistics stage from the current logistics node to the target logistics node;
displaying the first predicted time to the user through a display interface of a client;
Under the condition that the target package reaches a next logistics node, determining second physical flow information of the target package and second actual time of the target package reaching the next logistics node;
Inputting the second physical flow information and the second actual time into a second aging prediction model corresponding to a second physical flow stage to obtain a second predicted time for the target package to reach a target physical flow node, wherein the second physical flow stage is a physical flow stage from the next physical flow node to the target physical flow node;
and updating the first predicted time according to the second predicted time, obtaining the target predicted time for the target package to reach the target logistics node, and displaying the target predicted time to the user through a display interface of the client.
According to a fourth aspect of embodiments of the present disclosure, there is provided a package arrival time prediction apparatus, applied to cloud-side equipment, including:
the first determining module is configured to respond to a logistics inquiry request of a user, and in the case that the target package arrives at a current logistics node, determine first logistics information of the target package and first actual time of the target package arriving at the current logistics node;
The first input module is configured to input the first logistics information and the first actual time into a first time efficiency prediction model corresponding to a first logistics stage, and obtain a first predicted time for the target package to reach a target logistics node, wherein the first logistics stage is a logistics stage from the current logistics node to the target logistics node;
The display module is configured to display the first predicted time to the user through a display interface of a client;
A second determining module configured to determine second flow information of the target package and a second actual time of arrival of the target package at a next logistics node, if it is determined that the target package arrives at the next logistics node;
The second input module is configured to input the second stream information and the second actual time into a second aging prediction model corresponding to a second stream stage, so as to obtain a second predicted time for the target package to reach a target stream node, wherein the second stream stage is a stream stage from the next stream node to the target stream node;
and the updating module is configured to update the first predicted time according to the second predicted time, obtain the target predicted time for the target package to reach the target logistics node, and display the target predicted time to the user through the display interface of the client.
According to a fifth aspect of embodiments of the present specification, there is provided a computing device comprising:
A memory and a processor;
The memory is configured to store computer-executable instructions that, when executed by the processor, implement the steps of the parcel arrival time prediction method described above.
According to a sixth aspect of embodiments of the present description, there is provided a computer-readable storage medium storing computer-executable instructions which, when executed by a processor, implement the steps of the parcel arrival time prediction method described above.
According to a seventh aspect of embodiments of the present specification, there is provided a computer program, wherein the computer program, when executed in a computer, causes the computer to perform the steps of the above-described parcel arrival time prediction method.
One embodiment of the present disclosure provides a method for predicting arrival time of a package, where in a case where it is determined that a target package arrives at a current logistics node, first logistics information of the target package and a first actual time of arrival of the target package at the current logistics node are determined; inputting the first logistics information and the first actual time into a first time-efficient prediction model corresponding to a first logistics stage to obtain a first prediction time for the target package to reach a target logistics node, wherein the first logistics stage is a logistics stage from the current logistics node to the target logistics node; under the condition that the target package reaches a next logistics node, determining second physical flow information of the target package and second actual time of the target package reaching the next logistics node; inputting the second physical flow information and the second actual time into a second aging prediction model corresponding to a second physical flow stage to obtain a second predicted time for the target package to reach a target physical flow node, wherein the second physical flow stage is a physical flow stage from the next physical flow node to the target physical flow node; and updating the first predicted time according to the second predicted time to obtain the target predicted time for the target package to reach the target logistics node.
According to the method, an aging prediction model is set for each logistics stage in the logistics process of the target package, when the target package arrives at the current logistics node, the first prediction time of the target package arriving at the target logistics node is obtained by using the first time efficiency prediction model corresponding to the first logistics stage according to the first actual time and the first logistics information of the target package arriving at the current logistics node, and when the target package arrives at the next logistics node, the second prediction time of the target package arriving at the target logistics node is predicted in real time according to the second actual time and the second logistics information of the target package arriving at the next logistics node, the first prediction time is updated in real time by using the second prediction time, and finally the target prediction time of the target package arriving at the target logistics node is obtained, so that the real-time prediction and updating of the time of the target package arriving at the target logistics node are realized, the accuracy of the predicted arrival time is further ensured, and therefore a user can arrange a picking experience according to the target prediction time is ensured.
Drawings
Fig. 1 is a schematic application scenario of a package arrival time prediction method according to an embodiment of the present disclosure;
FIG. 2 is a flow chart of a method of parcel arrival time prediction provided in one embodiment of the present disclosure;
FIG. 3 is a block diagram of a method for predicting arrival time of a package according to one embodiment of the present disclosure;
FIG. 4 is a process flow diagram of a method for predicting arrival time of a package according to one embodiment of the present disclosure;
FIG. 5 is a schematic diagram of a package arrival time prediction apparatus according to one embodiment of the present disclosure;
FIG. 6 is a flow chart of another method of parcel arrival time prediction provided in one embodiment of the present disclosure;
FIG. 7 is a schematic diagram of another package arrival time prediction apparatus according to one embodiment of the present disclosure;
FIG. 8 is a block diagram of a computing device provided in one embodiment of the present description.
Detailed Description
In the following description, numerous specific details are set forth in order to provide a thorough understanding of the present description. This description may be embodied in many other forms than described herein and similarly generalized by those skilled in the art to whom this disclosure pertains without departing from the spirit of the disclosure and, therefore, this disclosure is not limited by the specific implementations disclosed below.
The terminology used in the one or more embodiments of the specification is for the purpose of describing particular embodiments only and is not intended to be limiting of the one or more embodiments of the specification. As used in this specification, one or more embodiments and the appended claims, the singular forms "a," "an," and "the" are intended to include the plural forms as well, unless the context clearly indicates otherwise. It should also be understood that the term "and/or" as used in one or more embodiments of the present specification refers to and encompasses any or all possible combinations of one or more of the associated listed items.
It should be understood that, although the terms first, second, etc. may be used in one or more embodiments of this specification to describe various information, these information should not be limited by these terms. These terms are only used to distinguish one type of information from another. For example, a first may also be referred to as a second, and similarly, a second may also be referred to as a first, without departing from the scope of one or more embodiments of the present description. The word "if" as used herein may be interpreted as "at … …" or "at … …" or "in response to a determination" depending on the context.
Furthermore, it should be noted that, user information (including, but not limited to, user equipment information, user personal information, etc.) and data (including, but not limited to, data for analysis, stored data, presented data, etc.) according to one or more embodiments of the present disclosure are information and data authorized by a user or sufficiently authorized by each party, and the collection, use, and processing of relevant data is required to comply with relevant laws and regulations and standards of relevant countries and regions, and is provided with corresponding operation entries for the user to select authorization or denial.
In one or more embodiments of the present description, a large model refers to a deep learning model with large scale model parameters, typically including hundreds of millions, billions, trillions, and even more than one billion model parameters. The large Model can be called as a Foundation Model, a training Model is performed by using a large-scale unlabeled corpus, a pre-training Model with more than one hundred million parameters is produced, the Model can adapt to a wide downstream task, and the Model has better generalization capability, such as a large-scale language Model (Large Language Model, LLM), a multi-modal pre-training Model (multi-modal pre-training Model) and the like.
When the large model is actually applied, the pretrained model can be applied to different tasks by only slightly adjusting a small number of samples, the large model can be widely applied to the fields of natural language processing (Natural Language Processing, NLP for short), computer vision and the like, and particularly can be applied to the tasks of the computer vision fields such as vision question and answer (Visual Question Answering, VQA for short), image description (IC for short), image generation and the like, and the tasks of the natural language processing fields such as emotion classification based on texts, text abstract generation, machine translation and the like, and main application scenes of the large model comprise digital assistants, intelligent robots, searching, online education, office software, electronic commerce, intelligent design and the like.
First, terms related to one or more embodiments of the present specification will be explained.
Bayesian Network (Bayesian Network): a probability map model for modeling uncertainty infers probability distributions of unknown variables by representing conditional dependencies between the variables.
Aging pre-estimation (DELIVERY TIME ESTIMAT ion): the time required for a parcel or shipment in a stream to reach another node is predicted.
Logistics link (Logistics Chain): various links from commodity production to end user delivery are involved, including links of purchasing, production, storage, transportation, and distribution. The domestic one-section logistics refers to the signing and receiving from the shipping end of the merchant to the shipping warehouse; 2) Two-section logistics refers to the delivery from a shipping warehouse to a consumer for signing, wherein the two-section logistics can be divided into a warehouse trunk connecting section, a trunk section and a last kilometer section. The trunk handover refers to the start of shipment from the shipping warehouse to the main line segment, the start of the main line segment to the end kilometer allocation node, and the end kilometer segment refers to the end kilometer allocation from the end kilometer distribution.
Conditional probability (Conditional Probability): given the precondition that one event occurs, the probability of another event occurs.
Chi-Square Test: a statistical test method for determining whether there is a correlation between two classification variables determines the significance of the correlation by comparing the degree of deviation between the observed and expected values.
Data sparsity (DATA SPARSITY): refers to situations where there are a large number of missing or zero values in the data set, resulting in uneven data distribution or an insufficient number of samples.
In the present specification, a parcel arrival time prediction method is provided, and the present specification relates to a parcel arrival time prediction apparatus, a computing device, and a computer-readable storage medium, which are described in detail in the following embodiments one by one.
Referring to fig. 1, fig. 1 is a schematic diagram illustrating an application scenario of a package arrival time prediction method according to an embodiment of the present disclosure.
An end side device 102 and a cloud side device 104 are included in fig. 1.
In the e-commerce field, a user can enter an e-commerce platform through the end side device 102 to shop, after the user places an order, a merchant of the e-commerce platform can ship goods placed by the user, and the shipping condition is synchronized to the cloud side device 104. The cloud-side device 104 may obtain the logistics situation of the commodity package, predict the expected delivery time of the commodity package, and display the expected delivery time through the display interface of the end-side device 102.
In implementation, after determining the shipment of the commodity package, the cloud side device 104 may determine an actual shipment time of the commodity package and first logistics information of the commodity package, where the first logistics information may include package information of the commodity package, weather information of the shipment, express company information of the package, whether the shipment time is a holiday or a weekend, and the like, and predict, according to the first logistics information and the actual shipment time, a first predicted time (i.e., a predicted arrival time) for the commodity package to reach the package receiving node by using a first time prediction model corresponding to a first logistics stage formed by the shipment node to the package receiving node. The first predicted time may be displayed to the user through the display interface of the end-side device 102, so that the user may know the predicted delivery time of the package of the commodity in time. And under the condition that the commodity package is determined to reach the next logistics node, determining second physical flow information of the commodity package and actual time of the commodity package reaching the next logistics node, and predicting second predicted time (namely predicted delivery time) of the commodity package reaching the package signing node by using a second aging prediction model corresponding to a second physical flow stage formed by the next logistics node to the package signing node according to the actual time and the second physical flow information. The second predicted time can be sent to the end-side device 102, so that the first predicted time originally displayed by the end-side device 102 is updated to the second predicted time, the predicted delivery time of the package displayed to the user is updated in real time according to the real-time logistics condition of the package, and the accuracy of the predicted delivery time is ensured.
As shown in fig. 1, in the display interface of the end-side device 102, the commodity package is displayed for 12 months and 1 day, where the 12 months and 1 day are the actual delivery time of the commodity package, and the cloud-side device 102 may predict that the estimated delivery time of the commodity package is 12 months and 5 days according to the actual delivery time and the first logistics information, and display the estimated delivery time to the user on the display interface of the end-side device 102. In the case that the commodity package arrives at the transfer station (i.e. the next logistics node), the display interface of the end device 102 displays to the user that the current state of the commodity package is that 12 months 3 days have arrived at the transfer station, then 12 months 3 days are the actual time for the commodity package to arrive at the transfer station, and the cloud device 104 can predict that the estimated delivery time of the commodity package is 12 months 6 days according to the actual time and the second physical flow information, and update the estimated delivery time on the display interface of the end device 102, so as to display the more accurate estimated delivery time to the user.
The end-side device 102 may include a browser, APP (Application), or web Application such as H5 (Hyper Text Markup Language, hypertext markup language version 5) Application, or a light Application (also referred to as applet, a lightweight Application), or cloud Application, etc., and the end-side device may be based on a software development kit (SDK, software Development Kit) of the corresponding service provided by the service end, such as a real-time communication (RTC, real Time Communication) based SDK development acquisition, etc. The end-side device may be deployed in an electronic device, need to run depending on the device or some APP in the device, etc. The electronic device may have a display screen and support information browsing, etc., as may be a personal mobile terminal such as a cell phone, tablet computer, personal computer, etc. Various other types of applications are also commonly deployed in electronic devices, such as human-machine conversation type applications, model training type applications, text processing type applications, web browser applications, shopping type applications, search type applications, instant messaging tools, mailbox clients, social platform software, and the like.
Cloud-side device 104 may be understood as a server that provides various services, including physical servers, cloud servers, such as a server that provides communication services for multiple clients, as well as servers that provide support for models used on clients for background training, as well as servers that process data sent by clients, and so on. It should be noted that, the cloud-side device 104 may be implemented as a distributed server cluster formed by a plurality of servers, or may be implemented as a single server. Cloud-side device 104 may also be a server of a distributed system, or a server that incorporates a blockchain. Cloud-side device 104 may also be a cloud server for cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communications, middleware services, domain name services, security services, content delivery networks (CDN, content Delivery Network), and basic cloud computing services such as big data and artificial intelligence platforms, or an intelligent cloud computing server or intelligent cloud host with artificial intelligence technology.
It should be noted that, the parcel arrival time prediction method provided in the embodiment of the present disclosure may be performed by the cloud side device 104, and in other embodiments of the present disclosure, the age prediction model may be deployed in the end side device 102, so that the end side device 102 may also have a similar function to the cloud side device 104, so as to perform the parcel arrival time prediction method provided in the embodiment of the present disclosure; in other embodiments, the method for predicting arrival time of a package provided in the embodiments of the present disclosure may be performed by the end-side device 102 and the cloud-side device 104 together.
Referring to fig. 2, fig. 2 shows a flowchart of a method for predicting arrival time of a package according to an embodiment of the present disclosure, which specifically includes the following steps.
Step 202: and under the condition that the target package arrives at the current logistics node, determining first logistics information of the target package and first actual time of the target package arriving at the current logistics node.
Specifically, the method for predicting the arrival time of the package provided in the embodiments of the present disclosure may be applied to a server, or may be applied to a platform for predicting the arrival time of the package, where the method may be used for predicting the time when a user signs in the package (i.e., the time when the package arrives at a target object flow node).
The target package can be understood as a commodity package purchased by a user at an e-commerce platform. The current logistics node can be understood as a logistics node that the target package arrives at the current moment. The target package can pass through a plurality of logistics nodes in the logistics process and finally reach the user address, for example, the target package can pass through a delivery node, a transfer station node and a dispatch post node in the logistics process and finally reach the signing node of the user address. The first logistics information of the target package may include package information of the target package and logistics information of the target package to the current logistics node, the package information may be package size information of the target package, express company information of package delivery, etc., and the logistics information may be time information (such as a day of the week, a weekend, etc.) of the target package reaching the current logistics node, weather information of the target package reaching the current logistics node, road condition information of the target package on a route reaching the current logistics node, etc.
Based on this, the package information and the logistics information of the target package and the first actual time of arrival of the target package at the current logistics node can be determined under the condition that the target package arrives at the current logistics node.
For example, the logistics process of the target package includes a delivery node, a logistics node 1, a logistics node 2 and a logistics node 3, and after the target package passes through the three logistics nodes, package signing is completed. Then, after the target package arrives at the logistics node 1 after being shipped by the shipment node, the logistics node 1 is the current logistics node, and at this time, the first logistics information of the target package and the first actual time for the target package to arrive at the logistics node 1 can be determined for 10 months and 19 days.
In practical application, the logistics process (i.e. the whole logistics link) of the target package can be divided into a domestic section, a shipping bin section, a trunk section and a last kilometer section.
Step 204: and inputting the first logistics information and the first actual time into a first time-efficient prediction model corresponding to a first logistics stage to obtain a first prediction time for the target package to reach a target logistics node, wherein the first logistics stage is a logistics stage from the current logistics node to the target logistics node.
Specifically, after the first logistics information and the first actual time are determined, a first time-efficiency prediction model corresponding to a first logistics stage from the current logistics node to the target logistics node can be utilized to obtain a first prediction time for the target package to reach the target logistics node.
The target package is subjected to a plurality of logistics nodes in a logistics process, the target logistics nodes are end logistics nodes of the logistics process of the target package, each logistics node in the plurality of logistics nodes to the target logistics node form a logistics stage, and each logistics stage corresponds to an aging prediction model. For example, when the target package needs to pass through the logistics nodes 1,2 and 3 to complete package signing in the logistics process, the target logistics node is the last logistics node (namely the logistics node 3) in the logistics process, and when the target package arrives at the logistics node 1, the logistics node 1 is the current logistics node, the logistics nodes 1 to 3 are the first logistics stage corresponding to the first time-efficiency prediction model, the first time-efficiency prediction model can be used for predicting the time when the target package arrives at the logistics node 3, and the logistics nodes 2 to 3 are the second logistics stage corresponding to the second time-efficiency prediction model. The first predicted time may be understood as the time required for the target package to reach the target logistics node from the current logistics node, for example, the first predicted time may be 3 days, 5 days, etc.
Based on the method, a first time-efficient prediction model corresponding to a first logistics stage from the current logistics node to the target logistics node can be determined, first logistics information and first actual time are input into the first time-efficient prediction model, and first prediction time for the target package to reach the target logistics node is obtained.
Along the above example, a first time-efficiency prediction model corresponding to the first logistics stage from the logistics node 1 to the logistics node 3 can be determined, and the first logistics information of the target package and the first actual time of the target package reaching the logistics node 1 for 10 months and 19 days are input into the first time-efficiency prediction model, so that the first prediction time required by the target package reaching the logistics node 3 from the logistics node 1 is 3 days, that is, the target logistics node is predicted to be reached after 3 days, or the target logistics node is predicted to be reached after 10 months and 22 days.
In practical application, when the target package arrives at the target object flow node, it can be understood that the target package is signed by the user, and then the package arrival time prediction method provided in the embodiment of the present disclosure can be used for the predicted arrival time or the predicted signing time of the target package.
Further, in order to ensure accuracy of the prediction time, the method may be implemented by segment modeling, that is, an independent prediction model is built for each first sub-logistics stage in the first logistics stages, and the specific implementation manner is as follows:
The first logistics stage comprises a plurality of first sub-logistics stages, and the first time-efficiency prediction model comprises a plurality of first time-efficiency prediction sub-models, wherein each first sub-logistics stage corresponds to one first time-efficiency prediction sub-model;
Correspondingly, inputting the first logistics information and the first actual time into a first time-efficiency prediction model corresponding to a first logistics stage to obtain a first predicted time for the target package to reach a target logistics node, including:
Inputting the first logistics information and the first actual time into a first time-efficient prediction sub-model corresponding to a current first sub-logistics stage, and obtaining the intermediate prediction time for the target package to reach a next logistics node, wherein the current first sub-logistics stage is formed according to the current logistics node and the next logistics node;
and under the condition that the next logistics node is determined to be the target logistics node, determining the intermediate predicted time for the target package to reach the next logistics node as the first predicted time for the target package to reach the target logistics node.
The first sub-logistics stage is understood to be a logistics stage formed by two adjacent logistics nodes in the first logistics stage, for example, the first logistics stage is a logistics stage formed by logistics nodes 1 to 3, and then the first logistics stage comprises two first sub-logistics stages, namely, a first sub-logistics stage from logistics node 1 to logistics node 2 and a first sub-logistics stage from logistics node 2 to logistics node 3. The next logistics node can be understood as the next logistics node of the current logistics node, for example, the target package arrives at the logistics node 1 at the current moment, and then the logistics node 1 is the current logistics node, and the next logistics node 2 of the logistics node 1 is the next logistics node.
Based on the method, a first time-efficient prediction sub-model corresponding to a first sub-logistics stage formed by the current logistics node and the next logistics node can be determined, first logistics information and first actual time are input into the first time-efficient prediction sub-model, the intermediate prediction time for the target package to reach the next logistics node is obtained, and under the condition that the next logistics node is determined to be the target logistics node. And if the next logistics node is the last target logistics node, determining the intermediate predicted time of the target package reaching the next logistics node as the first predicted time of the target package reaching the target logistics node.
In addition, inputting the first logistics information and the first actual time into a first time-efficient prediction sub-model corresponding to a current first sub-logistics stage, and obtaining the intermediate prediction time for the target package to reach a next logistics node further includes:
And under the condition that the next logistics node is not the target logistics node, taking the next logistics node as a current logistics node, taking the intermediate predicted time of the next logistics node as the first actual time of the current logistics node, continuously executing the first logistics information and the first actual time, inputting a first time-efficient predictor model corresponding to the current first sub-logistics stage, and obtaining the intermediate predicted time of the target package reaching the next logistics node.
Specifically, under the condition that the next logistics node is determined not to be the target logistics node, that is, the next logistics node is not the final logistics node, the next logistics node still exists behind the next logistics node, at this time, the next logistics node can be used as the current logistics node, the intermediate prediction time of the next logistics node is used as the first actual time of the current logistics node, the first logistics information and the first actual time are continuously executed, the first time-efficient prediction sub-model corresponding to the current first sub-logistics stage is input, and the step of obtaining the intermediate prediction time of the target package reaching the next logistics node is obtained until the next logistics node is the target logistics node.
Along the above example, the first logistics stage from the logistics node 1 to the logistics node 3 comprises a first sub-logistics stage A from the logistics node 1 to the logistics node 2 and a first sub-logistics stage B from the logistics node 2 to the logistics node 3, wherein the first sub-logistics stage corresponds to a first time-efficiency prediction sub-model A, and the first sub-logistics stage B corresponds to a first time-efficiency prediction sub-model B. The first actual time of the first logistics information and the target package reaching the logistics node 1 (current logistics node) may be input into the first time-efficient prediction sub-model a for 10 months and 19 days, the intermediate prediction time of the target package reaching the logistics node 2 (next logistics node) is obtained for 1 day (i.e. 10 months and 20 days), when the logistics node 2 is not the target logistics node (i.e. the last logistics node 3), then the intermediate prediction time and the first logistics information may be input into the first time-efficient prediction sub-model B corresponding to the first sub-logistics stage B, the intermediate prediction time of the target package reaching the logistics node 3 is obtained for 3 days (i.e. 10 months and 23 days), when the logistics node 3 is the target logistics node, then the intermediate prediction time may be taken as the first prediction time of the target package reaching the target logistics node for 10 months and 23 days, that is, the target package is predicted to reach for 10 months and 23 days.
In practical application, the first time-efficient predictor model may be a bayesian network model, and under the condition that the whole logistics process is divided into a domestic segment, a shipping bin segment, a trunk segment and a last kilometer segment, a bayesian network model corresponding to the domestic segment, a bayesian network model corresponding to the shipping bin segment, a bayesian network model corresponding to the trunk segment and a bayesian network model corresponding to the last kilometer segment may be established. Taking the domestic segment as an example, the bayesian network model corresponding to the domestic segment can represent the relationship between the occurrence time of the domestic segment logistics time and the occurrence time of the last logistics event.
In summary, the logistics process of the package is divided into a plurality of logistics stages, and an independent prediction model is built for each logistics stage, and through sectional modeling, the operation rule and the assessment target in each stage can be better captured, so that the prediction accuracy is improved. And a probability model of the occurrence time of the logistics event is established by using a Bayesian network, the influence of the occurrence time of the last logistics event (namely, the arrival of the package at the last logistics node) on the next logistics event (namely, the arrival of the package at the next logistics node) is considered, and the occurrence time of the next logistics event is predicted through conditional probability, so that the aging estimation is realized.
In practical application, the training step of the first time-efficient predictor model includes:
Determining the actual sample time of the target sample package reaching the current logistics node, the actual label time of the target sample package reaching the next logistics node and the sample logistics information of the target sample package;
Inputting the sample actual time, the tag actual time and the sample logistics information into a first time effect prediction sub-model to obtain the sample prediction time which is output by the first time effect prediction model and is obtained when the target sample package arrives at the next logistics node;
And training the first time-efficient predictive sub-model according to the sample prediction time and the tag actual time until the first time-efficient predictive sub-model meeting the training stop condition is obtained.
The training stopping condition may be understood as that the model loss value reaches a preset loss value threshold value and/or the training frequency reaches a preset frequency threshold value. The sample logistics information can include package information of the target sample package and logistics information of the target sample package to the current logistics node, the package information can be package size information of the target sample package, express company information of package dispatch and the like, and the logistics information can be time information (such as a day of the week, a weekend and the like) of the target sample package reaching the current logistics node, weather information of the target sample package reaching the current logistics node, road condition information of a route of the target sample package reaching the current logistics node and the like.
Specifically, when training the first time-efficient predictor model, the actual sample time of the target sample package reaching the current logistics node, the actual label time of the target sample package reaching the next logistics node, and the sample logistics information of the target sample package between the current logistics node and the next logistics node can be determined, the actual sample time, the actual label time and the sample logistics information are input into the first time-efficient predictor model, the predicted sample time of the target sample package reaching the next logistics node is obtained, and the first time-efficient predictor model is trained according to the predicted sample time and the actual label time until the first time-efficient predictor model meeting the training stop condition is obtained.
For example, for the first time-efficient predictor model corresponding to the logistics stage formed by the logistics node 1 and the logistics node 2, the first time-efficient predictor model may be trained according to the actual sample time when the target sample package arrives at the logistics node 1, the actual label time when the target sample package arrives at the logistics node 2, and the sample logistics information of the target sample package between the logistics node 1 and the logistics node 2.
It can be appreciated that, for the aging prediction model corresponding to each two consecutive logistics nodes, the training process can be used for training, which is not repeated in the embodiment of the present specification.
In practical application, the operation rule and the assessment rule of each logistics stage can be used as independent sectional modeling according to the first-order Markov assumption, and the aging prediction model of the whole logistics link can be obtained through conditional probability combination.
For example, assuming that three consecutive logistics links A, B and C are desired, in order to estimate the total aging from A to C, a model for each link can be built separately, i.e., model (M_A) estimates aging (t_A) from delivery to completion of link A, model (M_B) estimates aging (t_B) from completion of link A to completion of link B, model (M_C) estimates aging (t_C) from completion of link B to completion of link C (final destination), then the total aging (T) for the entire logistics link can be expressed as: [ T=t_A+t_B+t_C ]
Under the first order Markov assumption, (t_B) depends only on the implementation of (t_A) and not on the earlier link, and, (t_C) depends only on (t_B). The conditional probabilities for each segment, such as (P (t_b|t_a)) and (P (t_c|t_b)), can be calculated separately, and represent the probability distribution of the current link completing the aging given that the previous link completed the aging. Further, if the probability distribution of the entire link age needs to be calculated, the following calculation can be performed in order: the probability distribution (P (t_A)) of the time period (t_A) is obtained from the model (M_A). For each possible (t_a), a conditional probability distribution (P (t_b|t_a)) of (t_b) is calculated. For each possible (t_b), a conditional probability distribution (P (t_c|t_b)) of (t_c) is calculated. Then, using the Chain Rule (Chain Rule) and the total probability Rule (Law of Total Probability), we can calculate the probability distribution over the link time-course: [ P (T) = \sum\t_A } \sum\t_B } P (t_A) P (t_B|t_A) P (t_C|t_B).
By calculating the probability distribution of the time effect of the whole link, the possibility of different time effect results can be known, which is very important for decision making processes such as logistics management, inventory control, transportation planning and the like. For example, the risk of delayed delivery may be assessed according to a probability distribution, or the resource allocation may be optimized to reduce the likelihood of delays. The probability distribution provides more comprehensive information including not only the expected average age, but also the variability of the age and other possible outcomes. This helps to improve the accuracy of the aging prediction, especially in the face of uncertainty factors such as traffic conditions, weather changes, operating efficiency fluctuations, etc. By analyzing the probability distribution of the aging of the whole link, key links and factors influencing the aging can be identified, and a basis is provided for improving the operation efficiency and quality of the logistics link. For example, if a conditional probability distribution of a link is found to deviate from expectations, an in-depth investigation and optimization of the link may be required. Probability distributions can be used to perform scene modeling and sensitivity analysis to help understand how the aging behavior of a logistics link may change under different assumptions or conditions. This helps to address various potential project challenges and market changes. The service quality is guaranteed, the customer satisfaction is improved, and the accurate ageing probability distribution can help to set reasonable delivery date commitments and improve the customer satisfaction. By knowing the possible scope of aging, an enterprise can better manage customer expectations and take remedial action if necessary, such as informing in advance of possible delays or providing an alternative solution.
In summary, a probability model of occurrence time of a physical distribution event is established by using a Bayesian network, influence of occurrence time of a previous physical distribution event on a next physical distribution event is considered, and occurrence time of the next physical distribution event is predicted through conditional probability, so that aging estimation is realized.
In implementation, the determining the sample actual time for the sample package to reach the current logistics node, the tag actual time for the sample package to reach the next logistics node, and the sample logistics information of the sample package includes:
Determining training sample data corresponding to each sample package in a plurality of sample packages, wherein the training sample data comprises sample actual time of each sample package reaching the current logistics node, label actual time of the sample package reaching the next logistics node and sample logistics information of the sample package;
And determining a target sample package and target training sample data corresponding to the target sample package from the plurality of sample packages, wherein the target training sample data comprises sample actual time of the target sample package reaching the current logistics node, tag actual time of the target sample package reaching the next logistics node and sample logistics information of the sample package.
Specifically, before training the first time-efficient predictor model, training data may also be sampled to improve the acuity and coverage of the model. The specific implementation mode is as follows:
the determining the sample actual time of the sample package reaching the current logistics node, the tag actual time of the sample package reaching the next logistics node and the sample logistics information of the sample package includes:
Determining training sample data corresponding to each sample package in a plurality of sample packages, wherein the training sample data comprises sample actual time of each sample package reaching the current logistics node, label actual time of the sample package reaching the next logistics node and sample logistics information of the sample package;
And determining a target sample package and target training sample data corresponding to the target sample package from the plurality of sample packages, wherein the target training sample data comprises sample actual time of the target sample package reaching the current logistics node, tag actual time of the target sample package reaching the next logistics node and sample logistics information of the target sample package.
Specifically, a target sample package meeting the condition and target training sample data corresponding to the target sample package can be selected from a plurality of sample packages and training sample data corresponding to each sample package.
In one embodiment of the present disclosure, a target sample package meeting the conditions may be selected by a time weight, and the specific implementation manner is as follows:
Determining a target sample package and target training sample data corresponding to the target sample package from the plurality of sample packages, including:
determining a time weight corresponding to each sample package in the plurality of sample packages according to a preset time condition;
And determining a target sample package and target training sample data corresponding to the target sample package from the plurality of sample packages according to the time weight corresponding to each sample package.
The preset time condition may be understood as a time condition that the sample package needs to satisfy. For example, the preset time condition may be that the actual time of the label of the sample package is between 2015 and 2016, the time weight is a weight value a, the actual time of the label of the sample package is between 2016 and 2017, the time weight is a weight value b, the actual time of the label of the sample package is between 2017 and 2018, the time weight is c, and the like.
Based on the above, the time weight corresponding to each sample package can be determined according to the preset time condition, the sample package with the time weight meeting the preset time weight threshold is determined to be the target sample package, and the target training sample data corresponding to the target sample package is determined.
In practice, the time weight may be a decreasing function, such as a linear decreasing function (i.e., the time weight decreases linearly with increasing time difference), an exponential decreasing function, a piecewise function (i.e., different time weights are given in different time periods, such as the time weight of the last month of the sample package is highest), etc. It will be appreciated that the decreasing function may be selected based on the sensitivity of the actual demand to the historical data and the distribution characteristics of the historical data.
In conclusion, by setting the time weight, the real-time performance of the training data can be ensured, and the model performance is further improved.
Correspondingly, when training the first time-efficient predictor model, the time weight and the model loss value can be combined, and the specific implementation mode is as follows:
Training the first time-efficient predictor model according to the sample predicted time and the tag actual time until a first time-efficient predictor model meeting a training stop condition is obtained, including:
Determining a target time weight corresponding to the target sample package;
calculating a model loss value according to the sample prediction time and the tag actual time;
And training the first time-efficient predictive sub-model according to the model loss value and the target time weight until the first time-efficient predictive sub-model meeting the training stop condition is obtained.
Specifically, after determining a target sample package from a plurality of sample packages, determining a target time weight corresponding to the target sample package, calculating a model loss value according to a sample prediction time obtained by inputting target training sample data corresponding to the target sample package into a first time-efficient prediction sub-model and a tag actual time included in the target training sample data, multiplying the model loss value by the target time weight to obtain a target model loss value, and training the first time-efficient prediction sub-model according to the target model loss value until the first time-efficient prediction sub-model meeting a training stop condition is obtained.
In summary, the model is trained by employing a weighted training method during the model training phase. In calculating the loss function, the loss of each sample is multiplied by its temporal weight. This may result in the model focusing more on data that is more heavily weighted (i.e., more recent days).
In another embodiment of the present disclosure, selecting the target sample package may also be implemented by using a hierarchical sampling model, and the specific implementation manner is as follows:
Determining a target sample package and target training sample data corresponding to the target sample package from the plurality of sample packages, including:
And determining a target sample package and target training sample data corresponding to the target sample package from the plurality of sample packages according to a hierarchical sampling model.
Specifically, taking a case that a plurality of sample packages are taken as 100 sample packages as an example for explanation, then 100 samples comprise 100 groups of sample stream information, a hierarchical sampling model can be utilized to determine sample stream information 1 contained in the 100 groups of sample stream information, the sample stream information 1 can be understood as sample stream information with highest occurrence frequency in the 100 groups of sample stream information, the hierarchical sampling model screens 90 groups of sample stream information comprising the sample stream information 1 from the 100 groups of sample stream information according to the sample stream information 1, screens the 90 groups of sample stream information according to chi-square check and sparsity screening to obtain 70 groups of sample stream information from the 90 groups of sample stream information, screens the 70 groups of sample stream information according to sample stream information 2 contained in the 70 groups of sample stream information to obtain 50 groups of sample stream information wrapping the sample stream information 2, screens the 50 groups of sample stream information according to chi-square check and sparsity, finally obtains 3 groups of sample stream information meeting requirements, determines training data of the corresponding sample packages of the 3 groups of sample stream information as target samples, and trains the training model according to the chi-square check and sparsity, and trains the training model for the target samples corresponding to the target packages.
In summary, during data sampling, according to granularity and data washing attribute of the feature combination (namely, each group of sample stream information), the feature combination layering is adopted to sample and train the data by adopting a model, and according to fine granularity and single-quantity threshold of the feature combination, corresponding training data is selected to improve the acuity and coverage rate of the model.
In addition, after determining the target sample package and the target training sample data corresponding to the target sample package from the plurality of sample packages, the method further includes:
Inputting the sample actual time, the tag actual time and the sample logistics information into a first time-efficient prediction sub-model to obtain a sample predicted time, output by the first time-efficient prediction model, of the arrival of the sample package at the next logistics node
Calculating the time deviation degree between the sample predicted time and the tag actual time;
and determining invalid training sample data from the target training sample data according to the time deviation degree, and deleting the invalid training sample data.
Specifically, according to chi-square test, the time deviation degree between the sample prediction time and the tag actual time can be calculated, and target training sample data with the time deviation degree meeting a preset deviation degree threshold is taken as invalid training sample data and deleted.
In summary, the relevance between the flow direction characteristic combination and the occurrence time period is judged by using chi-square test, and the flow direction characteristic combination which does not affect the aging prediction is screened out by counting the deviation degree between the actual observed value and the theoretical inferred value, so that the performance of the model is further ensured.
In an embodiment of the present disclosure, after determining a target sample package and target training sample data corresponding to the target sample package from the plurality of sample packages, the method further includes:
determining occurrence frequency of sample stream information included in the target training sample data;
And determining invalid sample stream information from sample stream information included in the target training sample data according to the occurrence frequency, and deleting the invalid sample stream information.
Specifically, the sample stream information included in the target training sample data may be subjected to data sparsity screening, and according to the occurrence frequency, the sample stream information with the lowest occurrence frequency is used as invalid sample stream information and deleted.
In addition, the aging prediction can be realized by a machine learning method.
In one embodiment of the present description, the age prediction may be based on a regression model. Specifically, machine learning algorithms such as linear regression, decision tree regression, random forest regression and the like can be adopted, historical data are trained, ageing prediction models for all logistics links are established, and the models can predict the stay time of the package in each link according to package attributes, logistics attributes and other relevant characteristics, so that the ageing of the whole logistics link is calculated.
In another embodiment of the present description, the age prediction may be based on a sequence model. Specifically, a cyclic neural network or a long-short-term memory network and other sequence models can be used for modeling the time sequence data of the logistics event. By learning the time sequence relation among the sequence data, the occurrence time of the future logistics event is predicted, and the time dependence among the front event and the rear event is considered, so that the accuracy of time-dependent prediction is improved.
In another embodiment of the present description, the aging prediction may also be performed based on deep reinforcement learning. Specifically, a deep reinforcement learning method can be adopted, and a logistics link is regarded as a reinforcement learning environment through an aging optimization strategy of interactive learning optimization with the environment, and a transportation mode is adjusted and resource allocation is optimized according to the current state (such as package attribute, logistics attribute and aging of the current link) so as to maximize the accuracy of aging prediction and user satisfaction.
In another embodiment of the present description, the aging prediction may also be performed based on ensemble learning. Specifically, the prediction results of the multiple basic models can be combined by using an integrated learning method, such as random forests, gradient lifting trees, and the like, so as to obtain more accurate aging prediction results. The limitation of a single model can be made up by carrying out weighted fusion or voting decision on the prediction results of different models, and the accuracy and the robustness of prediction are improved.
Step 206: and under the condition that the target package reaches the next logistics node, determining second physical flow information of the target package and second actual time for the target package to reach the next logistics node.
The second stream information may include package information of the target package and stream information of the target package reaching the next stream node.
Specifically, in the case that the target package arrives at the next logistics node, the second physical time of arrival of the target package at the next logistics node and the second physical time of arrival of the target package at the next logistics node may be determined.
Step 208: and inputting the second physical stream information and the second actual time into a second aging prediction model corresponding to a second physical stream stage to obtain a second predicted time for the target package to reach a target physical stream node, wherein the second physical stream stage is a physical stream stage from the next physical stream node to the target physical stream node.
Specifically, a second aging prediction model corresponding to a second flow stage from the next flow node to the target flow node can be determined, second flow information and second actual time are input into the second aging prediction model, and second prediction time for the target package to reach the target flow node is obtained.
Similar to the previous, the second stream stage may also include a plurality of second sub-stream stages, and the second age prediction model may also include a plurality of second age prediction sub-models, one for each second sub-stream stage. The training process of the second aging predictor model is similar to that described above, and will not be repeated here.
Step 210: and updating the first predicted time according to the second predicted time to obtain the target predicted time for the target package to reach the target logistics node.
Specifically, the first prediction time can be updated according to the second prediction time, so that the target prediction time for the target package to reach the target logistics node is obtained, the predicted arrival time of the package is updated according to the latest logistics information and the actual arrival time of the package to reach the logistics node, and the accuracy and the flexibility of the aging prediction are ensured.
In one embodiment of the present specification, the method further comprises:
and responding to target event early warning information, and adjusting the target predicted time for the target package to reach the target logistics node.
The target event may include, for example, an event that can affect logistics aging, such as extreme weather, e-commerce platform promotion, traffic control, and the like.
Based on the target event early warning information, the target prediction time for the target package to reach the target logistics node can be adjusted.
In practical application, the method can be realized by configuring the target event calendar. In particular, referring to fig. 3, fig. 3 shows a schematic diagram of a method for predicting arrival time of a package according to an embodiment of the present disclosure. As shown in fig. 3, the parcel arrival time prediction method can be applied to a parcel arrival time prediction platform, the parcel arrival time prediction platform can build a stable guarantee closed loop, complete configuration of an event calendar in combination with a special event early warning mechanism, realize automatic switching of a holiday model, and provide estimated accuracy by monitoring and early warning of parcel hour achievement rate and timely adjusting an aging estimation algorithm or an expression side. And establishing a corresponding Bayesian network model for each logistics stage through which the package passes, so that multi-classification prediction is realized by using the Bayesian network model, and accuracy optimization is ensured. In the training process of the bayesian network model corresponding to each logistics stage, the package arrival time prediction platform comprises a data layer, the data layer comprises feature data (namely training data) of each dimension, the training data can comprise basic features (namely logistics information and package information of packages), the basic features comprise logistics network features (first kilometer, shipping warehouse, trunk line and last kilometer), operation features (cut-off time, time rate, sea and land transportation and operation rules), promotion features (major promotion level, major promotion stage, major promotion time, major promotion name), time features (year/month/week/day, holiday/weekend) and sales features (buyers, sellers, brands/categories, large and small pieces), worker pieces, real-time single quantity and project dynamics and the like. And, the training data can be preprocessed at the data preprocessing layer, and the preprocessing modes can include Ground Truth (true value) processing, type conversion, missing value filling, outlier processing, coefficient weighting, online log analysis, derivative feature processing, sampling optimization, a rule engine, great promotion alignment and the like. Specifically, at the application layer, training and model optimization can be performed on the bayesian network model corresponding to each logistics stage according to the preprocessed training data. The stability guarantee mechanism of the package arrival time prediction platform can be used for configuring an event calendar and checking timeliness in advance, carrying out early warning in the event, carrying out early warning and operation adjustment aiming at abnormality, and carrying out post-monitoring including parameter entering monitoring and index monitoring.
In conclusion, the configuration of an event calendar is completed by setting up a stable guarantee closed loop and combining a special event early warning mechanism, automatic switching of a holiday model is realized, and an aging estimation algorithm or an expression side is timely adjusted by monitoring and early warning the coverage hour achievement rate, so that the estimation accuracy is improved.
The method for predicting the arrival time of the package provided in the present specification will be further described with reference to fig. 4 by taking an application of the method for predicting the arrival time of the package in an e-commerce platform as an example. Fig. 4 is a flowchart of a process of a method for predicting arrival time of a package according to an embodiment of the present disclosure, which specifically includes the following steps.
Step 402: and inputting the first actual time and first logistics information of the target package reaching the logistics node 1 into an aging prediction model corresponding to a logistics stage formed by the logistics node 1 to the logistics node 2, and obtaining the intermediate prediction time of the target package reaching the logistics node 2.
In the embodiment of the specification, signing is taken as an example after the target package passes through the logistics nodes of the first section, the shipping bin, the trunk line and the last kilometer in China, wherein the first section to the shipping bin in China corresponds to an ageing prediction model, the shipping bin to the trunk line corresponds to an ageing prediction model, and the trunk line to the last kilometer corresponds to an ageing prediction model.
Specifically, the first actual time (i.e., shipping time) of the target package reaching the logistics node 1 (i.e., a period in China) and the first logistics information are input into an aging prediction model corresponding to a logistics stage formed by the logistics node 1 to the logistics node 2 (i.e., a shipping bin), so as to obtain the intermediate prediction time (i.e., the estimated warehousing time) of the target package reaching the logistics node 2.
In the training phase of the aging prediction model, multiple sets of training data can be obtained, wherein each set of training data comprises the actual time of a sample package reaching a section of the country (namely delivery time), the actual time of a label of the sample package reaching the shipping bin (namely estimated warehousing time), and sample logistics information of the sample package, such as the package size of the sample package, the package arrival weather, logistics node information and the like. The method comprises the steps of preprocessing a plurality of groups of training data to obtain target training data obtained after preprocessing, training an aging prediction model according to the target training data, specifically, inputting sample actual time, label actual time and sample logistics information contained in the target training data into the aging prediction model to obtain sample prediction time of a sample package output by the aging prediction model to reach a shipping bin, calculating a model loss value according to the sample prediction time and the label actual time, and training the aging prediction model according to the model loss value.
For example, in the training stage of the aging prediction model, training data of 100 sample packages in a logistics stage formed from a domestic section to a shipping bin (that is, 100 sets of training data are acquired), and the training data of each sample package includes the actual time of the sample package reaching the domestic section, the actual time of the sample package reaching a label of the shipping bin, and sample logistics information of the sample package. It will be appreciated that the sample stream information is a combination of features. For 100 groups of sample stream information, the time weight of each group of sample stream information in the 100 groups of sample stream information can be determined according to the actual time of the label of the sample package reaching the shipping bin, which corresponds to the 100 groups of sample stream information, for example, the actual time of the label of the sample package 1 reaching the shipping bin is 2015, and because the actual time of the label is far from the current time, the time weight of the sample stream information of the sample package 1 is lower. That is, the time weight of each set of sample stream information may be determined according to a preset time condition, for example, the preset time condition may be that the actual time of the label of the sample package is between 2015 and 2016, the time weight is a weight value a, the actual time of the label of the sample package is between 2016 and 2017, the time weight is a weight value b, the actual time of the label of the sample package is between 2017 and 2018, and the time weight is c … ….
Correspondingly, during model training, for example, the sample actual time, the label actual time and the sample logistics information of the sample package 1 can be input into an aging prediction model, the sample prediction time of the sample package reaching the shipping bin output by the aging prediction model is obtained, a model loss value is calculated according to the sample prediction time and the label actual time, and then the aging prediction model is trained according to the time weight of the sample logistics information of the sample package 1 and the model loss value.
In another embodiment of the present disclosure, for 100 sets of sample stream information, the 100 sets of sample stream information may be hierarchically screened according to a hierarchical sampling model. As shown in fig. 4, a hierarchical sampling model may be used to determine sample stream information 1 included in 100 sets of sample stream information (F1, F2, F3 … … Fn), where F1 may be understood as a set of sample stream information, sample stream information 1 may be understood as sample stream information with highest occurrence frequency in 100 sets of sample stream information, the hierarchical sampling model may screen 90 sets of sample stream information including the sample stream information 1 from the 100 sets of sample stream information according to the sample stream information 1, screen the 90 sets of sample stream information according to chi-square check and sparsity, screen 70 sets of sample stream information (F1, F2, F3 … … Fn-1) from the 90 sets of sample stream information, screen 50 sets of sample stream information wrapping the sample stream information 2 according to sample stream information 2 included in the 70 sets of sample stream information, screen the 50 sets of sample stream information according to chi-square check and sparsity, and continue to screen the 50 sets of sample stream information, and finally obtain 3 sets of sample stream information (F1, F2 and F3) meeting requirements, and training the training model according to the corresponding to the time-domain sample stream information.
When the method is implemented, during screening according to chi-square verification, the actual sample time, the actual label time and each group of sample logistics information can be input into an aging prediction model, the sample prediction time of the sample package output by the aging prediction model reaching the shipping bin is obtained, the time offset degree is calculated according to the sample prediction time and the actual label time, and the sample logistics information corresponding to the sample prediction time with the larger time offset degree is used as invalid feature combination to be deleted.
Specifically, for the aging prediction model corresponding to each logistics stage, the conditional probability distribution of the aging prediction model corresponding to the next logistics stage relative to the aging prediction model corresponding to the last logistics stage can be calculated, so that the conditional probability distribution of the whole logistics process is calculated, and the target aging prediction model of the whole logistics process is obtained.
Step 404: and inputting the intermediate prediction time of the target package reaching the logistics node 2 and the first logistics information into an aging prediction model corresponding to a logistics stage formed by the logistics node 2 to the logistics node 3 (namely a trunk line), and obtaining the intermediate prediction time (namely the estimated ex-warehouse time) of the target package reaching the logistics node 3.
Step 406: and inputting the intermediate predicted time of the target package reaching the logistics node 3 and the first logistics information into an aging prediction model corresponding to a logistics stage formed by the logistics node 3 to the logistics node 4 (namely, last kilometer), and obtaining the intermediate predicted time of the target package reaching the logistics node 4 (namely, the estimated trunk line reaching time).
Step 408: and inputting the intermediate predicted time of the target package reaching the logistics node 4 and the first logistics information into an aging prediction model corresponding to a logistics stage formed by the logistics node 4 to the signing node, and obtaining the first predicted time of the target package reaching the signing node (namely, the predicted signing time).
It will be appreciated that steps 402 to 408 described above are a process of predicting a predicted time to pick up a target package when the target package is shipped from a home country. After the target package actually arrives at the shipping bin, the pre-estimated signing process for the target package may be re-predicted according to the actual time and logistics information of the target package actually arriving at the shipping bin, as described above in steps 404 to 408.
Corresponding to the method embodiment, the present disclosure further provides an embodiment of a package arrival time prediction apparatus, and fig. 5 shows a schematic structural diagram of a package arrival time prediction apparatus provided in one embodiment of the present disclosure. As shown in fig. 5, the apparatus includes:
A first determining module 502 configured to determine, in a case where it is determined that a target package arrives at a current logistics node, first logistics information of the target package and a first actual time of arrival of the target package at the current logistics node;
the first input module 504 is configured to input the first logistics information and the first actual time into a first time-efficiency prediction model corresponding to a first logistics stage, and obtain a first predicted time for the target package to reach a target logistics node, where the first logistics stage is a logistics stage from the current logistics node to the target logistics node;
A second determining module 506 configured to determine second flow information of the target package and a second actual time of arrival of the target package at a next logistics node, if it is determined that the target package arrives at the next logistics node;
A second input module 508, configured to input the second stream information and the second actual time into a second aging prediction model corresponding to a second stream stage, to obtain a second predicted time for the target package to reach a target stream node, where the second stream stage is a stream stage from the next stream node to the target stream node;
An updating module 510 configured to update the first predicted time based on the second predicted time to obtain a target predicted time for the target package to reach the target logistics node.
In an alternative embodiment, the first logistics stage includes a plurality of first sub-logistics stages, and the first time-efficient prediction model includes a plurality of first time-efficient prediction sub-models, where each first sub-logistics stage corresponds to one first time-efficient prediction sub-model;
accordingly, the first input module 504 is further configured to:
Inputting the first logistics information and the first actual time into a first time-efficient prediction sub-model corresponding to a current first sub-logistics stage, and obtaining the intermediate prediction time for the target package to reach a next logistics node, wherein the current first sub-logistics stage is formed according to the current logistics node and the next logistics node;
and under the condition that the next logistics node is determined to be the target logistics node, determining the intermediate predicted time for the target package to reach the next logistics node as the first predicted time for the target package to reach the target logistics node.
In an alternative embodiment, the first input module 504 is further configured to:
Inputting the first logistics information and the first actual time into a first time-efficient predictor model corresponding to a current first sub-logistics stage, and obtaining the intermediate predicted time for the target package to reach a next logistics node, wherein the method further comprises the following steps:
And under the condition that the next logistics node is not the target logistics node, taking the next logistics node as a current logistics node, taking the intermediate predicted time of the next logistics node as the first actual time of the current logistics node, continuously executing the first logistics information and the first actual time, inputting a first time-efficient predictor model corresponding to the current first sub-logistics stage, and obtaining the intermediate predicted time of the target package reaching the next logistics node.
In an optional embodiment, the target package is subjected to a plurality of logistics nodes in a logistics process, the target logistics nodes are end logistics nodes of the logistics process of the target package, and each logistics node to the target logistics node in the plurality of logistics nodes form a logistics stage, and each logistics stage corresponds to an aging prediction model.
In an alternative embodiment, the apparatus further comprises a training module configured to:
Determining the actual sample time of the target sample package reaching the current logistics node, the actual label time of the target sample package reaching the next logistics node and the sample logistics information of the target sample package;
Inputting the sample actual time, the tag actual time and the sample logistics information into a first time-efficient prediction sub-model to obtain the sample prediction time which is output by the first time-efficient prediction sub-model and is obtained when the target sample package arrives at the next logistics node;
And training the first time-efficient predictive sub-model according to the sample prediction time and the tag actual time until the first time-efficient predictive sub-model meeting the training stop condition is obtained.
In an alternative embodiment, the training module is further configured to:
Determining training sample data corresponding to each sample package in a plurality of sample packages, wherein the training sample data comprises sample actual time of each sample package reaching the current logistics node, label actual time of the sample package reaching the next logistics node and sample logistics information of the sample package;
And determining a target sample package and target training sample data corresponding to the target sample package from the plurality of sample packages, wherein the target training sample data comprises sample actual time of the target sample package reaching the current logistics node, tag actual time of the target sample package reaching the next logistics node and sample logistics information of the target sample package.
In an alternative embodiment, the training module is further configured to:
determining a time weight corresponding to each sample package in the plurality of sample packages according to a preset time condition;
And determining a target sample package and target training sample data corresponding to the target sample package from the plurality of sample packages according to the time weight corresponding to each sample package.
In an alternative embodiment, the training module is further configured to:
Determining a target time weight corresponding to the target sample package;
calculating a model loss value according to the sample prediction time and the tag actual time;
And training the first time-efficient predictive sub-model according to the model loss value and the target time weight until the first time-efficient predictive sub-model meeting the training stop condition is obtained.
In an alternative embodiment, the training module is further configured to:
And determining a target sample package and target training sample data corresponding to the target sample package from the plurality of sample packages according to a hierarchical sampling model.
In an alternative embodiment, the training module is further configured to:
Inputting the sample actual time, the tag actual time and the sample logistics information into a first time-efficient prediction sub-model to obtain a sample predicted time, output by the first time-efficient prediction model, of the arrival of the sample package at the next logistics node
Calculating the time deviation degree between the sample predicted time and the tag actual time;
and determining invalid training sample data from the target training sample data according to the time deviation degree, and deleting the invalid training sample data.
In an alternative embodiment, the training module is further configured to:
determining occurrence frequency of sample stream information included in the target training sample data;
And determining invalid sample stream information from sample stream information included in the target training sample data according to the occurrence frequency, and deleting the invalid sample stream information.
In an alternative embodiment, the apparatus further comprises an adjustment module configured to:
and responding to target event early warning information, and adjusting the target predicted time for the target package to reach the target logistics node.
In summary, the device sets an aging prediction model for each logistics stage in the logistics process of the target package, when the target package arrives at the current logistics node, according to the first actual time and the first logistics information of the target package arriving at the current logistics node, the first time prediction model corresponding to the first logistics stage is utilized to obtain the first prediction time of the target package arriving at the target logistics node, when the target package arrives at the next logistics node, according to the second actual time and the second logistics information of the target package arriving at the next logistics node, the second prediction time of the target package arriving at the target logistics node is predicted in real time, the first prediction time is updated in real time by utilizing the second prediction time, and finally the target prediction time of the target package arriving at the target logistics node is obtained, so that the real-time prediction and updating of the time of the target package arriving at the target logistics node are realized, the accuracy of the predicted arrival time is further ensured, and therefore, a user can arrange a picking-up object according to the target prediction time, and a picking-up experience of the user is ensured.
The above is an exemplary scheme of a parcel arrival time prediction apparatus of the present embodiment. It should be noted that, the technical solution of the parcel arrival time prediction apparatus and the technical solution of the parcel arrival time prediction method belong to the same concept, and details of the technical solution of the parcel arrival time prediction apparatus, which are not described in detail, can be referred to the description of the technical solution of the parcel arrival time prediction method.
Referring to fig. 6, fig. 6 shows a flowchart of another method for predicting arrival time of a package according to an embodiment of the present disclosure, which specifically includes the following steps.
Step 602: responding to a logistics inquiry request of a user, and under the condition that a target package arrives at a current logistics node, determining first logistics information of the target package and first actual time of the target package arriving at the current logistics node;
step 604: inputting the first logistics information and the first actual time into a first time-efficient prediction model corresponding to a first logistics stage to obtain a first prediction time for the target package to reach a target logistics node, wherein the first logistics stage is a logistics stage from the current logistics node to the target logistics node;
Step 606: displaying the first predicted time to the user through a display interface of a client;
Step 608: under the condition that the target package reaches a next logistics node, determining second physical flow information of the target package and second actual time of the target package reaching the next logistics node;
Step 610: inputting the second physical flow information and the second actual time into a second aging prediction model corresponding to a second physical flow stage to obtain a second predicted time for the target package to reach a target physical flow node, wherein the second physical flow stage is a physical flow stage from the next physical flow node to the target physical flow node;
Step 612: and updating the first predicted time according to the second predicted time, obtaining the target predicted time for the target package to reach the target logistics node, and displaying the target predicted time to the user through a display interface of the client.
In summary, according to the method, by setting the time-efficiency prediction model for each logistics stage in the logistics process of the target package, when the target package arrives at the current logistics node, according to the first actual time and the first logistics information of the target package arriving at the current logistics node, the first time-efficiency prediction model corresponding to the first logistics stage is utilized to obtain the first prediction time of the target package arriving at the target logistics node, and when the target package arrives at the next logistics node, according to the second actual time and the second logistics information of the target package arriving at the next logistics node, the second prediction time of the target package arriving at the target logistics node is predicted in real time, the first prediction time is updated in real time by utilizing the second prediction time, and finally the target prediction time of the target package arriving at the target logistics node is obtained, so that the real-time prediction and updating of the time of the target package arriving at the target logistics node are realized, and the accuracy of the predicted arrival time is further ensured, so that a user can arrange a picking-up item according to the target prediction time, and the picking-up experience of the user is ensured.
The above is an exemplary scheme of a parcel arrival time prediction method of the present embodiment. It should be noted that, the technical solution of the parcel arrival time prediction method and the technical solution of the parcel arrival time prediction method belong to the same concept, and details of the technical solution of the parcel arrival time prediction method which are not described in detail can be referred to the description of the technical solution of the parcel arrival time prediction method.
Corresponding to the above method embodiment, the present disclosure further provides an embodiment of a package arrival time prediction apparatus, and fig. 7 shows a schematic structural diagram of another package arrival time prediction apparatus provided in one embodiment of the present disclosure. As shown in fig. 7, the apparatus includes:
a first determining module 702 configured to determine, in response to a logistic query request of a user, first logistic information of a target package and a first actual time of the target package reaching a current logistic node in a case where the target package is determined to reach the current logistic node;
the first input module 704 is configured to input the first logistics information and the first actual time into a first time-efficiency prediction model corresponding to a first logistics stage, and obtain a first predicted time for the target package to reach a target logistics node, where the first logistics stage is a logistics stage from the current logistics node to the target logistics node;
a presentation module 706 configured to present the first predicted time to the user via a presentation interface of a client;
A second determining module 708 configured to determine second flow information of the target package and a second actual time of arrival of the target package at a next logistics node, if it is determined that the target package arrives at the next logistics node;
A second input module 710, configured to input the second stream information and the second actual time into a second aging prediction model corresponding to a second stream stage, to obtain a second predicted time for the target package to reach a target stream node, where the second stream stage is a stream stage from the next stream node to the target stream node;
and the updating module 712 is configured to update the first predicted time according to the second predicted time, obtain a target predicted time for the target package to reach the target logistics node, and display the target predicted time to the user through the display interface of the client.
In summary, the device sets an aging prediction model for each logistics stage in the logistics process of the target package, when the target package arrives at the current logistics node, according to the first actual time and the first logistics information of the target package arriving at the current logistics node, the first time prediction model corresponding to the first logistics stage is utilized to obtain the first prediction time of the target package arriving at the target logistics node, when the target package arrives at the next logistics node, according to the second actual time and the second logistics information of the target package arriving at the next logistics node, the second prediction time of the target package arriving at the target logistics node is predicted in real time, the first prediction time is updated in real time by utilizing the second prediction time, and finally the target prediction time of the target package arriving at the target logistics node is obtained, so that the real-time prediction and updating of the time of the target package arriving at the target logistics node are realized, the accuracy of the predicted arrival time is further ensured, and therefore, a user can arrange a picking-up object according to the target prediction time, and a picking-up experience of the user is ensured.
The above is an exemplary scheme of a parcel arrival time prediction apparatus of the present embodiment. It should be noted that, the technical solution of the parcel arrival time prediction apparatus and the technical solution of the parcel arrival time prediction method belong to the same concept, and details of the technical solution of the parcel arrival time prediction apparatus, which are not described in detail, can be referred to the description of the technical solution of the parcel arrival time prediction method.
Fig. 8 illustrates a block diagram of a computing device 800 provided in accordance with one embodiment of the present description. The components of computing device 800 include, but are not limited to, memory 810 and processor 820. Processor 820 is coupled to memory 810 through bus 830 and database 850 is used to hold data.
Computing device 800 also includes access device 840, access device 840 enabling computing device 800 to communicate via one or more networks 860. Examples of such networks include public switched telephone networks (PSTN, public Switched Telephone Network), local area networks (LAN, local Area Network), wide area networks (WAN, wide Area Network), personal area networks (PAN, personal Area Network), or combinations of communication networks such as the internet. The access device 840 may include one or more of any type of network interface, wired or wireless, such as a network interface card (NIC, network interface controller), such as an IEEE802.11 wireless local area network (WLAN, wireless Local Area Network) wireless interface, a worldwide interoperability for microwave access (Wi-MAX, worldwide Interoperability for Microwave Access) interface, an ethernet interface, a universal serial bus (USB, universal Serial Bus) interface, a cellular network interface, a bluetooth interface, a Near Field Communication (NFC) interface, and so forth.
In one embodiment of the application, the above-described components of computing device 800, as well as other components not shown in FIG. 8, may also be connected to each other, such as by a bus. It should be understood that the block diagram of the computing device illustrated in FIG. 8 is for exemplary purposes only and is not intended to limit the scope of the present application. Those skilled in the art may add or replace other components as desired.
Computing device 800 may be any type of stationary or mobile computing device including a mobile computer or mobile computing device (e.g., tablet, personal digital assistant, laptop, notebook, netbook, etc.), mobile phone (e.g., smart phone), wearable computing device (e.g., smart watch, smart glasses, etc.), or other type of mobile device, or a stationary computing device such as a desktop computer or personal computer (PC, personal Computer). Computing device 800 may also be a mobile or stationary server.
Wherein the processor 820 is configured to execute computer-executable instructions that, when executed by the processor, perform the steps of the package arrival time prediction method described above.
The foregoing is a schematic illustration of a computing device of this embodiment. It should be noted that, the technical solution of the computing device and the technical solution of the parcel arrival time prediction method belong to the same concept, and details of the technical solution of the computing device, which are not described in detail, can be referred to the description of the technical solution of the parcel arrival time prediction method.
An embodiment of the present disclosure also provides a computer-readable storage medium storing computer-executable instructions that, when executed by a processor, implement the steps of the parcel arrival time prediction method described above.
The above is an exemplary version of a computer-readable storage medium of the present embodiment. It should be noted that, the technical solution of the storage medium and the technical solution of the parcel arrival time prediction method belong to the same concept, and details of the technical solution of the storage medium which are not described in detail can be referred to the description of the technical solution of the parcel arrival time prediction method.
An embodiment of the present disclosure further provides a computer program, where the computer program, when executed in a computer, causes the computer to perform the steps of the parcel arrival time prediction method described above.
The above is an exemplary version of a computer program of the present embodiment. It should be noted that, the technical solution of the computer program and the technical solution of the parcel arrival time prediction method belong to the same concept, and details of the technical solution of the computer program, which are not described in detail, can be referred to the description of the technical solution of the parcel arrival time prediction method.
The foregoing describes specific embodiments of the present disclosure. Other embodiments are within the scope of the following claims. In some cases, the actions or steps recited in the claims can be performed in a different order than in the embodiments and still achieve desirable results. In addition, the processes depicted in the accompanying figures do not necessarily require the particular order shown, or sequential order, to achieve desirable results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.
The computer instructions include computer program code that may be in source code form, object code form, executable file or some intermediate form, etc. The computer readable medium may include: any entity or device capable of carrying the computer program code, a recording medium, a U disk, a removable hard disk, a magnetic disk, an optical disk, a computer Memory, a Read-Only Memory (ROM), a random access Memory (RAM, random Access Memory), an electrical carrier signal, a telecommunications signal, a software distribution medium, and so forth. It should be noted that the content of the computer readable medium can be increased or decreased appropriately according to the requirements of the patent practice, for example, in some areas, according to the patent practice, the computer readable medium does not include an electric carrier signal and a telecommunication signal.
It should be noted that, for simplicity of description, the foregoing method embodiments are all expressed as a series of combinations of actions, but it should be understood by those skilled in the art that the embodiments are not limited by the order of actions described, as some steps may be performed in other order or simultaneously according to the embodiments of the present disclosure. Further, those skilled in the art will appreciate that the embodiments described in the specification are all preferred embodiments, and that the acts and modules referred to are not necessarily all required for the embodiments described in the specification.
In the foregoing embodiments, the descriptions of the embodiments are emphasized, and for parts of one embodiment that are not described in detail, reference may be made to the related descriptions of other embodiments.
The preferred embodiments of the present specification disclosed above are merely used to help clarify the present specification. Alternative embodiments are not intended to be exhaustive or to limit the invention to the precise form disclosed. Obviously, many modifications and variations are possible in light of the teaching of the embodiments. The embodiments were chosen and described in order to best explain the principles of the embodiments and the practical application, to thereby enable others skilled in the art to best understand and utilize the invention. This specification is to be limited only by the claims and the full scope and equivalents thereof.
Claims (10)
1. A method of parcel arrival time prediction, comprising:
Under the condition that the target package reaches the current logistics node, determining first logistics information of the target package and first actual time of the target package reaching the current logistics node;
Inputting the first logistics information and the first actual time into a first time-efficient prediction model corresponding to a first logistics stage to obtain a first prediction time for the target package to reach a target logistics node, wherein the first logistics stage is a logistics stage from the current logistics node to the target logistics node;
Under the condition that the target package reaches a next logistics node, determining second physical flow information of the target package and second actual time of the target package reaching the next logistics node;
Inputting the second physical flow information and the second actual time into a second aging prediction model corresponding to a second physical flow stage to obtain a second predicted time for the target package to reach a target physical flow node, wherein the second physical flow stage is a physical flow stage from the next physical flow node to the target physical flow node;
and updating the first predicted time according to the second predicted time to obtain the target predicted time for the target package to reach the target logistics node.
2. The parcel arrival time prediction method of claim 1, the first logistics stage comprising a plurality of first sub-logistics stages, the first time-efficient prediction model comprising a plurality of first time-efficient prediction sub-models, wherein each first sub-logistics stage corresponds to a first time-efficient prediction sub-model;
Correspondingly, inputting the first logistics information and the first actual time into a first time-efficiency prediction model corresponding to a first logistics stage to obtain a first predicted time for the target package to reach a target logistics node, including:
Inputting the first logistics information and the first actual time into a first time-efficient prediction sub-model corresponding to a current first sub-logistics stage, and obtaining the intermediate prediction time for the target package to reach a next logistics node, wherein the current first sub-logistics stage is formed according to the current logistics node and the next logistics node;
and under the condition that the next logistics node is determined to be the target logistics node, determining the intermediate predicted time for the target package to reach the next logistics node as the first predicted time for the target package to reach the target logistics node.
3. The parcel arrival time prediction method according to claim 2, wherein the inputting the first logistics information and the first actual time into a first time-efficient prediction sub-model corresponding to a current first sub-logistics stage, after obtaining the intermediate predicted time for the target parcel to arrive at a next logistics node, further comprises:
And under the condition that the next logistics node is not the target logistics node, taking the next logistics node as a current logistics node, taking the intermediate predicted time of the next logistics node as the first actual time of the current logistics node, continuously executing the first logistics information and the first actual time, inputting a first time-efficient predictor model corresponding to the current first sub-logistics stage, and obtaining the intermediate predicted time of the target package reaching the next logistics node.
4. The parcel arrival time prediction method of claim 2, the training step of the first time-efficient predictor model comprising:
Determining the actual sample time of the target sample package reaching the current logistics node, the actual label time of the target sample package reaching the next logistics node and the sample logistics information of the target sample package;
Inputting the sample actual time, the tag actual time and the sample logistics information into a first time-efficient prediction sub-model to obtain the sample prediction time which is output by the first time-efficient prediction sub-model and is obtained when the target sample package arrives at the next logistics node;
And training the first time-efficient predictive sub-model according to the sample prediction time and the tag actual time until the first time-efficient predictive sub-model meeting the training stop condition is obtained.
5. The parcel arrival time prediction method of claim 4, the determining the sample actual time for the target sample parcel to arrive at the current logistics node, the tag actual time for the target sample parcel to arrive at the next logistics node, and the sample logistics information for the target sample parcel, comprising:
Determining training sample data corresponding to each sample package in a plurality of sample packages, wherein the training sample data comprises sample actual time of each sample package reaching the current logistics node, label actual time of the sample package reaching the next logistics node and sample logistics information of the sample package;
And determining a target sample package and target training sample data corresponding to the target sample package from the plurality of sample packages, wherein the target training sample data comprises sample actual time of the target sample package reaching the current logistics node, tag actual time of the target sample package reaching the next logistics node and sample logistics information of the target sample package.
6. The parcel arrival time prediction method according to claim 5, wherein determining a target specimen parcel and target training specimen data corresponding to the target specimen parcel from the plurality of specimen parcels comprises:
determining a time weight corresponding to each sample package in the plurality of sample packages according to a preset time condition;
And determining a target sample package and target training sample data corresponding to the target sample package from the plurality of sample packages according to the time weight corresponding to each sample package.
7. A parcel arrival time prediction method is applied to cloud side equipment and comprises the following steps:
Responding to a logistics inquiry request of a user, and under the condition that a target package arrives at a current logistics node, determining first logistics information of the target package and first actual time of the target package arriving at the current logistics node;
Inputting the first logistics information and the first actual time into a first time-efficient prediction model corresponding to a first logistics stage to obtain a first prediction time for the target package to reach a target logistics node, wherein the first logistics stage is a logistics stage from the current logistics node to the target logistics node;
displaying the first predicted time to the user through a display interface of a client;
Under the condition that the target package reaches a next logistics node, determining second physical flow information of the target package and second actual time of the target package reaching the next logistics node;
Inputting the second physical flow information and the second actual time into a second aging prediction model corresponding to a second physical flow stage to obtain a second predicted time for the target package to reach a target physical flow node, wherein the second physical flow stage is a physical flow stage from the next physical flow node to the target physical flow node;
and updating the first predicted time according to the second predicted time, obtaining the target predicted time for the target package to reach the target logistics node, and displaying the target predicted time to the user through a display interface of the client.
8. A package arrival time prediction apparatus, comprising:
a first determining module configured to determine first logistics information of a target package and a first actual time of arrival of the target package at a current logistics node, if it is determined that the target package arrives at the current logistics node;
The first input module is configured to input the first logistics information and the first actual time into a first time efficiency prediction model corresponding to a first logistics stage, and obtain a first predicted time for the target package to reach a target logistics node, wherein the first logistics stage is a logistics stage from the current logistics node to the target logistics node;
A second determining module configured to determine second flow information of the target package and a second actual time of arrival of the target package at a next logistics node, if it is determined that the target package arrives at the next logistics node;
The second input module is configured to input the second stream information and the second actual time into a second aging prediction model corresponding to a second stream stage, so as to obtain a second predicted time for the target package to reach a target stream node, wherein the second stream stage is a stream stage from the next stream node to the target stream node;
And the updating module is configured to update the first predicted time according to the second predicted time to obtain a target predicted time for the target package to reach the target logistics node.
9. A computing device, comprising:
A memory and a processor;
The memory is configured to store computer-executable instructions that, when executed by the processor, implement the steps of the parcel arrival time prediction method of any of claims 1 to 7.
10. A computer readable storage medium storing computer executable instructions which when executed by a processor implement the steps of the parcel arrival time prediction method of any one of claims 1 to 7.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202410033829.5A CN117993545A (en) | 2024-01-09 | 2024-01-09 | Method, device, computing equipment and storage medium for predicting arrival time of package |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202410033829.5A CN117993545A (en) | 2024-01-09 | 2024-01-09 | Method, device, computing equipment and storage medium for predicting arrival time of package |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN117993545A true CN117993545A (en) | 2024-05-07 |
Family
ID=90894422
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202410033829.5A Pending CN117993545A (en) | 2024-01-09 | 2024-01-09 | Method, device, computing equipment and storage medium for predicting arrival time of package |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN117993545A (en) |
-
2024
- 2024-01-09 CN CN202410033829.5A patent/CN117993545A/en active Pending
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108320171A (en) | Hot item prediction technique, system and device | |
| US20220398610A1 (en) | Method of forecasting store demand based on artificial intelligence and system therefor | |
| KR20180117800A (en) | Method for providing asset portfolio recommendating service | |
| US11481674B2 (en) | Digital content communications system for account management and predictive analytics | |
| CN113763093A (en) | User portrait-based item recommendation method and device | |
| KR20220076012A (en) | System for providing artificial intelligence based price decision service for business-to-business market | |
| CN111667024B (en) | Content pushing method, device, computer equipment and storage medium | |
| CN112150214A (en) | Data prediction method and device and computer readable storage medium | |
| CN111415236B (en) | Block chain big data financial user tag estimation method and system based on cloud computing | |
| Wambura et al. | Robust anomaly detection in feature-evolving time series | |
| US20210004716A1 (en) | Real-time global ai platform | |
| CN117993545A (en) | Method, device, computing equipment and storage medium for predicting arrival time of package | |
| CN107424026A (en) | Businessman's reputation evaluation method and device | |
| CN115983759A (en) | Method and device for predicting demand quantity of goods, electronic equipment and storage medium | |
| CN114677183A (en) | New product sales prediction method and device, computer equipment and storage medium | |
| Wang | Exploration on the Construction of Cross‐Border E‐Commerce Logistics System Using Deep Learning | |
| Rizzo et al. | Uncertainty-bounded reinforcement learning for revenue optimization in air cargo: a prescriptive learning approach | |
| de Morais et al. | A data‐driven optimization approach to plan smart waste collection operations | |
| CN113159877A (en) | Data processing method, device, system and computer readable storage medium | |
| Keerthi et al. | Predictive analysis for modeling travel decision making | |
| Sison et al. | Survey of Machine Learning and Deep Learning Techniques for Travel Demand Forecasting | |
| US20230108482A1 (en) | System and method for large-scale accelerated parallel predictive modelling and control | |
| Gao et al. | Supply Chain Management and Backorder Products Prediction Utilizing Neural Network and Naive Bayes Machine Learning Techniques in Big Data Area: A Real-life Case Study | |
| CN117114352B (en) | Vehicle maintenance method, device, computer equipment and storage medium | |
| CN117993824A (en) | Delivery amount prediction method, device, equipment, medium and program product |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination |