CN110414731A - Method, apparatus, computer readable storage medium and the electronic equipment of Order splitting - Google Patents
Method, apparatus, computer readable storage medium and the electronic equipment of Order splitting Download PDFInfo
- Publication number
- CN110414731A CN110414731A CN201910667271.5A CN201910667271A CN110414731A CN 110414731 A CN110414731 A CN 110414731A CN 201910667271 A CN201910667271 A CN 201910667271A CN 110414731 A CN110414731 A CN 110414731A
- Authority
- CN
- China
- Prior art keywords
- order
- jockey
- feature
- prediction model
- path planning
- 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.)
- Granted
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
- G06Q10/063—Operations research, analysis or management
- G06Q10/0631—Resource planning, allocation, distributing or scheduling for enterprises or organisations
- G06Q10/06311—Scheduling, planning or task assignment for a person or group
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/08—Logistics, e.g. warehousing, loading or distribution; Inventory or stock management
- G06Q10/083—Shipping
- G06Q10/0835—Relationships between shipper or supplier and carriers
- G06Q10/08355—Routing methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/044—Recurrent networks, e.g. Hopfield networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
Abstract
Subject description discloses the method, apparatus of Order splitting, computer readable storage medium and electronic equipments, first according to order to be designated and the appointment order of each jockey, determine the route characteristic of each jockey's path planning, again for each order for including in each path planning, according to jockey and the corresponding information of order, it determines that the jockey corresponds to jockey's feature of the order, then according to the jockey's feature and route characteristic determined, determines the estimated delivery time of each order.Finally, the estimated delivery time based on each order, determines matching degree of each jockey respectively with the order to be designated, the order to be designated is distributed.By for each order in path planning, the mode of the estimated delivery time of each order is determined respectively, it avoids by the way that the prediction time-consuming in multiple sections to be added, lead to the case where predicting the accumulation of error, so that the estimated delivery time predicted is more acurrate, the allocation result determined is more accurate, improves dispatching efficiency.
Description
Technical field
This application involves logistics distribution technical field more particularly to the method, apparatus of Order splitting, computer-readable storage
Medium and electronic equipment.
Background technique
Currently, taking out dispatching platform in order to improve dispatching efficiency, generally according to the matching degree of order to be designated and jockey, In
Order to be designated is assigned to jockey by scheduling instance.Also, when determining the matching degree of order to be designated and jockey, it is also contemplated that
To after order to be designated is assigned to jockey, the influence that business generates is executed to jockey.Usually consider: being ordered by be designated
After being singly assigned to jockey, whether jockey can dispense whether the order to be designated and the jockey can dispense it on time on time
His order.
General platform is when determining the matching degree of order to be designated and jockey, it will usually first be ordered according to the appointment of the jockey
The single and order to be designated, re-starts path planning, determines the optimal distribution project of jockey, judged again later in the path
Whether each order in planning can dispense on time.Due to having re-started path planning, it is thus possible to occur having referred to originally
It sends order not overtime, but becomes overtime order in the path planning redefined, then cannot be by order to be designated
It is assigned to such jockey.Wherein, the picking position and delivery position of each order needed to be implemented in path planning comprising jockey,
And further comprise jockey's picking delivery sequence.
In the prior art, determine each order in path planning whether time-out process when, by taking in path planning
Goods yard set with delivery position, be used as task point.That is, task point is the ground that jockey completes that dispatching business needs to reach
Point.According to the dispatching sequence of task point each in path planning, time-consuming of the jockey between two task points is predicted respectively, thus really
The estimated delivery time of each order is made, then judges whether there is the order of time-out.
For example, it is assumed that the order of appointment for taking out jockey X is A and B, order to be designated is C after path planning, is obtained
The Distribution path of Fig. 1.Wherein, triangle indicates pick-up position, and circle indicates food delivery position, and the letter in figure indicates position pair
Which order what is answered is.Light figure is the position that jockey has arrived at, and dark color is the position that jockey not yet reaches.Platform can divide
Not Yu Ce time-consuming of the jockey between two task points, that is, the time-consuming in every section of path shown in figure, that is, the number in figure
1~5 corresponding path.Then the estimated delivery time of order A is the sum of the time-consuming in section 1~4, the estimated delivery time of order B
For the sum of the time-consuming in section 1~5, the estimated delivery time of order C is the sum of the time-consuming in section 1~3.
It but is the combination of multiple prediction periods, therefore pre- due to the estimated delivery time of order determining in the prior art
It surveys error to be easy to be amplified, causes the estimated delivery time for order not accurate enough, influence the accuracy rate of Order splitting.
Summary of the invention
This specification embodiment provides method, apparatus, computer readable storage medium and the electronic equipment of Order splitting, uses
Problems of the prior art are solved in part.
This specification embodiment adopts the following technical solutions:
The method for the Order splitting that this specification provides, comprising:
The road of the jockey is determined according to order to be designated and the appointment order of the jockey at least one jockey
Diameter planning；
The lane features extraction submodel for including in the first trained according to the path planning and in advance prediction model,
Determine route characteristic；
It is corresponding according to the corresponding information of the jockey and the order for each order for including in the path planning
Information determines that the jockey corresponds to the order by the jockey's feature extraction submodel for including in first prediction model
Jockey's feature, according to the jockey determined correspond to the order jockey's feature and the route characteristic, pass through described the
The output submodel for including in one prediction model determines the estimated delivery time of the order；
According to the estimated delivery time for each order for including in the path planning determined, the jockey and institute are determined
State the matching degree of order to be designated；
According to the matching degree of at least one described jockey and the order to be designated, the order to be designated is distributed.
Optionally, the lane features extraction submodel includes shot and long term memory network and attention layer；Correspondingly,
Lane features extraction for including in first trained according to the path planning and in advance prediction model
Model determines route characteristic, comprising:
For each task point in the path planning, according at least one of the type of the task point and coordinate,
Determine the feature vector of the task point；
According to dispatching sequence of each task point in the path planning, successively by the feature vector of each task point, input
In the shot and long term memory network, the corresponding output result of each task point is obtained；
By the attention layer, the corresponding output result of each task point is subjected to attention weighting, is obtained
Attention weighted results, using the attention weighted results as route characteristic.
Optionally, jockey's feature extraction submodel includes the first multi-layer perception (MLP)；Correspondingly,
It is described according to the corresponding information of the jockey and the corresponding information of the order, by being wrapped in first prediction model
The jockey's feature extraction submodel contained determines that the jockey corresponds to jockey's feature of the order, comprising:
According to the corresponding information of the jockey and the corresponding information of the order, jockey's feature vector is determined；
Jockey's feature vector is inputted in first multi-layer perception (MLP), the defeated of first multi-layer perception (MLP) is obtained
Out as a result, corresponding to jockey's feature of the order using the output result as the jockey.
Optionally, the output submodel by including in first prediction model determines that the estimated of the order is sent
Up to before the time, the method also includes:
Determine the dispatching state of the order for picking；
When the dispatching state of the order is non-picking, the method also includes:
According to the corresponding information of the order, the corresponding information of provider of order dispatching object is determined；
According to the corresponding information of the provider, pass through the provider's feature for including in the second prediction model of training in advance
Submodel is extracted, determines provider's feature；
Jockey's feature, provider's feature and the route characteristic for corresponding to the order according to the jockey, pass through
The output submodel for including in the second prediction model of training in advance, determines the estimated delivery time of the order.
Optionally, provider's feature extraction submodel includes the second multi-layer perception (MLP)；Correspondingly,
It is described according to the corresponding information of the provider, pass through the provider for including in the second prediction model of training in advance
Feature extraction submodel determines provider's feature, comprising:
According to the corresponding information of the provider, provider's feature vector is determined；
Provider's feature vector is inputted in second multi-layer perception (MLP), second multi-layer perception (MLP) is obtained
Output is as a result, as provider's feature.
Optionally, first prediction model of preparatory training, comprising:
The corresponding historical data of order is completed in acquisition in history；
Order is completed at least one, determination executes the jockey that order is completed in history, as specified jockey；
Be from the dispatching state that order is completed in history determine given time in the period of picking, and according to
The given time this road of the corresponding information of order, the specified corresponding information of jockey and the specified jockey is completed
Diameter planning, determines training sample；
The corresponding historical data of order is completed according to this, determines that the actual service time of order is completed in this, as institute
State the training sample corresponding actual service time；
According to the training sample determined, with the training sample corresponding actual service time for expected output, training institute
State the first prediction model.
Optionally, the training sample that the basis is determined is expected with the training sample corresponding actual service time
Output, training first prediction model, comprising:
At least two training sample point is determined by first prediction model at least two training samples
Not corresponding estimated delivery time；
According at least two training sample corresponding actual service time and at least two training sample
This corresponding estimated delivery time determines the damage of at least two training sample by preset first-loss function
The sum of lose；
With the minimum optimization aim of the sum of the loss, the parameter of first prediction model is adjusted.
The device of this specification offer Order splitting, comprising:
Path planning module, is configured for at least one jockey, according to order to be designated and the jockey
Order has been assigned, has determined the path planning of the jockey；
First determining module is configured in the first prediction model trained according to the path planning and in advance
The lane features extraction submodel for including, determines route characteristic；
Second determining module is configured for for each order for including in the path planning, according to the jockey
Corresponding information and the corresponding information of the order pass through the jockey's feature extraction submodule for including in first prediction model
Type determines that the jockey corresponds to jockey's feature of the order, according to jockey's feature of the jockey determined and the path
Feature determines the estimated delivery time of the order by the output submodel for including in first prediction model；
Third determining module is configured for according to the pre- of each order for including in the path planning determined
Delivery time is counted, determines the matching degree of the jockey Yu the order to be designated；
Distribution module is configured for the matching degree according to described at least one jockey and the order to be designated, point
With the order to be designated.
The computer readable storage medium that this specification provides, which is characterized in that the storage medium is stored with computer
Program, the computer program realizes the above order distribution method when being executed by processor.
The electronic equipment that this specification provides, including memory, processor and storage are on a memory and can be in processor
The computer program of upper operation, which is characterized in that the processor realizes the above order distribution method when executing described program.
This specification embodiment use at least one above-mentioned technical solution can reach it is following the utility model has the advantages that
When it needs to be determined that executing the jockey of the corresponding task of order to be designated, firstly, at least one jockey, root can be directed to
According to the order to be designated and the appointment order of the jockey, the path planning of the jockey is determined, and then according to preparatory training
The first prediction model in include lane features extraction submodel, the route characteristic of the path planning is determined, then, for this
The each order for including in path planning, it is pre- by first according to the corresponding information of the jockey and the corresponding information of the order
The jockey's feature extraction submodel for including in model is surveyed, determines that the jockey corresponds to jockey's feature of the order, that is to say, that this
When jockey's feature be actually jockey Yu order feature intersection, and then according to the jockey's feature and the path determined
Feature determines the estimated delivery time of the order by the output submodel in the first prediction model.Finally, according to determining
Path planning in include each order estimated delivery time, determine the matching degree of jockey Yu order to be designated, and be based on
The matching degree of at least one jockey and the order to be designated for determining, distribute the order to be designated.Since order to be designated exists
Distribute to after jockey may the order of appointment to the jockey have an impact, such as may cause the dispatching sequence for having assigned order
Variation, the path of dispatching generate variation, and this influence be difference is executed based on different jockeys to have assigned order and generated, and
And the influences for having assigned order to generate the difference of different jockeys are all not exactly the same, therefore can first determine should be to for server
Assign Order splitting to after the jockey, the path planning of the jockey passes through determining road to determine influence that path change generates
Diameter feature characterizes.And later, it is thus necessary to determine that by Order splitting to be designated to jockey after, jockey executes the estimated of each order and send
Up to the time, order to be designated how is distributed so that determination is subsequent.Then for every in the path planning of each jockey and jockey
A order determines jockey's feature according to information of both jockey and order, according to jockey's feature and route characteristic, determines
The estimated delivery time of each order in the path planning.What is determined at this time is not only the order to be designated, also there is the jockey's
Assigned order, thus can in the path planning based on each jockey each order estimated delivery time, determine each jockey with should
The matching degree of order to be designated, and the appointment order is distributed according to each matching degree determined.Compared to existing by multiple pre-
The combination for surveying the period, determines the estimated delivery time of order, and what this specification provided is individually determined estimated for each order
Delivery time, therefore predict that error is exactly the error of model, and the accumulation of prediction error is avoided, so that is predicted expects to send
More acurrate up to the time, the allocation result determined is more accurate, improves dispatching efficiency.
Detailed description of the invention
The drawings described herein are used to provide a further understanding of the present application, constitutes part of this application, this Shen
Illustrative embodiments and their description please are not constituted an undue limitation on the present application for explaining the application.In the accompanying drawings:
Fig. 1 is the schematic diagram that prior-art fragmentation determines estimated delivery time；
Fig. 2 is the process for the Order splitting that this specification embodiment provides；
The structure for the lane features extraction submodel for including in the first prediction model that Fig. 3 provides for this specification embodiment
Schematic diagram；
Fig. 4 is the framework schematic diagram for the first and second prediction model that this specification embodiment provides；
Fig. 5 is the structural schematic diagram of the device for the Order splitting that this specification embodiment provides；
Fig. 6 is the electronic equipment schematic diagram corresponding to Fig. 2 that this specification embodiment provides.
Specific embodiment
To keep the purposes, technical schemes and advantages of this specification clearer, it is embodied below in conjunction with this specification
Technical scheme is clearly and completely described in example and corresponding attached drawing.Obviously, described embodiment is only this Shen
Please a part of the embodiment, instead of all the embodiments.The embodiment of base in this manual, those of ordinary skill in the art exist
Every other embodiment obtained under the premise of creative work is not made, shall fall in the protection scope of this application.
Below in conjunction with attached drawing, the technical scheme provided by various embodiments of the present application will be described in detail.
Fig. 2 is the process of Order splitting that this specification embodiment provides, specifically may include one in following steps or
It is multiple:
S102: the jockey is determined according to order to be designated and the appointment order of the jockey at least one jockey
Path planning.
In the present specification, the Order splitting process can be specifically executed by the server of dispatching platform, due to needing from each
It is determined in jockey and executes the jockey that order to be designated corresponds to task, it is therefore desirable to first determined for the order to be designated to be assigned to and respectively ride
After hand, the Distribution path of each jockey.Then, server can be directed at least one jockey, according to order to be designated and the jockey
Appointment order, determine to jockey path planning.
Specifically, server can be directed at least one jockey, the appointment order of the jockey is first determined.Wherein this has been assigned
Order is the order for being already assigned to jockey but having not carried out completion, can be according to the order history number for being already assigned to jockey
According to, and the historical data of order completed determines, the appointment order of the jockey.Since the order of appointment of jockey is
Jockey needs to be implemented the task of completion, therefore has to consider in the path planning for determining jockey.
And then the corresponding task point of the order of appointment according to the order to be designated and the jockey, according to preset
Path optimization's algorithm determines the path planning of the jockey.Wherein, the task point includes the picking position and delivery position of order
It sets.Because being to carry out path planning according to the corresponding task point of several orders (that is, order to be designated and assigned order)
, therefore the path planning of the jockey determined in step s 102, dispatching task may be carrying out with current time jockey
When path planning it is not quite identical.
For example, it is assumed that the picking position of order to be designated, the picking position of order has been assigned positioned at two differences of jockey
It is intermediate, it is determined that in the path planning of the jockey gone out, jockey may need it is above-mentioned two assigned order it is previous
After the picking position picking for assigning order, the picking position picking of order to be designated is first gone, goes above-mentioned two assigned again later
The latter of order has assigned the picking position picking of order.
Certainly, such planning path may need to expend certain time due to the picking position picking in order to be designated,
Situations such as may cause and chain reaction occur, jockey is made to lead to order dispatching time-out when executing business according to the path planning goes out
It is existing.Then server needs by subsequent step, to determine the estimated of each order for including in the path planning of the jockey
Delivery time.
S104: lane features extraction for including in the first trained according to the path planning and in advance prediction model
Model determines route characteristic.
In the present specification, it due to the estimated delivery time for each order for including in the path planning, is advised with the path
Correlation is drawn, therefore server is after determining the path planning of the jockey, it can be by the first prediction model of training in advance
The lane features extraction submodel for including, determines the route characteristic of the path planning, so that subsequent step determines the path planning
In include each order estimated delivery time.
Specifically, the lane features extraction submodel includes: shot and long term memory network (Long in the present specification
Short-Term Memory, LSTM) and attention layer (Attention), as shown in Figure 3.
The structural representation for the lane features extraction submodel for including in the first prediction model that Fig. 3 provides for this specification
Figure, it is seen that input data first inputs after LSTM, by different time LSTM export as a result, input attention layer after, output
The output of attention weighting is as a result, as route characteristic.There is the characteristic that long-term memory is carried out to input information using LSTM,
Important information in each task point is retained, then by Attention layers of progress attention weighting after, can
To extract for determining the more important feature of order delivery time, route characteristic is determined.
Firstly, server can be for each task point in the path planning of at least one jockey, according to the task point
At least one of type and coordinate determine the feature vector of the task point.
Wherein, the type of task point includes: picking position and delivery position.And in the present specification, server is also
It can determine the information of the corresponding task object of task point, and according to the information of the type of task point, coordinate and task object, really
The feature vector of the fixed task point.Wherein, when the type of task point is picking position, the corresponding task object of task point
Information be trade company information (e.g., the type of trade company, the ID of trade company, trade company positive rating, prepare dispatching object history average time
Etc.), when the type of task point is delivery position, the information of the corresponding task object of task point be the information of user (e.g.,
User initiates the frequency of business, positive rating of user etc.).For example, the feature vector of some task point be (1,
39.9156343816,116.43911257741), wherein 1 indicates pick-up position, the longitude and latitude of rear two expressions task point
Degree.
Later, server can be according to each task point sequence of the dispatching in the path planning, successively by the spy of each task point
Vector is levied, is input in LSTM, the corresponding output result of each task point of LSTM output is obtained.In the present specification, by
The sequencing of task point quantity and task point in path planning is all determining, therefore server can be in order successively
After the feature vector for entering and leaving each task point, then perform the next step.
Finally, the corresponding output result of each task point can be carried out attention weighting by attention layer by server,
The power that gains attention weighted results, using attention weighted results as route characteristic.Wherein, since what is exported by LSTM is multiple
Therefore output is as a result, input the matrix that can be considered as an output result composition of attention layer, and corresponding attention layer
Multiplication cross can be carried out by the matrix of attention matrix and output result composition, that is, attention weighted results, and as the path
The route characteristic of planning, as shown in Attention layers in Fig. 3.
Wherein, due to LSTM output output result dimension be it is fixed, the line number of the attention matrix is also
Fixed, and server can determine attention weighted results in the matrix after attention weights.For example, data result is m
The matrix of × n, attention matrix are the matrix of n × p, then server can determine attention from multiplication cross result m × p matrix
Weighted results are the vector of m dimension.For example, can be determined by maximum pond (max-pooling) method, or from multiplication cross result
In every row select maximized element etc., this specification is not construed as limiting this.
S106: for each order for including in the path planning, according to the corresponding information of the jockey and the order
Corresponding information is determined that the jockey corresponds to and is somebody's turn to do by the jockey's feature extraction submodel for including in first prediction model
Jockey's feature of order, jockey's feature and the route characteristic according to the jockey determined corresponding to the order, passes through
The output submodel for including in first prediction model determines the estimated delivery time of the order.
In the present specification, since the order for including in each path planning is there is also difference, server can be directed to
The each order for including in path planning passes through described according to the corresponding information of the jockey and the corresponding information of the order
The jockey's feature extraction submodel for including in one prediction model determines that the jockey corresponds to jockey's feature of the order.And it can
Jockey feature and the route characteristic of the jockey determined with basis corresponding to the order, by the first prediction model
The output submodel for including determines the estimated delivery time of the order.
Jockey's feature extraction submodel and output submodel input and output are illustrated respectively.
Specifically, jockey's feature extraction submodel includes: the first multi-layer perception (MLP) (Multi-Layer
Perception, MLP).Server can determine jockey according to the corresponding information of the jockey and the corresponding information of the order first
Feature vector, the input as the first MLP.Wherein, the corresponding information of jockey can include: the information such as waybill number with jockey are ordered
Single corresponding information can include: advised in the path the corresponding picking position of order price, order to the corresponding delivery position of order
The new order quantity, etc. that the corresponding region of distance, order in drawing generates.Server can combine above-mentioned various information, as
The feature vector of jockey.
Later, server can input jockey's feature vector in the first MLP, obtain the output of the first MLP as a result,
And correspond to jockey's feature of the order using the output result as the jockey.
In addition, the output submodel is also possible to MLP model either other prediction models, this explanation in the present specification
Book is not construed as limiting this.For server in the route characteristic that will be determined in jockey's feature and step S104 as input, input should
After exporting submodel, the output result of output submodel can be obtained, that is, the estimated delivery time of the order.
Further, in the present specification, it is different from due to jockey in the time etc. that different user and trade company expend,
Such as trade company be fast food trade company when, the time of preparing for a meal is usually shorter, when the positive rating of user is higher, sensitivity of the user to distribution time
Degree may be lower.And above by jockey's feature extraction submodel and lane features extraction submodel output as a result, conduct
The input for exporting submodel has lacked the corresponding spy of trade company eventually by the method that output submodel determines estimated delivery time
Sign input, therefore the process of the estimated delivery time of above-mentioned determining order can only be used to determine that the dispatching state of order is to have taken
The estimated delivery time of the order of goods.Therefore for for dispatching state is the order of picking, it is contemplated that delivery time is
It is unrelated with trade company, therefore can not have to consider.
If the dispatching state of order is non-picking, server can also be further first according to the corresponding information of the order, really
Determine the corresponding information of provider of order dispatching object.Specifically can according to the corresponding information of the provider, determine provider's feature to
Amount.Wherein, the corresponding information of provider may include that quantity, the type of provider of the dispatching object that provider is not ready to (e.g., are taken out
Provider type can include: home cooking, fast food, barbecue etc.), history the rate of complaints of provider and the ID of provider etc.
Deng.
And then according to the corresponding information of provider, pass through the provider for including in the second prediction model of training in advance
Feature extraction submodel determines provider's feature.Wherein, provider's feature extraction submodel includes the 2nd MLP.Then server can
Provider's feature vector is inputted in the 2nd MLP, obtains the output of the 2nd MLP as a result, as provider's feature.
Finally, corresponding to jockey's feature, provider's feature and route characteristic of the order according to the jockey, pass through
The output submodel for including in the second prediction model of training in advance, determines the estimated delivery time of the order.So that final defeated
The estimated delivery time of order out is jockey's feature based on the information for containing order, provider's feature and for should
What the route characteristic of order was determined, it is seen that determine that the feature of input all corresponds to the order.So that output result
It is to be determined directly against order, reducing leads to the increased risk of error after multiple prediction results are overlapped.
It should be noted that in the present specification, the route characteristic in the first prediction model and the second prediction model mentions
Submodel and jockey's feature extraction submodel are taken, this can be shared due to inputting and inputting uniform reason, and first and second is pre-
The output submodel separately included in model is surveyed, is therefore different output submodels since input is different.Same provider is special
It is distinctive in the second prediction model that sign, which extracts submodel, therefore is also the distinct place of first and second prediction model.
Fig. 4 is the framework schematic diagram for the first and second prediction model that this specification provides.Wherein in visible dotted line frame
Lane features extraction submodel and jockey's feature extraction submodel be shared, and remaining submodel is all non-common.
S108: according to the estimated delivery time for each order for including in the path planning determined, determine that this is ridden
The matching degree of hand and the order to be designated.
S110: according to the matching degree of at least one described jockey and the order to be designated, the order to be designated is distributed.
In the present specification, at least one jockey, when each order for including in the path planning for determine the jockey
Estimated delivery time after, can determine whether the matching degree of the order to be designated and the jockey.As long as example, including in path planning
The estimated delivery time of each order be not later than the promise delivery time of each order, it is determined that the order to be designated and the jockey's
Matching.Then server finally can select a jockey to distribute the appointment order from respectively and in the matched jockey of order to be designated.
Since it is determined that the path planning in include each order estimated delivery time, be individually determined both for each order
, it is therefore expected that the problem of delivery time carries out cumulative caused error increase there is no multiple prediction results.Also, due to riding
When hand dispenses order to be designated, the change of other order distribution times may cause, therefore by determining each order on path planning
Estimated delivery time, can more accurately determine order to be designated and the matching degree of jockey.Although for example, order to be designated
It is expected that delivery time has not timed out, but other orders of jockey is caused to dispense time-out, the then matching of jockey and the order to be designated
It spends lower.
Based on the method for Order splitting shown in Fig. 2, since order to be designated may ride this after distributing to jockey
The order of appointment of hand has an impact, and such as may cause and the dispatching sequence of order has been assigned to change, and the path of dispatching generates variation,
And this influence is to execute difference based on different jockeys to have assigned order and generated, and assigned the difference of different jockeys
The influence that order generates is all not exactly the same, therefore server can be determined the Order splitting to be designated first to after the jockey,
The path planning of the jockey with determine path change generate influence, characterized by determining route characteristic.And later, it needs
Determine by Order splitting to be designated to jockey after, jockey executes the estimated delivery time of each order, how to divide so that determination is subsequent
With order to be designated.Then for each order in the path planning of each jockey and jockey, according to jockey and order two
The information of aspect determines jockey's feature, according to jockey's feature and route characteristic, determines the pre- of each order in the path planning
Count delivery time.What is determined at this time is not only the order to be designated, also has the appointment order of the jockey, therefore can be based on each
The estimated delivery time of each order in the path planning of jockey, determines the matching degree of each jockey Yu the order to be designated, and according to
Each matching degree determined distributes the appointment order.Compared to the existing combination by multiple prediction periods, the pre- of order is determined
Delivery time, the estimated delivery time being individually determined for each order that this specification provides are counted, therefore predicts that error is exactly
The error of model, and the accumulation of prediction error is avoided, so that the estimated delivery time predicted is more acurrate, the distribution determined
As a result more accurate, improve dispatching efficiency.
In addition, the estimated delivery time of order is closer to order for server in step S108 and step S110
Promise delivery time user experience it is preferable, but be too close to and will lead to when there is fortuitous event order dispatching time-out.
Then, server can determine planning path according to the difference of the promise delivery time of the estimated delivery time and order of each order
In each order and jockey matching degree.For example, determining order and jockey of the corresponding time difference of each order at 10 to 5 minutes
Matching degree highest, difference in 15 to 10 minutes take second place, etc., determine each order respectively with the matching degree of jockey, if difference
It then determines and mismatches for negative value.Later, according to the average value for the corresponding matching degree of each order for including in the Distribution path or
Summation selects jockey's distribution should order be reassigned according to sequence from big to small from each jockey.
Further, in the present specification, due to there are shared feelings in the submodel of the first and second prediction model
Condition, therefore can be trained jointly in first and second prediction model.
Specifically, server is available first is completed the corresponding historical data of order in history.
Secondly, order is completed at least one, determination executes the jockey that order is completed in history, as specified
Jockey.For example, the jockey for being then finally completed the order is the specified of the order when some order has replaced jockey when being not carried out
Jockey.
It then, is to determine given time, and root in the period of picking from the dispatching state that order is completed in history
According in the given time, the path of the corresponding information of order, the specified corresponding information of jockey and the specified jockey is completed in this
Planning, determines training sample.The data of determining training sample at this time, that is, needed in abovementioned steps S104 and step S106
The data in each submodel of the first prediction model are inputted respectively.
Later, the corresponding historical data of order is completed according to this, determines that the actual service time of order is completed in this, made
For the training sample corresponding actual service time.
Finally, according to the training sample determined, with the training sample corresponding actual service time for expected output, instruction
Practice first prediction model.
At this time during the first prediction model of training, the lane features extraction submodel and jockey's feature trained
Extracting submodel and being equivalent to also is that the second prediction model is trained.
Further, in the present specification, server is when determining training sample, can also from history this be completed
The dispatching state of order is to determine given time, and according to order pair is completed in this in the given time in the period of non-picking
The information answered, the specified corresponding information of jockey, the specified jockey path planning and this order is completed and corresponds to supplier
Information, determine training sample.Then at this point, in the training sample that server is determined, containing is to be suitable for the first prediction mould
Type and suitable for the second prediction model different training samples.Certainly, since the same order that is completed certainly exists
Dispatching state is non-picking and dispenses state as two kinds of situations of picking, but the actual service time is all that order is completed in this,
It therefore can be with the training sample corresponding actual service time for expected output, according to the difference of the training sample of input point
First and second prediction model is not trained.
In addition, in the present specification, when in order to avoid based on single sample training, the amplitude of parameter adjustment is too big, cause
The case where training effect reduces occurs.Server can also loss based on multiple samples and, to adjust the parameter of model.
Specifically, server adjust model parameter when, the seat of honour can be directed at least two training samples, by this first
Prediction model determines the corresponding estimated delivery time of at least two training sample.
Later, according at least two training sample corresponding actual service time and at least two training
The corresponding estimated delivery time of sample, by preset loss function, determine at least two training sample loss it
With.Wherein, the loss function be specially which kind of loss function this specification with no restrictions, can be set as needed.
Finally, adjusting the parameter of first prediction model with the minimum optimization aim of the sum of the loss.
Further, since the first and second prediction model can be trained simultaneously, thus determine that different predictions
The corresponding loss function of model can be not quite identical, such as the first and second prediction model respectively corresponds first-loss function
And second loss function.And it can be exported respectively according to the first and second prediction model simultaneously when calculating loss estimated
Delivery time and actual service time determine loss.Namely total losses the sum of the loss that is equal to each training sample, and each training
Sample may include the training sample for being applicable in the first and second prediction model respectively.Based on order allocation method shown in FIG. 1, originally
The structural schematic diagram of the also corresponding device that Order splitting is provided of specification embodiment, as shown in Figure 5.
Fig. 5 is the structural schematic diagram of the device for the Order splitting that this specification embodiment provides, and described device includes:
Path planning module 200 is configured for at least one jockey, according to order to be designated and the jockey
Appointment order, determine the path planning of the jockey；
First determining module 202 is configured for the first prediction mould trained according to the path planning and in advance
The lane features extraction submodel for including in type, determines route characteristic；
Second determining module 204 is configured for being ridden for each order for including in the path planning according to this
The corresponding information of hand and the corresponding information of the order pass through the jockey's feature extraction submodule for including in first prediction model
Type determines that the jockey corresponds to jockey's feature of the order, according to jockey's feature of the jockey determined and the path
Feature determines the estimated delivery time of the order by the output submodel for including in first prediction model；
Third determining module 206 is configured for according to each order for including in the path planning determined
Estimated delivery time, determine the matching degree of the jockey Yu the order to be designated；
Distribution module 208 is configured for the matching degree according to described at least one jockey and the order to be designated,
Distribute the order to be designated.
Optionally, the lane features extraction submodel includes shot and long term memory network and attention layer, correspondingly, first
Determining module 202 is configured for for each task point in the path planning, according to the type and seat of the task point
At least one of mark, determines the feature vector of the task point, sequentially according to dispatching of each task point in the path planning,
It successively by the feature vector of each task point, inputs in the shot and long term memory network, obtains the corresponding output of each task point
As a result, by the attention layer, the corresponding output result of each task point is subjected to attention weighting, is gained attention
Power weighted results, using the attention weighted results as route characteristic.
Optionally, jockey's feature extraction submodel includes the first multi-layer perception (MLP), correspondingly, corresponding according to the jockey
Information and the corresponding information of the order, determine jockey's feature vector, by jockey's feature vector input described more than first
In layer perceptron, obtains the output of first multi-layer perception (MLP) and be somebody's turn to do as a result, the output result is corresponded to as the jockey
Jockey's feature of order.
Optionally, the second determining module 204 is configured for described defeated by include in first prediction model
Submodel out before the estimated delivery time for determining the order, determines that the dispatching state of the order is picking, described device is also
Include: the 4th determining module 210, is configured for determining the provider of order dispatching object according to the corresponding information of the order
Corresponding information passes through the provider for including in the second prediction model of training in advance according to the corresponding information of the provider
Feature extraction submodel determines provider's feature, and jockey's feature, the provider's feature of the order are corresponded to according to the jockey
And the route characteristic determines the pre- of the order by the output submodel for including in the second prediction model of training in advance
Count delivery time.
Optionally, provider's feature extraction submodel includes the second multi-layer perception (MLP), correspondingly, the 4th determining module
210, it is configured for determining provider's feature vector according to the corresponding information of the provider, by provider's feature
Vector inputs in second multi-layer perception (MLP), obtains the output of second multi-layer perception (MLP) as a result, as provider's feature.
Optionally, described device further include: training module 212 is configured for acquisition and order pair is completed in history
Order is completed at least one in the historical data answered, and determination executes the jockey that order is completed in history, as specified
Jockey is to determine given time, and according to described in the period of picking from the dispatching state that order is completed in history
It advises in the given time path that the corresponding information of order, the specified corresponding information of jockey and the specified jockey is completed
It draws, determines training sample, the corresponding historical data of order is completed according to this, when determining that the actual service of order is completed in this
Between, as the training sample corresponding actual service time, according to the training sample determined, with the corresponding reality of training sample
Border delivery time is expected output, training first prediction model.
Optionally, training module 212 are configured for predicting at least two training samples by described first
Model determines the corresponding estimated delivery time of at least two training sample, according at least two training sample
Corresponding actual service time and the corresponding estimated delivery time of at least two training sample, by pre-
If loss function, determine the sum of the loss of at least two training sample, with it is described loss the sum of minimum optimization aim,
Adjust the parameter of first prediction model.
Based on the device of Order splitting shown in fig. 5, since order to be designated may ride this after distributing to jockey
The order of appointment of hand has an impact, and such as may cause and the dispatching sequence of order has been assigned to change, and the path of dispatching generates variation,
And this influence is to execute difference based on different jockeys to have assigned order and generated, and assigned the difference of different jockeys
The influence that order generates is all not exactly the same, therefore server can be determined the Order splitting to be designated first to after the jockey,
The path planning of the jockey with determine path change generate influence, characterized by determining route characteristic.And later, it needs
Determine by Order splitting to be designated to jockey after, jockey executes the estimated delivery time of each order, how to divide so that determination is subsequent
With order to be designated.Then for each order in the path planning of each jockey and jockey, according to jockey and order two
The information of aspect determines jockey's feature, according to jockey's feature and route characteristic, determines the pre- of each order in the path planning
Count delivery time.What is determined at this time is not only the order to be designated, also has the appointment order of the jockey, therefore can be based on each
The estimated delivery time of each order in the path planning of jockey, determines the matching degree of each jockey Yu the order to be designated, and according to
Each matching degree determined distributes the appointment order.Compared to the existing combination by multiple prediction periods, the pre- of order is determined
Delivery time, the estimated delivery time being individually determined for each order that this specification provides are counted, therefore predicts that error is exactly
The error of model, and the accumulation of prediction error is avoided, so that the estimated delivery time predicted is more acurrate, the distribution determined
As a result more accurate, improve dispatching efficiency.
This specification embodiment additionally provides computer readable storage medium, which is stored with computer program,
Computer program can be used for executing any of the method for the above order distribution.
Based on the method for Order splitting shown in Fig. 2, this specification embodiment also proposed electronic equipment shown in fig. 6
Schematic configuration diagram.Such as Fig. 6, in hardware view, which includes processor, internal bus, network interface, memory and non-
Volatile memory is also possible that hardware required for other business certainly.Processor is read from nonvolatile memory
Then corresponding computer program is run into memory, the method to realize the distribution of any one the above order.
Certainly, other than software realization mode, other implementations, such as logical device suppression is not precluded in this specification
Or mode of software and hardware combining etc., that is to say, that the executing subject of following process flow is not limited to each logic unit,
It is also possible to hardware or logical device.
In the 1990s, the improvement of a technology can be distinguished clearly be on hardware improvement (for example,
Improvement to circuit structures such as diode, transistor, switches) or software on improvement (improvement for method flow).So
And with the development of technology, the improvement of current many method flows can be considered as directly improving for hardware circuit.
Designer nearly all obtains corresponding hardware circuit by the way that improved method flow to be programmed into hardware circuit.Cause
This, it cannot be said that the improvement of a method flow cannot be realized with hardware entities module.For example, programmable logic device
(Programmable Logic Device, PLD) (such as field programmable gate array (Field Programmable Gate
Array, FPGA)) it is exactly such a integrated circuit, logic function determines device programming by user.By designer
Voluntarily programming comes a digital display circuit " integrated " on a piece of PLD, designs and makes without asking chip maker
Dedicated IC chip.Moreover, nowadays, substitution manually makes IC chip, this programming is also used instead mostly " is patrolled
Volume compiler (logic compiler) " software realizes that software compiler used is similar when it writes with program development,
And the source code before compiling also write by handy specific programming language, this is referred to as hardware description language
(Hardware Description Language, HDL), and HDL is also not only a kind of, but there are many kind, such as ABEL
(Advanced Boolean Expression Language)、AHDL(Altera Hardware Description
Language)、Confluence、CUPL(Cornell University Programming Language)、HDCal、JHDL
(Java Hardware Description Language)、Lava、Lola、MyHDL、PALASM、RHDL(Ruby
Hardware Description Language) etc., VHDL (Very-High-Speed is most generally used at present
Integrated Circuit Hardware Description Language) and Verilog.Those skilled in the art also answer
This understands, it is only necessary to method flow slightly programming in logic and is programmed into integrated circuit with above-mentioned several hardware description languages,
The hardware circuit for realizing the logical method process can be readily available.
Controller can be implemented in any suitable manner, for example, controller can take such as microprocessor or processing
The computer for the computer readable program code (such as software or firmware) that device and storage can be executed by (micro-) processor can
Read medium, logic gate, switch, specific integrated circuit (Application Specific Integrated Circuit,
ASIC), the form of programmable logic controller (PLC) and insertion microcontroller, the example of controller includes but is not limited to following microcontroller
Device: ARC 625D, Atmel AT91SAM, Microchip PIC18F26K20 and Silicone Labs C8051F320 are deposited
Memory controller is also implemented as a part of the control logic of memory.It is also known in the art that in addition to
Pure computer readable program code mode is realized other than controller, can be made completely by the way that method and step is carried out programming in logic
Controller is obtained to come in fact in the form of logic gate, switch, specific integrated circuit, programmable logic controller (PLC) and insertion microcontroller etc.
Existing identical function.Therefore this controller is considered a kind of hardware component, and to including for realizing various in it
The device of function can also be considered as the structure in hardware component.Or even, it can will be regarded for realizing the device of various functions
For either the software module of implementation method can be the structure in hardware component again.
System, device, module or the unit that above-described embodiment illustrates can specifically realize by computer chip or entity,
Or it is realized by the product with certain function.It is a kind of typically to realize that equipment is computer.Specifically, computer for example may be used
Think personal computer, laptop computer, cellular phone, camera phone, smart phone, personal digital assistant, media play
It is any in device, navigation equipment, electronic mail equipment, game console, tablet computer, wearable device or these equipment
The combination of equipment.
For convenience of description, it is divided into various units when description apparatus above with function to describe respectively.Certainly, implementing this
The function of each unit can be realized in the same or multiple software and or hardware when specification.
It should be understood by those skilled in the art that, the embodiment of the present invention can provide as method, system or computer program
Product.Therefore, complete hardware embodiment, complete software embodiment or reality combining software and hardware aspects can be used in the present invention
Apply the form of example.Moreover, it wherein includes the computer of computer usable program code that the present invention, which can be used in one or more,
The computer program implemented in usable storage medium (including but not limited to magnetic disk storage, CD-ROM, optical memory etc.) produces
The form of product.
The present invention be referring to according to the method for the embodiment of the present invention, the process of equipment (system) and computer program product
Figure and/or block diagram describe.It should be understood that every one stream in flowchart and/or the block diagram can be realized by computer program instructions
The combination of process and/or box in journey and/or box and flowchart and/or the block diagram.It can provide these computer programs
Instruct the processor of general purpose computer, special purpose computer, Embedded Processor or other programmable data processing devices to produce
A raw machine, so that being generated by the instruction that computer or the processor of other programmable data processing devices execute for real
The device for the function of being specified in present one or more flows of the flowchart and/or one or more blocks of the block diagram.
These computer program instructions, which may also be stored in, is able to guide computer or other programmable data processing devices with spy
Determine in the computer-readable memory that mode works, so that it includes referring to that instruction stored in the computer readable memory, which generates,
Enable the manufacture of device, the command device realize in one box of one or more flows of the flowchart and/or block diagram or
The function of being specified in multiple boxes.
These computer program instructions also can be loaded onto a computer or other programmable data processing device, so that counting
Series of operation steps are executed on calculation machine or other programmable devices to generate computer implemented processing, thus in computer or
The instruction executed on other programmable devices is provided for realizing in one or more flows of the flowchart and/or block diagram one
The step of function of being specified in a box or multiple boxes.
In a typical configuration, calculating equipment includes one or more processors (CPU), input/output interface, net
Network interface and memory.
Memory may include the non-volatile memory in computer-readable medium, random access memory (RAM) and/or
The forms such as Nonvolatile memory, such as read-only memory (ROM) or flash memory (flash RAM).Memory is computer-readable medium
Example.
Computer-readable medium includes permanent and non-permanent, removable and non-removable media can be by any method
Or technology come realize information store.Information can be computer readable instructions, data structure, the module of program or other data.
The example of the storage medium of computer includes, but are not limited to phase change memory (PRAM), static random access memory (SRAM), moves
State random access memory (DRAM), other kinds of random access memory (RAM), read-only memory (ROM), electric erasable
Programmable read only memory (EEPROM), flash memory or other memory techniques, read-only disc read only memory (CD-ROM) (CD-ROM),
Digital versatile disc (DVD) or other optical storage, magnetic cassettes, tape magnetic disk storage or other magnetic storage devices
Or any other non-transmission medium, can be used for storage can be accessed by a computing device information.As defined in this article, it calculates
Machine readable medium does not include temporary computer readable media (transitory media), such as the data-signal and carrier wave of modulation.
It should also be noted that, the terms "include", "comprise" or its any other variant are intended to nonexcludability
It include so that the process, method, commodity or the equipment that include a series of elements not only include those elements, but also to wrap
Include other elements that are not explicitly listed, or further include for this process, method, commodity or equipment intrinsic want
Element.In the absence of more restrictions, the element limited by sentence "including a ...", it is not excluded that including described want
There is also other identical elements in the process, method of element, commodity or equipment.
It will be understood by those skilled in the art that the embodiment of this specification can provide as the production of method, system or computer program
Product.Therefore, complete hardware embodiment, complete software embodiment or implementation combining software and hardware aspects can be used in this specification
The form of example.Moreover, it wherein includes the computer of computer usable program code that this specification, which can be used in one or more,
The computer program implemented in usable storage medium (including but not limited to magnetic disk storage, CD-ROM, optical memory etc.) produces
The form of product.
This specification can describe in the general context of computer-executable instructions executed by a computer, such as journey
Sequence module.Generally, program module include routines performing specific tasks or implementing specific abstract data types, programs, objects,
Component, data structure etc..This specification can also be practiced in a distributed computing environment, in these distributed computing environment
In, by executing task by the connected remote processing devices of communication network.In a distributed computing environment, program module
It can be located in the local and remote computer storage media including storage equipment.
All the embodiments in this specification are described in a progressive manner, same and similar portion between each embodiment
Dividing may refer to each other, and each embodiment focuses on the differences from other embodiments.Especially for system reality
For applying example, since it is substantially similar to the method embodiment, so being described relatively simple, related place is referring to embodiment of the method
Part explanation.
The foregoing is merely the embodiments of this specification, are not limited to this specification.For art technology
For personnel, this specification can have various modifications and variations.It is all made any within the spirit and principle of this specification
Modification, equivalent replacement, improvement etc., should be included within the scope of the claims of this specification.
Claims (10)
1. a kind of method of Order splitting characterized by comprising
The path rule of the jockey are determined according to order to be designated and the appointment order of the jockey at least one jockey
It draws；
The lane features extraction submodel for including in the first trained according to the path planning and in advance prediction model determines
Route characteristic；
For each order for including in the path planning, according to the corresponding information of the jockey and the corresponding letter of the order
Breath determines that the jockey corresponds to riding for the order by the jockey's feature extraction submodel for including in first prediction model
Hand feature, jockey's feature and the route characteristic according to the jockey determined corresponding to the order, passes through described first
The output submodel for including in prediction model determines the estimated delivery time of the order；
According to the estimated delivery time for each order for including in the path planning determined, determine the jockey and it is described to
Assign the matching degree of order；
According to the matching degree of at least one described jockey and the order to be designated, the order to be designated is distributed.
2. the method as described in claim 1, which is characterized in that the lane features extraction submodel includes shot and long term memory net
Network and attention layer；Correspondingly,
The lane features extraction submodel for including in first trained according to the path planning and in advance prediction model,
Determine route characteristic, comprising:
It is determined for each task point in the path planning according at least one of the type of the task point and coordinate
The feature vector of the task point；
According to dispatching sequence of each task point in the path planning, successively by the feature vector of each task point, described in input
In shot and long term memory network, the corresponding output result of each task point is obtained；
By the attention layer, the corresponding output result of each task point is subjected to attention weighting, is gained attention
Power weighted results, using the attention weighted results as route characteristic.
3. the method as described in claim 1, which is characterized in that jockey's feature extraction submodel includes the first Multilayer Perception
Machine；Correspondingly,
It is described according to the corresponding information of the jockey and the corresponding information of the order, by including in first prediction model
Jockey's feature extraction submodel determines that the jockey corresponds to jockey's feature of the order, comprising:
According to the corresponding information of the jockey and the corresponding information of the order, jockey's feature vector is determined；
Jockey's feature vector is inputted in first multi-layer perception (MLP), the output knot of first multi-layer perception (MLP) is obtained
Fruit corresponds to jockey's feature of the order using the output result as the jockey.
4. the method as described in claim 1, which is characterized in that output by including in first prediction model
Model, before the estimated delivery time for determining the order, the method also includes:
Determine the dispatching state of the order for picking；
When the dispatching state of the order is non-picking, the method also includes:
According to the corresponding information of the order, the corresponding information of provider of order dispatching object is determined；
According to the corresponding information of the provider, pass through the provider's feature extraction for including in the second prediction model of training in advance
Submodel determines provider's feature；
Jockey's feature, provider's feature and the route characteristic for corresponding to the order according to the jockey, by preparatory
The output submodel for including in the second trained prediction model determines the estimated delivery time of the order.
5. the method as described in right wants 4, which is characterized in that provider's feature extraction submodel includes the second Multilayer Perception
Machine；Correspondingly,
It is described according to the corresponding information of the provider, pass through the provider's feature for including in the second prediction model of training in advance
Submodel is extracted, determines provider's feature, comprising:
According to the corresponding information of the provider, provider's feature vector is determined；
Provider's feature vector is inputted in second multi-layer perception (MLP), the output of second multi-layer perception (MLP) is obtained
As a result, as provider's feature.
6. method as claimed in claim 4, which is characterized in that preparatory first prediction model of training, comprising:
The corresponding historical data of order is completed in acquisition in history；
Order is completed at least one, determination executes the jockey that order is completed in history, as specified jockey；
It is to determine given time, and according to described in the period of picking from the dispatching state that order is completed in history
It advises in the given time path that the corresponding information of order, the specified corresponding information of jockey and the specified jockey is completed
It draws, determines training sample；
The corresponding historical data of order is completed according to this, determines that the actual service time of order is completed in this, as the instruction
Practice the sample corresponding actual service time；
According to the training sample determined, with the training sample corresponding actual service time for expected output, training described the
One prediction model.
7. method as claimed in claim 6, which is characterized in that the training sample that the basis is determined, with training sample pair
The actual service time answered is expected output, training first prediction model, comprising:
For at least two training samples, by first prediction model, determine that at least two training sample is right respectively
The estimated delivery time answered；
According at least two training sample corresponding actual service time and at least two training sample point
Not corresponding estimated delivery time determines the sum of the loss of at least two training sample by preset loss function；
With the minimum optimization aim of the sum of the loss, the parameter of first prediction model is adjusted.
8. a kind of device of Order splitting, which is characterized in that described device includes:
Path planning module, is configured for at least one jockey, according to having referred to for order to be designated and the jockey
Order is sent, determines the path planning of the jockey；
First determining module is configured for according to the path planning and in the first prediction model of training includes in advance
Lane features extraction submodel, determine route characteristic；
Second determining module is configured for for each order for including in the path planning, corresponding according to the jockey
Information and the corresponding information of the order, by the jockey's feature extraction submodel for including in first prediction model, really
The fixed jockey corresponds to jockey's feature of the order, according to jockey's feature of the jockey determined and the route characteristic,
By the output submodel for including in first prediction model, the estimated delivery time of the order is determined；
Third determining module is configured for being sent according to the estimated of each order for including in the path planning determined
Up to the time, the matching degree of the jockey Yu the order to be designated are determined；
Distribution module is configured for distributing institute according to the matching degree of described at least one jockey and the order to be designated
State order to be designated.
9. a kind of computer readable storage medium, which is characterized in that the storage medium is stored with computer program, the calculating
The claims 1-7 any method is realized when machine program is executed by processor.
10. a kind of electronic equipment including memory, processor and stores the calculating that can be run on a memory and on a processor
Machine program, which is characterized in that the processor realizes the claims 1-7 any method when executing described program.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910667271.5A CN110414731B (en) | 2019-07-23 | 2019-07-23 | Order distribution method and device, computer readable storage medium and electronic equipment |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910667271.5A CN110414731B (en) | 2019-07-23 | 2019-07-23 | Order distribution method and device, computer readable storage medium and electronic equipment |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN110414731A true CN110414731A (en) | 2019-11-05 |
| CN110414731B CN110414731B (en) | 2021-02-02 |
Family
ID=68362726
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910667271.5A Active CN110414731B (en) | 2019-07-23 | 2019-07-23 | Order distribution method and device, computer readable storage medium and electronic equipment |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110414731B (en) |
Cited By (13)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110910019A (en) * | 2019-11-22 | 2020-03-24 | 拉扎斯网络科技（上海）有限公司 | Data processing method and device, electronic equipment and computer readable storage medium |
| CN111652477A (en) * | 2020-05-19 | 2020-09-11 | 拉扎斯网络科技（上海）有限公司 | Order processing and similarity calculation model obtaining method and device and electronic equipment |
| CN112036697A (en) * | 2020-07-28 | 2020-12-04 | 拉扎斯网络科技（上海）有限公司 | Task allocation method and device, readable storage medium and electronic equipment |
| CN112258131A (en) * | 2020-11-12 | 2021-01-22 | 拉扎斯网络科技（上海）有限公司 | Path prediction network training and order processing method and device |
| CN112541610A (en) * | 2020-08-13 | 2021-03-23 | 深圳优地科技有限公司 | Robot control method, device, electronic device and storage medium |
| CN112837128A (en) * | 2021-02-19 | 2021-05-25 | 拉扎斯网络科技（上海）有限公司 | Order assignment method and device, computer equipment and computer readable storage medium |
| CN112862133A (en) * | 2019-11-12 | 2021-05-28 | 北京三快在线科技有限公司 | Order processing method and device, readable storage medium and electronic equipment |
| CN112907011A (en) * | 2019-12-04 | 2021-06-04 | 北京三快在线科技有限公司 | Distribution scheduling method, device, storage medium and electronic equipment |
| CN113222202A (en) * | 2021-06-01 | 2021-08-06 | 携程旅游网络技术（上海）有限公司 | Reservation vehicle dispatching method, reservation vehicle dispatching system, reservation vehicle dispatching equipment and reservation vehicle dispatching medium |
| CN113222487A (en) * | 2020-01-21 | 2021-08-06 | 北京三快在线科技有限公司 | Scheduling path generation method, device, storage medium and electronic equipment |
| CN113642603A (en) * | 2021-07-05 | 2021-11-12 | 北京三快在线科技有限公司 | Data matching method and device, storage medium and electronic equipment |
| CN114330797A (en) * | 2020-09-27 | 2022-04-12 | 北京三快在线科技有限公司 | Distribution time length prediction method and device, storage medium and electronic equipment |
| CN115409452A (en) * | 2022-10-27 | 2022-11-29 | 浙江口碑网络技术有限公司 | Distribution information processing method, device, system, equipment and readable storage medium |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107545315A (en) * | 2016-06-24 | 2018-01-05 | 北京三快在线科技有限公司 | Order processing method and device |
| CN109543886A (en) * | 2018-11-06 | 2019-03-29 | 斑马网络技术有限公司 | Prediction technique, device, terminal and the storage medium of destination |
| CN109685276A (en) * | 2018-12-27 | 2019-04-26 | 拉扎斯网络科技（上海）有限公司 | Order processing method, apparatus, electronic equipment and computer readable storage medium |
| CN109726843A (en) * | 2017-10-30 | 2019-05-07 | 阿里巴巴集团控股有限公司 | The method, apparatus and terminal of allocation data prediction |
| CN109791731A (en) * | 2017-06-22 | 2019-05-21 | 北京嘀嘀无限科技发展有限公司 | A kind of method and system for estimating arrival time |
| WO2019100279A1 (en) * | 2017-11-23 | 2019-05-31 | Beijing Didi Infinity Technology And Development Co., Ltd. | System and method for estimating arrival time |
| WO2019109604A1 (en) * | 2017-12-05 | 2019-06-13 | Beijing Didi Infinity Technology And Development Co., Ltd. | Systems and methods for determining an estimated time of arrival for online to offline services |
-
2019
- 2019-07-23 CN CN201910667271.5A patent/CN110414731B/en active Active
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107545315A (en) * | 2016-06-24 | 2018-01-05 | 北京三快在线科技有限公司 | Order processing method and device |
| CN109791731A (en) * | 2017-06-22 | 2019-05-21 | 北京嘀嘀无限科技发展有限公司 | A kind of method and system for estimating arrival time |
| CN109726843A (en) * | 2017-10-30 | 2019-05-07 | 阿里巴巴集团控股有限公司 | The method, apparatus and terminal of allocation data prediction |
| WO2019100279A1 (en) * | 2017-11-23 | 2019-05-31 | Beijing Didi Infinity Technology And Development Co., Ltd. | System and method for estimating arrival time |
| WO2019109604A1 (en) * | 2017-12-05 | 2019-06-13 | Beijing Didi Infinity Technology And Development Co., Ltd. | Systems and methods for determining an estimated time of arrival for online to offline services |
| CN109543886A (en) * | 2018-11-06 | 2019-03-29 | 斑马网络技术有限公司 | Prediction technique, device, terminal and the storage medium of destination |
| CN109685276A (en) * | 2018-12-27 | 2019-04-26 | 拉扎斯网络科技（上海）有限公司 | Order processing method, apparatus, electronic equipment and computer readable storage medium |
Non-Patent Citations (2)
| Title |
|---|
| NIKLAS CHRISTOFFER PETERSEN等: ""Multi-output bus travel time prediction with convolutional LSTM neural network"", 《EXPERT SYSTEMS WITH APPLICATIONS》 * |
| Y DUAN等: ""Travel time prediction with LSTM neural network"", 《IEEE XPLORE》 * |
Cited By (19)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112862133A (en) * | 2019-11-12 | 2021-05-28 | 北京三快在线科技有限公司 | Order processing method and device, readable storage medium and electronic equipment |
| CN110910019A (en) * | 2019-11-22 | 2020-03-24 | 拉扎斯网络科技（上海）有限公司 | Data processing method and device, electronic equipment and computer readable storage medium |
| CN110910019B (en) * | 2019-11-22 | 2022-04-05 | 拉扎斯网络科技（上海）有限公司 | Data processing method and device, electronic equipment and computer readable storage medium |
| CN112907011A (en) * | 2019-12-04 | 2021-06-04 | 北京三快在线科技有限公司 | Distribution scheduling method, device, storage medium and electronic equipment |
| CN113222487A (en) * | 2020-01-21 | 2021-08-06 | 北京三快在线科技有限公司 | Scheduling path generation method, device, storage medium and electronic equipment |
| CN113222487B (en) * | 2020-01-21 | 2023-04-18 | 北京三快在线科技有限公司 | Scheduling path generation method, device, storage medium and electronic equipment |
| CN111652477A (en) * | 2020-05-19 | 2020-09-11 | 拉扎斯网络科技（上海）有限公司 | Order processing and similarity calculation model obtaining method and device and electronic equipment |
| CN111652477B (en) * | 2020-05-19 | 2024-01-23 | 拉扎斯网络科技（上海）有限公司 | Order processing and similarity calculation model obtaining method and device and electronic equipment |
| CN112036697A (en) * | 2020-07-28 | 2020-12-04 | 拉扎斯网络科技（上海）有限公司 | Task allocation method and device, readable storage medium and electronic equipment |
| CN112541610A (en) * | 2020-08-13 | 2021-03-23 | 深圳优地科技有限公司 | Robot control method, device, electronic device and storage medium |
| CN112541610B (en) * | 2020-08-13 | 2022-10-21 | 深圳优地科技有限公司 | Robot control method, device, electronic device and storage medium |
| CN114330797A (en) * | 2020-09-27 | 2022-04-12 | 北京三快在线科技有限公司 | Distribution time length prediction method and device, storage medium and electronic equipment |
| CN112258131B (en) * | 2020-11-12 | 2021-08-24 | 拉扎斯网络科技（上海）有限公司 | Path prediction network training and order processing method and device |
| CN112258131A (en) * | 2020-11-12 | 2021-01-22 | 拉扎斯网络科技（上海）有限公司 | Path prediction network training and order processing method and device |
| CN112837128A (en) * | 2021-02-19 | 2021-05-25 | 拉扎斯网络科技（上海）有限公司 | Order assignment method and device, computer equipment and computer readable storage medium |
| CN113222202A (en) * | 2021-06-01 | 2021-08-06 | 携程旅游网络技术（上海）有限公司 | Reservation vehicle dispatching method, reservation vehicle dispatching system, reservation vehicle dispatching equipment and reservation vehicle dispatching medium |
| CN113642603A (en) * | 2021-07-05 | 2021-11-12 | 北京三快在线科技有限公司 | Data matching method and device, storage medium and electronic equipment |
| CN115409452A (en) * | 2022-10-27 | 2022-11-29 | 浙江口碑网络技术有限公司 | Distribution information processing method, device, system, equipment and readable storage medium |
| CN115409452B (en) * | 2022-10-27 | 2024-02-23 | 浙江口碑网络技术有限公司 | Distribution information processing method, device, system, equipment and readable storage medium |
Also Published As
| Publication number | Publication date |
|---|---|
| CN110414731B (en) | 2021-02-02 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110414731A (en) | Method, apparatus, computer readable storage medium and the electronic equipment of Order splitting | |
| CN108364085B (en) | Takeout delivery time prediction method and device | |
| CN110070399A (en) | A kind of discount coupon method for pushing and device | |
| CN110231044A (en) | A kind of paths planning method and device | |
| CN108491951A (en) | A kind of prediction technique, device and electronic equipment for taking out distribution time | |
| CN107093096A (en) | A kind of Traffic prediction method and device | |
| CN107092973B (en) | The Forecasting Methodology and device of a kind of portfolio | |
| WO2020050873A1 (en) | System and method for ride order dispatching and vehicle repositioning | |
| CN110163432A (en) | A kind of method and device of order identification | |
| CN108537568A (en) | A kind of information recommendation method and device | |
| CN110400012A (en) | A kind of method and device of determining Distribution path | |
| CN109214739A (en) | order processing method and device | |
| CN109117824A (en) | Management method, device, electronic equipment and the storage medium of commodity | |
| CN109002949A (en) | A kind of method and device of air control strategy configuration and business air control | |
| CN110378726A (en) | A kind of recommended method of target user, system and electronic equipment | |
| CN107742245A (en) | A kind of merchant information recommends method, apparatus and equipment | |
| CN110032698A (en) | Information display method and device, information processing method and device | |
| CN109214731A (en) | Method and device for distributing logistics orders and computer system | |
| CN109428929A (en) | Determination method, server and the user client of the location information of target object | |
| CN110443541A (en) | A kind of pressure form processing method and device | |
| CN110148013A (en) | A kind of user tag distribution forecasting method, apparatus and system | |
| CN110766513A (en) | Information sorting method and device, electronic equipment and readable storage medium | |
| CN106469332A (en) | A kind of data processing method and device | |
| CN108023834A (en) | A kind of cloud resource auto-allocation method and device | |
| CN108920183A (en) | A kind of operational decision making method, device and equipment |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |