CN109598430A

CN109598430A - Dispense range generation method, device, electronic equipment and storage medium

Info

Publication number: CN109598430A
Application number: CN201811434820.6A
Authority: CN
Inventors: 汤浅伟; 李淳敏
Original assignee: Lazhasi Network Technology Shanghai Co Ltd
Current assignee: Rajax Network Technology Co Ltd; Lazhasi Network Technology Shanghai Co Ltd
Priority date: 2018-11-28
Filing date: 2018-11-28
Publication date: 2019-04-09
Anticipated expiration: 2038-11-28
Also published as: CN109598430B

Abstract

The present embodiments relate to a kind of dispatching range generation methods, comprising: obtains the order characteristic parameter that unit is respectively dispensed in the corresponding first dispatching range of target object, wherein target object is located at the initial dispatching unit in the first dispatching range；Centered on initially dispensing unit, expansion dispatching unit is obtained from dispatching unit according to dispatching range model and the order characteristic parameter for dispensing unit；The second dispatching range of target object is generated according to initial dispatching unit and expansion dispatching unit.The present invention passes through building dispatching range model, automation generates the intelligent distribution range of target object optimization, the method of dispatching range is confirmed instead of traditional geography information with around personal experience, common sense, target object and previous order analyzing, it is time saving and energy saving, improve dispatching efficiency.

Description

Dispense range generation method, device, electronic equipment and storage medium

Technical field

The present invention relates to field of communication technology, in particular to a kind of dispatching range generation method, device, electronic equipment and deposit Storage media.

Background technique

Dispatching range is very important for businessman for taking out, for dispensing model for the businessman of different categories The requirement enclosed is different, and the mode for determining that businessman dispenses range at present mainly includes according to each dimension in businessman's attribute information Numerical value carries out clustering to the vector of generation to generate vector, delimits the corresponding dispatching range of businessman.Although the above method The dispatching range of businessman can be determined according to the grid of delimitation, but do not illustrate how enterprising in existing dispatching scope Row optimization and adjustment.

Summary of the invention

Embodiment of the present invention is designed to provide a kind of dispatching range generation method, device, electronic equipment and storage Medium automatically delimit out one for target object and preferably dispenses range, reduce artificial, raising dispatching efficiency.

In order to solve the above technical problems, embodiments of the present invention provide a kind of dispatching range generation method, comprising: obtain Take the order characteristic parameter that unit is respectively dispensed in the corresponding first dispatching range of target object, wherein target object is located at first Dispense the initial dispatching unit in range；Centered on initially dispensing unit, according to dispatching range model and ordering for unit is dispensed Single characteristic parameter obtains expansion dispatching unit from dispatching unit；Target is generated according to initial dispatching unit and expansion dispatching unit Second dispatching range of object.

Embodiments of the present invention additionally provide a kind of generating means for dispensing range, comprising: data acquisition module obtains The order characteristic parameter of unit is respectively dispensed in the corresponding first dispatching range of target object, wherein target object is located at first and matches Send the initial dispatching unit in range；Mesh expansion module according to dispatching range model and is matched centered on initially dispensing unit The order characteristic parameter of unit is sent to obtain expansion dispatching unit from dispatching unit；Horizon generation module, it is single according to initial dispatching Member and expansion dispatching unit generate the second dispatching range of target object.

Embodiment of the present invention also discloses a kind of electronic equipment, including at least one processor；With at least one processing The memory of device communication connection；And respectively with processor and memory be communication connection communication component, communication component exists Data are sended and received under the control of processor；Wherein, memory is stored with the instruction that can be executed by least one processor, refers to It enables and being executed by least one processor to realize: obtaining the order for respectively dispensing unit in the corresponding first dispatching range of target object Characteristic parameter, wherein target object is located at the initial dispatching unit in the first dispatching range；Centered on initially dispensing unit, Expansion dispatching unit is obtained from dispatching unit according to dispatching range model and the order characteristic parameter for dispensing unit；According to initial It dispenses unit and expansion dispatching unit generates the second dispatching range of target object.

Embodiment of the present invention also discloses a kind of non-volatile memory medium, for storing computer-readable program, counts Calculation machine readable program is for executing dispatching range generation method as above for computer.

In terms of existing technologies, the main distinction and its effect are embodiment of the present invention: the present invention passes through building Range model is dispensed, it, can be automatic when inputting the data information of target object by constantly being trained to dispatching range model The intelligent distribution range that metaplasia optimizes at target object, around traditional experience, common sense, target object with individual Geography information and previous order analyzing confirm the method for dispatching range, compare conventional method, the present invention can be achieved automatic Metaplasia reduces manually at order intelligent distribution range, time saving and energy saving, and improves dispatching efficiency.

In addition, the method for embodiment of the present invention further include: building dispatching range model, dispatching range model includes state Value function, state value function is to dispense the order characteristic parameter of unit as parameter；According to machine learning algorithm, with target object institute The History Order characteristic parameter of object is trained dispatching range model in commercial circle.

In addition, machine learning algorithm is nitrification enhancement, centered on initially dispensing unit, according to dispatching range model Expansion dispatching unit is obtained from dispatching unit specifically: input in state value function each in the dispatching range of target object first The order characteristic parameter for dispensing unit obtains the score value of dispatching unit；According to the score value of dispatching range model and dispatching unit, obtain To the good working condition value matrix of the dispatching unit in the first dispatching range；Centered on initially dispensing unit, according to good working condition Value matrix judges dispatching unit, obtains expansion dispatching unit.Intelligent distribution range is generated by nitrification enhancement, It can be suitable for the simple scenario that the initial dispatching range of target object determines, it is accurate to calculate, the intelligence optimized the most Dispense range.

In addition, obtaining dispensing the complete of unit within the scope of the first dispatching according to the score value of dispatching range model and dispatching unit Whole state value matrix, specifically: according to the score value of dispatching range model and initial dispatching unit, obtain original state value matrix； It is the second dispatching unit set initially to dispense unit and the one or more dispatching units adjacent with initial dispatching unit, according to Dispense range model and second dispatching unit set dispatching unit score value and, obtain the second state value matrix；Match with second Sending unit set and the one or more dispatching units adjacent with the second dispatching unit set is that third dispenses unit set, according to Dispense range model and third dispatching unit set dispatching unit score value and, obtain third state value matrix；And so on, Until the dispatching unit of traversal the first dispatching range, good working condition value matrix is obtained.To gather as state, with the score value of set With as value of feedback, eventually form good working condition value matrix, convenient for by nitrification enhancement to initial dispatching range model into Row training.

In addition, being judged according to good working condition value matrix dispatching unit, being expanded centered on initially dispensing unit Dispatching unit is opened, specifically: score value and highest dispatching unit set are inquired in good working condition value matrix, by score value and highest Dispatching unit set in initial dispatching unit except dispatching unit, as expansion dispense unit.Choose score value and highest Dispatching unit set as intelligent distribution range, optimal dispatching range model can be obtained by training.

In addition, machine learning algorithm be depth practise algorithm, centered on initially dispensing unit, according to dispatching range model from Expansion dispatching unit order characteristic parameter is obtained in dispatching unit, specifically: target object first is inputted in state value function The order characteristic parameter that unit is respectively dispensed in dispatching range, obtains the score value of dispatching unit；Centered on initially dispensing unit, root Dispatching unit is judged according to the score value of dispatching unit, obtains expansion dispatching unit.Intelligence is generated by a little algorithms of depth Range can be dispensed, the hypodynamic complex application context of calculation of nitrification enhancement can be suitable for, to obtain optimal intelligence Dispense range.

In addition, being judged according to the score value of each dispatching unit dispatching unit, being obtained centered on initially dispensing unit Expansion dispatching unit, specifically: choose the highest dispatching unit conduct of score value in the dispatching unit adjacent with initial dispatching unit First expansion dispatching unit, initial dispatching unit and the first expansion dispatching unit collectively form the first expansion range；It chooses and the The highest dispatching unit of score value is as the second expansion dispatching unit, the first expansion range in the adjacent dispatching unit of one expansion range The second expansion range is collectively formed with the second expansion dispatching unit；And so on, the requirement until meeting the second dispatching range obtains To the n-th expansion dispatching unit and the n-th expansion range, wherein n is positive integer；By the first expansion dispatching unit, the second expansion dispatching The expansion of unit n-th dispatching unit is as expansion dispatching unit.When expanding outwardly, it is single that the highest dispatching of score value is chosen every time Member is matched as expansion dispatching unit convenient for obtaining the highest dispatching unit set of total score so as to obtain optimal intelligence Send range.

In addition, if obtained (n-a) expansion dispatching unit includes multiple, the highest n-th expansion range of selection total score As the dispatching range of target object, wherein a is the positive integer greater than 1 and less than n.

In addition, meeting the requirement of the second dispatching range specifically: the area of the second dispatching range is no more than preset area Maximum distance between threshold value, or initial dispatching unit and the expansion dispatching unit of outermost is no more than preset apart from threshold Value.The purpose of given threshold be in order to enable dispatching range extension be automatically stopped in suitable area, avoid infinitely expanding Calculation power deficiency problem caused by.

In addition, the order characteristic parameter of dispatching unit is obtained by following steps: extracting the first dispatching model in the database The order data enclosed, order data include the POI interface of the wide table of waybill, the wide table of order, the wide table of target object, electronic map；It is right Order data is screened to obtain the order characteristic parameter of each dispatching unit within the scope of the first dispatching.Order characteristic parameter is pair The most influential allocation datas such as the dispatching quality or dispatching efficiency of target object, are counted according to order characteristic parameter It calculates, is conducive to obtain optimal target object intelligent distribution range.

In addition, dispatching range generation method further include: obtain include predetermined keyword geography fence；It is enclosed according to geography Column judges dispatching unit in the second dispatching range, obtains dispensing unit in vain, wherein invalid subregion unit is located at ground It manages in fence or including geography fence；Invalid dispatching unit is deleted in the second dispatching range.It avoids being added in dispatching range Inactive area influences to dispense efficiency.

Detailed description of the invention

One or more embodiments are illustrated by the picture in corresponding attached drawing, these exemplary theorys The bright restriction not constituted to embodiment, the element in attached drawing with same reference numbers label are expressed as similar element, remove Non- to have special statement, composition does not limit the figure in attached drawing.

Fig. 1 is the flow chart of one of first embodiment dispatching range generation method according to the present invention；

Fig. 2-1 is the flow chart of one of second embodiment dispatching range generation method according to the present invention；

Fig. 2-2 is that the dispatching range of some object K1 in the commercial circle M in second embodiment according to the present invention generates signal Figure；

Fig. 2-3 is that the expansion dispatching unit according to the present invention in second embodiment obtains flow chart；

Fig. 2-4 is that the good working condition value matrix in second embodiment obtains flow chart according to the present invention；

Fig. 2-5 is that the dispatching range of the target object in second embodiment according to the present invention generates schematic diagram；

Fig. 3-1 is that one of third embodiment dispenses range generation method flow chart according to the present invention；

Fig. 3-2 is that the dispatching range of some object K1 in the commercial circle M in third embodiment according to the present invention generates signal Figure；

Fig. 3-3 is that the expansion dispatching unit according to the present invention in third embodiment obtains flow chart；

Fig. 3-4 is that the dispatching range of the target object in third embodiment according to the present invention generates schematic diagram；

Fig. 4-1 is that one of the 4th embodiment dispenses range generation method flow chart according to the present invention；

Fig. 4-2 is the generation method intelligent distribution model generated of the dispatching range in the 4th embodiment according to the present invention Enclose schematic diagram；

Fig. 5-1 is the generating means schematic diagram of the dispatching range in the 5th embodiment according to the present invention；

Fig. 5-2 is the mesh expansion module diagram in the 5th embodiment according to the present invention；

Fig. 5-3 is another mesh expansion module diagram in the 5th embodiment according to the present invention；

Fig. 6 is the electronic devices structure schematic diagram in sixth embodiment according to the present invention.

Specific embodiment

In order to make the object, technical scheme and advantages of the embodiment of the invention clearer, below in conjunction with attached drawing to the present invention Each embodiment be explained in detail.However, it will be understood by those skilled in the art that in each embodiment party of the present invention In formula, in order to make the reader understand this application better, many technical details are proposed.But even if without these technical details And various changes and modifications based on the following respective embodiments, the application technical solution claimed also may be implemented.

The first embodiment of the present invention is related to a kind of dispatching range generation method, present embodiment can be applied in terminal Side is such as applied in mobile phone, the terminal devices such as tablet computer, can also be applied in the server of network side.Present embodiment Embodiment dispatching range generation method as shown in Figure 1, this method comprises:

Step 103, the order characteristic parameter that unit is respectively dispensed in the corresponding first dispatching range of target object is obtained, In, target object is located at the initial dispatching unit in the first dispatching range；

Step 104, centered on initially dispensing unit, according to the order characteristic parameter of dispatching range model and dispatching unit Expansion dispatching unit is obtained from dispatching unit；

Step 105, the second dispatching range of target object is generated according to initial dispatching unit and expansion dispatching unit.

Specifically, we are formed centered on target object position to external radiation certain distance in step 103 The first dispatching range (i.e. initially dispatching range) of the target object, goal object can be dispatching website or trade company (such as dining room, sweets shop, pharmacy) etc.；Such as we are centered on the longitude and latitude of a certain trade company, to external radiation certain distance shape At a geographic range, this range is screened, such as rejecting highway, bridge etc. can not be sent to a little as order Region after formed a regular shape or irregular shape geographic range, the initial dispatching range as the trade company (the i.e. first dispatching range).Then the first dispatching range being divided into several dispatching units, (each dispatching unit can be the Some shape of the same size that one region averagely divides, can also be according to obtained by certain regular partition, such as according to geographical position Set and divided), it should be strongly noted that respectively dispatching unit can be grid, such as 140 meters × 140 meters of quadrangle lattice The cell of son or other shapes and size.

In present embodiment, order characteristic parameter is obtained according to specific reference to following methods: extracting first in the database Dispense the order data of range；Order data is screened, the order characteristic parameter of each dispatching unit is obtained.

It is illustrated by trade company of target object, order data related with the trade company in the past is had collected in database, In database extract first dispatching range order data, order data include the wide table of waybill (by trade company's coordinate, the pick-up time, Food delivery time, the database table for going out the compositions such as meal time, wide table refer to the more database table of field, typically refer to business master Inscribe the database table of relevant index, dimension, Attribute Association together), the wide table of order is (by the dining room class of History Order Not, the database table of the compositions such as canteen classification), the wide table of the trade company (number being made of canteen category, trade company's coordinate, businessman's label etc. According to library table), the POI interface of electronic map (Apache POI is the open source projects of Apache Software Foundation, POI provide API to The function that java applet is read and write to the document of Microsoft Office format)；Order data is arranged, is concluded After screening, the order characteristic parameter of each dispatching unit within the scope of the first dispatching is obtained.Order characteristic parameter can be pair The influential characteristic of the dispatching quality and efficiency of order, is calculated convenient for subsequent according to order characteristic parameter, with Obtain optimal dispatching range.

In practical applications, match in the orders such as POI interface to the wide table of waybill, the wide table of order, the wide table of trade company, electronic map Sending the order characteristic parameter obtained after data processing may include the ID of dispatching unit where trade company ID, waybill, trade company to dispatching (jockey arrived at the destination beginning upstairs or entered small the walking distance of unit, averagely dispatching duration, average AOI food delivery time Find the time of corresponding resident in area), average out the meal time, efficiently accomplish list, overtime trouble ticket etc. in overtime trouble ticket and preset duration.In addition For selling industry, wherein efficiently accomplish and singly refer to that jockey successfully send canteen to the order of user；Overtime trouble ticket is jockey not in regulation It is interior to send canteen to user, the order of caused food delivery time-out；Overtime trouble ticket refers in preset duration in preset duration, Qi Shouchao When food delivery order.

Further, the order for being screened to obtain each dispatching unit within the scope of the first dispatching to order allocation data is special Parameter, in particular to the ID by dispatching unit where the trade company ID and waybill in order allocation data are levied, filters out and falls into often The order of the trade company of a dispatching unit (grid) then as needed screens related order data as order characteristic parameter.

The expansion of present embodiment dispenses unit, refers to obtain the second dispatching model of target object more optimized It encloses, centered on the initial dispatching unit where target object, initial dispatching unit will constantly be added, to form the second dispatching model The target object first enclosed dispenses other dispatching units in range.

In the step 104 of present embodiment, centered on initially dispensing unit, according to dispatching range model from dispatching unit In obtain expansion dispatching unit, as shown in Figure 1, dispatching range model therein is obtained especially by following methods:

Step 101, the initial dispatching range model of building, the initial range model that dispenses includes state value function, state value letter Number is to dispense the order characteristic parameter of unit as parameter；

Step 102, according to machine learning algorithm, with the History Order characteristic parameter pair of the object of commercial circle where target object Initial dispatching range model is trained, and obtains optimization dispatching range model.

Here machine learning algorithm includes intensified learning or deep learning algorithm, intensified learning be a kind of self-recision and The mechanism of Machine Learning of feedback allows machine to possess the ability of self-teaching and self thought, such as Q-learning algorithm and intelligence It can body algorithm；And deep learning is concern is primarily with the extraction of feature, is reached by a large amount of labeled data come training pattern Accurately feature extraction and identification.

In the prior art, usual trade company dispatching range be according to the geography information around personal experience, common sense, trade company with And previous order analyzing, come what is divided, the dispatching range divided in this way relies primarily on human subjective's consciousness, it is likely that can will dive Client be divided to dispatching range outside and miss consumer, cause the loss of businessman, therefore in order to solve this problem, be trade company There is provided more excellent and more accurate dispatching range, the present embodiment building and optimization dispatching range model, and by a geographical model greatly It encloses and decomposes each small dispatching unit, the order feature by obtaining the dispatching unit that target object first dispenses range is joined Number, and dispensed according to the order characteristic parameter of trained dispatching range model and dispatching unit to obtain the expansion of target object Unit is gradually added into the expansions dispatching units for more meeting expanding policy, according to initially matching initially to dispense unit as starting point Unit and expansion dispatching unit is sent to generate the second dispatching range jointly, to automatically obtain the target object intelligent distribution model of optimization It encloses.

It should be noted that the dispatching range model of present embodiment can be depending on the demand of target object, specifically It is embodied on the different weighted value of each parameter of the state value function in dispatching range model, such as asked for more when certain trade company When order, the quantity on order parameter in state value function is just endowed biggish weighted value；If certain trade company concerns higher Dispatching quality, quantity on order is indifferent to, then the overtime trouble ticket number parameter in state value function is just endowed biggish weight Value.

Second embodiment of the invention discloses a kind of dispatching range generation method, as shown in Fig. 2-1, comprising:

Step 201, the order characteristic parameter that unit is respectively dispensed in the corresponding first dispatching range of target object is obtained, In, target object is located at the initial dispatching unit in the first dispatching range；

Step 202, target object first is inputted in state value function dispense the order feature that unit is respectively dispensed in range Parameter obtains the score value of dispatching unit；

Step 203, according to the score value of dispatching range model and dispatching unit, the dispatching unit within the scope of the first dispatching is obtained Good working condition value matrix；

Step 204, centered on initially dispensing unit, dispatching unit is judged according to good working condition value matrix, is obtained Unit is dispensed to expansion；

Step 205, the second dispatching range of target object is generated according to initial dispatching unit and expansion dispatching unit.

In present embodiment, first according to nitrification enhancement, with one of all objects in commercial circle where target object The History Order characteristic parameter of historical time section is trained the dispatching range model of building.Intensified learning is artificial intelligence The technology of field rate of exchange maturation, present embodiment are briefly described by taking Q-Learning as an example.

The target of Q-Learning is to learn a kind of strategy, tells when agency will take anything to take action.It Not the problem of not needing environmental model, can handle random transition and reward, without being adjusted.For any limited Markov Decision process (FMDP), Q-Learning have found that a kind of strategy is optimal, because since current state, all consecutive steps The desired value always rewarded be achievable maximum value.In the case where given unlimited exploration time and part randomized policy, Q- Learning can determine best movement selection strategy for any given FMDP." Q " is used to name return for providing reinforcing Reward function, and it may be said that represent " quality " of movement taken under given state.Q-learning includes one Agency, a series of states and a series of actions.By executing a movement, the state of agency can change S1 → S2 (Aa), executing this movement can make agency obtain a reward numerically, be calculated in Q-learning using feedback function The reward value of (state acts), in brief, Q-learning obtains change process by Status Change again and again In preferably (mass value highest) a series of actions, and by the score relation record of state and movement in Q-Table.

Specifically, using Q-learning nitrification enhancement, present embodiment is in the commercial circle M where target object KI For some object K1, as shown in Fig. 2-2, is inputted in the commercial circle M first in state value function and respectively dispense unit in a history The History Order characteristic parameter of period obtains the score value that unit is respectively dispensed in the commercial circle M；It is single with the initial dispatching where object K1 First A1 is as original state, value of feedback of the score value of the initial dispatching unit A1 where object K1 as original state, according to first The value of feedback of beginning state and original state obtains original state value matrix；Initially to dispense unit A1 and dispense unit with initial The second dispatching unit set that 4 A1 adjacent dispatching unit B 1-B4 are respectively constituted is the second state, i.e., the second dispatching unit collection Conjunction can have (A1, B1), (A1, B2), (A1, B3), (A1, B4) four dispatching unit sets, with this four dispatching unit sets Respective score value and value of feedback as the second state, obtain the second state value according to the value of feedback of the second state and the second state Matrix；With four dispatching unit sets (A1, B1), (A1, B2), (A1, B3), (A1, B4), and with this four dispatching unit collection Closing the third dispatching unit set that eight adjacent dispatching unit C1-C8 are respectively constituted is the third state, i.e. third dispenses unit Set can have (A1, B1, C1), (A1, B1, C2), (A1, B1, C8), (A1, B2, C3), (A1, B2, C2), (A1, B2, C3), (A1, B2, C4), (A1, B3, C4), (A1, B3, C5), (A1, B3, C6), (A1, B4, C6), (A1, B4, C7), (A1, B4, C8) This 12 dispatching unit sets, using this 12 respective score values of dispatching unit set and as the value of feedback of the third state, The second state value matrix is obtained according to the value of feedback of the third state and the third state；And so on, the whole of the guidance traversal commercial circle M Unit is dispensed, the corresponding good working condition matrix a of object K1 is finally obtained.

According to Q-learning nitrification enhancement, inquiry point in the above-mentioned good working condition matrix a corresponding to object K1 It is worth highest dispatching unit set, such as dispenses unit set (A1, B3, C6, D5Z26) corresponding score value and be 198, be highest numerical value in state value of feedback all in good working condition matrix a, then will just dispense unit set (A1, B3, C6, D5Z26) optimal dispatching range of this state as object A1, dispatching unit set (A1, B3, C6, D5Z26) included in each dispatching unit B 3, C6, D5Z26 i.e. be used as the expansion of target object K1 to dispense Unit.

Then by taking another object K2 in the commercial circle M where target object as an example, object is obtained also according to the above method The expansion of K2 dispenses unit, and obtains the optimal dispatching range of object K2.In machine-learning process, every time according to same Rule obtain the expansion dispatching unit and optimal dispatching range of an object of the commercial circle M, while abandoning other rules, from And all dispatching units of target object place commercial circle are utilized in the History Order characteristic parameter of a historical time section, to dispatching Range model carries out continuous optimization training, then when the order feature ginseng for inputting target object again in dispatching range model When number, the corresponding expansion dispatching unit of target object and optimal dispatching range can be generated according to same rule.

I.e. training complete after, centered on the initial dispatching unit where target object, according to dispatching range model from Expansion dispatching unit is obtained in dispatching unit, as Figure 2-3, specifically includes the following steps:

Step 2021, target object first is inputted in state value function dispense the order feature that unit is respectively dispensed in range Parameter obtains the score value of dispatching unit；

Step 2022, according to the score value of dispatching range model and dispatching unit, the dispatching list within the scope of the first dispatching is obtained The good working condition value matrix of member；

Step 2023, centered on initially dispensing unit, dispatching unit is judged according to good working condition value matrix, is obtained Unit is dispensed to expansion.

Further, the score value according to dispatching range model and dispatching unit of step 2022, obtains the first dispatching range The good working condition value matrix of interior dispatching unit specifically includes as in Figure 2-4:

Step 2022a obtains original state value matrix according to the score value of dispatching range model and initial dispatching unit；

Step 2022b, initially to dispense unit and the one or more dispatching units adjacent with initial dispatching unit are the Two dispatching unit sets, according to dispatching range model and second dispatching unit set dispatching unit score value and, obtain second State value matrix；

Step 2022c, with the second dispatching unit set and the one or more dispatchings adjacent with the second dispatching unit set Unit be third dispense unit set, according to dispatching range model and third dispatching unit set dispatching unit score value and, Obtain third state value matrix；

Step 2022d, and so on, until the dispatching unit of traversal the first dispatching range, obtain good working condition value matrix.

Such as each dispatching unit of the dispatching range of target object first is as seen in figs. 2-5, it is initial where target object Four dispatching units around dispatching unit P are respectively P1 (- 2), P2 (4), P3 (- 1), (digital representation in bracket should by P4 (1) Dispense the score value of unit), initial dispatching unit and adjacent each dispatching unit are respectively constituted by the second dispatching according to step 2022b Unit set, state value (it is each second dispatching unit set score value and) as shown in S1 in the following table 1, wherein the q1 table in S1 Show the one second dispatching unit set that initial dispatching unit P, dispatching unit P1 are constituted, (these three dispatchings are single for state value Member score value and) Q (S1, q1) be -2；Q2 in S1 indicate that initial dispatching unit P, dispatching unit P2 constituted another the Two dispatching unit sets, state value Q (S2, q2) they are 4, and so on, by initial dispatching unit and adjacent four The state value of dispatching unit is included in table；Each second dispatching unit set joined P5 (3), P6 (15), P7 (6), P8 (- 3), (2) eight P9 (- 6), P10 (10), P11 (- 3), P12 adjacent dispatching units；Unit is dispensed by each second according to step 2022c Set and adjacent each dispatching unit respectively constitute third dispatching unit set, state (point of each third dispatching unit set Value and) as shown in S2 in the following table 1, wherein q1 in S2 indicates initial dispatching unit P, dispatching unit P1, dispatching unit P6 composition Third dispense unit set, state value Q (S2, q1) be 13；Q2 in S2 indicate initial dispatching unit P, dispatching unit P1, Another third dispatching unit set that dispatching unit P7 is constituted, state value Q (S2, q2) is 4, and so on, The state value of second dispatching unit set and eight adjacent dispatching units is included in table；And so on, until Meet the requirement of dispatching range, obtains the n-th subregion set, state value (score value of each n-th subregion set) such as the following table 1 Shown in middle Sn, wherein n is positive integer.

It can be seen in table 1 that by initially dispensing unit P and dispensing unit P1, the third dispatching that dispatching unit P6 is collectively formed The score value and Q (S2, q1) of unit set are maximum, therefore dispatching unit P1, dispatching unit P6 are added as expansion dispatching unit Enter initial dispatching unit, finally matches what initial dispatching unit P, expansion dispatching unit P1, expansion dispatching unit P6 were collectively formed Send intelligent distribution range of the unit set as target object.

Table 1

	q1	q2	q3	q4	q5	q6	···
								S1	-2	4	-1	1
S2	13	4	2	-3	-1	1
								S3	4	-1	5	0	-2	3
S4	-3	3	-7	9	-4	0
								S5	2	10	1	-2	3	5
···	···	···	···	···	···	···	···
								Sn	N1	N2	N3	N4	N5	N6	···

It is noted that in the present embodiment, order characteristic parameter may include excellent with super in odd number, preset duration When order numbers, the outer overtime order numbers of preset duration.Specifically, excellent to match single order to be normally sent to, excellent more with odd number, this is matched The order of unit is sent to dispense quality higher；Overtime order numbers are more in preset duration, and the order dispatching efficiency of the dispatching unit is got over It is low；The outer overtime order numbers of preset duration are fewer, and the order dispatching efficiency of the dispatching unit is higher, these three order characteristic parameters are made For the factor that target object is concerned about the most, the dispatching quality dispensed to order, dispatching efficiency etc. have main influence power, therefore When expanding order dispatching range, by each dispatching unit carried it is excellent with odd number, it is overtime order numbers in preset duration, pre- If outer overtime order numbers of duration etc. train dispatching range model as training parameter.

Further, the state value function in present embodiment is feedback function, specifically, by will it is excellent with it is singular, In preset duration in outer this 3 kinds of order characteristic parameter input feedback functions of overtime order numbers of overtime order numbers, preset duration, obtain The score value of corresponding dispatching unit records the score value of each dispatching unit.

Present embodiment is trained initial dispatching range model by nitrification enhancement, to gather as state, Using the score value of set and as value of feedback, good working condition value matrix is eventually formed, score value and highest dispatching unit set are chosen As intelligent distribution range, optimal dispatching range can be obtained by training.

Third embodiment of the invention discloses a kind of dispatching range generation method, as shown in figure 3-1, comprising:

Step 301, the order characteristic parameter that unit is respectively dispensed in the corresponding first dispatching range of target object is obtained, In, target object is located at the initial dispatching unit in the first dispatching range；

Step 302, target object first is inputted in state value function dispense the order feature that unit is respectively dispensed in range Parameter obtains the score value of dispatching unit；

Step 303, centered on initially dispensing unit, dispatching unit is judged according to the score value of dispatching unit, is obtained Unit is dispensed to expansion；

Step 304, the second dispatching range of target object is generated according to initial dispatching unit and expansion dispatching unit.

In present embodiment, first according to deep learning algorithm, with one of all objects in commercial circle where target object The History Order characteristic parameter of historical time section is trained the dispatching range model of building.It is illustrated below to train Journey:

Such as equally by taking some object K1 in the commercial circle M where target object as an example, object K1 is located at initial dispatching unit In A1, as shown in figure 3-2, is inputted first in state value function and respectively dispense unit going through in historical time section in the commercial circle M History order characteristic parameter obtains the score value that unit is respectively dispensed in the commercial circle M；Access centered on initially dispensing unit A1, choose with The highest dispatching unit (such as B2) of score value is as the first expansion in four initial dispatching unit A1 adjacent dispatching unit B 1-B4 Unit is dispensed, by initially dispensing unit A1, the first expansion dispatching unit B 2 collectively forms the first expansion range, as shown in the figure to match Send unit set (A1, B2)；Then six the dispatching unit Bs 1, C2, C3, C4, B3 and B4 adjacent with the first expansion range are chosen Among the highest dispatching unit (such as C4) of score value as second expansion dispense unit, by first expansion range and second expansion match Unit C4 is sent to collectively form the second expansion range, i.e., dispatching unit set (A1, B2, C4) shown in figure；Choose again later with The highest dispatching unit of score value expands dispatching unit as third in seven adjacent dispatching units of second expansion range, with such Push away, until meet second dispatching range requirement, obtain the n-th expansion dispatching unit (such as Zm), by (n-1) expansion range and N-th expansion dispatching unit collectively forms the n-th expansion range (n is positive integer), then, the n-th expansion range is just as satisfaction second The corresponding optimal intelligent distribution range of object K1 of area requirement is dispensed, is initially dispensed included in the n-th expansion range single Other dispatching unit B 2, C4Zm except first A1 are just as expansion dispatching unit.

I.e. training complete after, centered on the initial dispatching unit where target object, according to dispatching range model from Expansion dispatching unit is obtained in dispatching unit, as shown in Fig. 3-3, specifically includes the following steps:

Step 3031, it chooses and initially dispenses in the adjacent dispatching unit of unit the highest dispatching unit of score value as first Expansion dispatching unit, initial dispatching unit and the first expansion dispatching unit collectively form the first expansion range；

Step 3032, it chooses and expands in the adjacent dispatching unit of range the highest dispatching unit of score value as second with first Expansion dispatching unit, the first expansion range and the second expansion dispatching unit collectively form the second expansion range；

Step 3033, and so on, the requirement until meeting the second dispatching range obtains the n-th expansion dispatching unit and n-th Range is expanded, wherein n is positive integer；

Step 3034, using first expansion dispatching unit, second expansion dispatching unit n-th expansion dispatching unit as Expansion dispatching unit.

It specifically, as shown in Figure 3-4, by taking target object B as an example, and is starting point with dispatching unit B 0 where target object B, The surrounding up and down four for comparing it dispenses the score value (non-identification division label in figure) of unit Bs 1, B2, B3, B4, selection point It is worth first expansion dispatching unit of the highest dispatching unit as target object B, it is assumed that the score value highest of dispatching unit B 3, therefore Dispatching unit B 3 is expanded as first and dispenses unit, then collectively forms first initially to dispense unit B 0 and dispatching unit B 3 Range is expanded, using the first expansion range as starting point, compares dispatching unit B 0 and dispenses adjacent around unit B 3 six and match Send the score value (non-identification division label in figure) of unit (B5, B6, B7, B1, B2, B4), it is assumed that the score value highest of dispatching unit B 7, Therefore second expansion dispatching unit of the unit B 7 as target object B will be dispensed, then initially to dispense unit B 0, dispatching unit B3, the region that is collectively formed of dispatching unit B 7 constitute the second expansion range, using the second expansion range as starting point, compare with this The score value ... of the adjacent multiple dispatching units of two expansion ranges, repeats above step, and so on, continuous adjustment and update Range is dispensed, so that dispatching range is in the state for constantly increasing and updating, the requirement until meeting the second dispatching range, with life At the optimal intelligent distribution range of target object B.

It should be further noted that if (n-a) expansion dispatching unit includes multiple, selection total score highest n-th Dispatching range of the range as target object is expanded, wherein a is the positive integer greater than 1 and less than n.Such as above-mentioned with target pair As in the preferred embodiment for B, if n=3, a=1, then n-a=2, it is assumed that the score value of B6 and B7 is identical, i.e., simultaneously Unit is dispensed to two second expansions, then needing to compare the total score of the third expansion range comprising B6 and B7, such as is wrapped Containing B6 third expansion range by with initially dispense unit B 0, dispatching unit B 3, dispatching unit B 6 and dispatching unit B 11 constituted, Its total score is 15；Third expansion range comprising B7 by initially to dispense unit B 0, dispatching unit B 3, dispatching and are matched at unit B 7 Unit B 12 is sent to constitute, total score 10, then choosing the second dispatching range that the former is target object B, i.e. selection dispatching is single First B3, dispatching unit B 6 and dispatching unit B 11 are as expansion dispatching unit.

Wherein, the requirement for meeting dispatching range can be the area for dispensing range no more than area threshold, or initially match The maximum distance of the expansion dispatching unit of unit and outermost is sent to be no more than distance threshold.The purpose of given threshold be in order to enable The extension of dispatching range is automatically stopped in suitable area, avoids dispatching is added in dispatching range that quality is low or inefficient Dispatching unit.

For example, the area threshold Z of setting is 10 square meters, when the gross area of the second dispatching range reaches 10 square meter, then stop Only extend；Or the distance threshold of setting is 5 kilometers, when the most long distance for the expansion dispatching unit for initially dispensing unit and outermost When from reaching 5 kilometers, it should stop extending.It in the present embodiment, is area threshold and apart from threshold respectively there are two preset thresholds Value according to the concrete condition of target object and can need to select corresponding preset threshold, to ensure to generate in practical applications Be the optimal dispatching range for meeting target object demand.

In the present embodiment, order characteristic parameter includes that blanket order amount, the order volume of the super preset duration of distribution time are (pre- If duration can be 30 minutes or 60 minutes), with other target objects of brand predict order volume.Specifically, in some scenarios, Blanket order amount is bigger, and trade company tends to select the dispatching unit further；The order volume of the super preset duration of distribution time is more, this is matched The order of unit is sent to dispense efficiency lower；More with other target objects of brand prediction order volume, trade company tends to select further The dispatching unit, these three order characteristic parameters match the quantity on order of prediction dispatching unit, order dispatching quality and order It send efficiency that there is main influence, therefore is accounted for when expansion dispenses range as preferred factor.

When present embodiment expands outwardly dispatching range again, the highest dispatching unit of score value is chosen every time as expansion dispatching Unit is then automatically stopped until being expanded to suitable range, so as to obtain optimal intelligent distribution range.

Four embodiment of the invention provides a kind of generation method of intelligent distribution range, as shown in Figure 4, comprising:

Step 401, the order characteristic parameter that unit is respectively dispensed in the corresponding first dispatching range of target object is obtained, In, target object is located at the initial dispatching unit in the first dispatching range；

Step 402, centered on initially dispensing unit, according to the order characteristic parameter of dispatching range model and dispatching unit Expansion dispatching unit is obtained from dispatching unit；

Step 403, the second dispatching range of target object is generated according to initial dispatching unit and expansion dispatching unit；

Step 404, the geography fence including predetermined keyword is obtained；

Step 405, dispatching unit in the second dispatching range is judged according to geography fence, obtains dispensing in vain single Member, wherein invalid subregion unit is located in geography fence or including geography fence；

Step 406, invalid dispatching unit is deleted in the second dispatching range.

In present embodiment, step 401- step 403 in foregoing embodiments to carry out explanation, and this will not be repeated here. In step 404, keyword can be rivers and lakes He Hai, scenic spot, park, cemetery, etc., and area above is not in order, Because of referred to herein as inactive area, judge whether first dispense in dispatching unit and target area in range each sub-regions unit Comprising rivers and lakes He Hai, scenic spot, park, cemetery, or judge each in dispatching unit and the target area in the first dispatching range Sub-regions unit whether be integrally located at rivers and lakes He Hai, scenic spot, park, within cemetery, if so, being picked in the second dispatching range Except the invalid subregion unit containing inactive area, intelligent distribution range as shown in the Fig. 4-2 is ultimately generated.

In present embodiment, invalid dispatching unit is deleted in the second dispatching range, by filtered second dispatching range The dispatching range of target object as generation avoids that inactive area is added in dispatching range, influences to dispense efficiency.

Fifth embodiment of the invention discloses a kind of generating means for dispensing range, as shown in fig. 5-1, comprising:

Data acquisition module 503 obtains the order feature that unit is respectively dispensed in the corresponding first dispatching range of target object Parameter, wherein target object is located at the initial dispatching unit in the first dispatching range；

Mesh expansion module 504, centered on initially dispensing unit, according to the order of dispatching range model and dispatching unit Characteristic parameter obtains expansion dispatching unit from dispatching unit；

Horizon generation module 505 generates the second dispatching of target object according to initial dispatching unit and expansion dispatching unit Range.

In addition, the generating means of the dispatching range of embodiment of the present invention further include:

Model construction module 501, building dispatching range model, dispatching range model includes state value function, state value letter Number is to dispense the order characteristic parameter of unit as parameter；

Model training module 502 is gone through according to machine learning algorithm with one of all objects in commercial circle where target object The History Order characteristic parameter of history period is trained dispatching range model.

Further, machine learning algorithm is nitrification enhancement, as before shown in 5-2, is had in mesh expansion module 504 Body includes:

Score value computing unit 5041 inputs in the dispatching range of target object first in state value function and respectively dispenses unit Order characteristic parameter obtains the score value of dispatching unit；

Matrix acquiring unit 5042 obtains within the scope of the first dispatching according to the score value of dispatching range model and dispatching unit Dispatching unit good working condition value matrix；

Expand judging unit 5043, centered on initially dispensing unit, according to good working condition value matrix to dispatching unit into Row judgement obtains expansion dispatching unit.

Intelligent distribution range is generated by nitrification enhancement in this way, the initial dispatching model of target object can be suitable for Determining simple scenario is enclosed, it is accurate to calculate, the intelligent distribution range optimized the most.

In addition, matrix acquiring unit 5042 obtains the first dispatching model according to the score value of dispatching range model and dispatching unit The good working condition value matrix of interior dispatching unit is enclosed, specifically: according to the score value of dispatching range model and initial dispatching unit, obtain Original state value matrix；Matched with initially dispensing unit and the one or more dispatching units adjacent with initial dispatching unit for second Send unit set, according to dispatching range model and second dispatching unit set dispatching unit score value and, obtain the second state Value matrix；Matched using the second dispatching unit set and the one or more dispatching units adjacent with the second dispatching unit set as third Send unit set, according to dispatching range model and third dispatching unit set dispatching unit score value and, obtain the third state Value matrix；And so on, until the dispatching unit of traversal the first dispatching range, obtain good working condition value matrix.Using gather as State eventually forms good working condition value matrix using the score value of set and as value of feedback, convenient for passing through nitrification enhancement to first Beginning dispatching range model is trained.

In addition, expansion judging unit 5043 is centered on initially dispensing unit, it is single to dispatching according to good working condition value matrix Member is judged, expansion dispatching unit is obtained, specifically: score value and highest dispatching unit are inquired in good working condition value matrix Set, it is single as expansion dispatching by the dispatching unit except the initial dispatching unit in score value and highest dispatching unit set Member.Score value and highest dispatching unit set are chosen as intelligent distribution range, optimal dispatching model can be obtained by training Boxing type.

In addition, machine learning algorithm is that depth habit algorithm can also wrap in mesh expansion module 504 as shown in Fig. 5-3 It includes:

Score value acquiring unit 5141 inputs in the dispatching range of target object first in state value function and respectively dispenses unit Order characteristic parameter obtains the score value of dispatching unit；

Expand acquiring unit 5142, centered on initially dispensing unit, according to dispatching unit score value to dispatching unit into Row judgement obtains expansion dispatching unit.

Intelligent distribution range is generated by a little algorithms of depth in this way, the calculation power of nitrification enhancement can be suitable for not The complex application context of foot, to obtain optimal intelligent distribution range.

In addition, acquiring unit 5142 is expanded centered on initially dispensing unit, according to the score value of each dispatching unit to dispatching Unit judged, expansion dispatching unit is obtained, specifically: choose in the dispatching unit adjacent with unit is initially dispensed score value most High dispatching unit collectively forms the first expansion as the first expansion dispatching unit, initial dispatching unit and the first expansion dispatching unit Open range；It is single as the second expansion dispatching to choose the highest dispatching unit of score value in the dispatching unit adjacent with the first expansion range Member, the first expansion range and the second expansion dispatching unit collectively form the second expansion range；And so on, match until meeting second The requirement for sending range obtains the n-th expansion dispatching unit and the n-th expansion range, and wherein n is positive integer；First expansion dispatching is single Member, the second expansion dispatching expansion of unit n-th dispatching unit are as expansion dispatching unit.When expanding outwardly, choose every time The highest dispatching unit of score value is as expansion dispatching unit, convenient for obtaining the highest dispatching unit set of total score, so as to Obtain optimal intelligent distribution range.

In addition, if (n-a) expansion dispatching unit obtained in expansion acquiring unit 5142 includes multiple, selection total score Dispatching range of the highest n-th expansion range as target object, wherein a is the positive integer greater than 1 and less than n.

In addition, meeting the requirement of the second dispatching range specifically: the area of the second dispatching range is no more than preset area Maximum distance between threshold value, or initial dispatching unit and the expansion dispatching unit of outermost is no more than preset apart from threshold Value.The purpose of given threshold be in order to enable dispatching range extension be automatically stopped in suitable area, avoid infinitely expanding Hypodynamic problem is calculated in operation caused by.

In addition, the order characteristic parameter of the dispatching unit in modules is obtained by following steps: taking out in the database The order data of the first dispatching range is taken, order data includes the wide table of waybill, the wide table of order, the wide table of target object, electronic map POI interface；Screened to obtain the order characteristic parameter of each dispatching unit within the scope of the first dispatching to order data.Order Characteristic parameter is the most influential allocation datas such as dispatching quality or dispatching efficiency to target object, according to order spy Parameter is levied to calculate, is conducive to obtain optimal target object intelligent distribution range.

In addition, the generating means of dispatching range further include:

Fence obtains module 506, obtains the geography fence including predetermined keyword；

Invalid judgment module 507 judges that it is invalid to obtain to dispatching unit in the second dispatching range according to geography fence Dispense unit, wherein invalid subregion unit is located in geography fence or including geography fence；

Invalid removing module 508 deletes invalid dispatching unit in the second dispatching range.By deleting rivers,lakes and seas, scape The logistics such as area, park, cemetery not can be carried out the inactive area of dispatching, advanced optimize order dispatching range.

Sixth embodiment of the invention is related to a kind of electronic equipment, and the electronic equipment of present embodiment is it may be said that terminal side is set Standby, such as mobile phone, the terminal devices such as tablet computer are also possible to the server of network side.

As shown in fig. 6, the electronic equipment: including at least a processor 601；And it is logical at least one processor 601 Believe the memory 602 of connection；And the communication component 603 with scanning means communication connection, communication component 603 is in processor 601 Control under send and receive data；Wherein, memory 602 is stored with the instruction that can be executed by least one processor 601, refers to It enables and being executed by least one processor 601 to realize:

Obtain the order characteristic parameter that unit is respectively dispensed in the corresponding first dispatching range of target object, wherein target pair As the initial dispatching unit being located in the first dispatching range；Centered on initially dispensing unit, according to dispatching range model and match The order characteristic parameter of unit is sent to obtain expansion dispatching unit from dispatching unit；It is single according to initial dispatching unit and expansion dispatching Member generates the second dispatching range of target object；

Obtain the geography fence including predetermined keyword；According to geography fence to each dispatching unit in the second dispatching range Judged with subregion unit, obtains invalid unit；Invalid unit is deleted in the second dispatching range.

Specifically, in Fig. 6 for being connected by bus.Memory 602 is deposited as a kind of non-volatile computer is readable Storage media can be used for storing non-volatile software program, non-volatile computer executable program and module.Processor 601 By non-volatile software program, instruction and the module of operation storage in the memory 602, thereby executing the various function of equipment It can apply and data processing, that is, realize the generation method of above-mentioned dispatching range.

Memory 602 may include storing program area and storage data area, wherein storing program area can store operation system Application program required for system, at least one function；It storage data area can the Save option list etc..In addition, memory 602 can be with It can also include nonvolatile memory, for example, at least disk memory, a flash memory including high-speed random access memory Device or other non-volatile solid state memory parts.In some embodiments, it includes relative to processing that memory 602 is optional The remotely located memory of device 601, these remote memories can pass through network connection to external equipment.The example of above-mentioned network Including but not limited to internet, intranet, local area network, mobile radio communication and combinations thereof.

One or more module stores in the memory 602, when being executed by one or more processor 601, holds The generation method of dispatching range in the above-mentioned any means embodiment of row.

The said goods can be performed the application embodiment provided by method, have the corresponding functional module of execution method and Beneficial effect, the not technical detail of detailed description in the present embodiment, reference can be made to method provided by the application embodiment.

In present embodiment, range model is dispensed by building, and constantly train to dispatching range model, when input mesh When marking the data information of object, the intelligent distribution range for generating target object optimization can be automated, instead of traditional with a The experience of people, common sense, the geography information around target object and previous order analyzing confirm the method for dispatching range, phase Than conventional method, the present invention can be achieved automation and generate order intelligent distribution range, reduce manually, time saving and energy saving, and improve Dispatching efficiency.

7th embodiment of the invention is related to a kind of non-volatile memory medium, for storing computer-readable program, The computer-readable program is used to execute above-mentioned all or part of embodiment of the method for computer.

That is, it will be understood by those skilled in the art that implement the method for the above embodiments be can be with Relevant hardware is instructed to complete by program, which is stored in a storage medium, including some instructions are to make It obtains an equipment (can be single-chip microcontroller, chip etc.) or processor (processor) executes each embodiment the method for the application All or part of the steps.And storage medium above-mentioned includes: USB flash disk, mobile hard disk, read-only memory (ROM, Read-OnlB MemorB), random access memory (RAM, Random Access MemorB), magnetic or disk etc. are various can store journey The medium of sequence code.

It will be understood by those skilled in the art that the respective embodiments described above are to realize specific embodiments of the present invention, And in practical applications, can to it, various changes can be made in the form and details, without departing from the spirit and scope of the present invention.

Embodiment of the present invention discloses a kind of generation method for dispensing range of A1., comprising:

Obtain the order characteristic parameter that unit is respectively dispensed in the corresponding first dispatching range of target object, wherein the mesh Mark the initial dispatching unit that object is located in the first dispatching range；

Centered on the initial dispatching unit, according to the order characteristic parameter of dispatching range model and the dispatching unit Expansion dispatching unit is obtained from the dispatching unit；

The second dispatching range of the target object is generated according to the initial dispatching unit and the expansion dispatching unit.

A2. dispatching range generation method according to a1, the method also includes:

The initial dispatching range model of building, the initial dispatching range model includes state value function, the state value letter Number is to dispense the order characteristic parameter of unit as parameter；

According to machine learning algorithm, with the History Order characteristic parameter of object in the commercial circle of the target object place to described Dispatching range model is trained.

A3. dispatching range generation method according to a1, the machine learning algorithm are nitrification enhancement,

It is described centered on the initial dispatching unit, according to dispatching range model expanded from the dispatching unit Unit order characteristic parameter is dispensed, specifically:

The order that the target object first dispenses each dispatching unit in range is inputted in the state value function Characteristic parameter obtains the score value of the dispatching unit；

According to the score value of the dispatching range model and the dispatching unit, obtain described in the first dispatching range Dispense the good working condition value matrix of unit；

Centered on the initial dispatching unit, the dispatching unit is sentenced according to the good working condition value matrix It is disconnected, obtain the expansion dispatching unit.

A4. dispatching range generation method according to a3, it is described single according to the dispatching range model and the dispatching The score value of member obtains the good working condition value matrix of the dispatching unit in the first dispatching range, specifically:

According to the score value of the dispatching range model and the initial dispatching unit, original state value matrix is obtained；

It is with the initial dispatching unit and the one or more dispatching unit adjacent with the initial dispatching unit Second dispatching unit set, according to point of the dispatching range model and the dispatching unit of the second dispatching unit set It is worth and obtains the second state value matrix；

It is single with the second dispatching unit set and the one or more dispatchings adjacent with the second dispatching unit set Member is that third dispenses unit set, according to the dispatching unit of the dispatching range model and third dispatching unit set Score value and, obtain third state value matrix；

And so on, the dispatching unit until traversing the first dispatching range obtains the good working condition value square Battle array.

A5. dispatching range generation method according to a4, it is described centered on the initial dispatching unit, according to described Good working condition value matrix judges the dispatching unit, obtains the expansion dispatching unit, specifically:

Score value and highest dispatching unit set are inquired in the good working condition value matrix, by the score value and highest The dispatching unit except the initial dispatching unit in unit set is dispensed, dispenses unit as the expansion.

A6. dispatching range generation method according to a1, the machine learning algorithm are that depth practises algorithm,

Centered on the initial dispatching unit, the dispatching unit is sentenced according to the score value of the dispatching unit It is disconnected, obtain the expansion dispatching unit.

A7. the dispatching range generation method according to A6, it is described centered on the initial dispatching unit, according to each institute The score value for stating dispatching unit judges the dispatching unit, obtains expansion dispatching unit, specifically:

The highest dispatching unit of score value in the dispatching unit adjacent with the initial dispatching unit is chosen to expand as first Unit is dispensed, the initial dispatching unit and the first expansion dispatching unit collectively form the first expansion range；

The highest dispatching unit of score value in the dispatching unit adjacent with the first expansion range is chosen to expand as second Unit is dispensed, the first expansion range and the second expansion dispatching unit collectively form the second expansion range；

And so on, the requirement until meeting the second dispatching range obtains the n-th expansion dispatching unit and the n-th expansion Range, wherein n is positive integer；

Using the first expansion dispatching unit, the second expansion dispatching expansion dispatching unit of unit n-th as described in Expansion dispatching unit.

A8. the dispatching range generation method according to A7, the dispatching range generation method further include:

If obtained (n-a) expansion dispatching unit includes multiple, the highest n-th expansion model of selection total score The dispatching range as the target object is enclosed, wherein a is the positive integer greater than 1 and less than n.

A9. the dispatching range generation method according to A7, the requirement for meeting the first dispatching range specifically: described The area of second dispatching range is no more than preset area threshold or the expansion of the initial dispatching unit and outermost dispenses Maximum distance between unit is no more than preset distance threshold.

A10. the dispatching range generation method according to any one of A1-A9, the order feature ginseng of the dispatching unit Number is obtained by following steps:

The order data of the first dispatching range is extracted in the database, and the order data includes the wide table of waybill, orders The POI interface of the wide table of Dan Kuanbiao, target object, electronic map；

The order data is screened to obtain the order feature ginseng of each dispatching unit in the first dispatching range Number.

A11. the dispatching range generation method according to A10, further includes:

Obtain the geography fence including predetermined keyword；

Each dispatching unit in the second dispatching range is judged according to the geography fence, is matched in vain Send unit, wherein the invalid subregion unit is located in the geography fence or including the geography fence；

The invalid dispatching unit is deleted in the second dispatching range.

Embodiment of the present invention additionally provides a kind of generating means for dispensing range of B12., comprising:

Data acquisition module obtains the order feature ginseng that unit is respectively dispensed in the corresponding first dispatching range of target object Number, wherein the target object is located at the initial dispatching unit in the first dispatching range；

Mesh expansion module, centered on the initial dispatching unit, according to dispatching range model and the dispatching unit Order characteristic parameter obtained from the dispatching unit expansion dispatching unit；

Horizon generation module generates the target object according to the initial dispatching unit and expansion dispatching unit Second dispatching range.

Embodiment of the present invention additionally provides C13. a kind of electronic equipment, including at least one processor；And

The memory and communication unit being connect at least one described processor communication；

Wherein, the memory is stored with the instruction that can be executed by least one described processor, and described instruction is described At least one processor is executed to realize:

Obtain the order characteristic parameter that unit is respectively dispensed in the corresponding first dispatching range of target object, wherein the mesh It marks in the initial dispatching unit that object is located in the first dispatching range；

C14. the electronic equipment according to C13, the processor are also used to execute:

Building dispatching range model, the dispatching range model includes state value function, and the state value function is to dispense The order characteristic parameter of unit is parameter；

C15. the electronic equipment according to 13, the machine learning algorithm are nitrification enhancement,

The processor executes centered on the initial dispatching unit, according to dispensing range model from the dispatching unit In obtain expansion dispatching unit order characteristic parameter, specifically:

C16. the electronic equipment according to C15, the processor execute described according to the dispatching range model and institute The score value for stating dispatching unit obtains the good working condition value matrix of the dispatching unit set in the first dispatching range, specifically Are as follows:

C17. the electronic equipment according to C16, the processor execution is described centered on the initial dispatching unit, The dispatching unit is judged according to the good working condition value matrix, obtains expansion dispatching unit, specifically:

C18. the electronic equipment according to C13, the machine learning algorithm are that depth practises algorithm,

The history that the dispatching unit in the dispatching of target object first range is inputted in the state value function is ordered Single characteristic parameter obtains the score value of the dispatching unit；

C19. the electronic equipment according to C18, the processor execution is described centered on the initial dispatching unit, The dispatching unit is judged according to the score value of each dispatching unit, obtains expansion dispatching unit, specifically:

C20. the electronic equipment according to C19, the processor are also used to execute:

C21. the electronic equipment according to C19, the requirement for meeting the first dispatching range specifically: described second matches Send range area be no more than preset area threshold or it is described it is initial dispatching unit and outermost expansion dispatching unit it Between maximum distance be no more than preset distance threshold.

C22. the order characteristic parameter of the electronic equipment according to any one of C13-C21, the dispatching unit passes through It obtains below:

C23. the electronic equipment according to C22, the processor is also with execution:

Obtain the geography fence including predetermined keyword；

It is invalid, which to obtain, is judged to each dispatching unit in the second dispatching range according to the geography fence Dispense unit, wherein the invalid subregion unit is located in the geography fence or including the geography fence；

The invalid dispatching unit is deleted in the second dispatching range.

Embodiment of the present invention additionally provides a kind of non-volatile memory medium of D24., for storing computer-readable journey Sequence, the computer-readable program are used to execute the dispatching range generation method as described in any one of A1 to A12 for computer.

Claims

1. a kind of generation method for dispensing range characterized by comprising

Obtain the order characteristic parameter that unit is respectively dispensed in the corresponding first dispatching range of target object, wherein the target pair As the initial dispatching unit being located in the first dispatching range；

Centered on the initial dispatching unit, according to the order characteristic parameter of dispatching range model and the dispatching unit from institute It states and obtains expansion dispatching unit in dispatching unit；

2. dispatching range generation method according to claim 1, which is characterized in that the method also includes:

The initial dispatching range model of building, the initial dispatching range model includes state value function, the state value function with The order characteristic parameter for dispensing unit is parameter；

According to machine learning algorithm, with the History Order characteristic parameter of object in the commercial circle of the target object place to the dispatching Range model is trained.

3. dispatching range generation method according to claim 1, which is characterized in that the machine learning algorithm is extensive chemical Algorithm is practised,

It is described centered on the initial dispatching unit, according to dispatching range model obtained from the dispatching unit expansion dispense Unit order characteristic parameter, specifically:

The order feature that the target object first dispenses each dispatching unit in range is inputted in the state value function Parameter obtains the score value of the dispatching unit；

According to the score value of the dispatching range model and the dispatching unit, the dispatching in the first dispatching range is obtained The good working condition value matrix of unit；

Centered on the initial dispatching unit, the dispatching unit is judged according to the good working condition value matrix, is obtained Unit is dispensed to the expansion.

4. dispatching range generation method according to claim 3, which is characterized in that described according to the dispatching range model With the score value of the dispatching unit, the good working condition value matrix of the dispatching unit in the first dispatching range is obtained, specifically Are as follows:

With the initial dispatching unit and the one or more dispatching unit adjacent with the initial dispatching unit for second Unit set is dispensed, according to the score value of the dispatching range model and the dispatching unit of the second dispatching unit set With obtain the second state value matrix；

It is with the second dispatching unit set and the one or more dispatching units adjacent with the second dispatching unit set Third dispenses unit set, according to point of the dispatching range model and the dispatching unit of third dispatching unit set It is worth and obtains third state value matrix；

And so on, the dispatching unit until traversing the first dispatching range obtains the good working condition value matrix.

5. dispatching range generation method according to claim 4, which is characterized in that described to be with the initial dispatching unit Center judges the dispatching unit according to the good working condition value matrix, obtains the expansion dispatching unit, specifically Are as follows:

Score value and highest dispatching unit set are inquired in the good working condition value matrix, by the score value and highest dispatching The dispatching unit except the initial dispatching unit in unit set, dispenses unit as the expansion.

6. dispatching range generation method according to claim 1, which is characterized in that the machine learning algorithm is depth habit Algorithm,

Centered on the initial dispatching unit, the dispatching unit is judged according to the score value of the dispatching unit, is obtained Unit is dispensed to the expansion.

7. dispatching range generation method according to claim 6, which is characterized in that described to be with the initial dispatching unit Center judges the dispatching unit according to the score value of each dispatching unit, obtains expansion dispatching unit, specifically:

The highest dispatching unit of score value in the dispatching unit adjacent with the initial dispatching unit is chosen to dispense as the first expansion Unit, the initial dispatching unit and the first expansion dispatching unit collectively form the first expansion range；

The highest dispatching unit of score value in the dispatching unit adjacent with the first expansion range is chosen to dispense as the second expansion Unit, the first expansion range and the second expansion dispatching unit collectively form the second expansion range；

Using the first expansion dispatching unit, the second expansion dispatching expansion dispatching unit of unit n-th as the expansion Dispense unit.

8. a kind of generating means for dispensing range characterized by comprising

Data acquisition module obtains the order characteristic parameter that unit is respectively dispensed in the corresponding first dispatching range of target object, In, the target object is located at the initial dispatching unit in the first dispatching range；

Mesh expansion module is ordered according to dispatching range model and the dispatching unit centered on the initial dispatching unit Single characteristic parameter obtains expansion dispatching unit from the dispatching unit；

Horizon generation module generates the second of the target object according to the initial dispatching unit and expansion dispatching unit Dispense range.

9. a kind of electronic equipment, including at least one processor；And

Wherein, the memory be stored with can by least one described processor execute instruction, described instruction by it is described at least One processor is executed to realize:

Obtain the order characteristic parameter that unit is respectively dispensed in the corresponding first dispatching range of target object, wherein the target pair As being located in the initial dispatching unit that described first dispenses in range；

10. a kind of non-volatile memory medium, for storing computer-readable program, the computer-readable program is by for based on Calculation machine executes the dispatching range generation method as described in any one of claims 1 to 7.