CN110110903A - A kind of distribution vehicle paths planning method based on neural evolution - Google Patents
A kind of distribution vehicle paths planning method based on neural evolution Download PDFInfo
- Publication number
- CN110110903A CN110110903A CN201910307290.7A CN201910307290A CN110110903A CN 110110903 A CN110110903 A CN 110110903A CN 201910307290 A CN201910307290 A CN 201910307290A CN 110110903 A CN110110903 A CN 110110903A
- Authority
- CN
- China
- Prior art keywords
- lorry
- time
- request
- place
- neural network
- 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
- 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
- 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"
- G06Q10/047—Optimisation of routes or paths, e.g. travelling salesman 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
- G06Q10/08—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
Landscapes
- Engineering & Computer Science (AREA)
- Business, Economics & Management (AREA)
- Human Resources & Organizations (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Economics (AREA)
- Strategic Management (AREA)
- General Physics & Mathematics (AREA)
- Tourism & Hospitality (AREA)
- Quality & Reliability (AREA)
- Operations Research (AREA)
- Marketing (AREA)
- Entrepreneurship & Innovation (AREA)
- Development Economics (AREA)
- General Business, Economics & Management (AREA)
- Data Mining & Analysis (AREA)
- Biophysics (AREA)
- Mathematical Physics (AREA)
- General Engineering & Computer Science (AREA)
- Computing Systems (AREA)
- Molecular Biology (AREA)
- General Health & Medical Sciences (AREA)
- Evolutionary Computation (AREA)
- Computational Linguistics (AREA)
- Software Systems (AREA)
- Biomedical Technology (AREA)
- Artificial Intelligence (AREA)
- Life Sciences & Earth Sciences (AREA)
- Health & Medical Sciences (AREA)
- Game Theory and Decision Science (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
- Traffic Control Systems (AREA)
Abstract
The invention belongs to Logistic Scheduling fields, more particularly to a kind of distribution vehicle paths planning method based on neural evolution, the first step establishes model, if there is M amount lorry in company, in dispatcher-controlled territory P and time range L within one day, the logistics request set R reached at any time is handled, each request r includes announcing the time, hand over picking place and handing over picking place time window.Second step, the request of each lorry handles which task depends on the corresponding priority of task, and the calculating of priority needs neural network, inputs the lorry to the information of a certain task, the task requests processing that the task is exported to the priority of the lorry, and then selects priority high.The corresponding cost function of each neural network obtains the optimal Logistic Scheduling scheme of cost function by neural evolution algorithm to Neural Network Optimization.The present invention helps to reduce the transportation cost of logistics company, and allows offline transfer is a large amount of to calculate work under dynamic scene, reduces hardware requirement.
Description
Technical field
The invention belongs to Logistic Scheduling fields, are related to a kind of distribution vehicle paths planning method based on neural evolution.
Background technique
The construction of traffic route is a stubborn problem, has been always the class of people's further investigation since over half a century
Topic.Partially due to the limitation on calculating, many logistics companies are applied with limitation to its client, it is common practice that allow client at some
It places an order before deadline, them is only just handled after this deadline and is requested.Logistics distribution is DYNAMIC DISTRIBUTION
The problem of formula, but problem is all adjusted to static version and creates route using centralized algorithm by most of existing technologies.By
The static version solved the problems, such as when new request response every time requires largely to calculate, so usual way is from calculating
One group of route start, when there is new request, update these routes using heuristic, be included in existing route insert
Enter, delete and switching purpose etc..
It is that static solution is usually concentrated and computation-intensive, especially needing to carry out many at any time again
In the case where calculating.On the other hand, it needs to respond in real time in more dynamic scene, such as tax services, with city express delivery
Service etc..Due to communicating and the fast development of information technology, information can obtain and handle in real time, the planning of vehicle dynamic route and
Logistics dynamic scheduling problem has become a research hotspot.
Summary of the invention
The technical problem to be solved by the present invention is to solve the constraint of having time window and while fetching and delivering the vehicle routing optimization of goods
Problem, and in particular to the creation of neural evolution algorithm and optimization multi-agent system.One is designed in dynamic logistics scheduling field
Distributed multi agent system with acceptable global performance is difficult.Firstly, the coordination problem between multiple agent.
Secondly, data (such as the client to be serviced) are not completely known, but to be held according to imperfect and uncertain information
Solution before the trade.Therefore, policymaker can not once solve the problems, such as entire, need to design a kind of incomplete using part
The decision-making mechanism of information.Finally, each request has the time window in a picking place and drop-off location and two places,
Complicate problem further.
Technical scheme is as follows:
A kind of distribution vehicle paths planning method based on neural evolution, steps are as follows:
Step 1: problem describes
The distributed nature of dynamic logistics scheduling problem allows the logical mappings of intelligent body: each lorry a is by an intelligence
Body surface shows.In certain dispatcher-controlled territory P and time range L within one day, lorry set M starts (parking lot in same parking lot
Place o, parking lot time window (eo,lo]), and with constant speed drive, one group of logistics reached at any time when lorry works of processing
Transport request set.One request r includes announcing time, delivery of cargo place ri, delivery of cargo place time window (eri,lri], drop-off location
rj, drop-off location time window (erj,lrj].Time window be it is half-open, lorry cannot before time window starts service request, but still
Can after the window terminates service request.It ignores the time required to delivery of cargo and delivery.Path between each place is considered as
Straight line.Lorry has handled all requests and has returned to parking lot and be considered as and solved the problems, such as, lorry does not allow to change its course during traveling.
It finds an optimal dynamic path planning method to go to minimize cost function, cost function is by following three value phase
Add:
(1) overall travel time of all lorries;
(2) total delay time (end time for being later than time window) in all picking places and delivery place;
(3) total time-out time (being later than scene end time) of all lorries.
This cost function does not include any probabilistic information about the following incoming request, but at the end of being applied to scene
Determine the request known, it may be assumed that
Wherein: TaIt is the overall travel time of lorry a.V is the set of picking place and drop-off location.tvIt is that lorry reaches v
The time (being equal to the time that v point is serviced) of point, lvIt is the end time of the place v time window.tkIt is that lorry a returns to parking lot
Time.
Step 2: the multi-agent system based on neural evolution is built and is optimized
Since the distributed nature of problem needs to consider the decision of each lorry to minimize above-mentioned cost function
Coordination between lorry.The decision of each lorry be by Processing with Neural Network priority it is high request realize, lorry it
Between coordination be by it is predetermined rule realize.Minimizing cost function is then by neural evolution algorithm evolution nerve
What the structure and weight of network were realized.
Coordination between 2.1 rule-based lorries
(1) when being announced new request, request can be all stored in a common request set R.Each lorry a
There is a privately owned request set Ra(indicate lorry a picking but without delivery);
(2) whenever lorry a has the appearance of one of following three kinds of situations: lorry does not receive request also in parking lot, and lorry arrival takes
Goods place picking is completed, and lorry reaches drop-off location delivery and completes.Lorry a will stop and check R ∪ R with fixed frequencyaIt is
No is empty;
(3) if being not sky, lorry a receives the request of highest priority by neural network, and (request when conflict in R is excellent
Prior to RaIn request);
(4) if lorry a can reach r in advanceiOr rjIf (r ∈ R, is ri;If r ∈ Ra, is rj), wherein mentioning
Preceding arrival means that lorry a is reached before the time window in the place starts and requires place, it will not go to and require place
And it waits in situ.Until being added to new request in L, return step (2) or time, which have arrived, requires place time window to open
Begin the time;
(5) if the time has arrived the place time window time started, lorry a starts to go to the place.After arrival, goods
Vehicle a picking or delivery, return step (2);
(6) when return step (2), L ∪ Ra is that empty and lorry a cannot terminate l in sceneoWhen preceding return parking lot, then
Lorry a returns to parking lot.
In addition, lorry a removes L or R before starting running, by the request r of processinga.If r ∈ L, lorry a pass through handle
Task r moves on to R from La, prevent other lorries from handling the request.If r ∈ Ra, that is, request to require freight to exist in r
On lorry a.Since lorry a is during hand over of goods, it cannot be interrupted and (not allow to change its course during Freight Transport), ask
Ask r will be to be processed, so lorry a is from RaMiddle deletion request.
2.2 lorry decisions neural network based
Present invention introduces neural networks, define following ten numerical value, characterization lorry a is in particular moment about request r's
Information, the input as neural network:
(1) the lorry quantity of these new requests to be subjected;
(2) lorry a is just freight quantity (one is requested the cargo of a corresponding unit);
(3) whether request r requires freight on lorry a;
(4) distance eriOr erjRemaining time subtract lorry a and reach r from now oniOr rj(this value is time
Non-negative);
(5)ri、rjAverage distance value, maximum range value and minimum between the corresponding place of delivery of lorry a freight
Distance value;
(6) place where lorry a reaches riAnd rjDistance;
(7)eriAnd erjWith the difference of current time (this value can be negative);
(8)lriAnd lrjWith the difference of current time (this value can be negative);
(9) terminate (parking lot time window end time l apart from sceneo) time (this value can be negative);
(10)riOr rjTo average distance, minimum distance and the maximum distance of all lorries in addition to lorry a.
And using request r for lorry a numerical priority value as unique output of neural network, lorry a passes through nerve
Each current can the executing of the task of network query function, selects the request of highest priority.
2.3 Evolutionary Neural Networks reduce cost function
An artificial neural network population is initialized, multiple neural networks, the corresponding cost of each neural network are obtained
Function optimizes the neural network using neural evolution algorithm, obtains optimal cost function.
On Scene case, each lorry goes to calculate the preferential of current each executable task by identical neural network
Grade, task of selecting priority high goes to execute, and meets rule-based coordination between lorry.In the same generation, each nerve net
Network all simulates whole process with identical Scene case, thus when obtaining the total travel of all lorries under each simulation process
Between, total time-out time of the total delay time and all lorries in all picking places and delivery place, and then calculate same
For cost function corresponding to each neural network.It is next that the low neural network of corresponding cost function has very big probability to enter
Generation, and the high neural network of corresponding cost function is then eliminated.By the evolution strategies such as intersect, make a variation, change previous generation cost
Connection between the neuron and neuron of the low neural network of function, changes the weight between their neurons, makes cost letter
Number decline, performance get a promotion.
Beneficial effects of the present invention:
The present invention allows offline transfer is a large amount of to calculate work, it is only necessary to seldom in the line computation time.It is needing quickly
It responds, can show relatively more preferable in the scene of more dynamical.
Detailed description of the invention
Fig. 1 is the flow chart of neural evolution algorithm.
Fig. 2 is the frame diagram for realizing neural evolution algorithm.
Specific embodiment
Below with reference to specification figure, the present invention is further described.
In order to realize the dynamic vehicle path planning of dynamic logistics scheduling, dispatching simulation device is needed to generate contextual data collection
(scene is actually to be made of some requests generated at random), builds multi-agent system, and this algorithm is deployed to simulation
In the intelligent body of device.Passage path planning module realizes Fitness analysis based on simulation, until individual most of in a generation
Fitness selects the neural network of best performance as the decision-making mechanism of intelligent body there is no terminating to evolve when significant change.
1) path planning module generates contextual data collection
Since the value of neural network input and the value of cost function cannot be derived analytically, need to introduce dispatching simulation
Device generates different scenes, is simulated in problem-instance.Simulator needs raw according to different spatio-temporal distributions
The request of Cheng Xin updates current dispatch situation (such as the operating condition of monitoring lorry and actual picking and time of delivery).
Simulator each of generates new request packet and includes existing time and picking and drop-off location with time window.These values are necessary
Be it is feasible, time window must reserve time enough and reach picking place by lorry, and picking simultaneously goes directly to place of delivery.This
A little values must also be real, because of the variation of the appearance requested (such as peak period) and space (densely inhabited district) at any time
And change.Model, which is generated, with the scene that can modify parameter generates given quantity (such as 800) feasible and real scene number
According to collection:
(1) discretization is carried out to time and space: dispatcher-controlled territory P is divided into lesser rectangle, time range L is divided
For lesser time interval;
(2) different rectangular areas and different time intervals, corresponding value are respectively represented by inputting a row and column
It respectively represents in time interval l, the picking place newly requested or place of delivery appear in the matrix A of the probability of region p, adjust
Degree simulator can generate a matrix B in each time interval ll, the row and column of the generator matrix respectively represents two differences
Rectangular area, corresponding value represents in time interval l, and the picking place i newly requested is in region p, drop-off location j in area
The probability that domain q occurs;
(3) by being uniformly distributed specific to specific when and where;
(4) the time t that request r is serviced the latest:
T=lo-(tij+tjo)(2)
Wherein, tijAnd tjoIt is ri、rj, parking lot o place and lorry the calculated running time of constant driving speed, T
It is current time, eriIt follows and is uniformly distributed in section [T, t];
(5) time window of picking place i is [eri,eri+x(lo- T)], wherein x (lo- T) it is when terminating preceding residue to scene
Between a part, x is in section [x1,x2] in follow and be uniformly distributed, x1And x2It is the section at any time that can be defined by a user and region
The parameter of variation;
(6) generation of drop-off location j time window is identical with aforementioned principles, only erjIn section [eri+tij,t+tij] in abide by
It follows and is uniformly distributed.
2) Decision of Neural Network module (being based on data set Evolutionary Neural Network)
The not no hidden layer of Decision of Neural Network module initialization one, only one arbitrarily inputs neuron and one is appointed
The artificial neural network population of meaning output neuron singular association.By intersecting, making a variation etc., evolution strategies increase neuron and mind
Through the connection between member, keep neural network complicated, performance gets a promotion.Artificial definition is needed to input, export and suitable
Response function.The algorithm has the function of feature selecting, and with the increase of extraneous features, performance is kept approximately constant.Because refreshing
The maximization of positive adaptation angle value is only supported through network decision module, if lorry returns to the time in parking lot considerably beyond scene knot
The time of beam, then assigning its fitness is zero, larger with one if all lorries return to parking lot before scene terminates
Normal number subtract cost function to calculate its fitness.Every generation successively uses identical scene.
In a large amount of individual fitness of certain generation all there is no terminating to evolve when significant change, select fitness highest
Decision-making mechanism of the neural network as intelligent body in multi-agent system.
Claims (1)
1. a kind of distribution vehicle paths planning method based on neural evolution, which is characterized in that steps are as follows: step 1: problem
Description
Each lorry a is indicated by an intelligent body;In certain dispatcher-controlled territory P and time range L within one day, lorry collection
It closes M to start in same parking lot, parking lot place o, parking lot time window (eo,lo], and with constant speed drive, one group is handled in lorry
The logistics transportation request set reached at any time when work;One request r includes announcing time, delivery of cargo place ri, delivery of cargo place
Time window (eri,lri], drop-off location rj, drop-off location time window (erj,lrj];Time window be it is half-open, lorry cannot be in the time
Service request before window starts, but still can after the window terminates service request;Ignore not the time required to delivery of cargo and delivery
Meter;Path between each place is considered as straight line;Lorry has handled all requests and has returned to parking lot and be considered as and solved the problems, such as, lorry is expert at
Do not allow to change its course during sailing;
It finds an optimal scheduling scheme to go to minimize cost function, cost function is added by following three value:
(1) overall travel time of all lorries;
(2) total delay time in all picking places and delivery place;
(3) total time-out time of all lorries;
This cost function does not include any probabilistic information about the following incoming request, but knows for determination at the end of scene
The request in road, it may be assumed that
Wherein: TaIt is the overall travel time of lorry a;V is the set of picking place and drop-off location;tvIt is that lorry reaches the place v
Time, lvIt is the end time of the place v time window;taIt is the time that lorry a returns to parking lot;
Step 2: the multi-agent system based on neural evolution is built and is optimized
Coordination between 2.1 rule-based lorries
(1) when being announced new request, request can be all stored in inside a central request list R, and each lorry a also has
One privately owned request set Ra;
(2) whenever lorry a has the appearance of one of following three kinds of situations: lorry does not receive request also in parking lot, and lorry is with reaching picking
Point picking is completed, and lorry reaches drop-off location delivery and completes;Lorry a will stop and with fixed frequency check R ∪ Ra whether be
It is empty;
(3) if being not sky, lorry a receives the request of highest priority by neural network;
(4) if lorry a can reach r in advanceiOr rj, mean lorry a in the time window in the place wherein reaching in advance
It is reached before starting and requires place, it will be waited;Until being added to new request in L, (2) of return step 2.1 or time are
The place time window time started is arrived;
(5) if the time has arrived the place time window time started, lorry a starts to go to the place;After arrival, lorry a
Picking or delivery, (2) of return step 2.1;
(6) when (2) of return step 2.1, L ∪ Ra is that empty and lorry a cannot be when returning to parking lot before scene terminates, then goods
Vehicle a returns to parking lot;
2.2 lorry decisions neural network based
Ten numerical value are defined, characterize information of the lorry a about request r, the input as neural network:
(1) quantity of the lorry of request r is waited;
(2) lorry a is just in freight quantity;
(3) it requests to require freight whether on lorry a in r;
(4) distance eriOr erjRemaining time subtract lorry a and reach r from now oniOr rjTime;
(5)ri、rjAverage distance value, maximum range value and minimum range between the corresponding place of delivery of lorry a freight
Value;
(6) place where lorry a reaches riOr rjDistance;
(7)eriAnd erjWith the difference of current time;
(8)lriAnd lrjWith the difference of current time;
(9) time terminated apart from scene;
(10)riOr rjTo average distance, minimum distance and the maximum distance of all lorries in addition to lorry a;
Numerical priority value unique output as neural network of the r for lorry a is requested, it is every that lorry a passes through neural computing
One can currently executing for task, selects the request of highest priority;
2.3 Evolutionary Neural Networks reduce cost function
An artificial neural network population is initialized, multiple neural networks are obtained, each neural network corresponds to a cost function,
The neural network is optimized using neural evolution algorithm, obtains optimal cost function.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910307290.7A CN110110903B (en) | 2019-04-17 | 2019-04-17 | Neural evolution-based distribution vehicle path planning method |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910307290.7A CN110110903B (en) | 2019-04-17 | 2019-04-17 | Neural evolution-based distribution vehicle path planning method |
Publications (2)
Publication Number | Publication Date |
---|---|
CN110110903A true CN110110903A (en) | 2019-08-09 |
CN110110903B CN110110903B (en) | 2022-09-30 |
Family
ID=67485670
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201910307290.7A Active CN110110903B (en) | 2019-04-17 | 2019-04-17 | Neural evolution-based distribution vehicle path planning method |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN110110903B (en) |
Cited By (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110674978A (en) * | 2019-09-10 | 2020-01-10 | 大连理工大学 | Task allocation and path planning method for unmanned workshop transportation system |
CN111103887A (en) * | 2020-01-14 | 2020-05-05 | 大连理工大学 | Multi-sensor-based multi-mobile-robot scheduling system design method |
CN112001557A (en) * | 2020-08-31 | 2020-11-27 | 物联云仓(成都)科技有限公司 | TMS system-based logistics distribution path optimization method, storage medium and computer equipment |
CN113919688A (en) * | 2021-10-09 | 2022-01-11 | 福州大学 | Logistics park approach vehicle dynamic scheduling method considering late arrival |
WO2022190219A1 (en) * | 2021-03-09 | 2022-09-15 | 日本電信電話株式会社 | Traveling plan generation device, traveling plan generation method, and program |
CN115438860A (en) * | 2022-09-06 | 2022-12-06 | 西安电子科技大学广州研究院 | Multi-agent path planning method based on evolutionary algorithm |
Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN107145971A (en) * | 2017-04-18 | 2017-09-08 | 苏州工业职业技术学院 | A kind of express delivery dispatching optimization method of dynamic adjustment |
CN107977739A (en) * | 2017-11-22 | 2018-05-01 | 深圳北斗应用技术研究院有限公司 | Optimization method, device and the equipment in logistics distribution path |
-
2019
- 2019-04-17 CN CN201910307290.7A patent/CN110110903B/en active Active
Patent Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN107145971A (en) * | 2017-04-18 | 2017-09-08 | 苏州工业职业技术学院 | A kind of express delivery dispatching optimization method of dynamic adjustment |
CN107977739A (en) * | 2017-11-22 | 2018-05-01 | 深圳北斗应用技术研究院有限公司 | Optimization method, device and the equipment in logistics distribution path |
Non-Patent Citations (3)
Title |
---|
周咏等: "冷链物流同时送取货车辆路径优化", 《数学的实践与认识》 * |
王力锋等: "物流运输快速配送路径规划仿真", 《计算机仿真》 * |
王蕾等: "基于改进蚁群算法的物流配送中多取货车辆的最优路径研究", 《公路交通科技(应用技术版)》 * |
Cited By (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110674978A (en) * | 2019-09-10 | 2020-01-10 | 大连理工大学 | Task allocation and path planning method for unmanned workshop transportation system |
CN111103887A (en) * | 2020-01-14 | 2020-05-05 | 大连理工大学 | Multi-sensor-based multi-mobile-robot scheduling system design method |
CN111103887B (en) * | 2020-01-14 | 2021-11-12 | 大连理工大学 | Multi-sensor-based multi-mobile-robot scheduling system design method |
CN112001557A (en) * | 2020-08-31 | 2020-11-27 | 物联云仓(成都)科技有限公司 | TMS system-based logistics distribution path optimization method, storage medium and computer equipment |
WO2022190219A1 (en) * | 2021-03-09 | 2022-09-15 | 日本電信電話株式会社 | Traveling plan generation device, traveling plan generation method, and program |
CN113919688A (en) * | 2021-10-09 | 2022-01-11 | 福州大学 | Logistics park approach vehicle dynamic scheduling method considering late arrival |
CN113919688B (en) * | 2021-10-09 | 2022-05-06 | 福州大学 | Logistics park approach vehicle dynamic scheduling method considering late arrival |
CN115438860A (en) * | 2022-09-06 | 2022-12-06 | 西安电子科技大学广州研究院 | Multi-agent path planning method based on evolutionary algorithm |
Also Published As
Publication number | Publication date |
---|---|
CN110110903B (en) | 2022-09-30 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
WO2021248607A1 (en) | Deep reinforcement learning-based taxi dispatching method and system | |
CN110110903A (en) | A kind of distribution vehicle paths planning method based on neural evolution | |
Teodorović | Swarm intelligence systems for transportation engineering: Principles and applications | |
Ghiani et al. | Real-time vehicle routing: Solution concepts, algorithms and parallel computing strategies | |
Dechampai et al. | A differential evolution algorithm for the capacitated VRP with flexibility of mixing pickup and delivery services and the maximum duration of a route in poultry industry | |
CN113811915A (en) | Unified order serving and fleet management for online shared travel platform | |
CN109194746A (en) | Heterogeneous Information processing method based on Internet of Things | |
CN108122048A (en) | A kind of transportation route dispatching method and its system | |
Kong et al. | Trajectory optimization for drone logistics delivery via attention-based pointer network | |
Jiang et al. | A multi-intersection vehicular cooperative control based on end-edge-cloud computing | |
Wang et al. | Optimization of ride-sharing with passenger transfer via deep reinforcement learning | |
Li et al. | A new fuzzy-based method for energy-aware resource allocation in vehicular cloud computing using a nature-inspired algorithm | |
CN113140108B (en) | Cloud traffic situation prediction method in internet-connected intelligent traffic system | |
Lu et al. | A multi-agent adaptive traffic signal control system using swarm intelligence and neuro-fuzzy reinforcement learning | |
CN117252318A (en) | Intelligent networking automobile group machine collaborative carpooling scheduling method and system | |
Castagna et al. | Demand-responsive rebalancing zone generation for reinforcement learning-based on-demand mobility | |
Chen et al. | A Multi-Agent Reinforcement Learning approach for bus holding control strategies. | |
Yu et al. | Real-time holding control for transfer synchronization via robust multiagent reinforcement learning | |
Yuan et al. | Reinforcement learning from optimization proxy for ride-hailing vehicle relocation | |
CN110046851A (en) | Unmanned vehicle logistics method for allocating tasks based on Multi-Paxos | |
Hu et al. | Autonomous valet parking optimization with two-step reservation and pricing strategy | |
Martynova et al. | Ant colony algorithm for rational transit network design of urban passenger transport | |
Liu et al. | Research on vehicle routing problem with time windows based on the dragonfly algorithm | |
Li et al. | Decentralized ride-sharing of shared autonomous vehicles using graph neural network-based reinforcement learning | |
CN107832887A (en) | A kind of shared automobile intelligent optimizing decision-making technique and system based on neutral net |
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 |