CN112766591A - Shared bicycle scheduling method - Google Patents
Shared bicycle scheduling method Download PDFInfo
- Publication number
- CN112766591A CN112766591A CN202110109869.XA CN202110109869A CN112766591A CN 112766591 A CN112766591 A CN 112766591A CN 202110109869 A CN202110109869 A CN 202110109869A CN 112766591 A CN112766591 A CN 112766591A
- Authority
- CN
- China
- Prior art keywords
- shared bicycle
- scheduling
- area
- dispatching
- sub
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 45
- 230000002787 reinforcement Effects 0.000 claims abstract description 29
- 230000007306 turnover Effects 0.000 claims abstract description 7
- 238000012549 training Methods 0.000 claims description 18
- 230000009471 action Effects 0.000 claims description 16
- 230000000694 effects Effects 0.000 claims description 12
- 238000007781 pre-processing Methods 0.000 claims description 4
- 238000011156 evaluation Methods 0.000 claims description 3
- 230000002159 abnormal effect Effects 0.000 claims description 2
- 238000004590 computer program Methods 0.000 claims 1
- 238000010276 construction Methods 0.000 claims 1
- 238000000638 solvent extraction Methods 0.000 claims 1
- 230000008569 process Effects 0.000 description 6
- 238000013528 artificial neural network Methods 0.000 description 5
- 238000013461 design Methods 0.000 description 5
- 230000006870 function Effects 0.000 description 4
- 230000009286 beneficial effect Effects 0.000 description 3
- 238000004364 calculation method Methods 0.000 description 3
- 230000008878 coupling Effects 0.000 description 3
- 238000010168 coupling process Methods 0.000 description 3
- 238000005859 coupling reaction Methods 0.000 description 3
- 238000010586 diagram Methods 0.000 description 3
- 239000011159 matrix material Substances 0.000 description 3
- 238000005457 optimization Methods 0.000 description 3
- 238000012360 testing method Methods 0.000 description 3
- 230000004913 activation Effects 0.000 description 2
- 238000004891 communication Methods 0.000 description 2
- 238000011161 development Methods 0.000 description 2
- 238000013507 mapping Methods 0.000 description 2
- 210000002569 neuron Anatomy 0.000 description 2
- 238000011084 recovery Methods 0.000 description 2
- 230000007613 environmental effect Effects 0.000 description 1
- 230000002452 interceptive effect Effects 0.000 description 1
- 238000011835 investigation Methods 0.000 description 1
- 238000012886 linear function Methods 0.000 description 1
- 239000000463 material Substances 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 230000008092 positive effect Effects 0.000 description 1
- 238000012545 processing Methods 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
- G06Q10/063—Operations research, analysis or management
- G06Q10/0631—Resource planning, allocation, distributing or scheduling for enterprises or organisations
- G06Q10/06315—Needs-based resource requirements planning or analysis
-
- 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
- G06Q30/00—Commerce
- G06Q30/06—Buying, selling or leasing transactions
- G06Q30/0645—Rental transactions; Leasing transactions
-
- 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
- G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
- G06Q50/40—Business processes related to the transportation industry
Landscapes
- Business, Economics & Management (AREA)
- Human Resources & Organizations (AREA)
- Engineering & Computer Science (AREA)
- Strategic Management (AREA)
- Economics (AREA)
- Theoretical Computer Science (AREA)
- Marketing (AREA)
- Physics & Mathematics (AREA)
- General Business, Economics & Management (AREA)
- General Physics & Mathematics (AREA)
- Entrepreneurship & Innovation (AREA)
- Development Economics (AREA)
- Accounting & Taxation (AREA)
- Tourism & Hospitality (AREA)
- Quality & Reliability (AREA)
- Game Theory and Decision Science (AREA)
- Operations Research (AREA)
- Finance (AREA)
- Educational Administration (AREA)
- Health & Medical Sciences (AREA)
- General Health & Medical Sciences (AREA)
- Primary Health Care (AREA)
- Traffic Control Systems (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
A dispatching method of a shared bicycle is characterized in that according to real-time use data of the shared bicycle, a dispatching decision is made by a dispatching model established based on a deep reinforcement learning algorithm, and a dispatching truck is guided to dispatch and transport the shared bicycle in an operation area. When the dispatching model makes a dispatching decision, the dispatching model collects the distribution data of the idle shared bicycle and also predicts the use requirements of the shared bicycle in the operation area by combining the historical use data of the shared bicycle. The dispatching model divides the operation area into a plurality of sub-areas, the sub-areas are used as the minimum space unit of the shared single-vehicle dispatching, and the division of the sub-areas is based on the urban road pattern. The invention combines the deep reinforcement learning with the shared bicycle scheduling problem, solves the problem of unbalanced supply and demand of the shared bicycle by using the advantages of the deep reinforcement learning, improves the operation efficiency, realizes real-time scheduling decision, and commands the scheduling truck which continuously runs in the network to perform bicycle scheduling so as to improve the vehicle turnover rate and improve the satisfaction degree of consumers.
Description
Technical Field
The invention belongs to the technical field of internet service, and particularly relates to a shared bicycle scheduling method for solving the problem of rebalance scheduling of the use requirements of a non-stop shared bicycle.
Background
The shared bicycle is beneficial to solving the problem of the last kilometer of urban traffic, is convenient for citizens to go out, relieves urban traffic pollution, and promotes the development of urban green traffic and slow traffic. However, the demands of consumers for the shared bicycle have time and space differences, so that the supply and the demands of the shared bicycle are unbalanced, the problems of low turnover rate and the like occur in the operation process of the shared bicycle, and the development of urban slow traffic is restricted.
At present, methods for scheduling shared bicycles at home and abroad comprise a threshold value setting scheduling method and an optimization model scheduling method. The threshold value setting scheduling method mainly relieves the phenomena of disordered parking and random release of the shared bicycle and excessive or insufficient bicycles at partial release points, can not thoroughly solve the problem of unbalanced supply and demand of the shared bicycle, and consumes a large amount of manpower and material resources. The optimization model scheduling method is high in accuracy, heuristic algorithms such as an ant colony algorithm and a tide algorithm are combined, but the method cannot perform scheduling decision in real time due to low calculation efficiency; the poor expansibility of the model causes that the method is difficult to be applied to complicated and changeable real scenes. In a word, the effect of sharing single-vehicle scheduling by using the existing scheduling method is poor, the calculation efficiency is low, and real-time sharing single-vehicle scheduling is difficult to realize under complex conditions.
Disclosure of Invention
The invention provides a shared bicycle scheduling method, and aims to realize real-time scheduling of shared bicycles and achieve supply and demand balance of shared bicycle usage.
The embodiment of the invention provides a shared bicycle scheduling method based on deep reinforcement learning. Dividing sub-areas according to factors such as land property of a target area, urban roads and the like to serve as basic units of scheduling; secondly, preprocessing the GPS data of the shared bicycle, predicting the vehicle using requirements of different sub-areas at different times based on historical data, and collecting the spatial distribution data of the idle bicycle to analyze the supply condition as a decision basis for scheduling the shared bicycle; then designing a neural network structure in the deep reinforcement learning model; designing a reinforcement learning model, mapping three elements in reinforcement learning, namely states, actions and returns to a regional shared bicycle scheduling task, and simultaneously determining hyper-parameters such as learning rate and the like in the model; and finally training the model based on historical data, and testing and evaluating the model under the actual condition.
The invention applies the deep reinforcement learning to solve the problem of shared bicycle scheduling, solves the problem of unbalanced supply and demand of shared bicycles by using the advantages of the deep reinforcement learning, improves the operation efficiency, realizes real-time scheduling decision, and commands the scheduling trucks continuously running in the network to perform bicycle scheduling so as to improve the vehicle turnover rate and improve the satisfaction degree of consumers.
Drawings
The above and other objects, features and advantages of exemplary embodiments of the present invention will become readily apparent from the following detailed description read in conjunction with the accompanying drawings. Several embodiments of the invention are illustrated by way of example, and not by way of limitation, in the figures of the accompanying drawings and in which:
FIG. 1 is a schematic flow diagram of a method according to an embodiment of the invention.
FIG. 2 is a diagram illustrating a design reinforcement learning model according to an embodiment of the present invention.
Fig. 3 is an example of area division in the embodiment of the present invention.
FIG. 4 is a diagram illustrating a design reinforcement learning model according to an embodiment of the present invention.
Detailed Description
In order to overcome the limitation of the conventional scheduling method of the shared bicycle, the invention provides the scheduling method of the shared bicycle based on deep reinforcement learning. The method comprises the steps of carrying out regional division on a throwing area of the shared bicycle, building a deep reinforcement learning model, and training based on historical demand data, so that the shared bicycle is dispatched in real time by using a dispatching truck, and the supply and demand balance of the shared bicycle in the region is achieved.
According to one or more embodiments, a shared bicycle scheduling method based on deep reinforcement learning comprises the following steps:
(1) and dividing the target area according to factors such as land property, urban roads and the like, wherein the divided sub-areas are used as basic space units for sharing the single-vehicle dispatching.
(2) The GPS historical data of the shared bicycle is preprocessed, the vehicle using requirements of different sub-areas at different time are predicted, and the space distribution data of the idle bicycle at different time periods are collected to analyze the supply condition and serve as the decision basis of the shared bicycle scheduling.
(3) And establishing a shared bicycle scheduling model based on a deep reinforcement learning (DQN) framework, and designing a neural network structure in the deep reinforcement learning model.
(4) And designing a reinforcement learning model. Mapping three elements in reinforcement learning, namely states, actions and returns, into a regional shared bicycle scheduling task, and determining interactive contents of an agent and the environment in the model and an action set executed by the agent; and determine other parameters such as the learning rate in the model.
(5) And training a deep reinforcement learning model according to the historical demand and supply data of the shared bicycle, and testing, evaluating and optimizing in an actual situation.
Preferably, in the step (1), the areas are divided according to urban land properties, specifically, school education areas, medical service areas, shopping areas in shopping malls, public transportation hub areas, residential areas in residential areas, enterprise areas, and the like, and each sub-area has an urban trunk as a boundary. Meanwhile, the minimum limit S of the area of the region is set in the region dividing processminAnd a maximum limit SmaxAnd the scheduling cost is increased because the area of the divided sub-regions is too small, and the scheduling effect of the balance of supply and demand of the shared bicycle is reduced because the area of the divided sub-regions is too large.
The sharing bicycle intelligent lock can send basic information such as positions to the server when a user order starts and ends, and meanwhile, the driving track can be recorded in the using process of the bicycle, so that historical demands can be analyzed, supply information can be obtained by analyzing GPS historical data of the sharing bicycle, and the demands can be predicted at the same time. Preferably, in the step (2), the processing and analyzing of the GPS data are as follows:
1) preprocessing is performed on the shared bicycle GPS data. The original data has data vacancy, error and the like. For the condition of data vacancy, if the order only contains starting point position or end point position information, counting the distribution of the starting point of the journey from the area or the distribution of the starting point of the journey to the area according to other order conditions of the area, and randomly generating corresponding end point or starting point position information according to the distribution; and if the starting point and the end point information in the order are completely missing, deleting the piece of data. In the case of data errors, for example, the order trip distance is too long, and the vehicle speed calculated from the GPS and time data is abnormal, it should be deleted.
2) And predicting the bicycle demand according to the shared bicycle historical data. The demands of the shared bicycle have time and space difference, so in space, demand prediction takes the subareas divided in the step (1) as basic units, namely, the demands are described from a region i to a region j; temporally, demand is predicted on a basic unit of time granularity Δ t, i.e., once per Δ t time, and differentiated to account for different periods of weekdays, weekends, holidays, significant activity, etc. The prediction method may use a weighted moving average method:
Ft＝ω1At-1+ω2At-2+……+ωnAt-n
wherein FtAs a predictor of the next phase, ω1Is the weight of the actual value in (t-1), At-1Is the actual value of the (t-1) period. The demand forecast value is changing continuously with the time, which is the basis of realizing real-time scheduling of the shared bicycle.
3) When an order on the uniquely numbered shared vehicle ends without a new order beginning, the vehicle is idle. The sum of the idle bicycles of each subarea, namely the supply number of each subarea can be obtained according to the GPS data of the shared bicycle.
Preferably, in the step (3), the neural network has a fully connected structure in order to sufficiently retain input information derived from the "environment".
Preferably, in the step (4), the reinforcement learning model building process is as follows:
1) the state elements in the model comprise demand information of each sub-area consumer using the shared bicycle, supply information of the shared bicycle and truck state information for scheduling the shared bicycle. As an example, if the study area is divided into N × N sub-areas and the status elements are set to a finite neighborhood area feature centered on the dispatch van, then the matrix dimension describing the status elements is (3 × N × N). Wherein the elements describing the demand information, supply information and dispatch truck status information are Dij,Sij,Vij. Wherein DijRepresenting the demand number of the corresponding sub-area at the next moment; sijRepresenting the supply number of the corresponding subarea at the current time, namely the number of idle bicycles in the subarea; vijAnd the number of the single vehicles in the dispatching truck of the corresponding subarea at the current time is represented.
2) The action elements in the model refer to dispatching the freight cars to move to the adjacent sub-areas and throwing or recovering a plurality of shared single cars. It is emphasized that the dispatching truck may also remain stationary or move but not engage in launch and recovery vehicle actions. The specific action number is related to the region division structure and the truck state.
3) The reward elements in the model are designed as follows. If the shared bicycle thrown by the dispatched truck is used in unit time (preset), the environment gives positive reward, and otherwise, the environment gives negative reward; if the shared vehicle recovered by the dispatch truck is not needed for a unit of time (default), the environment gives a positive reward and conversely a negative reward. Otherwise the reward is 0.
4) The setting of the hyper-parameters in the model can be debugged according to the training result, and the scheduling effect and the operation efficiency need to be comprehensively considered.
Preferably, in the step (5), historical data is collected for the target area, a model structure and hyper-parameters are set based on data such as demand and supply, model training is performed, and multi-pass training is performed according to a preset value. Meanwhile, evaluation indexes such as vehicle turnover rate are set to evaluate the training effect, and the hyper-parameters and the optimization model are adjusted according to the result.
One of the beneficial effects of the invention is that the scheduling model method of deep reinforcement learning has high execution efficiency and is suitable for complex areas; the expandability is strong, and the method is suitable for any shared bicycle throwing area; the problem that consumers are in 'no-vehicle availability' and too many shared single vehicles in partial areas is solved, and overall supply and demand balance is achieved.
According to one or more embodiments, a shared bicycle scheduling method based on deep reinforcement learning comprises the following steps:
(1) designing and constructing a training model according to a deep reinforcement learning method:
firstly, dividing a shared bicycle operation area into a plurality of sub-areas, taking the sub-areas as basic units for space scheduling of the shared bicycle, analyzing the demand and supply conditions of each sub-area, and arranging a plurality of trucks for scheduling the shared bicycle, wherein the training of the model aims to make decisions according to the demand, supply and truck state information so as to command the trucks to complete scheduling tasks such as moving, releasing or recovering vehicles, standing still and the like. The scheduling task is described as a Markov process, and a reinforcement learning model is constructed, wherein the reinforcement learning model comprises the design of states, actions and returns. The state elements comprise demand information, supply information and dispatching truck state information, and the matrix with the dimensionality of (3 multiplied by n) is used for describing the state; the action element is that the freight car is dispatched to move towards the adjacent subarea and a plurality of sharing single cars are put in or recovered; the design of the return element is based on the effect after scheduling, i.e. compared with the effect without scheduling action, the return is positive if the scheduling action has positive effect, otherwise, the return is negative, and the other condition is 0. And finally, designing a neural network structure according to the input and the output of the model, wherein a full-connection structure is generally adopted. Under the design, the truck can make a scheduling action beneficial to supply and demand balance according to the environmental information, the overall single-vehicle turnover rate is improved, and the problem of imbalance of supply and demand of the shared single vehicle is solved.
(2) Training a deep reinforcement learning model based on historical data:
the training times of each training process are preset. Meanwhile, partial hyper-parameters are modified according to the training effect so as to achieve the balance of the operation efficiency and the operation result. In the training process, different periods such as working days, weekends, holidays, major activities and the like are considered in a distinguishing manner, so that the robustness of the model is improved. And finally, testing in practical application, and setting relevant indexes for evaluation, wherein the indexes comprise the turnover rate of a single vehicle, the average service time of the single vehicle and the like.
The following is an example of the calculation.
The investigation region was divided into 5X 5 squares with each side being 0.5Km long. A single-car dispatch is performed in the network using one dispatch truck, the capacity of which is 20 single cars. The total number of the 25 sub-areas is used as a basic unit of scheduling, preprocessing is carried out according to GPS historical data of the shared bicycle, the vehicle using requirements of each sub-area at different time are predicted, and spatial distribution data of the idle bicycle are collected and used as a decision basis of scheduling of the shared bicycle. The unit time of the scheduling decision is 5 minutes, i.e. 5 minutes, a scheduling decision is made.
The neural network comprises a three-layer structure. The first layer is an input layer, and the input is unidimensionalized by using a Flatten () function in a module Keras; the second layer sets up 128 neurons. The activation function uses a linear rectification function (ReLU); the third layer is the output layer. The number of neurons is consistent with the number of actions of the agent, and a Linear function (Linear) is used as an activation function.
In the reinforcement learning model, the state information includes the states of demand, supply and truck, and the scale of the embodiment is small, so the state information is described by a matrix with the dimension of (3 × 5 × 5); since the truck capacity is 20, the number of trucks for loading or unloading is in the range of 0 to 20, and the moving directions of the trucks comprise east, south, west and north directions and five directions which are kept unchanged. Here the truck can only move the distance of one area within one dispatch period. Therefore, the total number of actions performed by the truck is 205, but the total number of actions performed by the truck is related to the position of the truck and the number of the single trucks in the truck. The reward elements in the model are designed as follows. When the action of launching the bicycle is executed, if n shared bicycles launched by the dispatched trucks are used within the dispatching time (5 minutes), the reward is n, and conversely, the reward is-n. When the recovery bicycle is executed, if n shared bicycles recovered by the dispatching truck need to be used within the dispatching time (5 minutes), the return is-n, and otherwise, the return is n. Specifically, if an original bicycle N in a sub-area is used for a dispatching truck to recover N bicycles in a period t, and a required number D exists in a period t +1, the return is as follows:
the hyper-parameters in the model are set as: learning rate parameter of 10-5The memory size of the intelligent agent is 50000, the parameters of the Bellman formula are 0.9, and the network updating rate is 10-2Training is divided into 20 courses, each training course comprises 10000 steps.
In the several embodiments provided in the present application, it should be understood that the disclosed system, apparatus and method may be implemented in other ways. For example, the above-described apparatus embodiments are merely illustrative, and for example, the division of the units is only one logical division, and other divisions may be realized in practice, for example, a plurality of units or components may be combined or integrated into another system, or some features may be omitted, or not executed. In addition, the shown or discussed mutual coupling or direct coupling or communication connection may be an indirect coupling or communication connection through some interfaces, devices or units, and may also be an electric, mechanical or other form of connection.
The integrated unit, if implemented in the form of a software functional unit and sold or used as a stand-alone product, may be stored in a computer readable storage medium. Based on such understanding, the technical solution of the present invention essentially or partially contributes to the prior art, or all or part of the technical solution can be embodied in the form of a software product stored in a storage medium and including instructions for causing a computer device (which may be a personal computer, a server, or a network device) to execute all or part of the steps of the method according to the embodiments of the present invention. And the aforementioned storage medium includes: a U-disk, a removable hard disk, a Read-Only Memory (ROM), a Random Access Memory (RAM), a magnetic disk or an optical disk, and other various media capable of storing program codes.
While the invention has been described with reference to specific embodiments, the invention is not limited thereto, and various equivalent modifications and substitutions can be easily made by those skilled in the art within the technical scope of the invention. Therefore, the protection scope of the present invention shall be subject to the protection scope of the claims.
Claims (10)
1. A shared bicycle dispatching method is characterized in that,
and according to the real-time use data of the shared bicycle, a scheduling decision is made by a scheduling model established based on a deep reinforcement learning algorithm, and a scheduling truck is guided to schedule and transport the shared bicycle in the operating area.
2. The method according to claim 1, wherein the scheduling model predicts the usage demand of the shared bicycle in the operation area in combination with historical usage data of the shared bicycle in addition to collecting distribution data of idle shared bicycles when making scheduling decisions.
3. The shared bicycle scheduling method according to claim 2, wherein the operation area is divided into a plurality of sub-areas, the sub-areas are used as the minimum space units of the shared bicycle scheduling, and the sub-areas are divided according to the urban road pattern.
4. The shared bicycle scheduling method of claim 3, wherein the sub-area division further considers urban land properties,
the classification of urban land includes: school education area, medical service area, shopping area, public transportation junction area, residential area, and enterprise area,
each subarea takes the city main road as a boundary, and simultaneously,
setting regions in a region partitioning processMinimum limit of area SminAnd a maximum limit Smax。
5. The shared bicycle scheduling method of claim 4, wherein the preprocessing of the shared bicycle history data comprises:
when the positioning data has data vacancy and only contains starting point position or end point position information, according to the travel data statistics of other bicycles in the sub-area of the bicycle, the corresponding end point or starting point position information is randomly generated according to the distribution,
when the information of the starting point and the end point of the positioning data is completely lacked, deleting the single-vehicle travel positioning data,
and if the positioning data has errors, including overlong travel distance and abnormal vehicle speed calculated according to the positioning data and the time data, deleting the single vehicle forming data.
6. The shared bicycle scheduling method of claim 5, wherein the method of predicting the shared bicycle usage demand includes,
describing the use requirement from the use requirement of the area i to the use requirement of the area j by taking the sub-area as a basic unit;
in time, the use demand is predicted once in time granularity delta t, namely every delta t time, and different periods of working days, weekends, holidays and major activities are considered in a distinguishing way,
using a weighted moving average method, Ft＝ω1At-1+ω2At-2+……+ωnAt-n (1)
Wherein FtAs a predictor of t-time, ω1Is the weight of the actual value of the (t-1) epoch, At-1Is the actual value of the (t-1) period.
7. The method of claim 6, wherein the construction elements of the scheduling model include:
the status elements in the scheduling model include demand information of each sub-area consumer using the shared bicycle, supply information of the shared bicycle, and truck status information for scheduling the shared bicycle,
the action elements in the dispatching model refer to dispatching the freight car to move to the adjacent sub-area and throwing or reclaiming a plurality of shared single cars,
the reward elements in the scheduling model are set as:
if the shared bicycle thrown by the dispatching truck is used in the preset unit time, the environment gives a positive reward, otherwise, the environment gives a negative reward,
if the shared bicycle recovered by the dispatching truck is not needed to be used within the preset unit time, the environment gives a positive reward, and conversely gives a negative reward.
8. The shared bicycle scheduling method according to claim 7, wherein a training effect of the scheduling model is evaluated using a turnover rate of a bicycle as an evaluation index, and a hyper-parameter of the scheduling model is adjusted according to the training effect to optimize the model.
9. A shared bicycle scheduling platform, comprising a server having a memory; and
a processor coupled to the memory, the processor configured to execute instructions stored in the memory, the processor to:
and according to the real-time use data of the shared bicycle, a scheduling decision is made by a scheduling model established based on a deep reinforcement learning algorithm, and a scheduling truck is guided to schedule and transport the shared bicycle in the operating area.
10. A storage medium on which a computer program is stored which, when executed by a processor, carries out the method of any one of claims 1 to 8.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110109869.XA CN112766591A (en) | 2021-01-27 | 2021-01-27 | Shared bicycle scheduling method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110109869.XA CN112766591A (en) | 2021-01-27 | 2021-01-27 | Shared bicycle scheduling method |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN112766591A true CN112766591A (en) | 2021-05-07 |
Family
ID=75706023
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110109869.XA Pending CN112766591A (en) | 2021-01-27 | 2021-01-27 | Shared bicycle scheduling method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112766591A (en) |
Cited By (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113743797A (en) * | 2021-09-08 | 2021-12-03 | 北京化工大学 | Pile-free shared bicycle operation scheduling strategy optimization method and system based on big data |
| CN114254856A (en) * | 2021-11-08 | 2022-03-29 | 武汉小安科技有限公司 | Shared single-vehicle operation and maintenance scheduling method and system based on honeycomb map |
| CN114331299A (en) * | 2022-03-11 | 2022-04-12 | 北京骑胜科技有限公司 | Data processing method and device, electronic equipment and computer readable storage medium |
| CN114970944A (en) * | 2022-03-29 | 2022-08-30 | 武汉大学 | Order matching and vehicle repositioning method based on multi-agent reinforcement learning |
| CN115936409A (en) * | 2023-02-27 | 2023-04-07 | 北京阿帕科蓝科技有限公司 | Work order information generation method and device and computer equipment |
| CN116363854A (en) * | 2023-05-31 | 2023-06-30 | 北京阿帕科蓝科技有限公司 | Shared travel vehicle dispatching method and device and computer equipment |
| CN116824861A (en) * | 2023-08-24 | 2023-09-29 | 北京亦庄智能城市研究院集团有限公司 | Method and system for scheduling sharing bicycle based on multidimensional data of urban brain platform |
| CN116934056A (en) * | 2023-09-14 | 2023-10-24 | 北京阿帕科蓝科技有限公司 | Vehicle scheduling method, device, computer equipment and storage medium |
| CN114254856B (en) * | 2021-11-08 | 2024-05-31 | 武汉小安科技有限公司 | Shared single-vehicle operation and maintenance scheduling method and system based on honeycomb diagram |
Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105095980A (en) * | 2014-05-04 | 2015-11-25 | 黄山市黄山风景区供水有限公司 | Mountain scenic area water use optimizing and scheduling method |
| CN107480807A (en) * | 2017-07-06 | 2017-12-15 | 中山大学 | Shared bicycle destination Forecasting Methodology and device based on space-time layered perception neural networks |
| CN108596727A (en) * | 2018-04-26 | 2018-09-28 | 重庆邮电大学 | A kind of management of shared bicycle and decision-making technique |
| CN108805402A (en) * | 2018-04-28 | 2018-11-13 | 北京机械设备研究所 | A kind of shared bicycle scheduling system based on ectoskeleton |
| CN111080116A (en) * | 2019-12-11 | 2020-04-28 | 东南大学 | Cooperative scheduling method for shared bicycle and public bicycle |
| CN111376954A (en) * | 2020-06-01 | 2020-07-07 | 北京全路通信信号研究设计院集团有限公司 | Train autonomous scheduling method and system |
| CN111598481A (en) * | 2020-05-26 | 2020-08-28 | 山东交通学院 | Shared bicycle flow system, automatic scheduling system and method based on sub-area division |
| CN111862579A (en) * | 2020-06-10 | 2020-10-30 | 深圳大学 | Taxi scheduling method and system based on deep reinforcement learning |
| CN112068515A (en) * | 2020-08-27 | 2020-12-11 | 宁波工程学院 | Full-automatic parking lot scheduling method based on deep reinforcement learning |
-
2021
- 2021-01-27 CN CN202110109869.XA patent/CN112766591A/en active Pending
Patent Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105095980A (en) * | 2014-05-04 | 2015-11-25 | 黄山市黄山风景区供水有限公司 | Mountain scenic area water use optimizing and scheduling method |
| CN107480807A (en) * | 2017-07-06 | 2017-12-15 | 中山大学 | Shared bicycle destination Forecasting Methodology and device based on space-time layered perception neural networks |
| CN108596727A (en) * | 2018-04-26 | 2018-09-28 | 重庆邮电大学 | A kind of management of shared bicycle and decision-making technique |
| CN108805402A (en) * | 2018-04-28 | 2018-11-13 | 北京机械设备研究所 | A kind of shared bicycle scheduling system based on ectoskeleton |
| CN111080116A (en) * | 2019-12-11 | 2020-04-28 | 东南大学 | Cooperative scheduling method for shared bicycle and public bicycle |
| CN111598481A (en) * | 2020-05-26 | 2020-08-28 | 山东交通学院 | Shared bicycle flow system, automatic scheduling system and method based on sub-area division |
| CN111376954A (en) * | 2020-06-01 | 2020-07-07 | 北京全路通信信号研究设计院集团有限公司 | Train autonomous scheduling method and system |
| CN111862579A (en) * | 2020-06-10 | 2020-10-30 | 深圳大学 | Taxi scheduling method and system based on deep reinforcement learning |
| CN112068515A (en) * | 2020-08-27 | 2020-12-11 | 宁波工程学院 | Full-automatic parking lot scheduling method based on deep reinforcement learning |
Cited By (14)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113743797A (en) * | 2021-09-08 | 2021-12-03 | 北京化工大学 | Pile-free shared bicycle operation scheduling strategy optimization method and system based on big data |
| CN113743797B (en) * | 2021-09-08 | 2023-08-04 | 北京化工大学 | Pile-free sharing bicycle operation scheduling strategy optimization method and system based on big data |
| CN114254856B (en) * | 2021-11-08 | 2024-05-31 | 武汉小安科技有限公司 | Shared single-vehicle operation and maintenance scheduling method and system based on honeycomb diagram |
| CN114254856A (en) * | 2021-11-08 | 2022-03-29 | 武汉小安科技有限公司 | Shared single-vehicle operation and maintenance scheduling method and system based on honeycomb map |
| CN114331299A (en) * | 2022-03-11 | 2022-04-12 | 北京骑胜科技有限公司 | Data processing method and device, electronic equipment and computer readable storage medium |
| CN114331299B (en) * | 2022-03-11 | 2022-07-22 | 北京骑胜科技有限公司 | Data processing method and device, electronic equipment and computer readable storage medium |
| CN114970944A (en) * | 2022-03-29 | 2022-08-30 | 武汉大学 | Order matching and vehicle repositioning method based on multi-agent reinforcement learning |
| CN115936409A (en) * | 2023-02-27 | 2023-04-07 | 北京阿帕科蓝科技有限公司 | Work order information generation method and device and computer equipment |
| CN116363854A (en) * | 2023-05-31 | 2023-06-30 | 北京阿帕科蓝科技有限公司 | Shared travel vehicle dispatching method and device and computer equipment |
| CN116363854B (en) * | 2023-05-31 | 2023-10-24 | 北京阿帕科蓝科技有限公司 | Shared travel vehicle dispatching method and device and computer equipment |
| CN116824861B (en) * | 2023-08-24 | 2023-12-05 | 北京亦庄智能城市研究院集团有限公司 | Method and system for scheduling sharing bicycle based on multidimensional data of urban brain platform |
| CN116824861A (en) * | 2023-08-24 | 2023-09-29 | 北京亦庄智能城市研究院集团有限公司 | Method and system for scheduling sharing bicycle based on multidimensional data of urban brain platform |
| CN116934056A (en) * | 2023-09-14 | 2023-10-24 | 北京阿帕科蓝科技有限公司 | Vehicle scheduling method, device, computer equipment and storage medium |
| CN116934056B (en) * | 2023-09-14 | 2024-03-15 | 北京阿帕科蓝科技有限公司 | Vehicle scheduling method, device, computer equipment and storage medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN112766591A (en) | Shared bicycle scheduling method | |
| CN108162771B (en) | Intelligent charging navigation method for electric automobile | |
| Brandstätter et al. | Overview of optimization problems in electric car-sharing system design and management | |
| Yi et al. | Data-driven optimal charging decision making for connected and automated electric vehicles: A personal usage scenario | |
| Li et al. | Benefit analysis of shared depot resources for multi-depot vehicle routing problem with fuel consumption | |
| CN108776852A (en) | Without stake vehicle dispatching method and system | |
| Li et al. | Public charging station localization and route planning of electric vehicles considering the operational strategy: A bi-level optimizing approach | |
| CN103745110B (en) | Method of estimating operational driving range of all-electric buses | |
| CN104442825A (en) | Method and system for predicting remaining driving mileage of electric automobile | |
| Fotouhi et al. | Electric vehicle energy consumption estimation for a fleet management system | |
| CN109934403A (en) | Charge load Analysis prediction technique in electric car resident region based on mathematical model | |
| Volodarets et al. | Optimization of vehicle operating conditions by using simulation modeling software | |
| EP4138051A1 (en) | Method of collecting data from fleet of vehicles | |
| CN117546218A (en) | Distributed multitasking machine learning for traffic prediction | |
| CN116798218A (en) | Urban low-carbon traffic big data detection method based on digital twinning | |
| Miao et al. | Highly Automated Electric Vehicle (HAEV)-based mobility-on-demand system modeling and optimization framework in restricted geographical areas | |
| CN106682759B (en) | Battery supply system for electric taxi and network optimization method | |
| CN111784027A (en) | Urban range electric vehicle charging demand prediction method considering geographic information | |
| CN113222248B (en) | Automatic taxi-driving charging pile selection method | |
| CN104616066B (en) | A kind of acquisition method and system of driving usage data | |
| CN115600710A (en) | Method for predicting regional demand of manned vehicle, electronic device and storage medium | |
| Kuehnel et al. | Shifts in perspective: Operational aspects in (non-) autonomous ride-pooling simulations | |
| Liu et al. | Real‐Time Return Demand Prediction Based on Multisource Data of One‐Way Carsharing Systems | |
| Arasu et al. | Energy optimal routing of a delivery vehicle fleet with diverse powertrains | |
| CN113869545A (en) | Method and system for predicting power consumption of unmanned tramcar |
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 | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20210507 |
|
| RJ01 | Rejection of invention patent application after publication |