CN109754638A - A kind of parking space allocation method based on distributed computing technology - Google Patents
A kind of parking space allocation method based on distributed computing technology Download PDFInfo
- Publication number
- CN109754638A CN109754638A CN201910124113.5A CN201910124113A CN109754638A CN 109754638 A CN109754638 A CN 109754638A CN 201910124113 A CN201910124113 A CN 201910124113A CN 109754638 A CN109754638 A CN 109754638A
- Authority
- CN
- China
- Prior art keywords
- individual
- user
- information
- matrix
- algorithm
- 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
Landscapes
- Traffic Control Systems (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The parking space allocation method based on distributed computing technology that the present invention relates to a kind of, including constructs database, user demand acquisition, data prediction, algorithm on operation line, send allocation result and terminal guidance;Constructs database: using the storage of table structure and its database progress user data based on Mysql and HBase database;User demand acquisition: collecting the parking request of user in normalization time slot, and user uses intelligent terminal, destination is chosen by way of searching for and clicking;Data prediction: receiving the stage of user demand in server, and each data information of user demand is collected in normalization time slot and user and the parking stall being related to are numbered.The beneficial effects of the present invention are: the present invention can provide a parking stall that is available and closing on target point for car owner, car owner can largely be reduced seeks ETB expected time of berthing, pool cost is sought in reduction, and has certain benefit in terms of Optimizing Urban Transportation overall situation and parking benefit.
Description
Technical field
The present invention relates to parking space allocation methods, and more specifically, it is related to a kind of parking stall based on distributed computing technology point
Method of completing the square.
Background technique
Currently, parking position difficulty, which is sought, has become large size city common fault, because the dispute that parking problem causes is commonplace.No matter
The megalopolis such as the ultra-large types city such as Beijing, Shanghai or Chongqing, Chengdu or even population only have tens of thousands of, ten tens of thousands of county towns and
The problem of small towns, parking difficulty, all brings many worries to people life and government's traffic administration.And parking problem is also brought
Multiple complications, such as the problems such as traffic congestion, it is to increase the cruise time of parking that parking problem, which directly affects, increase vehicle
Main traffic expense and living cost influence also to have indirectly, such as reduce people's lives Happiness Index, cause contradiction between car owner
Etc..
In addition, the stress-relieving activity of people is also increased with the continuous improvement of Chinese Urban Residents living standard, it is such as false
Day tourism, activity shopping, viewing performance, the daily routines such as film, a large amount of at this time and concentration home vehicle traveler is according to purpose
Ground information, the place for going selection distance nearest is stopped, the result is that all can not be real caused by the consistent parking target of these travelers
Existing set parking target, all may cause the confusion of regional traffic, brings serious traffic problems.
In China, can not accomplish in city parking position long-term plan and short time under the situation of enlarging large parking lot, if
Urban traffic pressure can largely be alleviated by counting practical and effective parking mechanism, simultaneously because car owning amount
Rapid growth, design parking mechanism be also an indispensable ring for the following intelligent transportation system.For existing traffic
Situation, researchers have contemplated many possible traffic congestion situations, have designed efficient and feasible parking machine in this scenario
System helps directly to reduce and seeks pool flow, reduces unnecessary traffic problems situation, with enlarging parking lot and progress traffic control
The methods of have same importance.And under the solution for assignment problem of stopping can be combined with problem and worst case most
The growth of excellent complexity scale and increase.Existing parking space allocation mostly uses greedy algorithm on the market, is vehicle allocation one
A nearest parking stall, how distributing parking stall there is no research, parking space allocation is uneven and the social effect of optimization to optimize and solve
Benefit.Under the overall background of construction wisdom traffic and big data era, how using big data technology it to be efficiently modified existing parking
Bit allocation method is the project for being worth research.
Patent 201810235965.7 " a kind of parking space allocation system based on big data " provides a kind of based on big number
According to parking space allocation system, it is hard that this method deploys server managed in bottom sensor and line for parking navigation etc.
Part facility realizes the real time monitoring and physical navigation of parking stall availability, realizes the management for parking stall different function, help
In the management and use of inner part of parking lot, facilitates the workload and work difficulty that reduce parking lot management personnel, be intelligent control
The quantum jump in conjunction with big data technology is made, is the important innovations of parking service in urban transportation.The invention is suitable for parking
Parking management, record and the fee collection of field.Patent 201810852256.3 " parking stall distribution method and device " provides one kind
Parking stall distribution method and device, this method by obtain user information, type of vehicle, predeterminated position, for user distribute one
Available parking stall in the corresponding period realizes the real time implementation management and prediction of parking stall, helps to improve the utilization rate on parking stall,
Parking stall is able to achieve in the ordering management of different periods and the disclosure of parking space information, is parking stall management and computerized algorithm technology
Good combination is of great significance to traditional parking lot in computer technology fusion.Both methods is mainly taken based on biography
The parking stall perception of sensor and distribution system on the line based on user demand and parking stall real time availability with realize to user
The parking stall of parking demand is distributed.But these methods only realize the real-time distribution for parking stall availability, it is not wired worthwhile
Method carries out the pool and distribution of whole parking stall, cannot plan the lower parking route of cost at the first time for user, can not lead to
Algorithm Optimized Matching information on line is crossed, and then achievees the purpose that optimize public parking resource, therefore stop without fundamentally improving
Parking stall allocative efficiency.
Summary of the invention
The purpose of the present invention is overcoming deficiency in the prior art, a kind of parking space allocation based on distributed computing technology is provided
Method.
This parking space allocation method based on distributed computing technology, including constructs database, user demand acquisition, data are pre-
Algorithm, transmission allocation result and terminal guidance in processing, operation line;
Constructs database: the design of database table structure, storage and inquiry for user information have conclusive
It influences, and then affects the operation of whole system success;This patent uses the table structure based on Mysql and HBase database
And its database carries out the storage of user data.
User demand acquisition: collecting the parking request of user in normalization time slot, and user uses intelligent terminal, passes through search
Destination is chosen with the mode clicked, and destination coordinate information and user's changing coordinates information are then passed through into BD09II (Baidu
Latitude and longitude coordinates) form return to server, server calculates user to the cost on the neighbouring parking stall in destination, and to
Family and parking stall are numbered, by the data information memory of each user into user message table.
Data prediction: receiving the stage of user demand in server, needs to collect user demand in normalization time slot
Each data information and user and the parking stall being related to are numbered, the information of collection includes user's changing coordinates number
Parking stall coordinate data according to, customer objective is closed on to coordinate data, customer objective, and by above-mentioned data information, is calculated
Out user arrive at each parking stall apart from cost numerical value.
Run algorithm on line: in the algorithm operation phase, the preprocessed data in normalization time slot is aggregated into cost by system
Matrix, algorithm in input line, operation obtains including to distribute the allocation matrix of information, and according to allocation matrix and user information
Number in table is decoded, and obtains the allocation result of each user.
Send allocation result: every allocation result is sent to corresponding user by the allocation result run according to algorithm
Intelligent terminal.
Terminal guidance: intelligent terminal carries out real-time navigation, guidance user to calculation according to the allocation result received, to user
The parking stall of method distribution.
The system modular structure of this method is as shown in Figure 1, the specific implementation steps are as follows:
Step 1, constructs database
According to algorithm need data carry out table structure, using include Customs Assigned Number, user's changing coordinates data,
Customer objective coordinate data and up to parking bit number table structure to store user information, with include parking stall compile
Number, parking stall coordinate, parking stall can with situation and up to Customs Assigned Number table structure to store parking space information, with including
Intermediate quantity of the table structure to store algorithm on line of Customs Assigned Number, parking bit number, running distance.
Step 2, user demand acquisition
In normalization time slot, system carries out information collection by intelligent terminal, and user uses intelligent terminal, passes through search
Destination is chosen with the mode clicked, then by destination coordinate information and user's changing coordinates information by way of BD09II
Server is returned to, the user information received is numbered and stored into user message table by server, and server is according to user
Destination information retrieves surrounding available parking places, and parking space information number is stored into parking stall table.
Step 3, data prediction
By user message table in step 2 and parking space information table, the reachable parking stall of each user and corresponding is calculated
Running distance stores Customs Assigned Number, parking bit number and corresponding running distance into scale among algorithm.
Algorithm in step 4, operation line
Step 4.1, algorithm principle
This patent carries out operation on line using the innovatory algorithm based on genetic algorithm, and genetic algorithm has good be distributed
Property, algorithm flow, which can be divided mainly into individual generation, fitness calculates, screen parent, crisscross inheritance and output condition determines.It should
The flow chart of algorithm is as shown in Figure 2.
The distributed implementation of step 4.2, algorithm
The Map stage: the individual matrix information being stored in HDFS is input in thread by the Mapping stage, on node
Population carry out non-dominated ranking operation, and by wheel disc gamble screening technique select current parent generation and by the parent of generation into
Row Shuffling shuffle operation is passed to reduce.The input of this stage is to be assigned to several individual informations of the node, exports and is
<id, (i, des (i), d (i))>, wherein the integer value that the number that the id as key value is individual is rounded downwards later divided by 10,
Key value that will be current is set as the number of reduce stage node, can so guarantee every thread needs of reduce stage
The number of individuals of processing is no more than ten, and value value is numbered by individual, the matrix information composition of individual adaptation degree and individual.
The Reduce stage: the input in Reduce stage is the key-value pair with same id node serial number as key value, i.e., <
Id, (i, des (i), d (i)) >, corresponding individual is assigned to the same Reduce by identical key value id and saved by the Reduce stage
On point, individual data are the individual matrix information obtained inside value value and individual adaptation degree in the Reduce stage.It is logical
It crosses and judges whether there is the individual for meeting output condition or population on node, merge output if meeting, to node if being unsatisfactory for
On population carry out crisscross inheritance and mutation operation, generate new individual, new individual id retains parent id, and number i is according to suitable
Sequence inherits parent number, the offspring individual of generation is saved into new individual information matrix, the Map's as next stage is defeated
Enter value.
Step 5 sends allocation result
The maximum individual of fitness is concentrated to appoint if there are multiple optimum individuals by finding the solution that algorithm obtains in step 4
It takes first, finding the corresponding parking bit number j of each car number i by decoding.Decoding process is traversal matching matrix,
The columns j in the i-th row for 1 is found, then Customs Assigned Number i assigned parking bit number is j.Every matching result is passed through into service
Device is sent to corresponding user's intelligent terminal.Matching result includes that Customs Assigned Number, parking bit number, running distance and parking stall are sat
Mark information.
Step 6, terminal guidance
After the matching result that user's intelligent terminal is sent in receiving step 5, by traffic navigation function, with matching result
Shown in parking stall be destination carry out driving navigation.
As preferred: the step 4.1 the following steps are included:
Step 4.1.1, individual generates
Use dijIndicate that Customs Assigned Number i reaches the running distance of parking bit number j, by current in scale among server algorithm
It includes all d that all match informations are generated according to number order in time slotijDistance matrix Dij。
Define binary decision variable (xij)i∈Nt, j ∈ Mt, to indicate the occupancy on parking stall after distributing, see public affairs
Formula (1):
Variable x in formula (1)ijThe matrix X of compositionijMeet the constraint item that each row and column at most only one value is not 0
Part, the value in the matrix for 1 represent corresponding vehicle i and have been matched to corresponding parking stall j, and physical significance is: a vehicle can
To be assigned to a parking stall, and a vehicle can only be at most distributed on a parking stall.
According to generated matrix DijAnd XijAnd its constraint condition, system model is obtained according to the target of algorithm, is seen
Formula (2):
xij∈ { 0,1 }, i ∈ Nt, j ∈ Mt
It is the decimal system and binary system that genetic algorithm gene, which encodes common type, and binary coding in general operation is fast,
Meet minimum character set principle, convenient for being analyzed with schemata theorem, but also has Hamming distance larger simultaneously, continuous function is discrete
Mapping error when change, the shortcomings that being easily trapped into local optimum;Decimal coded is also known as real coding, and real coding is hereditary calculation
A kind of coding mode generally used in method when solving continuous parameter optimization problem, precision with higher are indicating continuous
There is advantage, but there is also make a variation the disadvantages of unstable in terms of gradual change problem.This patent uses the shape of binary matrix coding
Formula, each position only has a bit in matrix, represents corresponding vehicle i with 1 and is assigned to parking stall j, use 0 indicates vehicle not
The parking stall being matched to generates the matching matrix for meeting constraint condition as follows:
According to data in scale among algorithm, the running distance read between user and parking stall in normalization time slot is believed
Breath generates Distance matrix Dij, according to matrix dimension, random generation includes the matching matrix of Different matching information, these matchings
Matrix is individual, and in interative computation later, individual total quantity is remained unchanged, and individual is exemplified below:
Step 4.1.2, fitness calculates
Fitness is calculated including linear calibration, dynamic linear calibration, power law calibration, logarithm calibration and index calibrating etc., respectively
The corresponding data characteristics of kind of calculation is also different, this algorithm data feature has individual numerical value big, and between individual gap compared with
Greatly, it is more suitable for dynamic linear calibration, therefore this algorithm is demarcated using dynamic linear.
After generating individual, each individual is one group of matching matrix for storing the match information of vehicle and parking stall, will be a
The allocation matrix and cost matrix of body carry out following Matrix Multiplication and calculate operation:
It obtains as follows only comprising the individual cost matrix of vehicle and the range information of corresponding parking stall:
Required for then having been obtained after being allocated according to the individual to all numerical value traversal summations in this individual cost matrix
Cost summation des (i) the individual cost value of maximum in population is subtracted by current individual cost by the method for Linear Program
Value, then adds the k power of several ξ one minimum, and k is represented the number of iterations, the fitness of the individual is denoted as with this result, sees formula
(3):
F '=fmax-f+ξk (3)
Step 4.1.3, parent is screened
After acquiring individual cost and fitness, the individual choice of parent is carried out in the way of random roulette: random
Roulette according to the current fitness value of individual, divides each individual and arrives different value ranges, then generate in total size again
One random number, the individual in the value range where random number is selected, and repetitive operation is until parent individuality number reaches total appearance
Amount, citing are shown in Table 1.
Table 1
Individual number | 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | 10 |
Cost summation | 30 | 60 | 45 | 28 | 35 | 50 | 55 | 40 | 33 | 32 |
Fitness | 30 | 0 | 15 | 32 | 25 | 10 | 5 | 20 | 27 | 28 |
Fitness summation is 192 in table, then random integers is taken between 1 to 192, if the integer taken is added in [1,30]
One No. 1 individual is into parent group, if the integer taken adds No. 3 individuals into parent group in [31,45], if taking
Integer [46,77] then add No. 4 individuals into parent group, and so on.No. 2 individual fitness are 0 in table,
So not eliminating worst individual in a disguised form between occupied area when selecting individual.Carrying out parent selection using the method can make
Outstanding individual is easier to enter parent by selection, also ensures that non-outstanding individual organic can be selected into parent, and can drop
The worst individual chooses probability in low current population.
Step 4.1.4, crisscross inheritance
After selecting parent individuality, crisscross inheritance operation is carried out.When crossing operation, if two parent individualities are same
Body is selected, then between any two without crossing operation.Choose two Different Individuals number parent individuality, it is each individual it is each
Row selects the hereditary information as filial generation using 50% probability, generates offspring individual until offspring individual number reaches parent individuality
Number.
After offspring individual generates, if a have the match information row for not meeting allocation matrix constraint condition in vivo, search
The corresponding cost matrix information of rope conflict row, find better than current matching information or increase the relatively small match information of cost into
Row substitution.Duplicate operates when conflict, is exemplified below:
When the offspring individual of generation is as shown above, the first row conflicts with fourth line generation, is unsatisfactory for the pact of matching matrix
When beam condition, its corresponding cost matrix row is found, under the premise of not generating new conflict, searches the matching for being better than current cost
Or increase the relatively small matching of cost.
Step 4.1.5, output condition determines
It is requested according to user in time slot, the minimum cost and Des (min) under unconfined condition is acquired by greedy algorithm,
If current optimum individual reaches the level of minimum cost Des (min), optimum individual is exported as optimal solution.
When the fitness of individual, which reaches, is expected or is unchanged, the highest individual of fitness is exported, is obtained
Pareto is dominant disaggregation.
The beneficial effects of the present invention are: the invention proposes a kind of parking space allocation method based on distributed computing technology, it should
Method can provide a parking stall that is available and closing on target point for car owner, and can largely reduce car owner seeks pool
Time, pool cost is sought in reduction, and has certain benefit in terms of Optimizing Urban Transportation overall situation and parking benefit.For height
Peak period data real-time and computationally intensive problem, this patent innovatively propose the distribution of the improvement based on genetic algorithm
Algorithm can carry out the distributed arithmetic of algorithm with MapReduce frame, materially increase the anti-pressure ability of system
With the accuracy of matching result.Meanwhile this distributed computing technology also mitigates the work load of back-stage management person, does not need
The adjustment and control that parameter is carried out according to road conditions and real time information, improve the effect of urban traffic control to a certain extent
Rate accelerates the construction of smart city.
Detailed description of the invention
Fig. 1 is parking distribution system modular construction figure;
Fig. 2 is genetic algorithm flow chart;
Fig. 3 is genetic algorithm MapReduce frame diagram;
Fig. 4 is user profile acquisition terminal figure;
Fig. 5 is intelligent terminal navigation picture.
Specific embodiment
The present invention is described further below with reference to embodiment.The explanation of following embodiments is merely used to help understand this
Invention.It should be pointed out that for those skilled in the art, without departing from the principle of the present invention, also
Can be with several improvements and modifications are made to the present invention, these improvement and modification also fall into the protection scope of the claims in the present invention
It is interior.
The parking space allocation method based on distributed computing technology, the specific implementation steps are as follows:
Step 1: constructs database
Devise three table structures in database table of the present invention, in Mysql: Users, Parkingslot,
Matchdistance, wherein table structure is corresponded with field in Fig. 2.What wherein Users table included is Customs Assigned Number, user
Changing coordinates data, customer objective ground coordinate data and up to parking bit number, include in Parkingslot table is parking stall
Number, parking stall coordinate, parking stall can use situation and reachable Customs Assigned Number, and include in Matchdistance table is user's volume
Number, parking bit number, corresponding running distance.
Step 2: user demand acquires
Assuming that there is car owner to need to find the parking position near destination, found first by the map search interface of Fig. 4
Destination, and select to go to, client is by the current coordinate of user and destination coordinate with BD09II number behind selection destination
Server is returned to according to format, server is Customs Assigned Number, by the use of reference numeral according to the request sequence received in time slot
Family and user information are stored into Users table, and the reachable parking stall of user according to serial number in time slot and is stored in
In Parkingslot table.
Step 3: data prediction
By the reachable and coordinate information being stored in step 2 in Users table and Parkingslot table, calculated by distance
Interface calculates user and reaches the running distance on each parking stall, and by corresponding Customs Assigned Number, parking bit number and driving away from
From in deposit Matchdistance table.Distance matrix is generated by the sum of Customs Assigned Number and parking bit number.By apart from square
Battle array generates a volume matrix comprising Different matching information and stores into the interim table of Hbase.
Step 4: algorithm on operation line
Improved adaptive GA-IAGA based on MapReduce is realized in MapReduce distributed computing platform, it is important to
The design of map function, reduce function and jobCreate function.Map function is mainly responsible for for searching HBase sublist
Body matching matrix.Reduce function, which is mainly responsible for the output result for summarizing map function and generates, final meets output condition
Optimal solution set.JobCreate () function is used to complete user's custom-configuring and being submitted in cluster about job run
Operation.
The input data of this algorithm is stored in HBase, therefore the InputFormat of Mapreduce operation is set as
TableInputFormat.It to be also stored in due to the offspring individual of generation in the interim table of HBase, Mapreduce operation
OutputFormat be set as TableOutputFormat.It is according to HBase table when Hadoop is using HBase table as inputting
Region data divide Split, i.e. the corresponding Split of each Region is, thus also correspond to a Mapper.By setting
Setting InputFormat is TableInputFormat, and each Region is divided into<key according to rowKey by Mapper, value>
Right, key corresponds to each rowKey of the sublist, and value corresponds to the data that the row is included.MatchMapper is inherited from
TableMapper<Text, DoubleWritable>, it thus can directly handle the data in HBase table.
MatchReducer is inherited from TableReducer<Text, and DoubleWritable>, it thus can be reduce function
It exports in result write-in HBase table.MatchDriver is responsible for configuring distributed operation cluster environment, generates Mapreduce operation
And submit execution.
The major function of Mapper is to input several body matrix and screen according to the sequence of fitness size with parent, so
After send in Reducer and handle.Realize that the key code of map function is as follows:
The major function of Reducer is to summarize the output of each Mapper as a result, and being ranked up according to fitness size
After export.Reducer in this system determines whether to meet output by comparing the fitness and individual cost value of individual
Condition.The key code of reduce function is as follows:
After realizing map function and reduce function, it is also necessary to be arranged Mapreduce operation operation information and
The calling of map and reduce function.The realization class name of Mapper, the realization class name of Reducer, InputFormat format,
Position of input data and output data of OutputFormat format and operation etc., key code is as follows:
After configuring customized operation, so that it may carry out the operation of algorithm, key code is as follows:
Job job=jobCreate ();
Flag=job.waitForCompletion (true);
Step 5: sending allocation result
In step 4, the allocation matrix obtained by operation result is decomposed line by line, and each Customs Assigned Number is corresponding
Parking bit number, be stored in Matchdistance table, and the match information for Customs Assigned Number being corresponded in table is sent to use
The intelligent terminal at family.
Step 6: terminal guidance
User carries out real-time navigation according to the current location of the parking stall match information and user received, and navigate function
It can be using being realized based on the navigation module in Baidu map developer tool, as shown in Figure 5.
The present invention establishes information storage when realizing the parking space allocation method based on distributed computing technology proposed
MySql database, the Hbase database with storage algorithm initial matrix.The system uses 4 PC machine, three models
Dell Optiplex7050MT, a bench-type Dell Inspiron 7557, intel core i7-6700CPU, 8G memory,
GTX 960GPU, 1T HDD, 256G SSD.Wherein an installation Window10 steerable system is as exploitation host, remaining 3 peaces
Linux CentOS6.4 operating system is filled as working cluster.
Claims (2)
1. a kind of parking space allocation method based on distributed computing technology, which is characterized in that adopted including constructs database, user demand
Collection, data prediction run algorithm on line, send allocation result and terminal guidance;
Constructs database: table structure and its database based on Mysql and HBase database is used to carry out depositing for user data
Storage;
User demand acquisition: collecting the parking request of user in normalization time slot, and user uses intelligent terminal, passes through search and point
The mode of choosing chooses destination, then returns to destination coordinate information and user's changing coordinates information by way of BD09II
To server, server calculates user to the cost on parking stall near destination, and user and parking stall are numbered, will
The data information memory of each user is into user message table;
Data prediction: receiving the stage of user demand in server, and each number of user demand is collected in normalization time slot
It is believed that ceasing and user and the parking stall being related to being numbered, the information of collection includes user's changing coordinates data, user
Parking stall coordinate data is closed on to destination coordinate data and customer objective, and by data information, calculates user's arrival
Each parking stall apart from cost numerical value;
Run algorithm on line: in the algorithm operation phase, the preprocessed data in normalization time slot is aggregated into cost matrix by system,
Algorithm in input line, operation obtains including the allocation matrix for distributing information, and according in allocation matrix and user message table
Number be decoded, obtain the allocation result of each user;
Send allocation result: every allocation result is sent to the intelligence of corresponding user by the allocation result run according to algorithm
It can terminal;
Terminal guidance: intelligent terminal carries out real-time navigation, guidance user to algorithm point according to the allocation result received, to user
The parking stall matched;
Realize that steps are as follows:
Step 1, constructs database
Table structure is carried out according to the data that algorithm needs, using including Customs Assigned Number, user's changing coordinates data, user
Destination coordinate data and up to parking bit number table structure to store user information, with include parking bit number, stop
Parking stall coordinate, parking stall can with situation and up to Customs Assigned Number table structure to store parking space information, with including user
The intermediate quantity of number, the table structure of stop bit number, running distance to store algorithm on line;
Step 2, user demand acquisition
In normalization time slot, system carries out information collection by intelligent terminal, and user uses intelligent terminal, passes through search and point
The mode of choosing chooses destination, then returns to destination coordinate information and user's changing coordinates information by way of BD09II
To server, the user information received is numbered and is stored into user message table by server, and server is according to customer objective
Available parking places around ground information retrieval, and parking space information number is stored into parking stall table;
Step 3, data prediction
By user message table in step 2 and parking space information table, the reachable parking stall of each user and corresponding driving are calculated
Distance stores Customs Assigned Number, parking bit number and corresponding running distance into scale among algorithm;
Algorithm in step 4, operation line
Step 4.1, algorithm principle
Using based on genetic algorithm innovatory algorithm carry out line on operation, genetic algorithm have it is good can distributivity, algorithm
Process, which is divided into individual generation, fitness calculates, screens parent, crisscross inheritance and output condition determines;
The distributed implementation of step 4.2, algorithm
The Map stage: the individual matrix information being stored in HDFS is input in thread by the Mapping stage, to the kind on node
Group carries out non-dominated ranking operation, and selects current parent's generation by the screening technique that wheel disc is gambled and carry out the parent of generation
Shuffling shuffle operation is passed to reduce;The input of this stage is to be assigned to several individual informations of the node, export for <
Id, (i, des (i), d (i)) >, wherein the integer value that the number that the id as key value is individual is rounded downwards later divided by 10,
Key value that will be current is set as the number of reduce stage node, can so guarantee every thread needs of reduce stage
The number of individuals of processing is no more than ten, and value value is numbered by individual, the matrix information composition of individual adaptation degree and individual;
The Reduce stage: the input in Reduce stage is key-value pair with same id node serial number as key value, i.e., < id,
(i, des (i), d (i)) >, corresponding individual is assigned to the same Reduce node by identical key value id by the Reduce stage
On, individual data are the individual matrix information obtained inside value value and individual adaptation degree in the Reduce stage;Pass through
Judge whether there is the individual for meeting output condition or population on node, merges output if meeting, on node if being unsatisfactory for
Population carry out crisscross inheritance and mutation operation, generate new individual, new individual id retains parent id, and number i is in sequence
Parent number is inherited, the offspring individual of generation is saved into new individual information matrix, the input of the Map as next stage
Value;
Step 5 sends allocation result
The maximum individual of fitness is concentrated to appoint and take if there are several optimum individuals by finding the solution that algorithm obtains in step 4
First, finding the corresponding parking bit number j of each car number i by decoding;Decoding process is traversal matching matrix, is looked for
The columns j for being 1 into the i-th row, then Customs Assigned Number i assigned parking bit number is j;Every matching result is passed through into server
It is sent to corresponding user's intelligent terminal;Matching result includes Customs Assigned Number, parking bit number, running distance and parking stall coordinate
Information;
Step 6, terminal guidance
After the matching result that user's intelligent terminal is sent in receiving step 5, by traffic navigation function, shown in matching result
Parking stall be destination carry out driving navigation.
2. the parking space allocation method according to claim 1 based on distributed computing technology, which is characterized in that the step
4.1 the following steps are included:
Step 4.1.1, individual generates
Use dijIndicate that Customs Assigned Number i reaches the running distance of parking bit number j, by current time slots in scale among server algorithm
It includes all d that interior all match informations are generated according to number orderijDistance matrix Dij;
Define binary decision variable (xij)i∈Nt, j ∈ Mt, to indicate the occupancy on parking stall after distributing, see formula
(1):
Variable x in formula (1)ijThe matrix X of compositionijMeet the constraint condition that each row and column at most only one value is not 0, it should
Value in matrix for 1 represents corresponding vehicle i and has been matched to corresponding parking stall j, indicates that a vehicle can be assigned to one
Parking stall, and a vehicle can only be at most distributed on a parking stall;
According to generated matrix DijAnd XijAnd its constraint condition, system model is obtained according to the target of algorithm, sees formula
(2):
Constraint condition
xij∈ { 0,1 }, i ∈ Nt, j ∈ Mt
Using the form of binary matrix coding, each position only has a bit in matrix, represents corresponding vehicle i with 1
It is assigned to parking stall j, use 0 indicates the parking stall that vehicle is not matched to, generate the matching matrix for meeting constraint condition as follows:
According to data in scale among algorithm, the user in normalization time slot and the running distance information between parking stall are read,
Generate Distance matrix Dij, according to matrix dimension, random generation includes the matching matrix of Different matching information, these matching matrixes
It is as individual, and in interative computation later, individual total quantity remains unchanged;
Step 4.1.2, fitness calculates
It includes that linear calibration, dynamic linear calibration, power law calibration, logarithm calibration and index calibrating, this algorithm are adopted that fitness, which calculates,
It is demarcated with dynamic linear;
After generating individual, each individual is one group of matching matrix for storing the match information of vehicle and parking stall, by individual
Allocation matrix and cost matrix carry out following Matrix Multiplication and calculate operation:
It obtains as follows only comprising the individual cost matrix of vehicle and the range information of corresponding parking stall:
In generation required for after being allocated according to the individual, has then been obtained to all numerical value traversal summations in this individual cost matrix
The individual cost value of maximum in population is subtracted current individual cost value, so by the method for Linear Program by valence summation des (i)
The k power of several ξ one minimum is added afterwards, k is represented the number of iterations, the fitness of the individual is denoted as with this result, see formula (3):
F '=fmax-f+ξk (3)
Step 4.1.3, parent is screened
After acquiring individual cost and fitness, the individual choice of parent is carried out in the way of random roulette: random wheel disc
Gambling according to the current fitness value of individual, divides each individual and arrives different value ranges, then generate one in total size again
Random number selects the individual in the value range where random number, and repetitive operation is until parent individuality number reaches total capacity;
Step 4.1.4, crisscross inheritance
After selecting parent individuality, crisscross inheritance operation is carried out;When crossing operation, if two parent individualities are same individual choosing
Out, then between any two without crossing operation;Choose the parent individuality of two Different Individuals number, every a line of each individual with
50% probability selects the hereditary information as filial generation, generates offspring individual until offspring individual number reaches parent individuality number
Mesh;
After offspring individual generates, if a have the match information row for not meeting allocation matrix constraint condition in vivo, punching is searched for
The prominent corresponding cost matrix information of row is found and is replaced better than current matching information or the increase relatively small match information of cost
Generation;
Step 4.1.5, output condition determines
It is requested according to user in time slot, the minimum cost and Des (min) under unconfined condition is acquired by greedy algorithm, if working as
Preceding optimum individual reaches the level of minimum cost Des (min), then exports optimum individual as optimal solution;
When the fitness of individual, which reaches, is expected or is unchanged, the highest individual of fitness is exported, Pareto is obtained and accounts for
Excellent disaggregation.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910124113.5A CN109754638B (en) | 2019-02-19 | 2019-02-19 | Parking space allocation method based on distributed technology |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910124113.5A CN109754638B (en) | 2019-02-19 | 2019-02-19 | Parking space allocation method based on distributed technology |
Publications (2)
Publication Number | Publication Date |
---|---|
CN109754638A true CN109754638A (en) | 2019-05-14 |
CN109754638B CN109754638B (en) | 2020-09-22 |
Family
ID=66406548
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201910124113.5A Active CN109754638B (en) | 2019-02-19 | 2019-02-19 | Parking space allocation method based on distributed technology |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN109754638B (en) |
Cited By (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111627246A (en) * | 2020-06-12 | 2020-09-04 | 桂林电子科技大学 | Intelligent parking space recommendation method based on image recognition and user preference |
CN113593289A (en) * | 2021-06-25 | 2021-11-02 | 桂林电子科技大学 | Method and system for sensing in-road parking conflict avoiding scheduling based on available parking position |
CN114664112B (en) * | 2022-03-16 | 2023-09-12 | 广州小鹏汽车科技有限公司 | Garage-oriented parking space recommendation method, server and storage medium |
Citations (15)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102831785A (en) * | 2012-09-14 | 2012-12-19 | 成都宇凡科技有限责任公司 | Parking space inquiry booking system and inquiry booking method thereof |
CN103400251A (en) * | 2013-08-02 | 2013-11-20 | 天津港中煤华能煤码头有限公司 | Bulk cargo wharf berth fast allocation system and method based on genetic algorithm |
CN104281894A (en) * | 2014-10-30 | 2015-01-14 | 大连海事大学 | Ship dispatch optimization method based on channel and berth resources |
CN106200577A (en) * | 2016-07-08 | 2016-12-07 | 青岛华通能源投资有限责任公司 | A kind of multi-storied garage dispatch control method |
CN107274713A (en) * | 2017-07-11 | 2017-10-20 | 东华大学 | A kind of parking induction method based on cluster |
CN107464005A (en) * | 2017-08-21 | 2017-12-12 | 中国人民解放军国防科技大学 | Expanded path planning method for vehicle reservation |
CN107633701A (en) * | 2017-10-27 | 2018-01-26 | 谢金伟 | A kind of parking space intelligent match control method |
US20180039974A1 (en) * | 2016-08-05 | 2018-02-08 | Denso International America, Inc. | System and method for selecting a parking space using a mobile device |
CN107680402A (en) * | 2017-10-27 | 2018-02-09 | 深圳市小猫信息技术有限公司 | A kind of information acquisition method, device, system, terminal device and storage medium |
CN107680403A (en) * | 2017-10-30 | 2018-02-09 | 中国联合网络通信集团有限公司 | Vehicle, which is led, stops method and system |
CN108182823A (en) * | 2017-12-14 | 2018-06-19 | 特斯联(北京)科技有限公司 | A kind of blocking wisdom management in garden parking stall and guide service system |
US20180218605A1 (en) * | 2017-02-01 | 2018-08-02 | Microsoft Technology Licensing, Llc | Automated parking lot space assignment |
CN108877286A (en) * | 2018-07-09 | 2018-11-23 | 胡佂宇 | A kind of parking stall based on block chain |
CN108921353A (en) * | 2018-07-06 | 2018-11-30 | 上海大学 | A kind of optimization method, device and the electronic equipment of parking lot dispatching |
CN109191906A (en) * | 2018-11-12 | 2019-01-11 | 广西民族师范学院 | A kind of parking lot distributed intelligence guidance system and method |
-
2019
- 2019-02-19 CN CN201910124113.5A patent/CN109754638B/en active Active
Patent Citations (15)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102831785A (en) * | 2012-09-14 | 2012-12-19 | 成都宇凡科技有限责任公司 | Parking space inquiry booking system and inquiry booking method thereof |
CN103400251A (en) * | 2013-08-02 | 2013-11-20 | 天津港中煤华能煤码头有限公司 | Bulk cargo wharf berth fast allocation system and method based on genetic algorithm |
CN104281894A (en) * | 2014-10-30 | 2015-01-14 | 大连海事大学 | Ship dispatch optimization method based on channel and berth resources |
CN106200577A (en) * | 2016-07-08 | 2016-12-07 | 青岛华通能源投资有限责任公司 | A kind of multi-storied garage dispatch control method |
US20180039974A1 (en) * | 2016-08-05 | 2018-02-08 | Denso International America, Inc. | System and method for selecting a parking space using a mobile device |
US20180218605A1 (en) * | 2017-02-01 | 2018-08-02 | Microsoft Technology Licensing, Llc | Automated parking lot space assignment |
CN107274713A (en) * | 2017-07-11 | 2017-10-20 | 东华大学 | A kind of parking induction method based on cluster |
CN107464005A (en) * | 2017-08-21 | 2017-12-12 | 中国人民解放军国防科技大学 | Expanded path planning method for vehicle reservation |
CN107633701A (en) * | 2017-10-27 | 2018-01-26 | 谢金伟 | A kind of parking space intelligent match control method |
CN107680402A (en) * | 2017-10-27 | 2018-02-09 | 深圳市小猫信息技术有限公司 | A kind of information acquisition method, device, system, terminal device and storage medium |
CN107680403A (en) * | 2017-10-30 | 2018-02-09 | 中国联合网络通信集团有限公司 | Vehicle, which is led, stops method and system |
CN108182823A (en) * | 2017-12-14 | 2018-06-19 | 特斯联(北京)科技有限公司 | A kind of blocking wisdom management in garden parking stall and guide service system |
CN108921353A (en) * | 2018-07-06 | 2018-11-30 | 上海大学 | A kind of optimization method, device and the electronic equipment of parking lot dispatching |
CN108877286A (en) * | 2018-07-09 | 2018-11-23 | 胡佂宇 | A kind of parking stall based on block chain |
CN109191906A (en) * | 2018-11-12 | 2019-01-11 | 广西民族师范学院 | A kind of parking lot distributed intelligence guidance system and method |
Non-Patent Citations (7)
Title |
---|
DIYA THOMAS 等: "A Genetic Algorithm Approach to Autonomous Smart Vehicle Parking system", 《 PROCEDIA COMPUTER SCIENCE》 * |
E. ALFONSETTI: "A Semi Distributed Approach for Min-Max Fair", 《ARXIV》 * |
卢凯 等: "停车总成本最小的停车位动态分配与诱导模型", 《华南理工大学学报(自然科学版)》 * |
李东萍: "基于分布式架构资源动态分配系统的设计与实现", 《中国优秀硕士学位论文全文数据库 信息科技辑》 * |
杨玉仲: "智能停车诱导系统的设计与实现", 《中国优秀硕士学位论文全文数据库 工程科技Ⅱ辑》 * |
王彬: "基于分布式架构的智慧停车管理服务系统", 《基于分布式架构的智慧停车管理服务系统》 * |
纪加木: "基于分布式技术的停车场管理系统的研究", 《中国优秀硕士学位论文全文数据库 信息科技辑》 * |
Cited By (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111627246A (en) * | 2020-06-12 | 2020-09-04 | 桂林电子科技大学 | Intelligent parking space recommendation method based on image recognition and user preference |
CN113593289A (en) * | 2021-06-25 | 2021-11-02 | 桂林电子科技大学 | Method and system for sensing in-road parking conflict avoiding scheduling based on available parking position |
CN113593289B (en) * | 2021-06-25 | 2023-02-28 | 桂林电子科技大学 | Method and system for sensing in-road parking conflict avoidance scheduling based on available parking position state |
CN114664112B (en) * | 2022-03-16 | 2023-09-12 | 广州小鹏汽车科技有限公司 | Garage-oriented parking space recommendation method, server and storage medium |
Also Published As
Publication number | Publication date |
---|---|
CN109754638B (en) | 2020-09-22 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
AU2015213280B9 (en) | Complex situation analysis system | |
CN103597503B (en) | Method and system for a pre-shopping reservation system with increased search efficiency | |
CN103412897B (en) | A kind of parallel data processing method based on distributed frame | |
Ota et al. | Stars: Simulating taxi ride sharing at scale | |
JP3952518B2 (en) | Multidimensional data processing method | |
US6795815B2 (en) | Computer based knowledge system | |
US8386479B2 (en) | Routing methods for multiple geographical entities | |
US20190102445A1 (en) | System and method for enabling multiple parents with weights in a multidimensional database environment | |
CN109754638A (en) | A kind of parking space allocation method based on distributed computing technology | |
Yang et al. | An improved artificial bee colony algorithm for optimal land-use allocation | |
CN102902782A (en) | Mass multisource heterogeneous spatial information data seamless integration management method | |
Zheng et al. | Online trichromatic pickup and delivery scheduling in spatial crowdsourcing | |
CN109447264B (en) | Virtual machine placement genetic optimization method based on VHAM-R model in cloud computing environment | |
US20170109631A1 (en) | System and method of multi-objective optimization for transportation arrangement | |
Lin et al. | A stacking model for variation prediction of public bicycle traffic flow | |
Yao et al. | Evaluation and development of sustainable urban land use plans through spatial optimization | |
CN111985716B (en) | Passenger traffic volume prediction system with passenger traffic information visualization function | |
CN112966862A (en) | Short-term quantitative rainfall forecasting method based on machine learning algorithm | |
CN102981913A (en) | Inference control method and inference control system with support on large-scale distributed incremental computation | |
US20150293971A1 (en) | Distributed queries over geometric objects | |
Liao et al. | Fusing geographic information into latent factor model for pick-up region recommendation | |
Panda et al. | Applications of machine learning in the planning of electric vehicle charging stations and charging infrastructure: A review | |
Niyomubyeyi et al. | An improved non-dominated sorting biogeography-based optimization algorithm for multi-objective land-use allocation: a case study in Kigali-Rwanda | |
CN115545543A (en) | Multi-disaster integrated natural disaster risk assessment and analysis system | |
Azri et al. | Classified and clustered data constellation: An efficient approach of 3D urban data management |
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 |