CN115830848A - Shared parking space intelligent distribution system and method based on LSTM model - Google Patents
Shared parking space intelligent distribution system and method based on LSTM model Download PDFInfo
- Publication number
- CN115830848A CN115830848A CN202211312801.2A CN202211312801A CN115830848A CN 115830848 A CN115830848 A CN 115830848A CN 202211312801 A CN202211312801 A CN 202211312801A CN 115830848 A CN115830848 A CN 115830848A
- Authority
- CN
- China
- Prior art keywords
- parking
- data
- time
- parking space
- vehicle
- 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
Images
Landscapes
- Traffic Control Systems (AREA)
Abstract
The application provides an LSTM model-based intelligent shared parking space distribution system and method, and the method comprises the following specific steps: collecting historical parking data and environmental data of a plurality of parking lots in a target area; respectively dividing historical data of a single parking lot into a plurality of data sets according to the environmental data; respectively constructing a residual parking space prediction model based on an LSTM model aiming at a plurality of data sets; collecting information of a vehicle to be parked, environmental parameters and real-time data of a parking lot in a target area, dividing the information, the environmental parameters and the real-time data into corresponding data sets, and predicting the number of parking lots through a corresponding residual parking lot prediction model; and allocating parking positions to the vehicles to be parked according to the predicted number of the remaining parking positions. According to the method and the device, the parking time of the target parking lot is calculated according to the predicted remaining number of the parking lots, the parking time of the vehicle to be parked is shortened, and the traffic jam condition of the target area is relieved.
Description
Technical Field
The invention relates to the field of shared parking space allocation, in particular to an LSTM (least Square) model-based intelligent shared parking space allocation system and method.
Background
With the improvement of living standard and the rapid development of urbanization, the quantity of retained urban motor vehicles is frequently high, the increasing rate of urban roads and shared parking spaces is far behind the increase of the quantity of retained motor vehicles, and the supply-demand relationship between the quantity of retained motor vehicles and the shared parking spaces becomes more tense as time goes on, so that how to allocate the relationship between the motor vehicles to be parked and the shared parking spaces on the existing basis becomes the most important ring for solving traffic jam, relieving parking difficulty and shortening parking time.
The existing shared parking space allocation methods mostly predict the remaining parking spaces of a target garage at a target time point through historical data and allocate parking spaces to vehicles to be parked through prediction results, and the methods relieve the problems of difficult parking and traffic jam to a certain extent; however, it does not consider that vehicles to be parked in many difficult parking areas often need to queue for long time to park, and also does not consider the difficulty of parking in parking lots with different idle rates, so that a parking space allocation scheme cannot be accurately given.
Disclosure of Invention
An object of the present invention is to provide an intelligent allocation method for shared parking spaces based on an LSTM model.
The invention is realized by the technical scheme, and the method comprises the following specific steps:
1) Data acquisition: collecting historical parking data and environmental data of a plurality of parking lots in a target area;
2) Data diversity: respectively dividing historical data of a single parking lot into a plurality of data sets according to the environmental data;
3) Constructing a model: respectively constructing a residual parking space prediction model based on an LSTM model aiming at a plurality of data sets;
4) And (3) predicting parking positions: collecting information of a vehicle to be parked, environmental parameters and real-time data of a parking lot in a target area, dividing the information into corresponding data sets in the step 2) according to the environmental parameters, and predicting the residual number of parking positions through a residual parking position prediction model corresponding to the data sets;
5) Parking space allocation: and allocating parking positions to the vehicles to be parked according to the predicted number of the remaining parking positions.
Further, the data acquisition in the step 1) comprises the following specific steps:
1-1) collecting historical parking data of m parking lots in an area to be allocated by taking days as a unit, wherein the historical parking data comprises: sampling time point parking lot parking space idle rate
Number of vehicles coming out of parking lot within single sampling period delta t
Maximum queuing length of parking lot entrance in single sampling period delta t
Obtaining a single parking lot
Wherein a belongs to M, t is sampling time, and M is the maximum sampling days of a single parking lot;
1-2) collecting environmental data in an area to be allocated by taking days as units, wherein the environmental data comprises: weather, temperature, holiday/weekday information.
Further, the specific method of data diversity in step 2) is:
2-1) dividing historical parking data into a plurality of data sets according to weather, temperature, holiday/workday data, wherein the data sets comprise:
a first data set of sunny-low-working day; a second data set of sunny-low-holiday, a third data set of sunny-medium-workday, a fourth data set of sunny-medium-holiday, a fifth data set of sunny-high-workday, a sixth data set of sunny-high-holiday, a seventh data set of rain/snow-low-workday, an eighth data set of rain/snow-low-holiday, a ninth data set of rain-medium-workday, a tenth data set of rain-medium-holiday, an eleventh data set of rain-high-workday, and a twelfth data set of rain-high-holiday;
2-2) respectively carrying out abnormal value elimination and normalization processing on the data of the data sets, and respectively dividing the processed data sets into training set data and test set data;
2-3) sampling time point parking lot parking space idle rate with training set and test set respectively in units of days
And sequencing according to the sampling time to respectively obtain the time sequence of the parking space idle rate of the parking lot.
Further, the specific steps of constructing and training the model in the step 3) are as follows:
3-1) respectively constructing a parking space idle rate prediction model based on an LSTM model for twelve data sets, wherein an Adam function is selected as a model optimizer, and initial parameters of the model comprise: initial learning rate l r The number of hidden layer units, the iteration times epoch, the minimum training batch and the error threshold;
3-2) respectively training twelve parking space idle rate prediction models based on the LSTM model by using twelve data sets: to train the parking idle rate eta of any t moment in the set t The parking space idle rate eta of the time points of the training set t-5 delta t, t-4 delta t, t-3 delta t, t-2 delta t and t-delta t is used as the output of the LSTM model t-5△t 、η t-4△t 、η t-3△t 、η t-2△t 、η t-△t Training the LSTM model for inputting the LSTM model;
3-3) detecting whether the prediction loss of the parking space idling rate prediction model is smaller than a threshold value or whether the iteration times are larger than or equal to the maximum iteration times, if so, outputting model parameters, outputting the trained parking space idling rate prediction model, and if not, returning to the step 3-2) to continue training.
Further, the step 4) of predicting the parking position comprises the following specific steps:
4-1) obtaining the information of the vehicle to be parked and calculating the time t required by the vehicle to go to the target area q Collecting weather, temperature, holiday/workday data of the day to be predicted, and dividing the data into corresponding data sets in the step 2) according to the weather, temperature, holiday/workday data;
4-2) setting the current time as t d Collecting t d -4△t、t d -3△t、t d -2△t、t d -△t、t d Temporal parking space idle rate
Inputting the parking space idle rate into the parking space idle rate prediction model trained in the step 3-3), and calculating the t d At time + Deltat
Make a prediction, then pass
To pair
Making prediction, and analogizing until t d +n△t≥t q Output t q Time-predicted parking space idle rate
Further, the specific steps of parking space allocation in step 5) are as follows:
5-1) passing t in step 4-2) q Time-predicted parking space idle rate
The time length t required for the vehicle to be parked to enter the parking space f And (3) calculating:
5-1-1) if the vehicle to be parked reaches the moment t q The predicted parking space idling rate of the jth parking lot is less than 100 percent, j belongs to m, and the time length required for parking the vehicle to be parked into the parking space is calculated
Wherein w is a time length influence factor required for parking, w 1 ～w s All are preset values of influence factors of the time length required by parking; t is t m The parking time is a preset value related to the number of layers of the parking lot and the number of single-layer parking berths;
5-1-2) if the vehicle to be parked reaches the moment t q If the predicted parking space idle rate of a certain parking lot is more than or equal to 100 percent, acquiring the queuing length of vehicles to enter the parking lot
For t q Maximum queue length of time of day
And (3) prediction is carried out:
if no vehicle is waiting in line for entering, that is
Then calculate the historical data M days and t q Closest to the sampling instant t q ' M maximum queue lengths
Taking the mean value of t q The mean of the maximum queue length at time' is t q Maximum queuing length at time
The predicted value of (c):
in the formula (I), the compound is shown in the specification,
is the t th day a q ' maximum queue length at sampling time;
if t is present d When a vehicle is queued to wait for entering the field at the moment, historical data M days and t are calculated d Closest samplingTime t' d M maximum queue lengths
And calculating t q Maximum queuing length at time
In the formula (I), the compound is shown in the specification,
t 'of day a' d Maximum queue length at the sampling time;
wherein s is a predetermined constant, and s is a linear,
is t 'in the a day' d Number of vehicles coming out of the field within a + Δ t sampling period;
calculating the minimum value of n by the above formula, and calculating the time length required for parking the vehicle to be parked into the parking space according to the value of n
5-2) repeating the step 5-1) to respectively calculate the time required by the vehicle to be parked in a plurality of parking lots of the area class to be distributed
And will be
And sequencing according to the size, distributing the three parking lots with the shortest time consumption to the driver of the vehicle to be parked according to the sequencing size, and displaying the time length required by the vehicle to be parked to park in the parking lots for the driver to select.
It is an object of the present invention to provide a shared parking space intelligent allocation system based on the LSTM model.
The purpose of the invention is realized by the technical scheme, which comprises the following modules:
a data acquisition module: the system comprises a data acquisition module, a data processing module and a data processing module, wherein the data acquisition module is used for acquiring historical parking data of a plurality of parking lots in a target area and environmental data;
a data diversity module: the system comprises a data acquisition unit, a data processing unit and a data processing unit, wherein the data acquisition unit is used for acquiring environmental data of a single parking lot in a target area;
constructing a model module: the system is used for respectively constructing a residual parking berth prediction model based on an LSTM model for a plurality of data sets;
parking position prediction module: the system comprises a data set, a parking space prediction model and a parking space prediction model, wherein the data set is used for collecting information of a vehicle to be parked, environmental parameters and real-time data of a parking lot in a target area, dividing the information into corresponding data sets according to the environmental parameters, and predicting the number of parking spaces through the residual parking space prediction model corresponding to the data sets;
a parking space assignment module: for assigning parking positions to the vehicle to be parked according to the predicted number of remaining parking positions.
Due to the adoption of the technical scheme, the invention has the following advantages:
1. according to the method and the device, historical data of a plurality of parking lots in the target area are collected, a residual parking space prediction model is built through the LSTM model, the residual number of parking spaces at the target time point is predicted, the parking time required by the target parking lot is calculated according to the residual number of the parking spaces, the parking lot with the shortest parking time required in the target area is distributed to the vehicle to be parked, the parking time of the vehicle to be parked is shortened, and the traffic jam condition of the target area is relieved.
2. According to the method and the device, the prediction accuracy of the prediction model is improved by classifying a plurality of historical data sets such as the temperature, weather, holiday/working day information and the like of the environment.
3. According to the method, the required parking time under the two conditions of idling and non-idling of the parking lot is calculated respectively through the predicted parking lot parking space idling rate, and the prediction of the required parking time when the parking lot is allowed to line up and enter the parking lot is improved through the collection of historical departure data and historical queuing length data.
Additional advantages, objects, and features of the invention will be set forth in part in the description which follows and in part will become apparent to those having ordinary skill in the art upon examination of the following or may be learned from practice of the invention. The objectives and other advantages of the invention will be realized and attained by the structure particularly pointed out in the written description and claims hereof.
Drawings
The drawings of the invention are illustrated below.
FIG. 1 is a flowchart of an intelligent allocation method for shared parking spaces based on an LSTM model according to the present invention.
Detailed Description
The invention is further illustrated by the following figures and examples.
Example 1:
as shown in fig. 1, the method for intelligently allocating shared parking spaces based on the LSTM model specifically includes the following steps:
1) Data acquisition: collecting historical parking data and environmental data of a plurality of parking lots in a target area; the method comprises the following specific steps:
1-1) collecting historical parking data of m parking lots in an area to be allocated by taking days as a unit, wherein the historical parking data comprises: sampling time point parking lot parking space idle rate
Number of vehicles leaving parking lot within single sampling period delta t
Maximum queuing length of parking lot entrance in single sampling period delta t
Obtaining a single parking lot
Wherein a belongs to M, t is sampling time, and M is the maximum sampling days of a single parking lot;
1-2) collecting environmental data in an area to be allocated by taking days as a unit, wherein the environmental data comprises: weather, temperature, holiday/weekday information.
In the embodiment of the invention, 10 parking lots in a Chongqing business district are taken as parking lots in a target area, and 20 minutes are taken as time intervals to collect parking data of the 10 parking lots from 7 am to 10 am in one day (according to historical experience, the parking lots from 10 am to 7 am in the business district have high parking space idle rate and smooth roads in the business district, and the time required for parking vehicles in the 10 parking lots is similar, so that the parking data from 10 pm to 7 am are removed to reduce data redundancy and calculated amount). As shown in table 1, each parking lot has 726 total parking spaces for the data of parking in 2022 year, 9 month and 24 days, and the parking lot is provided with a remaining parking space statistical system.
The weather of the region on the 24 th of 2022 and 9 months is as follows: light rain; the temperature is as follows: 15-20 ℃; saturday is Saturday and belongs to holidays.
TABLE 1 parking data in 2022, 9 months and 24 days in a parking lot in a target area
2) Data diversity: respectively dividing historical data of a single parking lot into a plurality of data sets according to the environmental data; the method comprises the following specific steps:
2-1) dividing historical parking data into a plurality of data sets according to weather, temperature, holiday/workday data, wherein the data sets comprise:
a first data set of sunny-low-working day; a second data set of sunny-low-holiday, a third data set of sunny-medium-workday, a fourth data set of sunny-medium-holiday, a fifth data set of sunny-high-workday, a sixth data set of sunny-high-holiday, a seventh data set of rain/snow-low-workday, an eighth data set of rain/snow-low-holiday, a ninth data set of rain-medium-workday, a tenth data set of rain-medium-holiday, an eleventh data set of rain-high-workday, and a twelfth data set of rain-high-holiday;
2-2) respectively carrying out abnormal value elimination and normalization processing on the data of the data sets, and respectively dividing the processed data sets into training set data and test set data;
2-3) sampling time point parking lot parking space idle rate with training set and test set respectively in units of days
And sequencing according to the sampling time to respectively obtain the time sequence of the parking space idle rate of the parking lot.
In the embodiment of the invention, the low temperature is below 10 ℃, the medium temperature is 10-30 ℃ and the high temperature is above 30 ℃. I.e. the data in table 1 is divided into a tenth data set.
3) Constructing a model: respectively constructing a residual parking space prediction model based on an LSTM model aiming at a plurality of data sets; the method comprises the following specific steps:
3-1) respectively constructing a parking space idle rate prediction model based on an LSTM model for twelve data sets, wherein an Adam function is selected as the model optimizer, and the initial parameters of the model comprise: initial learning rate l r The number of hidden layer units, the iteration times epoch, the minimum training batch and the error threshold;
3-2) respectively training twelve parking space idle rate prediction models based on the LSTM model by using twelve data sets: to train the parking idle rate eta of any t moment in the set t As the output of the LSTM model, a training set t-5 Deltat,Parking space idle rate eta at t-4 delta t, t-3 delta t, t-2 delta t and t-delta t t-5△t 、η t-4△t 、η t-3△t 、η t-2△t 、η t-△t Training the LSTM model for inputting the LSTM model;
3-3) detecting whether the prediction LOSS LOSS of the parking space idling rate prediction model is smaller than an error threshold or whether the iteration times is larger than or equal to the maximum iteration times, if so, outputting model parameters, outputting the trained parking space idling rate prediction model, and if not, returning to the step 3-2) to continue training.
In the embodiment of the invention, the LSTM model comprises an input gate, a forgetting gate and an output gate, each gate has different functions, the forgetting gate has the function of controlling a memory unit, the number of states reserved at the previous moment to the current moment is determined, a forgetting part is realized through a sigmoid layer, and the existence of the forgetting gate is also the guarantee that the LSTM can remember long-term memory; the input gate is mainly responsible for updating the cell state, and the updating is to forget a part and remember some new information; the output gate is the structure used to calculate the short term cell memory status, which is the summary of previous stage work and determines the output information.
In the embodiment of the invention, the neuron of an input layer of the LSTM model is 5, the neuron of an output layer is 1, the parameter of the LSTM model is trained by the model optimizer by adopting an Adam function, and the initial learning rate l r 0.001, the number of hidden layer units is 80; the number of iterations epoch is 600 and the minimum training batch is 20.
4) And (3) predicting parking positions: collecting information of a vehicle to be parked, environmental parameters and real-time data of a parking lot in a target area, dividing the information into corresponding data sets in the step 2) according to the environmental parameters, and predicting the residual number of parking positions through a residual parking position prediction model corresponding to the data sets; the method comprises the following specific steps:
4-1) obtaining the information of the vehicle to be parked and calculating the time t required for the vehicle to go to the target area q And collecting weather, temperature, holiday/workday data of the day to be predicted, and dividing the data into corresponding data in the step 2) according to the weather, temperature, holiday/workday dataData concentration;
4-2) setting the current time as t d Acquisition of t d -4△t、t d -3△t、t d -2△t、t d -△t、t d Temporal parking space idle rate
Inputting the parking space idle rate into the parking space idle rate prediction model trained in the step 3-3), and calculating the t d At time + Deltat
Make a prediction, then pass
To pair
Making prediction, and analogizing until t d +n△t≥t q Output t q Time-predicted parking space idle rate
5) Parking space allocation: distributing parking positions to the vehicles to be parked according to the predicted number of the remaining parking positions, and the method specifically comprises the following steps:
5-1) passing t in step 4-2) q Time-predicted parking space idle rate
The time length t required for the vehicle to be parked to enter the parking space f And (3) calculating:
5-1-1) if the vehicle to be parked reaches the moment t q The predicted parking space parking idle rate of the jth parking lot is less than 100%, j belongs to m, and the time length required for parking the vehicle to be parked into the parking space is calculated
Wherein w is a time length influence factor required for parking, w 1 ～w s All are preset values of influence factors of the time length required by parking; t is t m The preset value of the parking time is related to the number of layers of the parking lot and the number of single-layer parking berths.
In the embodiment of the present invention, taking the parking lot shown in the data in table 1 as an example, the parking lot is divided into three layers, the first layer has 143 parking spaces, the second layer has 293 parking spaces, and the third layer has 290 parking spaces; take the value of s as 4,w 1 ～w 4 0.8, 1, 1.4, 2,. Eta. 1 ～η 4 The values of (A) are respectively 80%, 60%, 40% and 10%; when parking area parking stall idle rate is higher, wait to park the vehicle and can find idle parking stall very easily, required parking is long shorter, when parking area parking stall idle rate is lower, waits to park the vehicle and finds the time of parking stall and will increase in proportion. t is t m The number of the parking spaces in a single parking lot layer, the number of the parking lots and the road condition of the vehicles going upstairs and downstairs are preset according to experience.
5-1-2) if the vehicle to be parked reaches the moment t q If the predicted parking space parking idle rate of a certain parking lot is more than or equal to 100 percent, acquiring the queuing length of the vehicles to be entered in the current parking lot
For t q Maximum queue length of time of day
And (3) prediction is carried out:
if no vehicle is waiting in line for entering, that is
Then calculate the historical data M days andt q closest to the sampling instant t q ' M maximum queuing lengths
Taking the mean value of t q The mean of the maximum queue length at time' is t q Maximum queuing length at time
The predicted value of (c):
in the formula (I), the compound is shown in the specification,
is the t th day a q ' maximum queue length at sampling time;
if the current t is d When a vehicle is queued to wait for entering the field at the moment, historical data M days and t are calculated d Closest to sample time t' d M maximum queue lengths
And calculating t q Maximum queuing length at time
In the formula (I), the compound is shown in the specification,
t 'of day a' d Maximum queue length at the sampling time;
wherein s is a predetermined constant,
is t 'in the a day' d The number of vehicles coming out of the field within a + Δ t sampling period;
calculating the minimum value of n by the above formula, and calculating the time length required for parking the vehicle to be parked into the parking space according to the value of n
5-2) repeating the step 5-1) to respectively calculate the time required by the vehicle to be parked in a plurality of parking lots of the area class to be distributed
And will be
And sequencing according to the sizes, distributing the three parking lots with the shortest time consumption to a driver of the vehicle to be parked according to the sequencing sizes, and displaying the time length required by the vehicle to be parked to the parking lots for the driver to select.
Example 2:
an intelligent shared parking space distribution system based on an LSTM model comprises the following modules:
a data acquisition module: the system comprises a data acquisition module, a data processing module and a data processing module, wherein the data acquisition module is used for acquiring historical parking data of a plurality of parking lots in a target area and environmental data;
a data diversity module: the system comprises a data acquisition module, a data processing module and a data processing module, wherein the data acquisition module is used for acquiring historical data of a single parking lot in a target area;
constructing a model module: the system is used for respectively constructing a residual parking berth prediction model based on an LSTM model for a plurality of data sets;
parking position prediction module: the system comprises a data set, a data set and a residual parking space prediction model, wherein the data set is used for acquiring information of a vehicle to be parked, environmental parameters and real-time data of a parking lot in a target area, dividing the information into corresponding data sets according to the environmental parameters, and predicting the number of parking spaces through the residual parking space prediction model corresponding to the data sets;
parking berth assignment module: for assigning parking positions to the vehicle to be parked according to the predicted number of remaining parking positions.
As will be appreciated by one skilled in the art, embodiments of the present application may be provided as a method, system, or computer program product. Accordingly, the present application may take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment combining software and hardware aspects. Furthermore, the present application may take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, and the like) having computer-usable program code embodied therein.
The present application is described with reference to flowchart illustrations and/or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the application. It will be understood that each flow and/or block of the flowchart illustrations and/or block diagrams, and combinations of flows and/or blocks in the flowchart illustrations and/or block diagrams, can be implemented by computer program instructions. These computer program instructions may be provided to a processor of a general purpose computer, special purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instruction means which implement the function specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
Finally, it should be noted that: although the present invention has been described in detail with reference to the above embodiments, it should be understood by those skilled in the art that: modifications and equivalents may be made to the embodiments of the invention without departing from the spirit and scope of the invention, which is to be covered by the claims.
Claims (7)
1. An intelligent allocation method for shared parking spaces based on an LSTM model is characterized by comprising the following specific steps:
1) Data acquisition: collecting historical parking data and environmental data of a plurality of parking lots in a target area;
2) Data diversity: respectively dividing historical data of a single parking lot into a plurality of data sets according to the environmental data;
3) Constructing a model: respectively constructing a residual parking space prediction model based on an LSTM model for a plurality of data sets;
4) And (3) predicting parking positions: collecting information of a vehicle to be parked, environmental parameters and real-time data of a parking lot in a target area, dividing the information into corresponding data sets in the step 2) according to the environmental parameters, and predicting the residual number of parking positions through a residual parking position prediction model corresponding to the data sets;
5) Parking space allocation: and allocating parking positions to the vehicles to be parked according to the predicted number of the remaining parking positions.
2. The intelligent allocation method for shared parking spaces based on neural networks as claimed in claim 1, wherein the data acquisition in step 1) specifically comprises the steps of:
1-1) collecting historical parking data of m parking lots in an area to be allocated by taking days as a unit, wherein the historical parking data comprises: sampling time point parking lot parking space idle rate
Number of vehicles leaving parking lot within single sampling period delta t
Maximum queuing length of parking lot entrance in single sampling period delta t
Obtaining a single parking lot
Wherein a belongs to M, t is sampling time, and M is the maximum sampling days of a single parking lot;
1-2) collecting environmental data in an area to be allocated by taking days as a unit, wherein the environmental data comprises: weather, temperature, holiday/weekday information.
3. The intelligent allocation method for shared parking spaces based on neural networks as claimed in claim 2, wherein the specific method of data diversity in step 2) is:
2-1) dividing historical parking data into a plurality of data sets according to weather, temperature, holiday/workday data, wherein the data sets comprise:
a first dataset for sunny-low-weekdays; a second data set of sunny-low-holiday, a third data set of sunny-medium-workday, a fourth data set of sunny-medium-holiday, a fifth data set of sunny-high-workday, a sixth data set of sunny-high-holiday, a seventh data set of rain/snow-low-workday, an eighth data set of rain/snow-low-holiday, a ninth data set of rain-medium-workday, a tenth data set of rain-medium-holiday, an eleventh data set of rain-high-workday, and a twelfth data set of rain-high-holiday;
2-2) respectively carrying out abnormal value elimination and normalization processing on the data of the data sets, and respectively dividing the processed data sets into training set data and test set data;
2-3) sampling time point parking lot parking space idle rate integrating training set and test set by taking day as unit
And sequencing according to the sampling time to obtain a time sequence of the parking space idle rate of the parking lot.
4. The intelligent allocation method for shared parking spaces based on neural networks as claimed in claim 3, wherein the specific steps of model construction and training in step 3) are as follows:
3-1) respectively constructing a parking space idle rate prediction model based on an LSTM model for twelve data sets, wherein an Adam function is selected as the model optimizer, and the initial parameters of the model comprise: initial learning rate l r The number of hidden layer units, the iteration times epoch, the minimum training batch and the error threshold;
3-2) respectively training twelve parking space idle rate prediction models based on the LSTM model by using twelve data sets: to train the parking idle rate eta of any t moment in the set t The parking space idle rate eta of the time points of the training set t-5 delta t, t-4 delta t, t-3 delta t, t-2 delta t and t-delta t is used as the output of the LSTM model t-5△t 、η t-4△t 、η t-3△t 、η t-2△t 、η t-△t Training the LSTM model for inputting the LSTM model;
3-3) detecting whether the prediction loss of the parking space idling rate prediction model is smaller than a threshold value or whether the iteration times are larger than or equal to the maximum iteration times, if so, outputting model parameters, outputting the trained parking space idling rate prediction model, and if not, returning to the step 3-2) to continue training.
5. The intelligent allocation method for shared parking spaces based on neural networks as claimed in claim 1, wherein the specific steps of the step 4) parking space prediction are as follows:
4-1) obtaining the information of the vehicle to be parked and calculating the time t required for the vehicle to go to the target area q Collecting weather, temperature, holiday/workday data of the day to be predicted, and dividing the data into corresponding data sets in the step 2) according to the weather, temperature, holiday/workday data;
4-2) setting the current time as t d Collecting t d -4△t、t d -3△t、t d -2△t、t d -△t、t d Parking space idling rate of time
Inputting the parking space idle rate into the parking space idle rate prediction model trained in the step 3-3), and calculating the t d At time + Deltat
Make a prediction, then pass
For is to
Making prediction, and analogizing until t d +n△t≥t q Output t q Time-predicted parking space idle rate
6. The intelligent allocation method for shared parking spaces based on neural networks as claimed in claim 1, wherein the parking space allocation in step 5) comprises the following specific steps:
5-1) passing t in step 4-2) q Time-predicted parking space idleRetention rate
The time length t required for the vehicle to be parked to enter the parking space f And (3) calculating:
5-1-1) if the vehicle to be parked reaches the moment t q The predicted parking space parking idle rate of the jth parking lot is less than 100%, j belongs to m, and the time length required for parking the vehicle to be parked into the parking space is calculated
Wherein w is a time length influence factor required for parking, w 1 ～w s All are preset values of influence factors of the time length required by parking; t is t m The parking time is a preset value related to the number of layers of the parking lot and the number of single-layer parking berths;
5-1-2) if the vehicle to be parked reaches the moment t q If the predicted parking space idle rate of a certain parking lot is more than or equal to 100 percent, acquiring the queuing length of vehicles to enter the parking lot
For t q Maximum queue length of time of day
And (3) predicting:
if no vehicle is waiting in line for entering, that is
Then calculate the historical data for M daysAnd t q Closest to the sampling instant t q ' M maximum queuing lengths
Is taken as the mean value of' q The average value of the maximum queuing length at the moment is t q Maximum queuing length at time
The predicted value of (c):
in the formula (I), the compound is shown in the specification,
is t 'of day a' q Maximum queue length at the sampling time;
if the current t is d When a vehicle is queued to enter the station at the moment, calculating the historical data M days and t d Closest to sample time t' d M maximum queue lengths
And calculating t q Maximum queuing length at time
In the formula (I), the compound is shown in the specification,
t 'day a' d Maximum queue length at the sampling time;
wherein s is a predetermined constant,
is t 'in the a day' d Number of vehicles coming out of the field within a + Δ t sampling period;
calculating the minimum value of n by the above formula, and calculating the time length required for parking the vehicle to be parked into the parking space according to the value of n
5-2) repeating the step 5-1) to respectively calculate the time required by the vehicle to be parked in a plurality of parking lots of the area class to be distributed
And will be
And sequencing according to the size, distributing the three parking lots with the shortest time consumption to the driver of the vehicle to be parked according to the sequencing size, and displaying the time length required by the vehicle to be parked to park in the parking lots for the driver to select.
7. An intelligent sharing parking space distribution system based on an LSTM model is characterized by comprising the following modules:
a data acquisition module: the system comprises a data acquisition module, a data processing module and a data processing module, wherein the data acquisition module is used for acquiring historical parking data of a plurality of parking lots in a target area and environmental data;
a data diversity module: the system comprises a data acquisition module, a data processing module and a data processing module, wherein the data acquisition module is used for acquiring historical data of a single parking lot in a target area;
constructing a model module: the system is used for respectively constructing a residual parking berth prediction model based on an LSTM model for a plurality of data sets;
parking position prediction module: the system comprises a data set, a data set and a residual parking space prediction model, wherein the data set is used for acquiring information of a vehicle to be parked, environmental parameters and real-time data of a parking lot in a target area, dividing the information into corresponding data sets according to the environmental parameters, and predicting the number of parking spaces through the residual parking space prediction model corresponding to the data sets;
parking berth assignment module: for assigning parking positions to the vehicle to be parked according to the predicted number of remaining parking positions.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211312801.2A CN115830848A (en) | 2022-10-25 | 2022-10-25 | Shared parking space intelligent distribution system and method based on LSTM model |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211312801.2A CN115830848A (en) | 2022-10-25 | 2022-10-25 | Shared parking space intelligent distribution system and method based on LSTM model |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN115830848A true CN115830848A (en) | 2023-03-21 |
Family
ID=85525440
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202211312801.2A Pending CN115830848A (en) | 2022-10-25 | 2022-10-25 | Shared parking space intelligent distribution system and method based on LSTM model |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115830848A (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116187591A (en) * | 2023-04-27 | 2023-05-30 | 松立控股集团股份有限公司 | Method for predicting number of remaining parking spaces in commercial parking lot based on dynamic space-time trend |
| CN116913123A (en) * | 2023-08-14 | 2023-10-20 | 合肥工业大学智能制造技术研究院 | Space-time integrated intelligent parking guidance method |
| CN117829377A (en) * | 2024-03-04 | 2024-04-05 | 德阳城市智慧之心信息技术有限公司 | Neural network-based remaining parking space prediction method, device, equipment and medium |
-
2022
- 2022-10-25 CN CN202211312801.2A patent/CN115830848A/en active Pending
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116187591A (en) * | 2023-04-27 | 2023-05-30 | 松立控股集团股份有限公司 | Method for predicting number of remaining parking spaces in commercial parking lot based on dynamic space-time trend |
| CN116913123A (en) * | 2023-08-14 | 2023-10-20 | 合肥工业大学智能制造技术研究院 | Space-time integrated intelligent parking guidance method |
| CN116913123B (en) * | 2023-08-14 | 2024-04-02 | 合肥工业大学智能制造技术研究院 | Space-time integrated intelligent parking guidance method |
| CN117829377A (en) * | 2024-03-04 | 2024-04-05 | 德阳城市智慧之心信息技术有限公司 | Neural network-based remaining parking space prediction method, device, equipment and medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN115830848A (en) | Shared parking space intelligent distribution system and method based on LSTM model | |
| CN110555990B (en) | Effective parking space-time resource prediction method based on LSTM neural network | |
| CN112330215B (en) | Urban vehicle demand prediction method, equipment and storage medium | |
| CN110390349A (en) | Bus passenger flow volume based on XGBoost model predicts modeling method | |
| CN108281033B (en) | Parking guidance system and method | |
| Li et al. | Deep learning based parking prediction on cloud platform | |
| CN110599236A (en) | Short-time parking demand prediction method based on GRU model | |
| CN111753910A (en) | Method and device for predicting drip order demand based on LSTM | |
| CN109376906B (en) | Travel time prediction method and system based on multi-dimensional trajectory and electronic equipment | |
| CN113538067B (en) | Inter-city network vehicle-closing demand prediction method and system based on machine learning | |
| WO2021115320A1 (en) | Traffic evaluation method and system | |
| CN111507762A (en) | Urban taxi demand prediction method based on multi-task co-prediction neural network | |
| Zhang et al. | PewLSTM: Periodic LSTM with Weather-Aware Gating Mechanism for Parking Behavior Prediction. | |
| CN110490365B (en) | Method for predicting network car booking order quantity based on multi-source data fusion | |
| CN110796317A (en) | Urban taxi scheduling method based on demand prediction | |
| CN115691165A (en) | Traffic signal lamp scheduling method, device and equipment and readable storage medium | |
| Petersen et al. | Multi-output deep learning for bus arrival time predictions | |
| CN114418606B (en) | Network vehicle order demand prediction method based on space-time convolution network | |
| CN114463868A (en) | Toll station traffic flow combination prediction method and system for traffic flow management and control | |
| CN113821547B (en) | Rapid and efficient short-time prediction method, system and storage medium for occupancy of parking lot | |
| Lai et al. | Taxi demand prediction with LSTM-based combination model | |
| CN112669595B (en) | Network taxi booking flow prediction method based on deep learning | |
| CN114549075A (en) | Regional parking evaluation method based on parking and video big data | |
| CN113838303A (en) | Parking lot recommendation method and device, electronic equipment and storage medium | |
| Xia et al. | A distributed EMDN-GRU model on Spark for passenger waiting time forecasting |
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 |