CN107239846A - parking lot berth prediction processing method and device - Google Patents
parking lot berth prediction processing method and device Download PDFInfo
- Publication number
- CN107239846A CN107239846A CN201610187894.9A CN201610187894A CN107239846A CN 107239846 A CN107239846 A CN 107239846A CN 201610187894 A CN201610187894 A CN 201610187894A CN 107239846 A CN107239846 A CN 107239846A
- Authority
- CN
- China
- Prior art keywords
- mrow
- parking lot
- berth
- prediction model
- msup
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
-
- G—PHYSICS
- G08—SIGNALLING
- G08G—TRAFFIC CONTROL SYSTEMS
- G08G1/00—Traffic control systems for road vehicles
- G08G1/14—Traffic control systems for road vehicles indicating individual free spaces in parking areas
Abstract
The invention discloses a kind of parking lot berth prediction processing method and device, wherein, this method includes:Obtain and integrate idle berth time series of the parking lot within the order history time and the characteristic in the parking lot in the order history time;Model training is carried out according to idle the berth time series and this feature data and obtains static prediction model;Processing is predicted to parking lot berth according to the static prediction model, the berth Forecasting Methodology for solving parking lot in correlation technique does not adapt to dynamic change and the problem of existing defects, can adapt to environment dynamic, improve user experience.
Description
Technical field
The present invention relates to technical field of intelligent traffic, in particular to a kind of parking lot berth prediction processing method and
Device.
Background technology
In smart city field, especially under wisdom traffic trip scene, parking difficulty has become can not avoid main
Problem.In the place that the crowd is dense such as CBD, transport hub, hospital, stadium, often occur parking position message it is inaccurate,
Then the serious reduction for bringing trip to experience.
In order to solve problem above, the industry specialists of many field of traffic all consider to utilize statistics or machine learning skill
Art, is recorded according to the history berth in target parking lot, is excavated the hidden patterns wherein contained, is then formed forecast model.
City manager can utilize model above, the issue berth person that offers Public Traveling in advance, be carried so as to provide pedestrian group
Supported for accurate data, facilitate smart city choice for traveling, while crowd density can also be monitored for public safety department,
Trend etc. is moved towards there is provided a more reliable standard.
Current subject matter is:Existing berth Forecasting Methodology, is required to carry out machine learning instruction using history data set
Practice, the prediction of new data is carried out with the result (often reacting for the combination of equation, formula, tree construction or the above) of training.And
This method tends not to good work in smart city field, because the crowd behaviour in the field, weather conditions, city are built
If situation is often continually changing and (is referred to as " concept drift ") over time, the model of the front construction caused is as time goes by
And constantly reduce predictablity rate;In addition, traditional data mining method needs just be predicted based on a large amount of historical samples, often
Secondary model modification be required for calculate full dose data, and the carrying cost of full dose data by with the passage of on-line time increasingly
Greatly, extra cost is caused to bear to system operation.
Not the problem of dynamic change and existing defects not being adapted to for the berth Forecasting Methodology in parking lot in correlation technique, also
Effective solution is not proposed.
The content of the invention
The invention provides a kind of parking lot berth prediction processing method and device, at least to solve to stop in correlation technique
Berth Forecasting Methodology do not adapt to dynamic change and the problem of existing defects.
According to an aspect of the invention, there is provided a kind of parking lot berth prediction processing method, including:
When obtaining and integrating idle berth value historical series and the order history of the parking lot within the order history time
The characteristic in the interior parking lot;
Model training is carried out according to idle the berth time series and the characteristic and obtains static prediction model;
Processing is predicted to parking lot berth according to the static prediction model.
Further, being predicted processing to parking lot berth according to the static prediction model includes:According to described quiet
State forecast model determines the available berth ratio in the parking lot;According to described with berth ratio and the berth in the parking lot
The available berth number in the parking lot in the capacity prediction following scheduled time.
Further, model training is carried out according to idle the berth time series and the characteristic and obtains static pre-
Surveying model includes:
Static prediction model W is iterated to calculate by equation below(k):
W(k)=W(k-1)+α(k-1)*XT*(Y-S(a))
A=X*W(k-1);
α is iterated to calculate by equation below(k):αk=α(k-1)*λ(k-1);
Wherein, perform number of times in circulation and be more than predefined parameter t, or the continuous 3 times values circulated, W(k-2)、W(k-1)、W(k)Approximately
Circulation is jumped out in the case of equal, and determines W(k)Training result be Whistory,α(k)For last step parameter αhistory, λ is instruction
Practice attenuation parameter, λ=0.95, k is more than or equal to 1;
Determine the available berth ratio in parking lot in the following manner according to the static prediction model:
Y1=exp (XTW(k))/(1+exp(XTW(k)));
Wherein, X is the vector form of time series and characteristic, and W (k) is the corresponding weight vectors of characteristic, and Y is
Within the order history time idle berth value historical series and in the order history time parking lot characteristic
Under, the ratio of the total Berth number in the parking lot, Y can be accounted for Berth number within the order history time1It is that parking lot is following pre-
The ratio of the total Berth number in the parking lot can be accounted in section of fixing time with Berth number.
Further, static state is being obtained according to idle berth time series and the characteristic progress model training
After forecast model, methods described also includes:Regularly update the dynamic prediction that the static prediction model obtains the parking lot
Model.
Further, regularly updating the dynamic prediction model that the static prediction model obtains the parking lot includes:
It is vector X by the observation of each new additionNew, the observation newly added is YNew, the observation to each new addition
Value is trained as follows:
Initialize W(k):By WhistoryIt is assigned to W(0),
W is iterated to calculate by equation below(k):
W(k)=W(k-1)+αhistory*XNew T*(YNew-S(XNew*W(k-1)));
W(k)Dynamic prediction model after being updated for training.
Further, the characteristic includes:Parking lot scale, whether be solid parking, whether configure ETC equipment,
Whether the periphery in parking lot has exhibition, weather, temperature.
According to another aspect of the present invention, a kind of parking lot berth prediction processing device is additionally provided, including:
Acquisition module, for obtaining and integrating idle berth time series of the parking lot within the order history time and described
The characteristic in the parking lot in the order history time;
Training module, obtains quiet for carrying out model training according to idle the berth time series and the characteristic
State forecast model;
Prediction module, for being predicted processing to parking lot berth according to the static prediction model.
Further, the prediction module includes:
Determining unit, the available berth ratio for determining the parking lot according to the static prediction model;
Predicting unit, for following pre- timing can be predicted with the berthing capacity in berth ratio and the parking lot according to described
The available berth number in the interior parking lot.
Further, the training module is additionally operable to:Static prediction model W is iterated to calculate by equation below(k):
W(k)=W(k-1)+α(k-1)*XT*(Y-S(a))
A=X*W(k-1);
α is iterated to calculate by equation below(k):αk=α(k-1)*λ(k-1);
Wherein, perform number of times in circulation and be more than predefined parameter t, or the continuous 3 times values circulated, W(k-2)、W(k-1)、W(k)Approximately
Circulation is jumped out in the case of equal, and determines W(k)Training result be Whistory,α(k)For last step parameter αhistory, λ is instruction
Practice attenuation parameter, λ=0.95, k is more than or equal to 1;
The determining unit is additionally operable to determine the available pool in parking lot in the following manner according to the static prediction model
Position ratio:Y1=exp (XTW(k))/(1+exp(XTW(k)));
Wherein, X is the vector form of time series and characteristic, and W (k) is the corresponding weight vectors of characteristic, and Y is
Within the order history time idle berth value historical series and in the order history time parking lot characteristic
Under, the ratio of the total Berth number in the parking lot, Y can be accounted for Berth number within the order history time1It is that parking lot is following pre-
The ratio of the total Berth number in the parking lot can be accounted in section of fixing time with Berth number.
Further, described device also includes:Update module, obtains described for regularly updating the static prediction model
The dynamic prediction model in parking lot.
By the present invention, using obtaining and integrate idle berth time series and institute of the parking lot within the order history time
State the characteristic in the parking lot in the order history time;Entered according to idle the berth time series and the characteristic
Row model training obtains static prediction model;Processing is predicted to parking lot berth according to the static prediction model, solved
The berth Forecasting Methodology in parking lot does not adapt to dynamic change and the problem of existing defects in correlation technique, can adapt to environment
Dynamic, improves user experience.
Brief description of the drawings
Accompanying drawing described herein is used for providing a further understanding of the present invention, constitutes the part of the application, this hair
Bright schematic description and description is used to explain the present invention, does not constitute inappropriate limitation of the present invention.In the accompanying drawings:
Fig. 1 is the flow chart of parking lot berth prediction processing method according to embodiments of the present invention;
Fig. 2 is the block diagram of parking lot berth prediction processing device according to embodiments of the present invention;
Fig. 3 is the block diagram one of parking lot berth according to the preferred embodiment of the invention prediction processing device;
Fig. 4 is the block diagram two of parking lot berth according to the preferred embodiment of the invention prediction processing device;
Fig. 5 is the flow chart of parking lot berth Forecasting Methodology according to embodiments of the present invention.
Embodiment
Describe the present invention in detail below with reference to accompanying drawing and in conjunction with the embodiments.It should be noted that not conflicting
In the case of, the feature in embodiment and embodiment in the application can be mutually combined.
The embodiments of the invention provide a kind of parking lot berth prediction processing method, Fig. 1 is according to embodiments of the present invention
The flow chart of parking lot berth prediction processing method, as shown in figure 1, including:
Step S102, obtain and integrate idle berth value historical series of the parking lot within the order history time and this make a reservation for
The characteristic in the parking lot in historical time;
Step S104, carries out model training according to idle the berth time series and this feature data and obtains static prediction mould
Type;
Step S106, processing is predicted according to the static prediction model to parking lot berth.
By above-mentioned steps, obtain and integrate idle berth time series of the parking lot within the order history time and this is pre-
Determine the characteristic in parking lot historical time Nei;Model training is carried out according to idle the berth time series and this feature data
Obtain static prediction model;Processing is predicted to parking lot berth according to the static prediction model, solved in correlation technique
The berth Forecasting Methodology in parking lot does not adapt to dynamic change and the problem of existing defects, can adapt to environment dynamic, improve
User experience.
Further, being predicted processing to parking lot berth according to the static prediction model includes:It is pre- according to the static state
Survey the available berth ratio that model determines the parking lot;According to the prediction of the berthing capacity in the available berth ratio and the parking lot not
Carry out the available berth number in the parking lot in the scheduled time.
Further, model training is carried out according to idle the berth time series and this feature data and obtains static prediction mould
Type includes:
Static prediction model W is iterated to calculate by equation below(k):
W(k)=W(k-1)+α(k-1)*XT*(Y-S(a))
A=X*W(k-1);
α is iterated to calculate by equation below(k):αk=α(k-1)*λ(k-1);
Wherein, perform number of times in circulation and be more than predefined parameter t, or the continuous 3 times values circulated, W(k-2)、W(k-1)、W(k)Approximately
Circulation is jumped out in the case of equal, and determines W(k)Training result be Whistory,α(k)For last step parameter αhistory, λ is instruction
Practice attenuation parameter, λ=0.95, k is more than or equal to 1;
Determine the available berth ratio in parking lot in the following manner according to the static prediction model:
Y1=exp (XTW(k))/(1+exp(XTW(k)));
Wherein, X is the vector form of time series and characteristic, and W (k) is the corresponding weight vectors of characteristic, and Y is
Idle berth within the order history time is worth in historical series and the order history time under the characteristic in the parking lot,
The ratio of the total Berth number in the parking lot, Y can be accounted in the order history time with Berth number1It is in the following predetermined amount of time in parking lot
The ratio of the total Berth number in the parking lot can be accounted for Berth number.
Further, static prediction is being obtained according to idle berth time series and this feature data progress model training
After model, this method also includes:Regularly update the dynamic prediction model that the static prediction model obtains the parking lot.
Further, regularly updating the dynamic prediction model that the static prediction model obtains the parking lot includes:
It is vector X by the observation of each new additionNew, the observation newly added is YNew, the observation to each new addition
Value is trained as follows:
Initialize W(k):By WhistoryIt is assigned to W(0),
W is iterated to calculate by equation below(k):
W(k)=W(k-1)+αhistory*XNew T*(YNew-S(XNew*W(k-1)));
W(k)Dynamic prediction model after being updated for training.
Further, this feature data include:Parking lot scale, whether be solid parking, whether configure ETC equipment, stop
Whether the periphery in parking lot has exhibition, weather, temperature.
The embodiment of the present invention additionally provides a kind of parking lot berth prediction processing device, and Fig. 2 is according to embodiments of the present invention
Parking lot berth prediction processing device block diagram, as shown in Fig. 2 including:
Acquisition module 22, for obtaining and integrating idle berth time series of the parking lot within the order history time and be somebody's turn to do
The characteristic in the parking lot in the order history time;
Training module 24, static state is obtained for carrying out model training according to idle the berth time series and this feature data
Forecast model;
Prediction module 26, for being predicted processing to parking lot berth according to the static prediction model.
Fig. 3 is the block diagram one of parking lot berth according to the preferred embodiment of the invention prediction processing device, as shown in figure 3,
Prediction module 26 includes:
Determining unit 32, the available berth ratio for determining the parking lot according to the static prediction model;
Predicting unit 34, for predicting the following scheduled time according to the berthing capacity of the available berth ratio and the parking lot
The available berth number in the interior parking lot.
Further, the training module is additionally operable to:Static prediction model W is iterated to calculate by equation below(k):
W(k)=W(k-1)+α(k-1)*XT*(Y-S(a))
A=X*W(k-1);
α is iterated to calculate by equation below(k):αk=α(k-1)*λ(k-1);
Wherein, perform number of times in circulation and be more than predefined parameter t, or the continuous 3 times values circulated, W(k-2)、W(k-1)、W(k)Approximately
Circulation is jumped out in the case of equal, and determines W(k)Training result be Whistory,α(k)For last step parameter αhistory, λ is instruction
Practice attenuation parameter, λ=0.95, k is more than or equal to 1;
The determining unit is additionally operable to determine the available berth in parking lot in the following manner according to the static prediction model
Ratio:Y1=exp (XTW(k))/(1+exp(XTW(k)));
Wherein, X is the vector form of time series and characteristic, and W (k) is the corresponding weight vectors of characteristic, and Y is
Idle berth within the order history time is worth in historical series and the order history time under the characteristic in the parking lot,
The ratio of the total Berth number in the parking lot, Y can be accounted in the order history time with Berth number1It is in the following predetermined amount of time in parking lot
The ratio of the total Berth number in the parking lot can be accounted for Berth number.
Fig. 4 is the block diagram two of parking lot berth according to the preferred embodiment of the invention prediction processing device, as shown in figure 4,
The device also includes:
Update module 42, the dynamic prediction model in the parking lot is obtained for regularly updating the static prediction model.
The embodiment of the present invention is further described with reference to specific embodiment.
To overcome existing berth Predicting Technique not adapt to the shortcoming of dynamic change, the embodiments of the invention provide one kind is right
The method that parking lot berth carries out dynamic prediction, the advantage of contrast Classical forecast technology is can be using real time data constantly more
New training forecast model, it is ensured that model can grow with each passing hour, possess enough real-time estimate precision, and can reduce history number
According to storage, all there is advantage in terms of predictablity rate and cost.
Fig. 5 is the flow chart of parking lot berth Forecasting Methodology according to embodiments of the present invention, as shown in figure 5, being broadly divided into 4
Individual step stage, including:
Step S502, data preparation stage:By parking lot berth historical time sequence data and parking lot relative token number
According to being integrated, specifically it is attached using time field, both the above data is attached, and further obtains shape
Such as within " certain year-certain moon-one day-certain period ", certain external condition (parking lot scale, whether solid parking, whether configure
ETC equipment, periphery whether have exhibition, weather how, temperature how) under certain parking lot remaining berth index.
Step S504, model static construction stage:The historical results integrated more than carry out model training, pass through machine
Learning method builds static prediction model Parking_Model0;
Step S506, model dynamic adjustment phase:Due to being in complication system as the field of smart city, with
Upper static models are reached the standard grade, predictablity rate will necessarily with the passage of time continuous decrease.Therefore need using continually
New data (be referred to as " berth real-time time sequence ") persistently adjust existing model, form the model for adapting to newest scene
Parking_Model1。
Step S508, model application stage:The data to be predicted newly entered are predicted using Parking_Model1,
According to the available berth data of current time, weather, surrounding enviroment, target parking lot recently, following short time, target are predicted
The available berth ratio in parking lot, multiplied by with the berthing capacity in target parking lot, you can draw the following short time parking lot
Berth number can be used.
Obtain and integrate the berth sequence data in order history time in parking lot and the characteristic in the parking lot;
Model training is carried out according to the berth sequence data and the characteristic and obtains static prediction model;
Prediction data is treated according to the static prediction model and is predicted processing.
Compared with the mode that other berths are predicted, in the embodiment of the present invention, model updates every time need not carry out full dose meter
Calculate, solve computation complexity and storage stress problems;The behind pushing factor that forecast model does not change over time loses essence
True property, solves precision problem.
The present embodiment lists a kind of relatively simple embodiment of the above method, related definition is provided first as follows:
Mapping function S (a) is defined, input parameter a is m dimensional vectors, and output quantity S (a) is also m dimensional vectors
Define two k dimensional vectors W(1), W(2)Apart from distance (W(1),W(2)),
Wherein W(1) jAnd W(2) jIt is two vectorial jth component values respectively:
Define two k dimensional vectors W(1), W(2)" approximately equal " condition:
distance(W(1),W(2)) mono- external definition of < very little floating number.
Historical data is made there are m bar training samples, every sample has n feature, its form is:
Wherein, XijRepresent the jth characteristic value of the i-th historic training data sample, YiRepresent the dependent variable of the i-th training data
Value, is idle rate of the parking lot at the correspondence moment here.
Specific implementation step is as follows:
During the idle berth of step 1, the parking lot correlated characteristic information that smart city operator is provided and parking lot
Between sequence be associated, be combined and count according to the period, formed parking position distribution isochronous surface, summary information make
For the training sample set of subsequent step, form is as follows:
After the completion of step 2, step 1, it is trained using above training sample set construction machine learning model, that is, ask
Go out parameter vector W each component value.Specifically:
Loop initialization counter k=0;Remember W(k)Weight vectors W after being circulated for kth wheel.
Initialize training pace parameter alpha=0.1;It is that k takes turns the step parameter after circulation to remember α (k).
Initialization training attenuation parameter λ=0.95.
N+1 dimensional weights vector W is initialized:To each weight component WjGenerate the random number between (0,1) and tax
Value, now W is designated as W(0)。
Into circulation, circulation jumps out condition and is:Circulation performs number of times and is more than external parameter t, or follows for continuous 3 times before
Ring, W(k-2)、W(k-1)、W(k)All each other " approximately equal ":
W is iterated to calculate by equation below(k):
W(k)=W(k-1)+α(k-1)*XT*(Y-S(X*W(k-1))
α is iterated to calculate by equation below(k):
αk=α(k-1)*λ(k-1)
Return to last W(k)It is used as training result Whistory,α(k)It is used as last step parameter αhistory;Static models
Parking_Model0, which is built, to be completed.
After the completion of step 3, step 2, have been built up completing static prediction model, as new data endlessly enters
(being consistent in form and step 1), enters Mobile state adjustment:
Further, it is vector X to make each artificial mark observational record newly addedNew, the mark value newly observed is YNew。
Observation to each new addition is trained as follows:
Loop initialization counter k=0;Remember W(k)Weight vectors W after being circulated for kth wheel.
N+1 dimensional weights vector W is initialized:By WhistoryW is assigned to, now W is designated as W(0)。
Into circulation, circulation jumps out condition and is:Circulation performs number of times and is more than external parameter t, or follows for continuous 3 times before
Ring, W(k-2)、W(k-1)、W(k)All each other " approximately equal ":
W is iterated to calculate by equation below(k):
W(k)=W(k-1)+αhistory*XNew T*(YNew-S(XNew*W(k-1)));
Return to last W(k)As the result of training, dynamic model Parking_Model1, which is built, to be completed.
Step 4, the available berth ratio for determining according to the dynamic prediction model parking lot in the following manner:
Y1=exp (XTW(k))/(1+exp(XTW(k)))。
After online implementing, for new observation data (being designated as vectorial X), the current newest model trained is on the one hand substituted into
Parameter vector W in Parking_Model1, can obtain the berth desired ratio and expected Berth number in target parking lot;It is another
Aspect, it is possible to use current data removes continuous updating dynamic model Parking_Model1, reaches the effect of real-time estimate.
Obviously, those skilled in the art should be understood that above-mentioned each module of the invention or each step can be with general
Computing device realize that they can be concentrated on single computing device, or be distributed in multiple computing devices and constituted
Network on, alternatively, the program code that they can be can perform with computing device be realized, it is thus possible to they are stored
Performed in the storage device by computing device, and in some cases, can be shown to be performed different from order herein
The step of going out or describe, they are either fabricated to each integrated circuit modules respectively or by multiple modules in them or
Step is fabricated to single integrated circuit module to realize.So, the present invention is not restricted to any specific hardware and software combination.
The preferred embodiments of the present invention are the foregoing is only, are not intended to limit the invention, for the skill of this area
For art personnel, the present invention can have various modifications and variations.Within the spirit and principles of the invention, that is made any repaiies
Change, equivalent substitution, improvement etc., should be included in the scope of the protection.
Claims (10)
1. a kind of parking lot berth prediction processing method, it is characterised in that including:
Obtain and integrate parking lot in the idle berth value historical series and the order history time within the order history time
The characteristic in the parking lot;
Model training is carried out according to idle the berth time series and the characteristic and obtains static prediction model;
Processing is predicted to parking lot berth according to the static prediction model.
2. according to the method described in claim 1, it is characterised in that parking lot berth is carried out according to the static prediction model
Prediction processing includes:
The available berth ratio in the parking lot is determined according to the static prediction model;
According to the parking lot in the berthing capacity prediction following scheduled time with berth ratio and the parking lot
Berth number can be used.
3. method according to claim 2, it is characterised in that according to idle the berth time series and the characteristic
Obtaining static prediction model according to progress model training includes:
Static prediction model W is iterated to calculate by equation below(k):
<mrow>
<mtable>
<mtr>
<mtd>
<mrow>
<msup>
<mi>W</mi>
<mrow>
<mo>(</mo>
<mi>k</mi>
<mo>)</mo>
</mrow>
</msup>
<mo>=</mo>
<msup>
<mi>W</mi>
<mrow>
<mo>(</mo>
<mi>k</mi>
<mo>-</mo>
<mn>1</mn>
<mo>)</mo>
</mrow>
</msup>
<mo>+</mo>
<msup>
<mi>&alpha;</mi>
<mrow>
<mo>(</mo>
<mi>k</mi>
<mo>-</mo>
<mn>1</mn>
<mo>)</mo>
</mrow>
</msup>
<mo>*</mo>
<msup>
<mi>X</mi>
<mi>T</mi>
</msup>
<mo>*</mo>
<mrow>
<mo>(</mo>
<mi>Y</mi>
<mo>-</mo>
<mi>S</mi>
<mo>(</mo>
<mi>a</mi>
<mo>)</mo>
<mo>)</mo>
</mrow>
</mrow>
</mtd>
</mtr>
<mtr>
<mtd>
<mrow>
<mi>S</mi>
<mrow>
<mo>(</mo>
<mi>a</mi>
<mo>)</mo>
</mrow>
<mo>=</mo>
<mfrac>
<mn>1</mn>
<mrow>
<mn>1</mn>
<mo>+</mo>
<mi>exp</mi>
<mrow>
<mo>(</mo>
<mo>-</mo>
<mi>a</mi>
<mo>)</mo>
</mrow>
</mrow>
</mfrac>
</mrow>
</mtd>
</mtr>
<mtr>
<mtd>
<mrow>
<mi>a</mi>
<mo>=</mo>
<mi>X</mi>
<mo>*</mo>
<msup>
<mi>W</mi>
<mrow>
<mo>(</mo>
<mi>k</mi>
<mo>-</mo>
<mn>1</mn>
<mo>)</mo>
</mrow>
</msup>
</mrow>
</mtd>
</mtr>
</mtable>
<mo>;</mo>
</mrow>
α is iterated to calculate by equation below(k):αk=α(k-1)*λ(k-1);
Wherein, perform number of times in circulation and be more than predefined parameter t, or the continuous 3 times values circulated, W(k-2)、W(k-1)、W(k)Approximately equal
In the case of jump out circulation, and determine W(k)Training result be Whistory,α(k)For last step parameter αhistory, λ declines for training
Subtract parameter, λ=0.95, k is more than or equal to 1;
Determine the available berth ratio in parking lot in the following manner according to the static prediction model:
Y1=exp (XTW(k))/(1+exp(XTW(k)));
Wherein, X is the vector form of time series and characteristic, W(k)It is the corresponding weight vectors of characteristic, Y is pre-
Determine idle berth value historical series in historical time and in the order history time under the characteristic in the parking lot,
The ratio of the total Berth number in the parking lot, Y can be accounted in the order history time with Berth number1It is the parking lot following scheduled time
The ratio of the total Berth number in the parking lot can be accounted for Berth number in section.
4. method according to claim 3, it is characterised in that according to idle the berth time series and the feature
Data progress model training is obtained after static prediction model, and methods described also includes:
Regularly update the dynamic prediction model that the static prediction model obtains the parking lot.
5. method according to claim 4, it is characterised in that regularly update the static prediction model and obtain the parking
The dynamic prediction model of field includes:
It is vector X by the observation of each new additionNew, the observation newly added is YNew, the observation to each new addition enters
The following training of row:
Initialize W(k):By WhistoryIt is assigned to W(0),
W is iterated to calculate by equation below(k):
W(k)=W(k-1)+αhistory*XNew T*(YNew-S(XNew*W(k-1)));
W(k)Dynamic prediction model after being updated for training.
6. method according to any one of claim 1 to 5, it is characterised in that the characteristic includes:Advise in parking lot
Mould, whether be solid parking, whether configure ETC equipment, whether the periphery in parking lot has exhibition, weather, temperature.
7. a kind of parking lot berth prediction processing device, it is characterised in that including:
Acquisition module, for obtaining and integrating idle berth time series of the parking lot within the order history time and described predetermined
The characteristic in the parking lot in historical time;
Training module, obtains static pre- for carrying out model training according to idle the berth time series and the characteristic
Survey model;
Prediction module, for being predicted processing to parking lot berth according to the static prediction model.
8. device according to claim 7, it is characterised in that the prediction module includes:
Determining unit, the available berth ratio for determining the parking lot according to the static prediction model;
Predicting unit, for that can be predicted according to described with the berthing capacity in berth ratio and the parking lot in the following scheduled time
The available berth number in the parking lot.
9. device according to claim 8, it is characterised in that the training module is additionally operable to:
Static prediction model W is iterated to calculate by equation below(k):
<mrow>
<mtable>
<mtr>
<mtd>
<mrow>
<msup>
<mi>W</mi>
<mrow>
<mo>(</mo>
<mi>k</mi>
<mo>)</mo>
</mrow>
</msup>
<mo>=</mo>
<msup>
<mi>W</mi>
<mrow>
<mo>(</mo>
<mi>k</mi>
<mo>-</mo>
<mn>1</mn>
<mo>)</mo>
</mrow>
</msup>
<mo>+</mo>
<msup>
<mi>&alpha;</mi>
<mrow>
<mo>(</mo>
<mi>k</mi>
<mo>-</mo>
<mn>1</mn>
<mo>)</mo>
</mrow>
</msup>
<mo>*</mo>
<msup>
<mi>X</mi>
<mi>T</mi>
</msup>
<mo>*</mo>
<mrow>
<mo>(</mo>
<mi>Y</mi>
<mo>-</mo>
<mi>S</mi>
<mo>(</mo>
<mi>a</mi>
<mo>)</mo>
<mo>)</mo>
</mrow>
</mrow>
</mtd>
</mtr>
<mtr>
<mtd>
<mrow>
<mi>S</mi>
<mrow>
<mo>(</mo>
<mi>a</mi>
<mo>)</mo>
</mrow>
<mo>=</mo>
<mfrac>
<mn>1</mn>
<mrow>
<mn>1</mn>
<mo>+</mo>
<mi>exp</mi>
<mrow>
<mo>(</mo>
<mo>-</mo>
<mi>a</mi>
<mo>)</mo>
</mrow>
</mrow>
</mfrac>
</mrow>
</mtd>
</mtr>
<mtr>
<mtd>
<mrow>
<mi>a</mi>
<mo>=</mo>
<mi>X</mi>
<mo>*</mo>
<msup>
<mi>W</mi>
<mrow>
<mo>(</mo>
<mi>k</mi>
<mo>-</mo>
<mn>1</mn>
<mo>)</mo>
</mrow>
</msup>
</mrow>
</mtd>
</mtr>
</mtable>
<mo>;</mo>
</mrow>
α is iterated to calculate by equation below(k):αk=α(k-1)*λ(k-1);
Wherein, perform number of times in circulation and be more than predefined parameter t, or the continuous 3 times values circulated, W(k-2)、W(k-1)、W(k)Approximately equal
In the case of jump out circulation, and determine W(k)Training result be Whistory,α(k)For last step parameter αhistory, λ declines for training
Subtract parameter, λ=0.95, k is more than or equal to 1;
The determining unit is additionally operable to determine the available parking stall ratio in parking lot in the following manner according to the static prediction model
Rate:
Y1=exp (XTW(k))/(1+exp(XTW(k)));
Wherein, X is the vector form of time series and characteristic, W(k)It is the corresponding weight vectors of characteristic, Y is pre-
Determine idle berth value historical series in historical time and in the order history time under the characteristic in the parking lot,
The ratio of the total Berth number in the parking lot, Y can be accounted in the order history time with Berth number1It is the parking lot following scheduled time
The ratio of the total Berth number in the parking lot can be accounted for Berth number in section.
10. device according to claim 9, it is characterised in that described device also includes:
Update module, the dynamic prediction model in the parking lot is obtained for regularly updating the static prediction model.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201610187894.9A CN107239846B (en) | 2016-03-29 | 2016-03-29 | Parking lot berth prediction processing method and device |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201610187894.9A CN107239846B (en) | 2016-03-29 | 2016-03-29 | Parking lot berth prediction processing method and device |
Publications (2)
Publication Number | Publication Date |
---|---|
CN107239846A true CN107239846A (en) | 2017-10-10 |
CN107239846B CN107239846B (en) | 2021-11-02 |
Family
ID=59983709
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201610187894.9A Active CN107239846B (en) | 2016-03-29 | 2016-03-29 | Parking lot berth prediction processing method and device |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN107239846B (en) |
Cited By (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN108776902A (en) * | 2018-05-08 | 2018-11-09 | 西安艾润物联网技术服务有限责任公司 | According to method, storage medium, server and the managing system of car parking in the historical data management parking lot of parking lot operation |
CN109146129A (en) * | 2018-07-03 | 2019-01-04 | 蔚来汽车有限公司 | Charging pile group recommended method can use stake prediction, result acquisition methods and controller |
CN110660219A (en) * | 2019-09-29 | 2020-01-07 | 百度在线网络技术(北京)有限公司 | Parking lot parking prediction method and device |
CN110674962A (en) * | 2018-07-03 | 2020-01-10 | 百度在线网络技术(北京)有限公司 | Vehicle journey prediction processing method and device and storage medium |
CN110888798A (en) * | 2019-10-14 | 2020-03-17 | 西安理工大学 | Software defect prediction method based on graph convolution neural network |
WO2020211606A1 (en) * | 2019-04-18 | 2020-10-22 | 孙进 | Parking space allocation method, parking space allocation system, and storage medium |
CN112435500A (en) * | 2020-12-01 | 2021-03-02 | 深圳市顺易通信息科技有限公司 | Method and device for counting remaining parking spaces of parking lot and terminal equipment |
CN113223291A (en) * | 2021-03-19 | 2021-08-06 | 青岛亿联信息科技股份有限公司 | System and method for predicting number of free parking spaces in parking lot |
Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20040032342A1 (en) * | 2002-08-14 | 2004-02-19 | Dunning Anne E. | Method and system for projecting dynamic parking availability based on an ongoing survey for remote lots with high demand |
CN102496076A (en) * | 2011-11-30 | 2012-06-13 | 广州市交通规划研究所 | Macroscopic, mid-scope and microscopic multilevel urban parking demand prediction model integrated system |
CN102855773A (en) * | 2012-09-13 | 2013-01-02 | 东南大学 | Short-term prediction method for occupancy of effective parking space of parking lot |
CN103093643A (en) * | 2013-01-09 | 2013-05-08 | 东南大学 | Public parking lot berth quantity confirming method |
-
2016
- 2016-03-29 CN CN201610187894.9A patent/CN107239846B/en active Active
Patent Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20040032342A1 (en) * | 2002-08-14 | 2004-02-19 | Dunning Anne E. | Method and system for projecting dynamic parking availability based on an ongoing survey for remote lots with high demand |
CN102496076A (en) * | 2011-11-30 | 2012-06-13 | 广州市交通规划研究所 | Macroscopic, mid-scope and microscopic multilevel urban parking demand prediction model integrated system |
CN102855773A (en) * | 2012-09-13 | 2013-01-02 | 东南大学 | Short-term prediction method for occupancy of effective parking space of parking lot |
CN103093643A (en) * | 2013-01-09 | 2013-05-08 | 东南大学 | Public parking lot berth quantity confirming method |
Cited By (10)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN108776902A (en) * | 2018-05-08 | 2018-11-09 | 西安艾润物联网技术服务有限责任公司 | According to method, storage medium, server and the managing system of car parking in the historical data management parking lot of parking lot operation |
CN109146129A (en) * | 2018-07-03 | 2019-01-04 | 蔚来汽车有限公司 | Charging pile group recommended method can use stake prediction, result acquisition methods and controller |
CN110674962A (en) * | 2018-07-03 | 2020-01-10 | 百度在线网络技术(北京)有限公司 | Vehicle journey prediction processing method and device and storage medium |
WO2020211606A1 (en) * | 2019-04-18 | 2020-10-22 | 孙进 | Parking space allocation method, parking space allocation system, and storage medium |
CN110660219A (en) * | 2019-09-29 | 2020-01-07 | 百度在线网络技术(北京)有限公司 | Parking lot parking prediction method and device |
CN110888798A (en) * | 2019-10-14 | 2020-03-17 | 西安理工大学 | Software defect prediction method based on graph convolution neural network |
CN110888798B (en) * | 2019-10-14 | 2022-11-04 | 西安理工大学 | Software defect prediction method based on graph convolution neural network |
CN112435500A (en) * | 2020-12-01 | 2021-03-02 | 深圳市顺易通信息科技有限公司 | Method and device for counting remaining parking spaces of parking lot and terminal equipment |
CN113223291A (en) * | 2021-03-19 | 2021-08-06 | 青岛亿联信息科技股份有限公司 | System and method for predicting number of free parking spaces in parking lot |
CN113223291B (en) * | 2021-03-19 | 2023-10-20 | 青岛亿联信息科技股份有限公司 | System and method for predicting number of idle parking spaces in parking lot |
Also Published As
Publication number | Publication date |
---|---|
CN107239846B (en) | 2021-11-02 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN107239846A (en) | parking lot berth prediction processing method and device | |
CN108197739B (en) | Urban rail transit passenger flow prediction method | |
CN102629418B (en) | Fuzzy kalman filtering-based traffic flow parameter prediction method | |
CN103971170B (en) | The method and apparatus that a kind of change being used for characteristic information is predicted | |
CN104183119B (en) | Based on the anti-arithmetic for real-time traffic flow distribution forecasting method pushed away of section OD | |
CN105321345B (en) | A kind of road traffic flow prediction method filtered based on ARIMA models and kalman | |
CN106647787A (en) | Satellite onboard autonomous task planning method and system | |
CN108491951A (en) | A kind of prediction technique, device and electronic equipment for taking out distribution time | |
CN109543295A (en) | The meteorological element data processing method and device of numerical weather forecast | |
CN104462685B (en) | Transmission Line Design method based on grid GIS and Floyd algorithms | |
Choi et al. | Bayesian nonparametric feature construction for inverse reinforcement learning | |
CN106127329A (en) | Order forecast method and device | |
CN105933857B (en) | A kind of mobile terminal locations prediction technique and device | |
CN105551239B (en) | travel time prediction method and device | |
CN103971289B (en) | The processing method and processing device of mine ore drawing data | |
CN104217258A (en) | Method for power load condition density prediction | |
CN107045785A (en) | A kind of method of the short-term traffic flow forecast based on grey ELM neutral nets | |
Khazaei et al. | Improving the performance of water balance equation using fuzzy logic approach | |
Macian-Sorribes et al. | Definition of efficient scarcity-based water pricing policies through stochastic programming | |
CN108106624A (en) | A kind of more people's Dispatch by appointment paths planning methods and relevant apparatus | |
CN107392446B (en) | A kind of step power station scheduling scheme evaluation method based on sensitivity analysis | |
CN110413722B (en) | Address selection method, device and non-transient storage medium | |
CN103745602A (en) | Traffic flow predicating method based on sliding window average | |
CN107526815A (en) | The determination method and electronic equipment of Move Mode in the range of target area | |
CN103889053B (en) | Automatic establishing method of self-growing-type fingerprint |
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 |