CN117829377A - Neural network-based remaining parking space prediction method, device, equipment and medium - Google Patents
Neural network-based remaining parking space prediction method, device, equipment and medium Download PDFInfo
- Publication number
- CN117829377A CN117829377A CN202410239050.9A CN202410239050A CN117829377A CN 117829377 A CN117829377 A CN 117829377A CN 202410239050 A CN202410239050 A CN 202410239050A CN 117829377 A CN117829377 A CN 117829377A
- Authority
- CN
- China
- Prior art keywords
- data
- parking space
- parking
- feature
- user
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 38
- 238000013528 artificial neural network Methods 0.000 title claims abstract description 29
- 238000012360 testing method Methods 0.000 claims abstract description 30
- 238000012549 training Methods 0.000 claims description 54
- 238000012795 verification Methods 0.000 claims description 21
- 238000004590 computer program Methods 0.000 claims description 9
- 238000013508 migration Methods 0.000 claims description 9
- 230000005012 migration Effects 0.000 claims description 9
- 238000012216 screening Methods 0.000 claims description 7
- 238000013075 data extraction Methods 0.000 claims description 6
- 238000005070 sampling Methods 0.000 claims description 6
- 238000005457 optimization Methods 0.000 claims description 5
- 238000007781 pre-processing Methods 0.000 claims description 4
- 238000004422 calculation algorithm Methods 0.000 claims description 3
- 238000011156 evaluation Methods 0.000 claims description 3
- 238000005259 measurement Methods 0.000 claims description 3
- 238000012545 processing Methods 0.000 abstract description 2
- 238000007726 management method Methods 0.000 description 6
- 230000000694 effects Effects 0.000 description 4
- 230000005540 biological transmission Effects 0.000 description 3
- 238000004364 calculation method Methods 0.000 description 3
- 238000010586 diagram Methods 0.000 description 3
- 238000013439 planning Methods 0.000 description 3
- 238000000605 extraction Methods 0.000 description 2
- 230000006870 function Effects 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 230000008569 process Effects 0.000 description 2
- 238000012800 visualization Methods 0.000 description 2
- 238000013459 approach Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000010276 construction Methods 0.000 description 1
- 238000013480 data collection Methods 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 238000013210 evaluation model Methods 0.000 description 1
- 230000004927 fusion Effects 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 239000003550 marker Substances 0.000 description 1
- 230000007246 mechanism Effects 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 238000013468 resource allocation Methods 0.000 description 1
- 238000012163 sequencing technique Methods 0.000 description 1
- 230000003068 static effect Effects 0.000 description 1
- 230000000007 visual effect Effects 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/213—Feature extraction, e.g. by transforming the feature space; Summarisation; Mappings, e.g. subspace methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
- G06Q50/10—Services
Landscapes
- Engineering & Computer Science (AREA)
- Business, Economics & Management (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- General Physics & Mathematics (AREA)
- Strategic Management (AREA)
- Tourism & Hospitality (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Human Resources & Organizations (AREA)
- Economics (AREA)
- General Business, Economics & Management (AREA)
- Artificial Intelligence (AREA)
- Life Sciences & Earth Sciences (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Evolutionary Biology (AREA)
- General Engineering & Computer Science (AREA)
- Evolutionary Computation (AREA)
- Bioinformatics & Computational Biology (AREA)
- Marketing (AREA)
- Primary Health Care (AREA)
- General Health & Medical Sciences (AREA)
- Health & Medical Sciences (AREA)
- Development Economics (AREA)
- Game Theory and Decision Science (AREA)
- Entrepreneurship & Innovation (AREA)
- Operations Research (AREA)
- Quality & Reliability (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
- Traffic Control Systems (AREA)
Abstract
The invention provides a method, a device, equipment and a medium for predicting a residual parking space based on a neural network, and relates to the technical field of digital data processing, wherein the method comprises the following steps: obtaining a prediction request of a target parking lot, obtaining historical parking data of the target parking lot and user consumption data, and generating a user portrait corresponding to the target parking lot; taking the time characteristics, the user images, the holiday characteristics and the charging characteristics of the target parking lot as characteristic data, extracting effective characteristics and generating screened characteristic data; constructing a data set of a time sequence model by taking a parking space of a target parking lot as first dimension data, a time period to be predicted as second dimension data and screened feature data as third dimension data, and predicting the residual parking space through the time sequence model by using a test set; and sending the predicted result of the residual parking space to the user terminal. By combining the feature data of the user portrait, the accuracy of the prediction of the residual parking space is improved.
Description
Technical Field
The invention relates to the technical field of digital data processing, in particular to a method, a device, equipment and a medium for predicting a residual parking space based on a neural network.
Background
As the urban process is accelerated, the quantity of the reserved automobiles is increased, and the problem of difficult urban parking is increasingly outstanding, the residual parking spaces of the urban large-scale mall parking lot are predicted so as to better plan and manage the parking lot. However, the remaining space of the parking space is varied, someone enters and exits, and assuming that there is no parking space, a function for predicting the remaining parking space needs to be designed, such as grasping features of the existing vehicle owners, such as past parking habits, whether to live nearby, whether to find out from consumption ordering information to eat temporarily, so as to infer the flow of the clients, predict the remaining parking space in a big data manner, and dynamically adjust the prediction in combination with the current situation of the reserved parking space, so that the problem needs to be solved. The existing prediction method has low accuracy and cannot meet the actual requirements.
Disclosure of Invention
In view of the above, the invention provides a method for predicting the remaining parking space based on a neural network, so as to solve the technical problem of lower accuracy of the prediction of the remaining parking space in the prior art. The method comprises the following steps:
obtaining a prediction request of a target parking lot, obtaining historical parking data and user consumption data of the target parking lot, associating the historical parking data with the user consumption data, and generating a user portrait corresponding to the target parking lot, wherein the user portrait comprises consumption characteristics and parking habits of a user;
taking time features, user images, holiday features and charging features of a target parking lot as feature data, screening the feature data through an integrated learning model, and generating screened feature data after extracting effective features;
taking a parking space of a target parking lot as first dimension data, taking a time period to be predicted as second dimension data, taking screened characteristic data as third dimension data, constructing a data set of a time sequence model, dividing the data in the data set into a training set, a verification set and a test set, training the time sequence model by using the training set, verifying by the verification set, adjusting super parameters of the time sequence model until the time sequence model meets target errors, and predicting the rest parking spaces by using the test set by using the time sequence model;
and sending the predicted result of the residual parking space to the user terminal.
The invention also provides a device for predicting the residual parking space based on the neural network, which is used for solving the technical problem of lower accuracy of the prediction of the residual parking space in the prior art. The device comprises:
the user portrait acquisition module is used for acquiring a prediction request of a target parking lot, acquiring historical parking data and user consumption data of the target parking lot, correlating the historical parking data with the user consumption data, and generating a user portrait corresponding to the target parking lot, wherein the user portrait comprises consumption characteristics and parking habits of a user;
the feature data extraction module is used for taking time features, user images, holiday features and charging features of the target parking lot as feature data, screening the feature data through the integrated learning model, and generating screened feature data after extracting effective features;
the parking space prediction module is used for taking a parking space of a target parking lot as first dimension data, a time period to be predicted as second dimension data and screened characteristic data as third dimension data to construct a data set of a time sequence model, dividing the data in the data set into a training set, a verification set and a test set, training the time sequence model by using the training set, verifying by the verification set, adjusting parameters of the time sequence model until the time sequence model meets target errors, and predicting the rest parking spaces by using the test set by using the time sequence model;
and the prediction result sending module is used for sending the prediction result of the residual parking space to the user terminal.
The invention also provides computer equipment, which comprises a memory, a processor and a computer program stored on the memory and capable of running on the processor, wherein the processor realizes the random residual parking space prediction method based on the neural network when executing the computer program, so as to solve the technical problem of lower residual parking space prediction accuracy in the prior art.
The invention also provides a computer readable storage medium which stores a computer program for executing the method for predicting the remaining parking space based on the neural network, so as to solve the technical problem of lower accuracy of predicting the remaining parking space in the prior art.
Compared with the prior art, the beneficial effects that above-mentioned at least one technical scheme that this specification adopted can reach include at least:
and introducing the living consumption habit of the user into a prediction model to realize personalized prediction of the residual parking space. Factors such as shopping time and consumption habit of different users influence parking requirements, so that the requirements of each user can be met more accurately by personalized prediction; the feature data is screened through the integrated learning model, so that the prediction accuracy is improved, the user portrait and the time dimension are added into the feature data of the time sequence model, the prediction is more accurate, and the model prediction accuracy is improved.
Drawings
In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings that are needed in the embodiments will be briefly described below, and it is obvious that the drawings in the following description are only some embodiments of the present application, and that other drawings may be obtained according to these drawings without inventive effort for a person skilled in the art.
FIG. 1 is a flowchart of a method for predicting a remaining parking space based on a neural network according to an embodiment of the present invention;
FIG. 2 is a block diagram of a computer device according to an embodiment of the present invention;
fig. 3 is a structural block diagram of a residual parking space prediction device based on a neural network according to an embodiment of the present invention.
Detailed Description
Embodiments of the present application are described in detail below with reference to the accompanying drawings.
Other advantages and effects of the present application will become apparent to those skilled in the art from the present disclosure, when the following description of the embodiments is taken in conjunction with the accompanying drawings. It will be apparent that the described embodiments are only some, but not all, of the embodiments of the present application. The present application may be embodied or carried out in other specific embodiments, and the details of the present application may be modified or changed from various points of view and applications without departing from the spirit of the present application. It should be noted that the following embodiments and features in the embodiments may be combined with each other without conflict. All other embodiments, which can be made by one of ordinary skill in the art based on the embodiments herein without making any inventive effort, are intended to be within the scope of the present application.
In an embodiment of the present invention, a method for predicting a remaining parking space based on a neural network is provided, as shown in fig. 1, where the method includes:
step S101: obtaining a prediction request of a target parking lot, obtaining historical parking data and user consumption data of the target parking lot, associating the historical parking data with the user consumption data, and generating a user portrait corresponding to the target parking lot, wherein the user portrait comprises consumption characteristics and parking habits of a user;
step S102: taking time features, user images, holiday features and charging features of a target parking lot as feature data, screening the feature data through an integrated learning model, and generating screened feature data after extracting effective features;
step S103: taking a parking space of a target parking lot as first dimension data, taking a time period to be predicted as second dimension data, taking screened characteristic data as third dimension data, constructing a data set of a time sequence model, dividing the data in the data set into a training set, a verification set and a test set, training the time sequence model by using the training set, verifying by the verification set, adjusting super parameters of the time sequence model until the time sequence model meets target errors, and predicting the rest parking spaces by using the test set by using the time sequence model;
step S104: and sending the predicted result of the residual parking space to the user terminal.
Specifically, the historical parking lot data includes parking space occupancy, date, time, and the like. The user consumption data includes shopping frequency, consumption time period, shopping place, etc. The time features include time-related features such as date, day of week, time period, etc. of extraction. The user features include introducing personalized consumption features in the user representation, such as shopping time, shopping places, etc. Holiday features include whether the marker is a holiday or a special event day.
Specifically, user privacy protection is an important problem of current social internet applications, and privacy protection measures are taken when collecting user data, such as data anonymization and encryption, so as to ensure the safety of the user data and to process sensitive information in compliance.
In the specific implementation, in order to predict the occupation condition of a parking space through a time sequence model, a data set of the time sequence model is constructed by taking a parking space of a target parking lot as first dimension data, a time period to be predicted as second dimension data and screened feature data as third dimension data, the data in the data set is divided into a training set, a verification set and a test set, after the training set is used for training the time sequence model, the super parameters of the time sequence model are verified through the verification set and adjusted until the time sequence model meets target errors, and the time sequence model is used for predicting the residual parking space:
taking the parking position of the target parking lot as first dimension data, taking the time period to be predicted as second dimension data and taking the screened characteristic data as third dimension data, respectively decomposing the first dimension data, the second dimension data and the third dimension data into time steps, and constructing a data set through the time steps; designating parameters of components of the time sequence model, and constructing the time sequence model; the method comprises the steps that a marked training set is generated, a marked training set is used for training a time sequence model, and super parameters of the time sequence model are adjusted through a back propagation and optimization algorithm until a preset error value is reached; and (3) inputting the test set as an input sequence into a time sequence model, generating an output sequence through the learned weight and state information, and predicting the residual parking space in a future time period according to the output sequence.
Specifically, the super parameters are learning rate, hidden layer size, etc.
In the specific implementation, in order to add the weather features to the feature data and then increase the variety of the feature data, the time features, the user images, the holiday features and the charging features of the target parking lot are used as the feature data through the following steps:
extracting weather data related to a target parking lot in a set time period as weather features; and taking the time characteristics, the user images, the holiday characteristics, the charging characteristics of the target parking lot and the weather characteristics as characteristic data.
In particular, because weather may affect parking requirements. And fusing various data sources, such as parking lot data, user consumption data, weather data and the like, into the prediction model. The comprehensive data fusion can more comprehensively understand the influence factors of the parking requirements, and improves the accuracy of prediction.
Specifically, the activities of the user on the social media may be related to the parking requirements of the user, and the social media data is utilized to capture shopping information, activity participation information and other information of the user, so that the user portraits are further enriched, and the prediction accuracy is improved.
In some embodiments, a set period of time between the previous hour of the current time and the next hour of the current time may be used as the extraction weather feature.
In specific implementation, in order to screen the feature data through the integrated learning model to improve the accuracy of the time sequence model prediction, the time feature, the user image, the holiday feature and the charging feature of the target parking lot are taken as the feature data, the feature data is screened through the integrated learning model, and the screened feature data is generated after effective features are extracted:
preprocessing the feature data, removing noise in the feature data, unifying and standardizing the length of the feature data, and generating preprocessed feature data; calculating the gradient of the sample of the preprocessed feature data, taking the sample as a non-important sample if the gradient of the sample is smaller than a threshold value, sampling the non-important sample, retaining a part of the samples, and generating the sampled feature data; extracting sample features in the sampled feature data, calculating the importance degree of the sample features, and sequencing the sample features according to the importance degree; selecting a plurality of sample features with high importance as feature parameters according to the required speed and precision to construct an integrated learning model, dividing the preprocessed feature data into a training set and a testing set, and training the integrated learning model by using the training set; verifying the training error of the integrated learning model by using the test set, calculating the training error by using the error measurement index, and adjusting the super parameters of the integrated learning model until the training error meets the evaluation standard; and (3) inputting the test set as a sequence into the integrated learning model, generating and outputting the filtered characteristic data.
Specifically, the important samples with larger gradients are reserved through a gradient sampling method, and meanwhile, the number of samples is reduced through sampling samples with smaller gradients, so that the calculation cost is reduced while the prediction accuracy is not reduced.
Specifically, an error metric such as Root Mean Square Error (RMSE) or Mean Absolute Error (MAE) is used.
In the specific implementation, in order to introduce a user participation mechanism, a user is allowed to provide actual parking information feedback, so that a prediction model is further optimized, the adaptability and the user satisfaction degree of a prediction method are enhanced, and the feedback of the user is updated to a verification set through the following steps:
collecting actual parking data of a user; comparing the actual parking data with the predicted remaining parking space data, and calculating the error of the actual parking data and the predicted remaining parking space data according to the comparison result; and according to the error, adjusting parameters of the time sequence model, and updating the actual parking data into a verification set after converting the data format.
In the implementation, in order to provide a visual preview of the remaining parking space for the user in the application program, and simultaneously provide intelligent recommended parking spaces based on user habits and prediction results, the user is helped to find a proper parking space more easily, and the navigation data are calculated and pushed to the user terminal through the following steps:
calling an open source map, and calculating the distance between the user terminal and the predicted remaining parking space in real time according to the predicted positioning information of the remaining parking space and the open source map; and selecting a plurality of residual parking spaces with short distance, generating navigation data from the user to the residual parking spaces according to the positioning information of the selected residual parking spaces, and pushing the navigation data to the user terminal.
Specifically, the prediction results are displayed to the user through the application program, so that the user is helped to plan the parking and shopping journey better.
In specific implementation, in order to utilize mobility learning for a new market or a new place, knowledge of a residual parking space prediction model of an existing market is migrated to the new place, so that a reliable prediction model is built more quickly, and the mobility learning is realized through the following steps:
when the rest parking spaces of other parking lots are predicted, sharing the data set of the time sequence model; directly applying the super parameters in the trained time sequence model to the rest parking space prediction of the parking lot at other positions; and constructing a migration strategy according to the common points and the differences of the user portraits of different parking lots, and adjusting super parameters in the time sequence model according to the migration strategy until the prediction accuracy of the parking lots at other positions is met.
Specifically, the prediction result of the residual parking space can also be used for planning and managing the parking resources of the market, the reasonable planning of the distribution of the parking space is facilitated by predicting the future parking demand, the utilization efficiency of the parking resources is improved, and the sustainable planning of the parking resources is carried out.
Specifically, the migration learning method is used to learn new knowledge by using existing knowledge. To migrating knowledge from an associated task to a new, different task, thereby reducing learning time and computing resource consumption on the new task. This approach can quickly adapt and learn new knowledge or skills in the face of complex problems using previous experience and techniques.
Specifically, a mobile application or website may be built and the predicted results presented to the user in the form of charts or text. Meanwhile, the parking condition of the parking lot is monitored through a management interface of the market management team, so that the market management team can allocate resources according to the prediction result of the residual parking spaces.
Specifically, the parking space remaining prediction system can be integrated with other fields of the smart city, such as traffic management, energy management and the like, so as to form a more comprehensive smart city solution.
In this embodiment, a computer device is provided, as shown in fig. 2, including a memory 201, a processor 202, and a computer program stored in the memory and capable of running on the processor, where the processor implements any of the above-mentioned remaining parking space prediction methods based on a neural network when executing the computer program.
In particular, the computer device may be a computer terminal, a server or similar computing means.
In this embodiment, a computer-readable storage medium is provided, in which a computer program for executing any of the above-described neural network-based remaining parking space prediction methods is stored.
In particular, computer-readable storage media include both permanent and non-permanent, removable and non-removable media. The readable storage medium may implement information storage by any method or technique. The information may be computer readable instructions, data structures, modules of a program, or other data. Examples of computer-readable storage media include, but are not limited to, phase-change memory (PRAM), static Random Access Memory (SRAM), dynamic Random Access Memory (DRAM), other types of Random Access Memory (RAM), read Only Memory (ROM), electrically Erasable Programmable Read Only Memory (EEPROM), flash memory or other memory technology, compact disc read only memory (CD-ROM), digital Versatile Disks (DVD) or other optical storage, magnetic cassettes, magnetic tape disk storage or other magnetic storage devices, or any other non-transmission medium, which can be used to store information that can be accessed by a computing device. Computer-readable storage media, as defined herein, does not include transitory computer-readable media (transmission media), such as modulated data signals and carrier waves.
Based on the same inventive concept, the embodiment of the invention also provides a device for predicting the remaining parking space based on the neural network, as described in the following embodiment. Because the principle of solving the problem of the residual parking space predicting device based on the neural network is similar to that of the residual parking space predicting method based on the neural network, the implementation of the residual parking space predicting device based on the neural network can be referred to the implementation of the residual parking space predicting method based on the neural network, and repeated parts are omitted. As used below, the term "unit" or "module" may be a combination of software and/or hardware that implements the intended function. While the means described in the following embodiments are preferably implemented in software, implementation in hardware, or a combination of software and hardware, is also possible and contemplated.
Fig. 3 is a block diagram of a device for predicting remaining parking space based on neural network according to an embodiment of the present invention, as shown in fig. 3, including: the configuration of the user figure acquisition module 301, the feature data extraction module 302, the parking space prediction module 303, and the prediction result transmission module 304 will be described below.
The user portrait acquisition module 301 is configured to acquire a prediction request of a target parking lot, acquire historical parking data and user consumption data of the target parking lot, and associate the historical parking data with the user consumption data to generate a user portrait corresponding to the target parking lot, where the user portrait includes consumption characteristics and parking habits of a user;
the feature data extraction module 302 is configured to use a time feature, a user image, a holiday feature, and a charging feature of a target parking lot as feature data, screen the feature data through an integrated learning model, extract effective features, and generate screened feature data;
the parking space prediction module 303 is configured to construct a dataset of a time sequence model by using a parking space of a target parking lot as first dimension data, a time period to be predicted as second dimension data, and filtered feature data as third dimension data, divide the dataset into a training set, a verification set and a test set, train the time sequence model by using the training set, verify the time sequence model by using the verification set, adjust parameters of the time sequence model until the time sequence model meets a target error, and predict the remaining parking space by using the test set through the time sequence model;
and the prediction result sending module 304 is configured to send the prediction result of the remaining parking space to the user terminal.
In one embodiment, the parking space prediction module includes:
the characteristic data preprocessing unit is used for preprocessing the characteristic data, removing noise in the characteristic data, unifying and standardizing the length of the characteristic data and generating preprocessed characteristic data;
the characteristic sampling unit is used for calculating the gradient of the sample of the preprocessed characteristic data, taking the sample as a non-important sample if the gradient of the sample is smaller than a threshold value, sampling the non-important sample, retaining a part of the samples, and generating the sampled characteristic data;
the feature ordering unit is used for extracting sample features in the sampled feature numbers, calculating the importance degree of the sample features, and ordering the sample features according to the importance degree;
the model training unit is used for selecting a plurality of sample features with high importance as feature parameters according to the required speed and precision to construct an integrated learning model, dividing the preprocessed feature data into a training set and a testing set, and training the integrated learning model by using the training set;
the evaluation model unit is used for verifying the training error of the integrated learning model by using the test set, calculating the training error by using the error measurement index, and adjusting the parameters of the integrated learning model until the training error meets the evaluation standard;
and the residual parking space prediction unit is used for inputting the test set as a sequence into the integrated learning model, generating and outputting the screened characteristic data.
In one embodiment, the feature data extraction module comprises:
the data set constructing unit is used for taking the parking position of the target parking lot as first dimension data, taking the time period to be predicted as second dimension data and taking the screened characteristic data as third dimension data, respectively decomposing the first dimension data, the second dimension data and the third dimension data into time steps, and constructing a data set through the time steps;
the time sequence model building unit is used for specifying parameters of components of the time sequence model and building the time sequence model;
the training time sequence model unit is used for marking the training set to generate a marked training set, training the time sequence model by using the marked training set, and adjusting super parameters of the time sequence model through a back propagation and optimization algorithm until a preset error value is reached;
and the characteristic data output unit is used for inputting the test set as a sequence into the time sequence model, generating and outputting the filtered characteristic data through the learned weight and state information.
In one embodiment, the feature data extraction module comprises:
the weather data acquisition unit is used for extracting weather data related to the target parking lot in a set time period as weather features;
and the characteristic data collection unit is used for taking the time characteristic, the user image, the holiday characteristic, the charging characteristic of the target parking lot and the weather characteristic as characteristic data.
In one embodiment, the apparatus further comprises:
and the continuous optimization module is used for collecting user feedback and actual prediction conditions and continuously optimizing the model to enable the model to better meet the actual demands of the user.
In one embodiment, the continuous optimization module includes:
the method comprises the steps of collecting an actual data unit, which is used for collecting actual parking data of a user;
the error calculation unit is used for comparing the actual parking data with the predicted remaining parking space data and calculating the error of the actual parking data and the predicted remaining parking space data according to the comparison result;
and the updating verification set unit is used for adjusting parameters of the time sequence model according to the errors and updating the actual parking data into the verification set after converting the data format.
In one embodiment, the apparatus further comprises:
and the visualization and intelligent recommendation module is used for providing the visualized parking space remaining condition for the user in the application program, and simultaneously providing intelligent recommended parking spaces based on the habit and the prediction result of the user, so that the user can find the proper parking spaces more easily.
In one embodiment, the visualization and intelligent recommendation module includes:
the distance calculating unit is used for calling the open source map, and calculating the distance between the user terminal and the predicted remaining parking space in real time according to the predicted positioning information of the remaining parking space and the open source map;
the navigation data pushing unit is used for selecting a plurality of residual parking spaces with short distance, generating navigation data from a user to the residual parking spaces according to the positioning information of the selected residual parking spaces, and pushing the navigation data to the user terminal.
In one embodiment, the apparatus further comprises:
and the mobility learning module is used for migrating knowledge of the parking space prediction model of the existing market to the new place by using mobility learning for the new market or the new place, so that a reliable prediction model can be built more quickly.
In one embodiment, the mobility learning module includes:
the shared data set unit is used for sharing the training set and the testing set of the time sequence model when the parking lot at other positions is subjected to residual parking space prediction;
the migration prediction unit is used for directly applying the super parameters in the trained time sequence model to the prediction of the residual parking spaces of the parking lots at other positions;
and the migration strategy construction unit is used for constructing a migration strategy according to the common points and the differences of the user portraits of different parking lots, and adjusting super parameters in the time sequence model according to the migration strategy until the prediction accuracy of the parking lots at other positions is met.
The embodiment of the invention realizes the following technical effects:
before prediction, screening the characteristic data, and calculating gradient and sample characteristics to reduce the data calculation amount and screening the characteristic data by an ensemble learning model to improve the prediction precision; the user portrait and the time dimension are added into the characteristic data of the time sequence model, so that the prediction is more accurate, and the accuracy of model prediction is improved; the living consumption habit of the user is introduced into the prediction model, so that personalized parking space remaining prediction is realized, and factors such as shopping time, consumption habit and the like of different users influence parking requirements, so that the requirements of each user can be more accurately met by personalized prediction; the method not only can predict the residual situation of the parking space, but also can update the prediction model in real time, and can be adjusted continuously according to actual parking data, and the dynamic property can cope with the continuously changing traffic flow and user habit in the city; the parking space remaining prediction system based on the consumer life consumption habit of the user is more intelligent, personalized and practical, better parking experience is provided for the user, and more accurate resource allocation decision support is provided for a market management team.
It will be apparent to those skilled in the art that the modules or steps of the embodiments of the invention described above may be implemented in a general purpose computing device, they may be concentrated on a single computing device, or distributed across a network of computing devices, they may alternatively be implemented in program code executable by computing devices, so that they may be stored in a storage device for execution by computing devices, and in some cases, the steps shown or described may be performed in a different order than what is shown or described, or they may be separately fabricated into individual integrated circuit modules, or a plurality of modules or steps in them may be fabricated into a single integrated circuit module. Thus, embodiments of the invention are not limited to any specific combination of hardware and software.
The above description is only of the preferred embodiments of the present invention and is not intended to limit the present invention, and various modifications and variations can be made to the embodiments of the present invention by those skilled in the art. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention should be included in the protection scope of the present invention.
Claims (10)
1. The method for predicting the remaining parking space based on the neural network is characterized by comprising the following steps of:
obtaining a prediction request of a target parking lot, obtaining historical parking data and user consumption data of the target parking lot, associating the historical parking data with the user consumption data, and generating a user portrait corresponding to the target parking lot, wherein the user portrait comprises consumption characteristics and parking habits of a user;
taking the time feature, the user image, the holiday feature and the charging feature of the target parking lot as feature data, screening the feature data through an integrated learning model, extracting effective features and generating screened feature data;
taking a parking space of the target parking lot as first dimension data, taking a time period to be predicted as second dimension data, taking the screened characteristic data as third dimension data, constructing a data set of a time sequence model, dividing the data in the data set into a training set, a verification set and a test set, using the training set to train the time sequence model, verifying and adjusting super parameters of the time sequence model through the verification set until the time sequence model meets a target error, and predicting the residual parking space by using the test set through the time sequence model;
and sending the prediction result of the residual parking space to a user terminal.
2. The method for predicting a remaining parking space based on a neural network according to claim 1, wherein a data set of a time series model is constructed by taking a parking space of the target parking space as first dimension data, a time period to be predicted as second dimension data, and the screened feature data as third dimension data, the data in the data set is divided into a training set, a verification set and a test set, the training set is used for training the time series model, and super parameters of the time series model are verified and adjusted through the verification set until the time series model meets a target error, and the time series model is used for predicting the remaining parking space by using the test set, including:
taking the parking position of the target parking lot as first dimension data, the time period to be predicted as second dimension data and the screened characteristic data as third dimension data, respectively decomposing the first dimension data, the second dimension data and the third dimension data into time steps, and constructing the data set through the time steps;
designating component parameters of the time sequence model, and constructing a time sequence model;
generating a marked training set after marking the training set, training the time sequence model by using the marked training set, and adjusting super parameters of the time sequence model by a back propagation and optimization algorithm until a preset error value is reached;
and inputting the test set as an input sequence into the time sequence model, generating an output sequence through the learned weight and state information, and predicting the residual parking space in the future time period according to the output sequence.
3. The neural network-based remaining parking space prediction method according to claim 2, wherein taking a time feature, the user image, a holiday feature, and a charging feature of the target parking lot as feature data, comprises:
extracting weather data related to the target parking lot in a set time period as weather features;
and taking the time characteristic, the user portrait, the holiday characteristic, the charging characteristic of the target parking lot and the weather characteristic as the characteristic data.
4. The neural network-based remaining parking space prediction method of claim 1, wherein the time feature, the user image, the holiday feature and the charging feature of the target parking lot are taken as feature data, the feature data are screened through an ensemble learning model, valid features are extracted, and screened feature data are generated, and the method comprises the following steps:
preprocessing the characteristic data, removing noise in the characteristic data, unifying and standardizing the length of the characteristic data, and generating preprocessed characteristic data;
calculating the gradient of a sample of the preprocessed feature data, taking the sample as a non-important sample if the gradient of the sample is smaller than a threshold value, sampling the non-important sample, and reserving a part of samples to generate sampled feature data;
extracting sample features in the sampled feature data, calculating the importance degree of the sample features, and sorting the sample features according to the importance degree;
selecting a plurality of sample features with high importance as feature parameters according to the required speed and precision to construct an integrated learning model, dividing the preprocessed feature data into a training set and a testing set, and training the integrated learning model by using the training set;
verifying the training error of the integrated learning model by using the test set, calculating the training error by using an error measurement index, and adjusting the super parameters of the integrated learning model until the training error meets an evaluation standard;
and inputting the test set as a sequence to an integrated learning model, generating and outputting the screened characteristic data.
5. The neural network-based remaining parking space prediction method according to any one of claims 1 to 4, further comprising:
collecting actual parking data of a user;
comparing the actual parking data with the predicted remaining parking space data, and calculating the error of the actual parking data and the predicted remaining parking space data according to a comparison result;
and according to the error, adjusting parameters of the time sequence model, and updating the actual parking data into the verification set after converting the data format.
6. The neural network-based remaining parking space prediction method according to any one of claims 1 to 4, further comprising:
calling an open source map, and calculating the distance between the user terminal and the predicted remaining parking space in real time according to the predicted positioning information of the remaining parking space and the open source map;
and selecting a plurality of residual parking spaces with short distance, generating navigation data from the user to the residual parking spaces according to the selected positioning information of the residual parking spaces, and pushing the navigation data to the user terminal.
7. The neural network-based remaining parking space prediction method according to any one of claims 1 to 4, further comprising:
when the rest parking spaces of other parking lots are predicted, sharing a training set and a testing set of the time sequence model;
directly applying the super parameters in the trained time sequence model to the rest parking space prediction of the parking lot at other positions;
and constructing a migration strategy according to the common points and the differences of the user portraits of different parking lots, and adjusting the super parameters in the time sequence model according to the migration strategy until the prediction accuracy of the parking lots at other positions is met.
8. The utility model provides a surplus parking stall prediction device based on neural network which characterized in that includes:
the user portrait acquisition module is used for acquiring a prediction request of a target parking lot, acquiring historical parking data and user consumption data of the target parking lot, associating the historical parking data with the user consumption data, and generating a user portrait corresponding to the target parking lot, wherein the user portrait comprises consumption characteristics and parking habits of a user;
the feature data extraction module is used for taking the time feature, the user image, the holiday feature and the charging feature of the target parking lot as feature data, screening the feature data through an integrated learning model, and generating screened feature data after extracting effective features;
the parking space prediction module is used for taking a parking space of the target parking lot as first dimension data, taking a time period to be predicted as second dimension data and taking the screened characteristic data as third dimension data to construct a data set of a time sequence model, dividing the data in the data set into a training set, a verification set and a test set, using the training set to train the time sequence model, verifying and adjusting super parameters of the time sequence model through the verification set until the time sequence model meets a target error, and predicting the residual parking space by using the test set through the time sequence model;
and the prediction result sending module is used for sending the prediction result of the residual parking space to a user terminal.
9. A computer device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, characterized in that the processor implements the neural network-based residual parking space prediction method of any one of claims 1 to 7 when the computer program is executed.
10. A computer-readable storage medium, characterized in that the computer-readable storage medium stores a computer program that performs the neural network-based remaining parking space prediction method of any one of claims 1 to 7.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202410239050.9A CN117829377A (en) | 2024-03-04 | 2024-03-04 | Neural network-based remaining parking space prediction method, device, equipment and medium |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202410239050.9A CN117829377A (en) | 2024-03-04 | 2024-03-04 | Neural network-based remaining parking space prediction method, device, equipment and medium |
Publications (1)
Publication Number | Publication Date |
---|---|
CN117829377A true CN117829377A (en) | 2024-04-05 |
Family
ID=90522919
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202410239050.9A Pending CN117829377A (en) | 2024-03-04 | 2024-03-04 | Neural network-based remaining parking space prediction method, device, equipment and medium |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN117829377A (en) |
Citations (10)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110503104A (en) * | 2019-08-29 | 2019-11-26 | 桂林电子科技大学 | A kind of remaining parking stall quantitative forecasting technique in short-term based on convolutional neural networks |
CN112650941A (en) * | 2019-10-12 | 2021-04-13 | 沈海峰 | Intelligent parking lot recommendation method based on sorting and collaborative algorithm |
CN112669071A (en) * | 2020-12-30 | 2021-04-16 | 中兴智能交通股份有限公司 | System and method for analyzing targeted advertisement marketing based on vehicle parking data |
CN113032670A (en) * | 2021-03-15 | 2021-06-25 | 清华大学 | Parking lot recommendation method and device, computer equipment and storage medium |
CN113822458A (en) * | 2020-11-18 | 2021-12-21 | 京东城市(北京)数字科技有限公司 | Prediction method, training method, device, electronic equipment and readable storage medium |
CN115830848A (en) * | 2022-10-25 | 2023-03-21 | 重庆电子工程职业学院 | Shared parking space intelligent distribution system and method based on LSTM model |
CN116167536A (en) * | 2022-12-07 | 2023-05-26 | 江苏巨楷科技发展有限公司 | Intelligent parking management method based on time period learning optimization |
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 |
CN116665479A (en) * | 2023-05-26 | 2023-08-29 | 成都赛力斯科技有限公司 | Parking stall recommendation method and device, electronic equipment and storage medium |
CN116665482A (en) * | 2023-06-05 | 2023-08-29 | 西华大学 | Parking space recommending method and device based on intelligent parking |
-
2024
- 2024-03-04 CN CN202410239050.9A patent/CN117829377A/en active Pending
Patent Citations (10)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110503104A (en) * | 2019-08-29 | 2019-11-26 | 桂林电子科技大学 | A kind of remaining parking stall quantitative forecasting technique in short-term based on convolutional neural networks |
CN112650941A (en) * | 2019-10-12 | 2021-04-13 | 沈海峰 | Intelligent parking lot recommendation method based on sorting and collaborative algorithm |
CN113822458A (en) * | 2020-11-18 | 2021-12-21 | 京东城市(北京)数字科技有限公司 | Prediction method, training method, device, electronic equipment and readable storage medium |
CN112669071A (en) * | 2020-12-30 | 2021-04-16 | 中兴智能交通股份有限公司 | System and method for analyzing targeted advertisement marketing based on vehicle parking data |
CN113032670A (en) * | 2021-03-15 | 2021-06-25 | 清华大学 | Parking lot recommendation method and device, computer equipment and storage medium |
CN115830848A (en) * | 2022-10-25 | 2023-03-21 | 重庆电子工程职业学院 | Shared parking space intelligent distribution system and method based on LSTM model |
CN116167536A (en) * | 2022-12-07 | 2023-05-26 | 江苏巨楷科技发展有限公司 | Intelligent parking management method based on time period learning optimization |
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 |
CN116665479A (en) * | 2023-05-26 | 2023-08-29 | 成都赛力斯科技有限公司 | Parking stall recommendation method and device, electronic equipment and storage medium |
CN116665482A (en) * | 2023-06-05 | 2023-08-29 | 西华大学 | Parking space recommending method and device based on intelligent parking |
Non-Patent Citations (1)
Title |
---|
韩蕊: "《阿里巴巴B2B电商算法实战》", 31 July 2020, 机械工业出版社, pages: 30 - 31 * |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
Zhu et al. | Parallel transportation management and control system and its applications in building smart cities | |
Long et al. | Mapping block-level urban areas for all Chinese cities | |
Liu et al. | Understanding public transit patterns with open geodemographics to facilitate public transport planning | |
Wang et al. | Trajectory analysis for on-demand services: A survey focusing on spatial-temporal demand and supply patterns | |
CN105893537B (en) | The determination method and apparatus of geography information point | |
Nuzzolo et al. | Individual utility‐based path suggestions in transit trip planners | |
CN107239846B (en) | Parking lot berth prediction processing method and device | |
Wismans et al. | Improving a priori demand estimates transport models using mobile phone data: a Rotterdam-region case | |
CN112069635B (en) | Method and device for deploying battery changing cabinet, medium and electronic equipment | |
Ma et al. | Public transportation big data mining and analysis | |
Wepulanon et al. | A real-time bus arrival time information system using crowdsourced smartphone data: a novel framework and simulation experiments | |
Mor et al. | Who is a tourist? Classifying international urban tourists using machine learning | |
CN106850777B (en) | Method and device for pushing information | |
Cheranchery et al. | Improving ridership and reducing subsidy for premium bus service in Kolkata Metro City | |
Gourbesville | Which models for decision support systems? Proposal for a methodology | |
Martí et al. | Charging stations and mobility data generators for agent-based simulations | |
CN117829377A (en) | Neural network-based remaining parking space prediction method, device, equipment and medium | |
Rahman et al. | MDLpark: available parking prediction for smart parking through mobile deep learning | |
Sun et al. | The Reliability Model for Bike‐Sharing Dispatch Based on Hotspot Detection and Hypothesis Test: A Case Study in Beijing | |
Wei | Energy Footprinting and Human-Centric Building Co-Optimization with Multi-Task Deep Reinforcement Learning | |
Rahaman | Context-aware mobility analytics and trip planning | |
Skhosana et al. | A Real-Time Machine Learning-Based Public Transport Bus-Passenger Information System. | |
Hu | Dynamic origin-destination estimation with location-based social networking data: exploring urban travel demand sensor | |
Shi et al. | A spatiotemporal analysis of metro‐stop based accessibility to bus services in Shanghai, China | |
Lelke et al. | Generating Standardized Agent-Based Transport Models in Germany |
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 |