Urban parking space scheduling method and system based on dynamic demand prediction

Abstract

The present invention discloses a method for scheduling urban parking spaces based on dynamic demand prediction, the method comprising: receiving historical parking data and real-time external data, and preprocessing the data; constructing a parking demand prediction model based on a deep learning model; calling a pre-constructed parking space scheduling model to output a parking space allocation strategy based on the parking demand prediction output by the parking demand prediction model; executing the allocation strategy and collecting feedback data in real time to adjust the parameters of the demand prediction model and the parking space scheduling model in real time. By receiving historical parking data and real-time external data and preprocessing them, using a deep learning model to construct a parking demand prediction model, combining the parking space scheduling model to generate a dynamic parking space allocation strategy, and optimizing the model parameters by collecting feedback data in real time, real-time dynamic adjustment of parking space allocation is achieved. The accuracy of parking demand prediction and parking space utilization are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of parking space scheduling, and in particular to a method and system for scheduling urban parking spaces based on dynamic demand prediction. Background Art

[0002] With the acceleration of urbanization and the continuous growth of motor vehicle ownership, the problem of urban parking is becoming increasingly serious. As an important part of the urban transportation system, the management efficiency of parking lots directly affects the smoothness of urban traffic and the quality of life of residents. However, traditional parking management methods usually rely on manual experience or simple rule settings, and cannot effectively cope with the complex dynamic parking needs of cities. This method leads to problems such as uneven distribution of parking resources, low utilization of parking spaces, and long waiting time for vehicles. In recent years, intelligent parking systems have gradually received attention and application. The intelligent parking lot management method using Internet of Things technology, sensor technology and artificial intelligence algorithms has significantly improved the efficiency of parking management. In particular, data-driven parking demand prediction and urban parking space scheduling methods provide an innovative solution to alleviate the contradiction between parking supply and demand. However, the current intelligent parking system still has the following technical challenges: parking management involves multi-dimensional data, including vehicle entry and exit data within the parking area, parking space usage data, and external real-time traffic flow, weather changes and other dynamic factors. The acquisition, cleaning, fusion and analysis of these data have not yet formed a systematic method, making it difficult to give full play to the value of the data. Existing parking demand prediction methods mostly use traditional statistical models or simple machine learning algorithms, which cannot fully capture the nonlinear changes and complex time series characteristics of parking demand, resulting in insufficient accuracy and real-time performance of the prediction results. Current parking scheduling usually uses fixed rules or static optimization models, which cannot dynamically adjust the parking allocation strategy according to real-time parking demand changes, which can easily lead to resource waste or increased vehicle waiting time. Many intelligent parking systems lack real-time feedback mechanisms and cannot adjust and optimize the prediction model and scheduling model based on actual execution results, making it difficult for the system to continuously improve performance. Summary of the invention

[0003] In view of this, the present invention proposes a method and system for scheduling urban parking spaces based on dynamic demand prediction, which can predict parking demand from multiple aspects and dynamically allocate parking spaces based on real-time parking data. The present invention provides the following technical solutions:

[0004] A method for scheduling urban parking spaces based on dynamic demand prediction, the method comprising: receiving historical parking data and real-time external data, and preprocessing the data; using the preprocessed data to build a parking demand prediction model based on a deep learning model; calling a pre-built parking space scheduling model to output a parking space allocation strategy based on the parking demand prediction output by the parking demand prediction model; executing the parking space allocation strategy and collecting feedback data of the allocation strategy in real time, so as to adjust the parameters of the demand prediction model and the parking space scheduling model in real time according to the feedback data.

[0005] Optionally, the method for receiving historical parking data and real-time external data includes: receiving historical parking data sent by a management system of a parking area, the historical parking data including vehicle entry and exit time data, vehicle type data, parking duration data and average daily traffic flow data of multiple parking areas; acquiring real-time external data outside the parking area through multiple open API interfaces in real time, the real-time external data including traffic flow data and influencing factor data, the traffic flow data including real-time traffic flow data, vehicle speed and congestion data of multiple road sections, and temporary traffic flow change data.

[0006] Optionally, the method for preprocessing data includes: performing data cleaning and data standardization on the historical parking data and real-time external data; extracting statistical features from the data after data cleaning and data standardization; performing standardization processing on numerical features in the statistical features; and analyzing the historical parking data in chronological order to extract time series features.

[0007] Optionally, the method of using preprocessed data to construct a parking demand prediction model based on a deep learning model includes: generating an input feature matrix based on preprocessed historical parking data, statistical features and time series features; selecting a deep learning network model to construct an LSTM model; receiving the input feature matrix through the input layer of the LSTM model; performing data processing on the input feature matrix through the LSTM layer of the LSTM model, and outputting a parking demand prediction through the output layer; dividing the historical parking data into a training set, a validation set and a test set, and training and validating the LSTM model; and performing iterative optimization until the final LSTM model is output.

[0008] Optionally, after outputting the final LSTM model, the method further includes: generating a real-time input feature matrix from the preprocessed real-time external data; and inputting the real-time input feature matrix into the LSTM model to enable the LSTM model to output a parking demand prediction based on the real-time external data.

[0009] Optionally, the method of calling a pre-built berth scheduling model to output a parking space allocation strategy based on the parking demand prediction output by the parking demand prediction model includes: obtaining berth status information of the current parking area; defining optimization objectives and constraints for berth allocation; calling a pre-built berth scheduling model to generate a berth allocation strategy based on the input parking demand prediction and berth status information, wherein the berth allocation strategy includes the allocation result of each berth, the berth reservation ratio and dynamic adjustment rules for different time periods, and priority allocation rules for specific types of vehicles.

[0010] Optionally, the method of executing the parking space allocation strategy and collecting feedback data of the allocation strategy in real time to adjust the parameters of the demand prediction model and the parking space scheduling model in real time according to the feedback data includes: executing the parking space allocation strategy and completing parking; collecting behavior feedback data of vehicles in the parking spaces; calculating the difference between the predicted parking demand and the actual demand according to the behavior feedback data, and adjusting the weight parameters of the demand prediction model according to the difference to optimize subsequent prediction results;

[0011] The present invention further discloses an urban parking space scheduling system based on dynamic demand prediction, comprising:

[0012] A data processing module is used to receive historical parking data and real-time external data, and pre-process the data;

[0013] A model building module, used to build a parking demand prediction model based on a deep learning model using the preprocessed data;

[0014] A strategy output module, used to call a pre-built parking space scheduling model to output a parking space allocation strategy based on the parking demand prediction output by the parking demand prediction model;

[0015] The model optimization module is used to execute the parking space allocation strategy and collect feedback data of the allocation strategy in real time, so as to adjust the parameters of the demand prediction model and the parking space scheduling model in real time according to the feedback data.

[0016] The present invention further discloses a computer-readable storage medium, wherein the storage medium stores a computer program, and when the computer program is executed by a processor, the urban parking space scheduling method based on dynamic demand prediction is implemented.

[0017] The present invention further discloses an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements a method for scheduling urban parking spaces based on dynamic demand prediction when executing the program.

[0018] According to the technical solution of the present invention, by receiving and preprocessing historical parking data and real-time external data, a parking demand prediction model is constructed using a deep learning model, a dynamic berth allocation strategy is generated in combination with a berth scheduling model, and model parameters are optimized by real-time collection of feedback data, thereby achieving real-time dynamic adjustment of berth allocation. This improves the accuracy of parking demand prediction and berth utilization, reduces vehicle waiting time, and significantly improves the intelligent level of parking management and user experience. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] For the purpose of illustration and not limitation, the present invention is now described in conjunction with the embodiments of the present invention and the accompanying drawings, in which:

[0020] Figure 1 is a flow chart of a method for scheduling urban parking spaces based on dynamic demand prediction in an embodiment of the present invention;

[0021] Figure 2 is a schematic diagram of the structure of an urban parking space scheduling system based on dynamic demand prediction in an embodiment of the present invention;

[0022] Figure 3 It is a schematic diagram of the structure of an electronic device in an embodiment of the present invention. DETAILED DESCRIPTION

[0023] In order to enable those skilled in the art to better understand the solution of the present application, the technical solution in the implementation mode of the present application will be clearly and completely described below in conjunction with the drawings in the implementation mode of the present application. Obviously, the described implementation mode is only a part of the implementation mode of the present application, not all the implementation modes. Based on the implementation mode in the present application, all other implementation modes obtained by ordinary technicians in the field without creative work should fall within the scope of protection of the present application.

[0024] It should be noted that the terms "first", "second", etc. in the specification of the present application and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchanged where appropriate, so that the embodiments of the present application described here. In addition, the terms "including" and "having" and any of their variations are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0025] In the present application, the terms "upper", "lower", "left", "right", "front", "back", "top", "bottom", "inner", "outer", "middle", "vertical", "horizontal", "lateral", "longitudinal" and the like indicate positions or positional relationships based on the positions or positional relationships shown in the drawings. These terms are mainly used to better describe the present application and its implementation methods, and are not used to limit the indicated devices, elements or components to have a specific orientation, or to be constructed and operated in a specific orientation.

[0026] In addition, some of the above terms may be used to express other meanings in addition to indicating orientation or positional relationship. For example, the term "on" may also be used to express a certain dependency or connection relationship in some cases. For those of ordinary skill in the art, the specific meanings of these terms in this application can be understood according to the specific circumstances. In addition, the term "plurality" should mean two or more than two.

[0027] It should be noted that, in the absence of conflict, the embodiments of the present application and the features in the embodiments can be combined with each other. The embodiments of the present application will be described in detail below with reference to the accompanying drawings.

[0028] refer to Figure 1 This embodiment discloses a method for scheduling urban parking spaces based on dynamic demand prediction, including:

[0029] S100: Receive historical parking data and real-time external data, and pre-process the data. The historical parking data is the historical parking data received from the management system of the parking area, specifically including vehicle entry and exit time data, vehicle type data, parking duration data and daily average vehicle flow data of multiple parking areas. Exemplarily, when the source of the historical parking data is a parking lot, the historical parking data includes: parking time data, which is the event of vehicles entering and leaving the parking lot; parking duration data, which is the parking occupancy time of the vehicle; parking space utilization rate, which is the occupancy of multiple parking spaces in the parking lot within a certain period of time; vehicle type data, such as sedans, trucks, SUVs, special vehicles, etc. The vehicle type data corresponds to the type of parking space in the current parking lot. The collection of historical parking data can be obtained by exporting and analyzing the log of the parking lot management system, and the historical parking data can be synchronized from the central database regularly.

[0030] Real-time external data is real-time external data outside the parking area obtained in real time through multiple open API interfaces. Real-time external data includes traffic flow data and influencing factor data. Specifically, traffic flow data includes real-time traffic flow data, vehicle speed and congestion data of multiple road sections, and temporary traffic flow change data. In this embodiment, real-time external data also includes external data that can affect the driver's travel intention and choice and affect the traffic information of a specific area, such as real-time weather conditions, holiday schedules, and schedules of large-scale events (such as sports events, concerts, etc.). Specifically, real-time external data includes traffic flow data, which is road network traffic information data provided in real time by the urban transportation system; weather data, which is weather conditions obtained in real time. Different weather conditions will affect the behavior of vehicles; event data, which is information about activities or emergencies near the parking area, such as sports events, concerts, etc. The collection of real-time external data obtains urban traffic flow and weather information in real time through an open API.

[0031] After obtaining the above historical parking data and real-time external data, the above data is preprocessed. Specifically, missing values ​​or outliers in the data are deleted. For example, abnormally high or low parking time data are deleted, unrecognizable vehicle types or repeated license plate number record data are deleted, and incomplete data are interpolated or filled. The average value of the data or the median can be used to fill. Then the data format is standardized, for example, the time field is unified into a standard format, and different vehicle type data are uniformly encoded. Additional features are generated from the original data, for example, hourly segment features are constructed: time is divided into peak and non-peak periods; berth idle time features: calculate the continuous idle time of each berth; event heat index: combine event type and time to generate an impact coefficient. Further, the above numerical features are normalized and standardized, statistical features are extracted from the data after data cleaning and data standardization, and the numerical features in the statistical features are standardized. The historical parking data is analyzed in chronological order and time series features are extracted: trend analysis to identify the long-term trend of berth demand; periodic analysis to extract the periodic demand pattern of each day or period. Store the preprocessed data in a database or memory for later use.

[0032] S200: Use the preprocessed data to build a parking demand prediction model based on the deep learning model. Specifically, generate an input feature matrix based on the preprocessed historical parking data, statistical features and time series features. The historical parking data includes time-related features, i.e., the hour, day of the week, holiday tags, etc. in the historical parking data in step S100, and statistical features, i.e., the historical vehicle flow, parking space occupancy rate and vehicle type data in step S100. The time series features include the time series trend and periodicity characteristics of parking space demand.

[0033] A deep learning network model is selected to construct an LSTM model. This embodiment uses a long short-term memory network model to construct an LSTM model to capture the long-term dependency and dynamic changes in berth demand. The input feature matrix of the LSTM model is further received through the input layer of the LSTM model; the hidden layer of the LSTM model includes an LSTM layer, which is used to perform data processing on the input feature matrix and output the parking demand prediction through the output layer. The output layer of the LSTM model is used to output the number of nodes corresponding to the predicted time step, for example, to predict the parking demand in the future time step. For example, it is necessary to predict the parking demand of a parking lot in the next 15 minutes. The output layer has one node, which predicts the demand in the next 15 minutes from the current time. The predicted output value is 10, which means that 10 cars are expected to need to park in the next 15 minutes.

[0034] For the training of the LSTM model, the historical parking data described in step S100 is divided into a training set, a validation set and a test set, and the LSTM model is trained and verified. Among them, the training set is used for learning model parameters, accounting for 70%, the validation set is used for model tuning and preventing overfitting, accounting for 15%, and the test set is used to evaluate the actual performance of the model, accounting for 15%. Through iterative optimization, until the final LSTM model is output. After the final LSTM model is output, the real-time input feature matrix is ​​generated from the preprocessed real-time external data; the real-time input feature matrix is ​​input into the LSTM model to realize the LSTM model to output parking demand prediction based on real-time external data. For the verification of the LSTM model, this embodiment uses K-fold cross-validation to evaluate the generalization ability of the model, and evaluates the model performance by calculating the accuracy indicators of predicted parking demand, such as MAE (mean absolute error), MSE (mean square error), etc. Exemplarily, the entire data set is evenly divided into K mutually exclusive subsets of similar size; the model is trained with data from K-1 subsets each time, and the remaining subset is used as a test set; this process is repeated K times, each time a different subset is selected as a test set, and the remaining subset is used as a training set; finally, the average value of these K test results is calculated as the final performance indicator of the model.

[0035] S300: Call the pre-built berth scheduling model to output a parking berth allocation strategy based on the parking demand prediction output by the parking demand prediction model. Specifically, the control system based on the parking area obtains the berth status information of the current parking area, and the berth status information includes the total number of berths, the number of occupied berths, the number of remaining available berths, the berth utilization rate in different time periods and its historical change trend. Further define the optimization objectives and constraints of berth allocation. For example, the optimization objectives include maximizing the berth utilization rate and minimizing the waiting time for vehicle parking. Constraints include berth type matching, parking time and vehicle entry time prediction range, and retaining a certain proportion of emergency berths.

[0036] The pre-built parking scheduling model is called to generate a parking allocation strategy based on the input parking demand forecast and parking status information, wherein the parking allocation strategy includes the allocation result of each parking space, the parking reservation ratio and dynamic adjustment rules for different time periods, and the priority allocation rules for specific types of vehicles. Exemplarily, the parking allocation strategy includes the allocation result of each parking space including the target vehicle type and the expected parking time, the parking reservation ratio and dynamic adjustment rules for different time periods, and the priority allocation rules for special vehicles.

[0037] After obtaining the parking space allocation strategy, the generated parking space allocation strategy is sent to the parking lot management system to guide the implementation process of the actual parking space allocation, and to collect feedback data on the parking space allocation in real time, including whether the vehicle is parked in the designated parking space, whether there are conflicts or unused parking spaces, etc.

[0038] This embodiment exemplifies a specific process of berth scheduling, which is exemplified as follows:

[0039] Input the parking demand forecast result: the demand forecast for the next hour shows that 40 ordinary parking spaces, 10 large vehicle parking spaces, and 5 charging vehicle parking spaces are needed;

[0040] Current parking space status information: 50 ordinary parking spaces, 10 large vehicle parking spaces, and 6 charging parking spaces are available.

[0041] Optimization goal: Ensure that vehicles of each parking type can be allocated parking spaces as needed and minimize the waiting time of vehicles at the entrance.

[0042] Calling algorithm: Output strategy after the berth scheduling model runs:

[0043] 40 ordinary berths are allocated to the predicted normal traffic flow, and 10 ordinary berths are reserved to cope with unexpected demand;

[0044] 10 parking spaces for large vehicles are allocated according to demand and are not reserved;

[0045] All five charging berths are allocated, with no reservations;

[0046] Execute strategies and collect feedback: Allocate parking spaces through the parking management system and collect actual parking conditions in real time.

[0047] S400: Execute the parking space allocation strategy and collect feedback data of the allocation strategy in real time, so as to adjust the parameters of the demand prediction model and the parking space scheduling model in real time according to the feedback data. Specifically, it includes executing the parking space allocation strategy and completing parking;

[0048] Collect behavioral feedback data of vehicles in parking spaces in real time. Specifically, record the entry and exit time of each vehicle and the parking space used in real time; monitor the occupancy status of each parking space in the parking lot and record the number of free and occupied parking spaces; record the length of time vehicles are parked in the parking lot to evaluate the efficiency of the parking space allocation strategy; record the time vehicles wait for parking, including the time from entering the parking lot to obtaining a parking space.

[0049] The difference between the predicted parking demand and the actual demand is calculated according to the behavior feedback data, and the weight parameters of the demand prediction model are adjusted according to the difference to optimize the subsequent prediction results. Specifically, the adjustment of the demand prediction model includes: calculating the difference between the predicted parking demand and the actual demand through feedback data, and then adjusting the weight parameters of the model (such as the learning rate and time series feature weight in the LSTM model) to make the prediction results more accurate; adjusting the model in the short term based on real-time feedback (such as parking space occupancy rate and waiting time) to better adapt to the changes in instantaneous demand; gradually optimizing the demand prediction model through accumulated historical feedback data to improve the accuracy of long-term prediction. Further, the present embodiment also discloses that the adjustment of the parking scheduling model includes: fine-tuning the parking scheduling strategy according to the actual feedback data (such as the parking space idle state and the vehicle waiting time). For example, the priority of the allocation strategy may be adjusted according to the vacant spaces or congestion that often occur in certain areas, or the proportion of parking space allocation in areas with high traffic volume may be adjusted; the effect after the execution of the strategy is analyzed in real time, and the key parameters of the scheduling model (such as parking priority, vehicle guidance rules, time window, etc.) are adjusted to optimize the parking allocation strategy.

[0050] The demand forecasting and berth scheduling models can gradually improve their accuracy through the continuous collection of feedback data. For example, through multiple training and optimization, the model can better adapt to various influencing factors such as different seasons, holidays, special events, etc., so as to make better predictions and scheduling in future practical applications.

[0051] In summary, this implementation method receives and preprocesses historical parking data and real-time external data, builds a parking demand prediction model based on deep learning, combines the parking scheduling model to generate a dynamic parking allocation strategy, and performs real-time execution and feedback optimization to achieve closed-loop adjustment. By dynamically matching parking demand and parking resources, the parking utilization rate is significantly improved, the vehicle waiting time is shortened, and the real-time response capability of the system is enhanced. The time series prediction model based on deep learning significantly improves the prediction accuracy. Combined with the real-time integration of external dynamic data, the parking allocation strategy is optimized, making the system more intelligent and efficient, effectively alleviating the problem of urban parking difficulties, and providing innovative solutions for smart transportation and smart city construction.

[0052] refer to Figure 2This embodiment further discloses a city parking space scheduling system based on dynamic demand prediction, including:

[0053] The data processing module 21 is used to receive historical parking data and real-time external data and pre-process the data; it is also used to receive historical parking data sent by the management system of the parking area, the historical parking data includes vehicle entry and exit time data, vehicle type data, parking duration data and daily average vehicle flow data of multiple parking areas; it obtains real-time external data outside the parking area in real time through multiple open API interfaces, the real-time external data includes traffic flow data and influencing factor data, the traffic flow data includes real-time vehicle flow data, vehicle speed and congestion data of multiple road sections, and temporary traffic flow change data; it is also used to clean and standardize the historical parking data and real-time external data; extract statistical features from the data after data cleaning and data standardization; perform standardization processing on the numerical features in the statistical features; analyze the historical parking data in chronological order to extract time series features.

[0054] The model building module 22 is used to use the preprocessed data to build a parking demand prediction model based on the deep learning model; it is also used to generate an input feature matrix based on the preprocessed historical parking data, statistical features and time series features; select a deep learning network model to build an LSTM model; receive the input feature matrix through the input layer of the LSTM model; perform data processing on the input feature matrix through the LSTM layer of the LSTM model, and output the parking demand prediction through the output layer; divide the historical parking data into a training set, a validation set and a test set, and train and verify the LSTM model; perform iterative optimization until the final LSTM model is output; it is also used to generate a real-time input feature matrix from the preprocessed real-time external data; input the real-time input feature matrix into the LSTM model to realize that the LSTM model outputs a parking demand prediction based on the real-time external data.

[0055] The strategy output module 23 is used to call the pre-built berth scheduling model to output a parking space allocation strategy based on the parking demand prediction output by the parking demand prediction model; it is also used to obtain the berth status information of the current parking area; define the optimization objectives and constraints of berth allocation; call the pre-built berth scheduling model to generate a berth allocation strategy based on the input parking demand prediction and berth status information, and the berth allocation strategy includes the allocation result of each berth, the berth reservation ratio and dynamic adjustment rules for different time periods, and the priority allocation rules for specific types of vehicles.

[0056] The model optimization module 24 is used to execute the parking space allocation strategy and collect feedback data of the allocation strategy in real time, so as to adjust the parameters of the demand prediction model and the parking space scheduling model in real time according to the feedback data; it is also used to execute the parking space allocation strategy and complete parking; collect behavior feedback data of vehicles in parking spaces in real time; calculate the difference between the predicted parking demand and the actual demand according to the behavior feedback data, and adjust the weight parameters of the demand prediction model according to the difference, so as to optimize subsequent prediction results.

[0057] Figure 3 A schematic diagram of the physical structure of an electronic device provided in an embodiment of the present invention, such as Figure 3 As shown, the electronic device 50 includes: a processor 501 (processor), a memory 502 (memory) and a bus 503;

[0058] The processor 501 and the memory 502 communicate with each other via the bus 503 ; the processor 501 is used to call program instructions in the memory 502 to execute the methods provided by the above-mentioned method implementation methods.

[0059] This embodiment provides a non-transitory computer-readable storage medium, which stores computer instructions. The computer instructions enable a computer to execute the methods provided by the above-mentioned method embodiments.

[0060] A person skilled in the art can understand that all or part of the steps for implementing the above-mentioned method implementation method can be completed by hardware related to program instructions, and the aforementioned program can be stored in a computer-readable storage medium, which, when executed, executes the steps of the above-mentioned method implementation method; and the aforementioned storage medium includes: ROM, RAM, magnetic disk or optical disk, etc., various storage media that can store program codes.

[0061] The device implementation described above is merely illustrative, wherein the units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, i.e., they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the present implementation scheme. Those of ordinary skill in the art may understand and implement it without creative effort.

[0062] Through the description of the above implementation methods, those skilled in the art can clearly understand that each implementation method can be implemented by means of software plus a necessary general hardware platform, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solution is essentially or the part that contributes to the prior art can be embodied in the form of a software product, and the computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, a disk, an optical disk, etc., including a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods of each implementation method or some parts of the implementation method.

[0063] The above specific implementations do not constitute a limitation on the protection scope of the present invention. It should be understood by those skilled in the art that various modifications, combinations, sub-combinations and substitutions may occur depending on design requirements and other factors. Any modification, equivalent substitution and improvement made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A method for scheduling urban parking spaces based on dynamic demand prediction, characterized in that: The method comprises: Receive historical parking data and real-time external data, and pre-process the data; Use the preprocessed data to build a parking demand prediction model based on a deep learning model; Calling a pre-built parking space scheduling model to output a parking space allocation strategy based on the parking demand prediction output by the parking demand prediction model; The parking space allocation strategy is executed and feedback data of the allocation strategy is collected in real time, so as to adjust the parameters of the demand prediction model and the parking space scheduling model in real time according to the feedback data.

2. The method for scheduling urban parking spaces based on dynamic demand prediction according to claim 1, characterized in that: The method for receiving historical parking data and real-time external data includes: Receiving historical parking data sent by a management system of a parking area, the historical parking data including vehicle entry and exit time data, vehicle type data, parking duration data, and daily average vehicle flow data of a plurality of parking areas; Real-time external data outside the parking area is acquired in real time through multiple open API interfaces, wherein the real-time external data includes traffic flow data and influencing factor data, wherein the traffic flow data includes real-time vehicle flow data, vehicle speed and congestion data of multiple road sections, and temporary traffic flow change data.

3. The method for scheduling urban parking spaces based on dynamic demand prediction according to claim 1, characterized in that: The method for preprocessing data comprises: Performing data cleaning and data standardization on the historical parking data and real-time external data; Extract statistical features from the data after data cleaning and data standardization; Performing standardization processing on the numerical features in the statistical features; The historical parking data are analyzed in chronological order to extract time series features.

4. The method for scheduling urban parking spaces based on dynamic demand prediction according to claim 3 is characterized in that: The method of using the preprocessed data to construct a parking demand prediction model based on a deep learning model includes: Generate an input feature matrix based on the preprocessed historical parking data, statistical features, and time series features; Use deep learning network model to build LSTM model; Receiving the input feature matrix through the input layer of the LSTM model; The LSTM layer of the LSTM model processes the input feature matrix and outputs the parking demand prediction through the output layer; Dividing the historical parking data into a training set, a validation set, and a test set, and training and validating the LSTM model; Through iterative optimization, the final LSTM model is output.

5. The method for scheduling urban parking spaces based on dynamic demand prediction according to claim 4 is characterized in that: After outputting the final LSTM model, the method further includes: Generate a real-time input feature matrix from preprocessed real-time external data; The real-time input feature matrix is ​​input into the LSTM model to enable the LSTM model to output parking demand prediction based on real-time external data.

6. The method for scheduling urban parking spaces based on dynamic demand prediction according to claim 1, characterized in that: The method of calling a pre-built parking space scheduling model to output a parking space allocation strategy based on the parking demand prediction output by the parking demand prediction model includes: Get the parking space status information of the current parking area; Define the optimization objectives and constraints for berth allocation; The pre-built berth scheduling model is called to generate a berth allocation strategy based on the input parking demand forecast and berth status information. The berth allocation strategy includes the allocation result of each berth, the berth reservation ratio and dynamic adjustment rules for different time periods, and the priority allocation rules for specific types of vehicles.

7. The method for scheduling urban parking spaces based on dynamic demand prediction according to claim 6, characterized in that: The method of executing the parking space allocation strategy and collecting feedback data of the allocation strategy in real time to adjust the parameters of the demand prediction model and the parking space scheduling model in real time according to the feedback data includes: Execute parking allocation strategies and complete parking; Collect behavioral feedback data of vehicles in the parking spaces in real time; The difference between the predicted parking demand and the actual demand is calculated based on the behavior feedback data, and the weight parameters of the demand prediction model are adjusted based on the difference to optimize subsequent prediction results.

8. An urban parking space scheduling system based on dynamic demand prediction, characterized in that: include: A data processing module is used to receive historical parking data and real-time external data, and pre-process the data; A model building module, used to build a parking demand prediction model based on a deep learning model using the preprocessed data; A strategy output module, used to call a pre-built parking space scheduling model to output a parking space allocation strategy based on the parking demand prediction output by the parking demand prediction model; The model optimization module is used to execute the parking space allocation strategy and collect feedback data of the allocation strategy in real time, so as to adjust the parameters of the demand prediction model and the parking space scheduling model in real time according to the feedback data.

9. A computer-readable storage medium, characterized in that: The storage medium stores a computer program, and when the computer program is executed by a processor, the method according to any one of claims 1 to 7 is implemented.

10. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the program, the method described in any one of claims 1 to 7 is implemented.

Images