Intelligent parking management method and system based on big data

Abstract

The invention relates to the technical field of intelligent parking, and discloses an intelligent parking management method and system based on big data, and the method comprises the following steps: deploying an intelligent parking space lock at each parking space of a parking lot, and capturing vehicle information according to a high-definition camera preset at a key point; based on a preset reservation rule, generating a dynamic pricing strategy by using the intelligent demand prediction and dynamic pricing model; the optimal parking space is recommended for the user by using the personalized parking space sorting model, and the parking space occupation state is updated in real time; when a vehicle arrives at the entrance, license plate information is captured, reservation records are inquired, after identity authentication and vehicle classification are completed, intelligent parking space locks of corresponding parking spaces are controlled to be unlocked, and corresponding parking space distribution schemes are executed; and when the vehicle leaves, the departure time is automatically recorded, and the parking fee is calculated according to a dynamic pricing strategy. The user experience is improved, the efficient and convenient parking process is ensured, and the operation efficiency and benefits of the parking lot are ensured.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of intelligent parking, in particular to an intelligent parking management method and system based on big data. BACKGROUND

[0002] With the acceleration of urbanization and the continuous increase in the number of motor vehicles, urban parking demand has risen sharply, leading to the increasingly prominent problem of "parking difficulty"; traditional parking lots generally have problems such as opaque parking space information, lack of unified scheduling, and users' inability to quickly find idle parking spaces, which not only reduces parking efficiency but also exacerbates urban traffic pressure; especially during peak hours, vehicles frequently circle to find parking spaces, leading to road congestion and increased carbon emissions, further deteriorating the urban travel environment.

[0003] The existing parking management may have the following shortcomings: 1. Parking management often relies on simple rules or historical data for parking space allocation and pricing, which cannot accurately predict parking space demand in real time, leading to uneven resource allocation, causing some parking spaces to be overcrowded while others are idle and wasted; 2. Parking management is usually based on fixed rules for parking space recommendation, ignoring users' individual needs, and in the face of dynamically changing parking demand, parking allocation may be inefficient, leading to poor user experience. SUMMARY

[0004] In order to solve the problem that parking management is usually based on fixed rules for parking space recommendation, ignoring users' individual needs, and in the face of dynamically changing parking demand, parking allocation may be inefficient, leading to poor user experience, the present application provides an intelligent parking management method and system based on big data.

[0005] In a first aspect, the present application provides an intelligent parking management method based on big data, which comprises the following steps: S1. Deploying an intelligent parking lock at each parking space in the parking lot, and using a pre-installed sensor to sense the parking space occupancy state, capturing vehicle information through a pre-installed high-definition camera at key points, and then transmitting it to the cloud platform in real time through Internet of Things technology; S2. Using the cloud platform to process the parking space occupancy state in real time, and displaying the current state of all parking spaces through the management background and App end, generating a dynamic pricing strategy using an intelligent demand prediction and dynamic pricing model based on pre-installed reservation rules; S3. The user inputs reservation information through the App end, and according to the real-time parking space state, dynamic pricing strategy and user historical preferences, uses a personalized parking sorting model to recommend the optimal parking space for the user, and when the user confirms, the parking space is locked and a reservation record is generated, and the parking space occupancy state is updated in real time; S4, when the vehicle arrives at the entrance, capture the license plate information, query the reservation record, complete the identity authentication and vehicle classification, control the smart parking lock of the corresponding parking space to open, and execute the corresponding parking space allocation scheme, provide real-time travel route to the user through indoor navigation technology; S5, when the vehicle leaves, automatically record the departure time, calculate the parking fee according to the dynamic pricing strategy, verify the reservation performance, and complete the fee settlement and payment.

[0006] Optionally, the cloud platform is used to process the parking space occupation state in real time, and the current state of all parking spaces is displayed through the management background and the App end. Based on the preset reservation rules, a dynamic pricing strategy is generated by using intelligent demand prediction and dynamic pricing model, including the following steps: S21, the cloud platform continuously collects data from the parking space sensor and the camera, and associates with the historical database, time, calendar and external data sources of weather to obtain multi-dimensional data, and then displays the current state of all parking spaces through the management background and the App end; S22, based on the preset reservation rules, the multi-dimensional data is feature extracted to generate a feature vector for predicting parking space demand, and the XGBoost model and the logistic regression model are trained respectively to predict the parking space demand; S23, using the Stacking meta-learner, the prediction results of the XGBoost model and the logistic regression model are integrated to obtain the final parking space demand prediction; S24, according to the final parking space demand prediction, the time period, region and user behavior are comprehensively considered to generate a real-time adjusted dynamic pricing strategy.

[0007] Optionally, based on the preset reservation rules, the multi-dimensional data is feature extracted to generate a feature vector for predicting parking space demand, and the XGBoost model and the logistic regression model are trained respectively to predict the parking space demand, including the following steps: S221, according to the preset reservation rules, the feature vector related to the parking space demand is extracted from the multi-dimensional data; S222, input the extracted feature vector into the trained XGBoost model, and output the prediction value of the parking space demand in the future preset time period; S223, input the extracted feature vector into the trained logistic regression model, calculate the influence of each feature on the parking space demand probability, and output the prediction probability of the parking space demand in the future time period.

[0008] Optionally, the expression for inputting the extracted feature vector into the trained XGBoost model to output the prediction value of the parking space demand in the future preset time period is:

[0009] In the formula, a demand value representing a parking space; an initial prediction value of the XGBoost model; a number of decision trees; a decision tree; a prediction value of the a prediction value of the decision tree for a feature vector; a prediction value of the decision tree for a feature vector; a learning rate.

[0010] Optionally, the user inputs reservation information through the App, and according to the real-time parking space state, dynamic pricing strategy and user historical preference, the optimal parking space is recommended for the user by using the personalized parking space ranking model, when the user confirms, the parking space is locked and a reservation record is generated, and the real-time parking space occupancy state is updated, including the following steps: S31, the user submits a reservation request through the App, and inputs vehicle information, estimated entry time, estimated stay duration and destination preference on the App; S32, the basic attribute information, real-time occupancy state and dynamic pricing strategy of each parking space in the parking lot are obtained; S33, according to the vehicle information, estimated entry time, estimated stay duration, destination preference, and the basic attribute information, real-time occupancy state and dynamic pricing strategy of each parking space, the occupied parking spaces and the parking spaces with unmatched sizes are deleted; for each candidate parking space, the business scenario features including location features, price features, user preference matching degree and real-time availability are constructed; S34, using the personalized parking space ranking model, a recommendation score is generated for each parking space according to the business scenario features, and the parking space list with the highest priority is selected and recommended to the user; S35, the user selects the most suitable parking space from the parking space list, and locks the parking space through the distributed lock mechanism, and generates a reservation record; S36, when the user confirms the reservation, the state of the parking space is updated from idle to reserved, and is synchronously updated to the parking space state on the App.

[0011] Optionally, using the personalized parking space ranking model, a recommendation score is generated for each parking space according to the business scenario features, and the parking space list with the highest priority is selected and recommended to the user, including the following steps: S341, the feature vector of each candidate parking space in the parking lot is taken as the input of the personalized parking space ranking model; S342, a recommendation score is generated for each candidate parking space through the personalized parking space ranking model; S343, by calculating the score difference of each pair of parking spaces, and combining the relative importance difference to calculate the ranking loss, the recommendation score of each candidate parking space is optimized; S344, ranking the parking spaces in descending order according to the optimized recommendation scores, selecting the top one with the highest score as the recommendation list, and presenting it to the user, while displaying the recommendation reasons for each parking space in the parking lot. K

[0012] Optionally, the expression of the ranking loss is calculated by calculating the score difference of each pair of parking spaces and combining the relative importance difference of the parking spaces as follows:

[0013] In the formula, denotes the ranking loss of the parking space and the parking space ; denotes the recommendation score of the parking space ; denotes the recommendation score of the parking space ; denotes the importance difference of the parking space and the parking space ; denotes the exponential function of the score difference of the parking space and the parking space .

[0014] Optionally, when the vehicle leaves, the leaving time is automatically recorded, the parking fee is calculated according to the dynamic pricing strategy, the reservation performance is checked, and the fee settlement and payment are completed, including the following steps: S51, when the vehicle leaves, the leaving vehicle is identified by license plate recognition, the leaving time of the vehicle is recorded, and the parking fee is automatically calculated based on the classification result of the vehicle and in combination with the parking duration and the dynamic pricing strategy; wherein the classification result of the vehicle includes a reservation vehicle, a monthly rental vehicle and a temporary vehicle; S52, for the reservation vehicle, the fee is calculated according to the charging rules of the App, and the performance is checked, whether the vehicle enters and leaves on time is checked according to the standard of the entry and leaving time and the reservation period, and the overtime or default behavior is handled; S53, the monthly rental vehicle and the temporary vehicle are settled according to the respective charging rules, wherein the monthly rental vehicle is settled according to the monthly fixed fee, and the temporary vehicle is calculated according to the parking duration and the dynamic pricing strategy; S54, the non-contact payment of all vehicles is completed through automatic deduction or code scanning payment.

[0015] In the second aspect, the present application also provides a smart parking management system based on big data, which is realized based on the smart parking management method based on big data.

[0016] In summary, the present application has at least one of the following beneficial technical effects: ​1、The application realizes real-time monitoring of parking space status through intelligent parking lock, sensor and camera, optimizes parking resource allocation by combining big data analysis and dynamic pricing strategy; at the same time, it improves user experience by using personalized parking recommendation and non-contact payment technology, ensures efficient and convenient parking process, and synchronizes data in real time through cloud platform to improve parking lot operation efficiency and revenue.

[0017] 2、The application integrates multi-dimensional data through cloud platform, and combines XGBoost and logistic regression model to predict parking demand, integrates prediction results by using Stacking meta-learner, generates real-time adjusted dynamic pricing strategy, accurately predicts parking demand, optimizes resource allocation, and adjusts pricing in real time according to time period, region and user behavior, effectively improves operation efficiency and user experience of parking lot.

[0018] 3、The application uses personalized parking sorting model and real-time data processing to recommend the best parking space for users, improving parking experience and resource utilization efficiency; according to real-time parking status, dynamic pricing strategy and user historical preferences, the parking space is accurately screened, and the distributed lock mechanism is used to ensure parking lock and reservation; when leaving, the parking time is automatically recorded and the fee is settled according to the dynamic pricing strategy, supporting non-contact payment, simplifying the payment process and ensuring efficient and convenient parking management. BRIEF DESCRIPTION OF DRAWINGS

[0019] Figure 1 is a flowchart of the method in the application. DETAILED DESCRIPTION

[0020] The embodiments of the application will be described in detail below, and examples of the embodiments are shown in the accompanying drawings.

[0021] In the description of the present specification, the description of the terms "some embodiments", "one embodiment", "some embodiments", "illustrative embodiments", "example", "specific example", or "some examples" means that the specific features, structures, materials or characteristics described in conjunction with the described embodiments or examples are included in at least one embodiment or example of the present application. In the present specification, the illustrative description of the above terms does not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described can be combined in any one or more embodiments or examples in a suitable manner.

[0022] The first embodiment of the present application discloses a smart parking management method based on big data, referring to Figure 1 The smart parking management method comprises the following steps: S1, deploy intelligent parking lock at each parking space in the parking lot, and use pre-installed sensors to sense the parking occupancy state, capture vehicle information through pre-installed high-definition cameras at key points, and then transmit the information to the cloud platform in real time through Internet of Things technology.

[0023] It should be explained that the intelligent parking lock is a parking management device controlled by electronic equipment, which is usually composed of motor, sensor and communication module. The parking lock can be remotely controlled to realize automatic locking and unlocking of the parking space. The parking lock will automatically lock the parking space to prevent other vehicles from occupying it. Only after authorization (such as through App or cloud platform control), the vehicle can be unlocked and driven out. Common sensor types include geomagnetic sensor, infrared sensor and ultrasonic sensor. The geomagnetic sensor detects whether the parking space is occupied by the change of geomagnetic field. The infrared sensor or ultrasonic sensor can determine whether the parking space is occupied by vehicle through ranging technology. The high-definition camera installed at the key position (such as the front end of the parking space and the entrance of the parking lot) is used to monitor the situation in the parking lot in real time, which can capture the appearance and license plate information of the vehicle with high quality. Internet of Things connects different devices and sensors through network, so that devices can exchange information in real time. In the management of parking lot, Internet of Things technology is used to transmit the data collected by various sensors (parking lock, occupancy sensor, camera, etc.) to the cloud platform in real time.

[0024] S2, use the cloud platform to process the parking occupancy state in real time, and display the current state of all parking spaces through the management background and App, generate dynamic pricing strategy based on the pre-installed reservation rules, and use intelligent demand prediction and dynamic pricing model.

[0025] Preferably, the cloud platform processes the parking occupancy state in real time, and displays the current state of all parking spaces through the management background and App, generates dynamic pricing strategy based on the pre-installed reservation rules, and uses intelligent demand prediction and dynamic pricing model, which includes the following steps: S21, the cloud platform continuously collects data from parking sensors and cameras, and associates with historical database, time, calendar and external data sources such as weather, to obtain multi-dimensional data, and then displays the current state of all parking spaces through the management background and App.

[0026] It should be explained that the role of historical database is to provide past parking usage patterns to help identify periodic changes or trends (such as parking demand fluctuations during peak hours or holidays); combined with external factors such as time, calendar and weather, the cloud platform can further optimize the prediction of parking demand; for example, according to the prediction of parking demand peak on sunny days or holidays, the allocation of parking spaces can be adjusted in advance or the reservation function can be provided.

[0027] The cloud platform displays the processed multi-dimensional data in real time through the management backend, facilitating data monitoring and management decisions for operations personnel. Through the app, users can also quickly view the real-time status of each parking space in the parking lot, whether it is vacant, occupied, or out of service, helping users save time in finding parking spaces and improve the utilization efficiency of parking lot resources.

[0028] S22. Based on the preset reservation rules, feature extraction is performed on multi-dimensional data to generate feature vectors for predicting parking space demand, and parking space demand is predicted using the trained XGBoost model and logistic regression model respectively.

[0029] Preferably, based on preset reservation rules, feature extraction is performed on multi-dimensional data to generate a feature vector predicting parking space demand, and parking space demand is predicted using a trained XGBoost model and a logistic regression model, respectively, including the following steps: S221. Based on the preset reservation rules, extract feature vectors related to parking space demand from multi-dimensional data; S222. Input the extracted feature vector into the trained XGBoost model and output the predicted value of parking space demand in the future preset time period. S223. Input the extracted feature vectors into the trained logistic regression model, calculate the impact of each feature on the probability of parking space demand, and output the predicted probability of parking space demand in future periods.

[0030] Preferably, the extracted feature vector is input into the trained XGBoost model, and the expression for outputting the predicted value of parking space demand within a preset future time period is:

[0031] In the formula, This indicates the required value for parking spaces; This represents the initial prediction value of the XGBoost model; Indicates the number of decision trees; Representing a decision tree; Indicates the first Decision trees for feature vectors The predicted value; This represents the learning rate.

[0032] It should be explained that the extracted feature vectors are input into the trained logistic regression model to calculate the impact of each feature on the probability of parking space demand. The expression for outputting the predicted probability of parking space demand in future time periods is as follows:

[0033] In the formula, Indicates that given a feature vector the probability that the demand for parking exceeds a pre-set threshold under the given conditions; denotes the base of the natural logarithm; denotes the bias of the logistic regression model; , , denotes the contribution weight of each feature to the probability of demand for parking; , , denotes the feature vector (e.g. parking occupancy, time, weather, etc.).

[0034] S23, using the Stacking meta-learner, integrate the prediction results of the XGBoost model and the logistic regression model to obtain the final prediction of parking demand.

[0035] S24, according to the final prediction of parking demand, comprehensively consider the time period, region and user behavior, and generate a real-time adjusted dynamic pricing strategy.

[0036] It needs to be explained that Stacking (Stacking Generalization) is an ensemble learning method that improves the accuracy of prediction by integrating the results of multiple prediction models; unlike other ensemble methods (such as Bagging or Boosting), the characteristic of Stacking is to take the prediction results of different models (base learners) as input, and pass them to a new model (usually called meta-learner), which is responsible for making the final prediction; In this scenario, XGBoost and logistic regression are used as base learners, and their prediction results are taken as input to a meta-learner for integration to obtain a more accurate prediction of parking demand; The explanations of the XGBoost model and the logistic regression model are as follows: 1. The XGBoost model is a decision tree-based model that can capture complex non-linear relationships between features, making it particularly suitable for handling structured data. Its output is a quantitative prediction of parking demand; 2. The logistic regression model is a linear model that outputs the probability of demand for parking, making it suitable for explaining the influence of each feature on demand.

[0037] Through Stacking, the prediction results of these two models are taken as new input and given to a meta-learner (e.g. a linear regression or other machine learning model) to optimize the combination of the outputs of these two models, ultimately obtaining a more accurate prediction of parking demand.

[0038] The task of the meta-learner is to learn how to extract useful information from the outputs of the base learners (XGBoost and logistic regression) and synthesize them; for example, if XGBoost's predictions are more accurate than logistic regression's predictions at certain times, the meta-learner will rely more on XGBoost's outputs.

[0039] The prediction results integrated by the Stacking model will provide the demand situation for parking spaces in a certain period or a certain area in the future; such predictions are usually based on time periods (e.g., morning peak, evening peak), areas (e.g., parking spaces near business centers or residential areas), and user behavior patterns (e.g., reservation frequency, parking habits, etc.); Dynamic pricing is a pricing strategy that adjusts prices based on real-time supply and demand relationships, in this context, parking fees are adjusted in real-time based on the prediction of parking demand; for example, when it is predicted that parking demand will exceed supply in a certain period, parking fees may automatically increase; conversely, if demand is lower than expected, prices can be appropriately reduced.

[0040] Integrating time periods, areas, and user behavior: 1. Time period: Parking demand varies by time period; for example, parking demand increases significantly during the morning and evening peak periods on weekdays, and prices may need to be increased; during the night or off-peak periods, parking demand is lower, and prices may be reduced; 2. Area: The difference in parking demand in different areas also affects pricing; for example, parking demand in the city center is usually higher, and prices should be appropriately increased; while in suburban or remote areas, parking demand is lower, and pricing can be relatively lower; 3. User behavior: The parking management end will adjust prices based on users' historical parking habits or behavior; for example, frequent users can enjoy discounts, or pricing can be dynamically adjusted through certain membership systems to attract users to park; Based on the prediction of parking demand, the pricing strategy will be adjusted in real-time to optimize the revenue of the parking lot, while also considering the parking experience of users.

[0041] Specific examples are as follows: Background scenario: underground parking lot of an office building The underground parking lot has 200 parking spaces and is located in the business center, with tight parking during the morning and evening peak periods on weekdays; Each parking space is equipped with an intelligent parking lock, sensors (magnetic, ultrasonic), high-definition cameras, and Internet of Things devices, and is connected to a cloud platform; Users can view parking space information and make reservations through the App; The equipment deployment is as follows: Each parking space is equipped with an intelligent parking lock with a communication module and a motor that can be remotely controlled to lift and lower; sensors detect vehicle parking (e.g., a geomagnetic sensor senses changes in the geomagnetic field to determine whether it is occupied); a camera captures the license plate number, such as identifying the license plate "Shanghai AX XXXX", and all data is uploaded to the cloud platform in real time through the Internet of Things gateway; The data aggregation and display are as follows: 1. Historical database display: the average parking space utilization rate of this parking lot is 92% from 8:00 to 10:00 on weekdays and 45% on weekends; 2. Weather data: today's forecast is thunder showers, and the number of people driving is expected to increase; 3. Time: it is currently Wednesday 7:45 and the peak is approaching; 4. After processing, the data is uploaded to the cloud platform background and App, showing that there are only 30 idle parking spaces.

[0042] The characteristic variables are as follows: Table 1. Characteristic variable table

[0043] According to the expression for outputting the predicted value of parking space demand in the future preset period, the predicted value is calculated as = 0.18; substituting the expression for the predicted probability of parking space demand in the future period, the probability p = 0.9961; Meta-learner input: XGBoost output 0.18, logistic regression output 0.9961; The meta-learner is a linear regression model, and the weights are as follows: XGBoost weight w 1 = 0.3; Logistic regression weight w 2 = 0.7; Bias term w 0 = 0.05; Using the Stacking meta-learner, the prediction results of the XGBoost model and the logistic regression model are integrated, and the final parking space demand prediction is 0.8013, which belongs to the high demand warning state.

[0044] According to the prediction result (high demand) + other dimensions, a dynamic pricing strategy is generated; Table 2. Dynamic pricing strategy

[0045] In the App, it prompts "peak period, parking is tight, the price has been dynamically adjusted.

[0046] S3, the user inputs reservation information through the App end, and according to the real-time parking space state, dynamic pricing strategy and user historical preference, uses the personalized parking space sorting model to recommend the optimal parking space for the user, when the user confirms, the parking space is locked and a reservation record is generated, and the parking space occupancy state is updated in real time.

[0047] Preferably, the user inputs reservation information through the App end, and according to the real-time parking space state, dynamic pricing strategy and user historical preference, uses the personalized parking space sorting model to recommend the optimal parking space for the user, when the user confirms, the parking space is locked and a reservation record is generated, and the parking space occupancy state is updated in real time, including the following steps: S31, the user submits a reservation request through the App end, and inputs vehicle type information, estimated entry time, estimated stay duration and destination preference on the App end.

[0048] S32, the basic attribute information, real-time occupancy state and dynamic pricing strategy of each parking space in the parking lot are obtained.

[0049] S33, according to the vehicle type information, estimated entry time, estimated stay duration, destination preference, and the basic attribute information, real-time occupancy state and dynamic pricing strategy of each parking space, delete the occupied parking spaces and the parking spaces with unmatched sizes; for each candidate parking space, construct a business scenario feature including location feature, price feature, user preference matching degree and real-time availability.

[0050] S34, using the personalized parking space sorting model, generate a recommendation score for each parking space according to the business scenario feature, and select the highest priority parking space list to recommend to the user.

[0051] Preferably, using the personalized parking space sorting model, generating a recommendation score for each parking space according to the business scenario feature, and selecting the highest priority parking space list to recommend to the user includes the following steps: S341, taking the feature vector of each candidate parking space in the parking lot as the input of the personalized parking space sorting model; S342, generating a recommendation score for each candidate vehicle through the personalized parking space sorting model; S343, calculating the score difference of each pair of parking spaces, and calculating the sorting loss by combining their relative importance difference to optimize the recommendation score of each candidate vehicle; S344, according to the optimized recommendation score, descendingly sorting the parking spaces, selecting the top K score parking spaces as the recommendation list, and presenting them to the user, while displaying the recommendation reasons for each parking space in the parking lot.

[0052] Preferably, the expression for calculating the sorting loss by combining the score difference of each pair of parking spaces and their relative importance difference is:

[0053] In the formula, indicates the ranking loss of the parking space and the parking space . indicates the recommendation score of the parking space . indicates the recommendation score of the parking space . indicates the importance difference of the parking space and the parking space . indicates the exponential function of the score difference of the parking space and the parking space .

[0054] S35, the user selects the most suitable parking space from the parking space list, and locks the parking space through the distributed lock mechanism, and generates a reservation record; S36, when the user confirms the reservation, the state of the parking space is updated from idle to reserved, and the parking space state is synchronously updated to the App end.

[0055] It needs to be explained that the feature data of the candidate vehicle (such as parking space distance, price, owner score, etc.) is collected and converted into a feature vector; then, the personalized parking space ranking model (based on gradient boosting tree learning algorithm) is used for training, and the feature of each vehicle and the preference relationship of the user are modeled during the training process, and the ranking accuracy is optimized; the trained model is constructed through multiple rounds of trees, and the recommendation score of each candidate vehicle is gradually improved, and finally a ranking result is output, the vehicle with higher score ranks higher, for the user to refer or recommend.

[0056] S4, when the vehicle arrives at the entrance, the license plate information is captured, the reservation record is queried, the identity authentication and vehicle classification are completed, the intelligent parking lock of the corresponding parking space is opened, and the corresponding parking allocation scheme is executed, and the indoor navigation technology is used to provide the user with a real-time travel route.

[0057] It needs to be explained that the indoor navigation technology is a solution that provides precise positioning and path guidance in indoor environments based on wireless communication, sensors, computer vision and other technologies; by using technologies such as Bluetooth, Wi-Fi, ultra-wideband (UWB), magnetic field, infrared, indoor navigation can determine the position of the user or object inside the building in real time, and provide navigation path, similar to outdoor GPS navigation system; widely used in complex environments such as shopping centers, airports, museums, etc., helping users to efficiently find the destination, and also can be used in asset tracking, security monitoring and other scenes.

[0058] S5, when the vehicle leaves, the leaving time is automatically recorded, the parking fee is calculated according to the dynamic pricing strategy, the reservation performance is checked, and the fee settlement and payment are completed.

[0059] Preferably, when the vehicle leaves, the leaving time is automatically recorded, the parking fee is calculated according to the dynamic pricing strategy, the reservation performance is checked, and the fee settlement and payment are completed, including the following steps: S51, when the vehicle leaves, the leaving vehicle is identified by license plate recognition, the leaving time of the vehicle is recorded, and the parking fee is automatically calculated based on the classification result of the vehicle and combined with the parking duration and the dynamic pricing strategy; wherein the classification result of the vehicle includes a reservation vehicle, a monthly rental vehicle and a temporary vehicle; S52, the fee of the reservation vehicle is calculated according to the charging rules of the App end, and the performance is checked, whether the vehicle enters and leaves on time is checked according to the standard of the entering and leaving time and the reservation period, and the overtime or default behavior is handled; S53, the monthly rental vehicle and the temporary vehicle are settled according to their respective charging rules, wherein the monthly rental vehicle is settled according to the monthly fixed fee, and the temporary vehicle is calculated according to the parking duration and the dynamic pricing strategy; S54, the non-contact payment of all vehicles is completed through automatic deduction or code scanning payment.

[0060] The specific example is as follows: the underground parking lot of an office building has deployed intelligent parking lock, sensor, camera and Internet of Things technology, and the real-time state of all parking spaces can be monitored and managed through the cloud platform; now combined with user reservation, parking space sorting, pricing strategy and other functions, it is further expanded to the process of user reservation, parking lock, license plate recognition and automatic payment.

[0061] Mr. Li submits a reservation request, and the reservation information is as follows: 1, vehicle type information: small car (length: 4.5 meters, width: 1.8 meters); 2, estimated entering time: 8:30 AM; 3, estimated stay duration: 2 hours; 4, destination preference: close to elevator and exit; Mr. Li inputs the above information through the App end and submits the reservation request; The basic information of all parking spaces has been stored in the database, including the size (length and width) of the parking space, the location (close to the elevator, close to the exit, etc.), the price (dynamic pricing), and the real-time occupancy state (whether idle).

[0062] Parking space A: close to the exit, 5 meters long, 2 meters wide, price 5 yuan per hour, current occupancy state: idle; Parking space B: close to the elevator, 4.5 meters long, 1.8 meters wide, price 6 yuan per hour, current occupancy state: occupied; Parking space C: far from the exit, 4.5 meters long, 1.8 meters wide, price 4 yuan per hour, current occupancy status: free; According to Mr. Li's vehicle information (small car, 4.5 meters long, 1.8 meters wide), parking space B and parking space C meet the requirements; parking space B has been occupied, excluded; parking space C is still free and can be used as a candidate parking space; At the same time, Mr. Li prefers to be close to the elevator and the exit, so the location advantage of parking space C needs to be further evaluated; Then it will be sorted based on the following characteristics: 1. Location characteristics Parking space C is far from the elevator but close to the exit; therefore, although Mr. Li prefers to be close to the elevator, the location characteristics score of parking space C is low; give a score for location characteristics, such as 0.6 (if the full score is 1, the parking space close to the elevator can get a higher score); 2. Price characteristics The price of parking space C is 4 yuan per hour, cheaper than the 6 yuan per hour of parking space B, and Mr. Li wants to save money, so the price characteristics of parking space C are a plus; Give a score for price characteristics, such as 0.8 (the parking space with lower price gets a higher score); 3. User preference matching degree Mr. Li prefers to be close to the elevator, while parking space C is far from the elevator, so parking space C scores low in user preference matching degree; Given the low preference matching degree, score 0.4 for this (not completely in line with the preference, so a lower score); 4. Real-time availability Parking space C is currently free, which means it scores high in availability.

[0063] The real-time availability score can be 1 (the highest score for a free parking space); To calculate the final recommendation score of parking space C, weight and sum the characteristic scores; assume the weights of each characteristic are as follows: The location characteristic weight is 0.3, the price characteristic weight is 0.3, the user preference matching degree weight is 0.2, and the real-time availability weight is 0.2, so the recommendation score of parking space C is 0.7; Parking space C scores 0.7 after scoring, which is the optimal recommendation according to Mr. Li's needs and preferences; after screening the parking spaces that meet the size, parking space C becomes the only available and price-demanding parking space, although it is far from the elevator, but considering the price and real-time availability, it is still the recommended parking space.

[0064] According to the demand and preference of Li, the parking space C is recommended as the optimal choice through the calculation of the personalized parking space ranking model, the detailed information of the parking space is displayed, including the price, location, real-time available state and the like, and is provided to Li for final confirmation; if Li selects the parking space, the parking space is locked through the distributed lock mechanism and a reservation record is generated, and the parking space state is updated to "reserved".

[0065] The second embodiment of the present application also discloses a big data-based intelligent parking management system, which is implemented based on the big data-based intelligent parking management method.

[0066] It should be noted that the calculation formula and the parameters participating in the operation in the present application are all pre-processed by dimensionless processing, and the process of dimensionless processing is known in the industry and will not be described here.

[0067] Although the embodiments of the present application have been shown and described above, it can be understood that the above-mentioned embodiments are exemplary and cannot be understood as limiting the present application, and those skilled in the art can make changes, modifications, replacements and modifications to the above-mentioned embodiments within the scope of the present application.

Claims

1. A big data-based intelligent parking management method, characterized in that, The intelligent parking management method comprises the following steps: S1, deploying an intelligent parking lock at each parking space in the parking lot, using a pre-installed sensor to sense the parking space occupation state, capturing vehicle information through a pre-installed high-definition camera, and then transmitting the information to a cloud platform in real time through Internet of Things technology; S2, using the cloud platform to process the parking space occupation state in real time, and displaying the current state of all parking spaces through the management background and App end, generating a dynamic pricing strategy based on the pre-installed reservation rules, using intelligent demand prediction and dynamic pricing model; S3, the user inputs the reservation information through the App end, and according to the real-time parking space state, dynamic pricing strategy and user historical preference, uses the personalized parking space sorting model to recommend the optimal parking space for the user, when the user confirms, the parking space is locked and a reservation record is generated, and the parking space occupation state is updated in real time; S4, when the vehicle arrives at the entrance, the license plate information is captured, the reservation record is queried, the identity authentication and vehicle classification are completed, the intelligent parking lock of the corresponding parking space is opened, and the corresponding parking allocation scheme is executed, and the indoor navigation technology is used to provide the user with a real-time travel route; S5, when the vehicle leaves, the leaving time is recorded automatically, the parking fee is calculated according to the dynamic pricing strategy, the reservation performance is verified, and the fee settlement and payment are completed. 2.The big data-based intelligent parking management method of claim 1, wherein, The cloud platform processes the parking space occupation state in real time, and displays the current state of all parking spaces through the management background and App end, generates a dynamic pricing strategy based on the pre-installed reservation rules, and uses intelligent demand prediction and dynamic pricing model, which comprises the following steps: S21, the cloud platform continuously collects data from the parking space sensors and cameras, and associates with historical databases, time, calendar and external data sources such as weather, to obtain multi-dimensional data, and then displays the current state of all parking spaces through the management background and App end; S22, based on the pre-installed reservation rules, the multi-dimensional data is feature extracted to generate a feature vector for predicting parking demand, and the XGBoost model and the logistic regression model are used for parking demand prediction; S23, using the Stacking meta-learner, the prediction results of the XGBoost model and the logistic regression model are integrated to obtain the final parking demand prediction; S24, according to the final parking demand prediction, the time period, region and user behavior are comprehensively considered to generate a real-time adjusted dynamic pricing strategy. 3.The big data-based intelligent parking management method of claim 2, wherein, The cloud platform processes the parking space occupation state in real time, and displays the current state of all parking spaces through the management background and App end, generates a dynamic pricing strategy based on the pre-installed reservation rules, and uses intelligent demand prediction and dynamic pricing model, which comprises the following steps: S221, according to the pre-set reservation rules, the feature vector related to the parking demand is extracted from the multi-dimensional data; S222, the extracted feature vector is input into the trained XGBoost model to output the prediction value of the parking demand in the future preset time period; S223, the extracted feature vector is input into the trained logistic regression model to calculate the influence of each feature on the parking demand probability, and output the prediction probability of the parking demand in the future time period. 4.The big data-based intelligent parking management method of claim 3, wherein, The expression for inputting the extracted feature vector into the trained XGBoost model and outputting the predicted value of the parking space demand in the future preset period is: ； wherein, represents a demand value of a parking space; represents an initial prediction value of an XGBoost model; represents the number of decision trees; represents a decision tree; represents a prediction value of the th decision tree for a feature vector ; and represents a learning rate. 5.The big data based smart parking management method according to claim 1, wherein, The user inputs reservation information through the App end, and according to the real-time parking space state, the dynamic pricing strategy and the user historical preference, the optimal parking space is recommended for the user by using the personalized parking space sorting model, when the user confirms, the parking space is locked and a reservation record is generated, and the parking space occupancy state is updated in real time, including the following steps: S31, the user submits a reservation request through the App end, and inputs the vehicle type information, the expected entry time, the expected stay duration and the destination preference on the App end; S32, the basic attribute information, the real-time occupancy state and the dynamic pricing strategy of each parking space in the parking lot are obtained; S33, according to the vehicle type information, the expected entry time, the expected stay duration, the destination preference, and the basic attribute information, the real-time occupancy state and the dynamic pricing strategy of each parking space, the occupied parking spaces and the parking spaces with unmatched sizes are deleted; for each candidate parking space, the business scenario features including the location features, the price features, the user preference matching degree and the real-time availability are constructed; S34, the personalized parking space sorting model is used to generate a recommendation score for each parking space according to the business scenario features, and the parking space list with the highest priority is selected and recommended to the user; S35, the user selects the most suitable parking space from the parking space list, and locks the parking space through the distributed lock mechanism to generate a reservation record; S36, when the user confirms the reservation, the state of the parking space is updated from idle to reserved, and is synchronously updated to the parking space state on the App end. 6.The big data-based intelligent parking management method of claim 5, wherein, The personalized parking space sorting model is used to generate a recommendation score for each parking space according to the business scenario features, and the parking space list with the highest priority is selected and recommended to the user, including the following steps: S341, the feature vector of each candidate parking space in the parking lot is taken as the input of the personalized parking space sorting model; S342, a recommendation score is generated for each candidate vehicle through the personalized parking space sorting model; S343, the score difference of each pair of parking spaces is calculated, and the sorting loss is calculated by combining the relative importance difference of the parking spaces, so as to optimize the recommendation score of each candidate vehicle; S344. Sort parking spaces in descending order based on the optimized recommendation scores, and select the first... K The highest-scoring parking spaces are selected as a recommended list and presented to the user, along with the reason for recommending each parking space in the parking lot. 7.The big data-based intelligent parking management method of claim 6, wherein, The expression for calculating the score difference of each pair of parking spaces and the relative importance difference of the parking spaces is: ； wherein represents the ranking loss of the parking space and the parking space ; represents the recommendation score of the parking space ; represents the recommendation score of the parking space ; represents the importance difference of the parking space and the parking space ; represents the exponential function of the score difference of the parking space and the parking space . 8.The big data based intelligent parking management method according to claim 1, wherein, When the vehicle leaves, the leaving time is automatically recorded, the parking fee is calculated according to the dynamic pricing strategy, the reservation performance is checked, and the fee settlement and payment are completed, including the following steps: S51, when the vehicle leaves, the leaving vehicle is identified through the license plate recognition, the leaving time of the vehicle is recorded, the parking fee is automatically calculated based on the classification result of the vehicle and in combination with the parking duration and the dynamic pricing strategy; wherein the classification result of the vehicle includes the reservation vehicle, the monthly rental vehicle and the temporary vehicle; S52, the reservation vehicle is calculated according to the charging rules on the App end, and the performance is checked, whether the vehicle enters and leaves on time according to the specification of the reservation period is checked, and the overtime or default behavior is handled; S53, the monthly car and the temporary car are settled according to respective billing rules, wherein the monthly car is settled according to a monthly fixed fee, and the temporary car is calculated for a parking time length and a dynamic pricing strategy; S54, the non-sensing payment of all vehicles is completed through automatic deduction or code scanning payment.

9. A big data-based intelligent parking management system, characterized in that, The system is based on the big data-based intelligent parking management method of any one of claims 1-8.

Images