Parking lot planning method based on parking lot real-time operation big data

Abstract

The invention discloses a parking lot planning method based on real-time operation big data of a parking lot, and the method can dynamically optimize the use and management of the parking lot based on the real-time operation big data of the parking lot. Firstly, in combination with vehicle access time and direction data and operator mobile phone signaling data, a trip starting point and a trip ending point of regional traffic flow are accurately calculated, and accurate analysis of traffic flow modes and parking requirements is facilitated. And secondly, a space-time state model of the parking lot is constructed through real-time flow data, the utilization rate and the turnover rate of the parking spaces in different time periods can be reflected in real time, and the parking demand and the vehicle flow trend in the future are predicted, so that a scientific basis is provided for resource optimization configuration of the parking lot. And finally, performing path planning by using the real-time vehicle position, the destination and the predicted parking lot state data, dynamically adjusting the vehicle driving route, and optimizing the optimal path for the vehicle to arrive at the destination.

Description

Technical Field

[0001] The present invention belongs to the field of parking lot planning, and in particular, relates to a parking lot planning method based on real-time parking lot operation big data. Background Art

[0002] With the acceleration of urbanization and the rapid increase in the number of motor vehicles, parking lot management has become an increasingly important issue in urban traffic management. This is especially true in urban central areas and busy commercial areas, where parking demand often surges during peak hours, resulting in limited parking resources being unable to meet the parking needs of a large number of vehicles. Traditional parking lot management methods usually rely on manual patrols and static information releases, which are not only inefficient, but also fail to provide real-time parking space status information, making it difficult to timely and effectively reflect the actual use of parking lots. This information lag causes car owners to spend a long time looking for vacant parking spaces in parking lots, which not only increases the parking time of car owners, but also easily causes anxiety and dissatisfaction among drivers. At the same time, the ineffective driving of vehicles in parking lots increases traffic congestion in parking lots, aggravates traffic pressure in local areas, and causes unnecessary fuel consumption and vehicle emissions, which has an adverse impact on the environment. In addition, most traditional parking lot management systems lack the ability to accurately predict vehicle flow, parking space utilization, and parking demand, and cannot be dynamically adjusted and optimized based on real-time data. As a result, it is difficult to improve the operating efficiency of parking lots, and parking space resources cannot be reasonably utilized. Especially when parking demand fluctuates greatly, it is easy to cause idle parking spaces and parking space shortages during peak demand periods, further reducing the overall utilization efficiency of parking lots. This management lag and uneven resource allocation have seriously restricted the improvement of urban parking lot management level and also affected the operating efficiency of the overall urban transportation system.

[0003] In view of this, the present invention is proposed. Summary of the invention

[0004] The technical problem to be solved by the present invention is to overcome the deficiencies of the prior art and provide a parking lot planning method based on real-time parking lot operation big data, thereby solving the problems raised in the above-mentioned background technology.

[0005] In order to solve the above technical problems, the basic concept of the technical solution adopted by the present invention is:

[0006] A parking lot planning method based on real-time parking lot operation big data includes the following steps:

[0007] S1, obtains vehicle entry and exit time and direction data, and combines it with operator mobile phone signaling data to calculate the starting and ending point information of regional vehicle flow;

[0008] S2, obtaining the real-time flow data of the parking lot from the global parking lot management system and establishing a spatiotemporal state model of the parking lot based on the obtained real-time flow data;

[0009] S3, using the spatiotemporal state model to analyze parking space utilization, parking space turnover rate and parking demand in different time periods, and predict future parking demand and vehicle flow trends;

[0010] S4, based on the real-time location of the vehicle, the destination and the predicted parking lot status, the path planning is performed to optimize the optimal path for the vehicle to reach the destination.

[0011] Optionally, the step of calculating the starting point and end point information of regional vehicle flow in step S1 is:

[0012] S101, install license plate recognition cameras on the road sections to capture the license plate numbers, entry and exit time and direction data of passing vehicles in real time and obtain mobile phone signaling data in the area. The mobile phone signaling data includes user location, movement trajectory and timestamp;

[0013] S102, clean the vehicle data and mobile phone signaling data, remove noise and abnormal data and use the formula Match the correlation score R(V i , U j ), and according to the correlation score R(V i , U j ) determines the association between the vehicle and the corresponding mobile phone user, where: It is the difference between the vehicle identification time and the mobile phone user positioning time, indicating the time difference. is the Euclidean distance between the vehicle identification location and the mobile phone user's location, indicating the spatial difference. is the signal strength of the mobile phone user, which is used to measure the reliability of the mobile phone signal. α, β, and γ are weight parameters used to adjust the importance of time difference, space difference, and signal strength in matching. V i represents the i-th car, is the time when the vehicle is identified, is the coordinates of the location where the vehicle is identified, U j represents the jth mobile phone user, The time when the mobile phone user is located. The location coordinates of the mobile phone user. Signal strength for mobile phone users;

[0014] S103, calculating the starting point and end point of the regional vehicle flow according to the cleaned and matched data, and generating an OD matrix OD ij , which reflects the travel mode and flow of vehicles in a specific time period, and its expression is: Among them, T represents the time range for calculating the 0D matrix, N ij (t) represents the number of vehicles actually observed from the starting point i to the end point j within time t, w ij (t) is the weight function, which represents the importance of going from starting point i to end point j within time t.

[0015] Optionally, in step S2, the spatiotemporal state model is a spatiotemporal autoregressive model and a grid model to form a STAR-Grid model, wherein the steps of establishing the STAR-Grid model are:

[0016] S201, collecting real-time data and historical data in the parking lot and cleaning and preprocessing the collected data;

[0017] S202, converting the plan of the parking lot into grid units according to the preprocessed data, each grid representing a parking space or a specific area, storing the number of parking spaces in the area and current status information, and discretizing the time axis of the grid into fixed time intervals, then recording the state changes of the parking spaces and the entry and exit of vehicles in each time interval;

[0018] S203, converting the status record of each grid unit into time series data, describing the change of parking space status over time and constructing a time autoregressive model based on the spatiotemporal data, and then training the STAR-Grid model using the historical data obtained in step S1.

[0019] Optionally, when the spatiotemporal autoregressive model and the grid model are combined for use, first determine the size of each grid according to the layout of the parking lot and the size of the parking spaces, then use OpenCV to divide the parking lot plan into grids, then calculate the coordinates of the upper left corner and the lower right corner of each grid, where each grid unit includes four coordinate values, current status, number of parking spaces, and timestamp, then use a database table to store the grid information, and then define spatial features for each grid unit and use geographic information system data to assist in extracting spatial features, then extract the historical parking data of each grid to form a time series, then identify the order of the autoregressive term, the number of differences, and the order of the moving average term of the model through the ACF and PACF graphs, and construct an ARIMA model based on the order of the autoregressive term, the number of differences, and the order of the moving average term, and finally, use the ARIMA model to predict the future parking space status and output the future parking space occupancy rate and demand of each grid.

[0020] Optionally, when training the STAR-Grid model, it is necessary to obtain historical data from the parking management system and collect external data that affects parking demand, integrate the historical data and external data into a data set, and divide the data set into a training set and a validation set. The training set is used when training the STAR-Grid model, and the validation set is used to verify the performance of the STAR-Grid model after the training is completed.

[0021] Optionally, the steps of using the spatiotemporal state model to analyze parking space utilization, parking space turnover rate and parking demand in different time periods and predict future parking demand and vehicle flow trends are as follows:

[0022] S301, inputting the parking space status data of the current parking lot and the latest external data into the trained STAR-Grid model, and the STAR-Grid model uses the current input data to predict the parking space status changes of each grid unit in the future time period;

[0023] S302, using the data output by the model, calculate the parking space utilization rate and parking space turnover rate in each future time period and identify the peak and off-peak periods, predict the overall parking demand change trend of the parking lot, and finally analyze the flow trend of vehicles in and out of the parking lot based on the parking space status prediction results.

[0024] Optionally, the steps of performing path planning based on the real-time location, destination and predicted parking lot status of the vehicle and optimizing the optimal path for the vehicle to reach the destination are as follows:

[0025] S401, obtaining the real-time location data of the vehicle and the destination information of the vehicle and obtaining the parking lot status data through the vehicle-mounted GPS system;

[0026] S402, constructing a road network model in the parking lot according to the plan of the parking lot, dividing the parking lot into a plurality of nodes and edges, and setting constraint conditions for path planning;

[0027] S403: Use Dijkstra algorithm to plan the route according to the road network model and the number of vehicles in the parking lot. Then, set the real-time position of the vehicle as the starting point and the destination as the end point. Update the edge attributes in the road network model according to the predicted parking lot status.

[0028] S404, using the Dijkstra algorithm to search for the optimal path from the current position of the vehicle to the destination in the parking lot road network model. During the path planning process, the algorithm calculates the total weight of different paths according to the weight of each edge and selects the path with the smallest weight;

[0029] S405, during the process of searching for a route, monitor the changes in the status of the vehicle and the parking lot in real time, and dynamically update the input of the road network model and the route planning algorithm according to the new data;

[0030] S406, outputting the calculated optimal path to the navigation system of the vehicle or the parking lot management system so that the vehicle can travel according to the planned path, and the vehicle is guided to travel along the optimal path by the vehicle's onboard navigation system.

[0031] Optionally, each node represents a location in the parking lot, and each edge represents a road between two nodes.

[0032] Optionally, before obtaining the optimal path, it is necessary to collect the status information of each parking space in real time through ground sensors and obtain the type of vehicles entering the parking lot and the needs of the owners;

[0033] Based on the real-time monitoring of parking space status and vehicle information, the owner's needs and vehicle type are matched with the available parking spaces. Based on the matching of vehicle type and parking space type, the space utilization of the parking lot is optimized. For example, electric vehicles are given priority in being allocated to parking spaces close to charging piles.

[0034] Optionally, the steps of using the Dijkstra algorithm in step S404 to search for the optimal path from the current position of the vehicle to the destination in the parking lot road network model are:

[0035] S4041, create a set Q, including all nodes in the parking lot road network model, select the node u closest to the starting point from the set Q, mark u as visited by v, and remove it from Q;

[0036] S4042, traverse all adjacent nodes v of node u, and for each adjacent node, calculate the total path distance d(u) from the starting point to node v, the formula is: d(v)=d(u)+w(u,v), where d(u) is the distance value of node u, and w(u,v) is the weight of edge (u,v);

[0037] S4043, if the calculated total distance d(v) of the new path is less than the current distance value of node v, then update the distance value and predecessor node of node v, that is, update d(v) to the new value, and record u as the predecessor node of v;

[0038] S4044, continue to select the next unvisited node closest to the starting point from the set Q, and repeat steps S4041 to S4043 until the set Q is empty or the target node v is marked as visited;

[0039] S4045, starting from the target end point, backtracking along the predecessor node step by step until backtracking to the starting point, forming an optimal path from the starting point to the end point, and finally, outputting all the nodes included in the optimal path in sequence as the recommended route for the vehicle.

[0040] After adopting the above technical solution, the present invention has the following beneficial effects compared with the prior art. Of course, any product implementing the present invention does not necessarily need to achieve all the advantages described below at the same time:

[0041] The present invention can dynamically optimize the use and management of parking lots based on the real-time operation big data of parking lots. First, by combining the vehicle entry and exit time and direction data and the operator's mobile phone signaling data, the starting point and end point of the regional traffic flow are accurately calculated, which is helpful for accurately analyzing the traffic pattern and parking demand. Secondly, the spatiotemporal state model of the parking lot is constructed through real-time flow data, which can reflect the utilization rate and parking space turnover rate of parking spaces in different time periods in real time, and predict future parking demand and vehicle flow trends, which provides a scientific basis for the optimal allocation of parking resources. Finally, the real-time vehicle position, destination and predicted parking lot status data are used to carry out path planning, dynamically adjust the vehicle driving route, and optimize the optimal path for the vehicle to reach the destination. Through this comprehensive dynamic management strategy, the search time of vehicles in the parking lot can be significantly reduced, the traffic congestion in the parking lot can be reduced, the parking efficiency and user experience can be improved, and the optimal allocation and efficient management of parking resources can be achieved.

[0042] The specific implementation modes of the present invention are further described in detail below with reference to the accompanying drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] The drawings described below are only some embodiments. For those skilled in the art, other drawings can be obtained based on these drawings without creative work.

[0044] In the figure:

[0045] Figure 1 Flowchart of parking lot planning methodology.

[0046] It should be noted that these drawings and textual descriptions are not intended to limit the conceptual scope of the present invention in any way, but are intended to illustrate the concept of the present invention for those skilled in the art by referring to specific embodiments. DETAILED DESCRIPTION

[0047] The present invention will now be described in further detail with reference to the accompanying drawings.

[0048] See also Figure 1 As shown, in this embodiment, a parking lot planning method based on real-time parking lot operation big data is provided, including the following steps:

[0049] S1, obtains vehicle entry and exit time and direction data, and combines it with the operator's mobile phone signaling data to calculate the starting and ending point information of regional vehicle flow; by obtaining vehicle entry and exit time, direction data and mobile phone signaling data, the starting and ending points of regional vehicle flow can be accurately analyzed. This data can help identify vehicle flow patterns and understand which areas have dense traffic and which areas have sparse traffic.

[0050] S2, obtaining the real-time flow data of the parking lot from the global parking lot management system and establishing a spatiotemporal state model of the parking lot based on the obtained real-time flow data;

[0051] S3, using the spatiotemporal state model to analyze parking space utilization, parking space turnover rate and parking demand in different time periods, and predict future parking demand and vehicle flow trends;

[0052] S4, based on the real-time location of the vehicle, the destination and the predicted parking lot status, the path planning is performed to optimize the optimal path for the vehicle to reach the destination.

[0053] In this embodiment, the steps of calculating the starting point and end point information of regional vehicle flow in step S1 are:

[0054] S101, install license plate recognition cameras on the road sections to capture the license plate numbers, entry and exit time and direction data of passing vehicles in real time and obtain mobile phone signaling data in the area. The mobile phone signaling data includes user location, movement trajectory and timestamp;

[0055] S102, clean the vehicle data and mobile phone signaling data, remove noise and abnormal data and use the formula Match the correlation score R(V i , U j ), and according to the correlation score R(V i , U j ) determines the association between the vehicle and the corresponding mobile phone user, where: It is the difference between the vehicle identification time and the mobile phone user positioning time, indicating the time difference. It is the Euclidean distance between the vehicle identification location and the mobile phone user positioning location, indicating the spatial difference. is the signal strength of the mobile phone user, which is used to measure the reliability of the mobile phone signal. α, β, γ are weight parameters used to adjust the importance of time difference, space difference and signal strength in matching. i represents the i-th car, Indicates the time when the vehicle was identified. The coordinates of the location where the vehicle was identified. j represents the jth mobile phone user, Indicates the time when the mobile phone user is located. Indicates the location coordinates of the mobile phone user. Indicates the signal strength for mobile phone users;

[0056] S103, calculating the starting point and end point of the regional traffic flow based on the cleaned and matched data, and generating OD Matrix OD ij , which reflects the travel mode and flow of vehicles in a specific time period, and its expression is: Where T represents the time range for calculating the OD matrix. For example, T can be a day, an hour, or a morning or evening rush hour. ij (t) represents the number of vehicles actually observed from the starting point i to the end point j in time t. This value is obtained directly from the data. ij (t) represents the weight function, which indicates the importance of traveling from starting point i to end point j within time t. The weight can be adjusted according to various factors, such as: Time factor: Traffic flow may be different in different time periods, such as the difference between peak and non-peak hours. Distance factor: The distance from the starting point to the end point. Generally, shorter distances may have a larger weight because closer ODs have a more significant impact on local traffic. Other factors: such as road conditions and weather conditions, can affect the setting of weights. Calculating the starting and ending points of regional traffic flow and generating an OD matrix (travel starting point-end point matrix) can accurately reflect the travel patterns and flow of vehicles in a specific time period.

[0057] In this embodiment, the spatiotemporal state model in step S2 is a STAR-Grid model formed by a spatiotemporal autoregressive model and a grid model. By integrating the spatiotemporal autoregressive model and the grid model (STAR-Grid model), it is possible to simultaneously consider the data changes in the time and space dimensions and more accurately analyze the dynamic changes in the parking lot. Among them, the steps of establishing the STAR-Grid model are:

[0058] S201, collecting real-time data and historical data in the parking lot and cleaning and preprocessing the collected data;

[0059] S202, converting the plan of the parking lot into grid units according to the preprocessed data, each grid representing a parking space or a specific area, storing the number of parking spaces in the area and current status information, and discretizing the time axis of the grid into fixed time intervals, then recording the state changes of the parking spaces and the entry and exit of vehicles in each time interval;

[0060] S203, convert the state record of each grid unit into time series data, describe the change of parking space status over time and build a time autoregressive model based on the spatiotemporal data, and then use the historical data obtained in step S1 to train the STAR-Grid model. Based on the accurate prediction results, parking lot managers can adjust the parking space allocation strategy in advance, optimize resource allocation, and avoid insufficient or wasted parking spaces.

[0061] In this embodiment, when the spatiotemporal autoregressive model and the grid model are combined for use, the size of each grid is first determined according to the layout of the parking lot and the size of the parking spaces. Then, the parking lot plan is divided into grids using OpenCV. Then, the coordinates of the upper left corner and the lower right corner of each grid are calculated, wherein each grid unit includes four coordinate values, the current state, the number of parking spaces, and a timestamp. Then, a database table is used to store the grid information. Then, spatial features are defined for each grid unit and geographic information system data is used to assist in extracting spatial features. Then, the historical parking data of each grid is extracted to form a time series. Then, the order of the autoregressive term, the number of differences, and the order of the moving average term of the model are identified through ACF and PACF graphs, and an ARIMA model is constructed according to the order of the autoregressive term, the number of differences, and the order of the moving average term. Finally, the ARIMA model is used to predict the future parking space status, and the future parking space occupancy rate and demand of each grid are output.

[0062] It should be noted that by defining the spatial characteristics of each grid and extracting relevant features with the help of geographic information system (GIS) data, not only the dimension of the data is enhanced, but also the prediction accuracy of the model is improved. After extracting the historical parking data to form a time series, the p, d, and q parameters of the ARIMA model are identified through the autocorrelation function (ACF) and partial autocorrelation function (PACF) diagrams to ensure the scientificity and rationality of the model parameters. Finally, the constructed ARIMA model is used to accurately predict the future parking space status and output the future parking space occupancy rate and demand of each grid, thereby effectively improving the intelligent level of parking management and resource allocation efficiency.

[0063] In this embodiment, when training the STAR-Grid model, it is necessary to obtain historical data from the parking lot management system and collect external data that affects parking demand, including weather conditions, holidays, large-scale events, and traffic flow. The historical data and external data are integrated into a data set, and the data set is divided into a training set and a validation set. The training set is used when training the STAR-Grid model, and the validation set is used to verify the performance of the STAR-Grid model after the training is completed. The STAR-Grid model uses historical data and real-time data, combined with external factors (such as weather, holidays, etc.), to more accurately predict future parking demand and parking space usage, thereby improving the accuracy and reliability of the prediction.

[0064] In this embodiment, the steps of using the spatiotemporal state model to analyze parking space utilization, parking space turnover rate and parking demand in different time periods and predicting future parking demand and vehicle flow trends are as follows:

[0065] S301, input the parking space status data of the current parking lot and the latest external data into the trained STAR-Grid model. The STAR-Grid model uses the current input data to predict the changes in the parking space status of each grid unit in the future time period; by analyzing the changing trends of parking space utilization and parking space turnover rate, it predicts future parking demand and helps parking lot managers plan and respond to changes in parking demand in advance.

[0066] S302, using the data output by the model, calculate the parking space utilization rate and parking space turnover rate in each time period in the future and identify the peak and off-peak periods, predict the overall parking demand change trend of the parking lot, and finally analyze the flow trend of vehicles in and out of the parking lot based on the parking space status prediction results. The prediction results can identify the peak and off-peak periods of the parking lot, and managers can formulate different management strategies based on this information, such as adjusting parking space prices, adding guide signs, etc. Secondly, by predicting the vehicle flow trend of the parking lot in real time, it helps managers dynamically adjust management strategies, optimize vehicle entry and exit routes, reduce congestion in the parking lot, and improve parking efficiency.

[0067] In this embodiment, the steps of performing path planning and optimizing the optimal path for the vehicle to reach the destination are as follows:

[0068] S401, obtaining the real-time location data of the vehicle and the destination information of the vehicle and obtaining the parking lot status data through the vehicle-mounted GPS system;

[0069] S402, constructing a road network model in the parking lot according to the plan of the parking lot, dividing the parking lot into multiple nodes and edges, and setting constraints for path planning, including allowing only one-way traffic, avoiding congested sections, giving priority to parking spaces near exits, etc. The optimization goal is to minimize the total driving time or distance from the current location of the vehicle to the destination;

[0070] S403, using the Dijkstra algorithm to perform path planning based on the road network model of the parking lot and the number of vehicles, then setting the real-time position of the vehicle as the starting point and the destination as the end point, and updating the edge attributes in the road network model based on the predicted parking lot status;

[0071] S404, using the Dijkstra algorithm to search for the optimal path from the current position of the vehicle to the destination in the parking lot road network model. During the path planning process, the algorithm calculates the total weight of different paths according to the weight of each edge and selects the path with the smallest weight;

[0072] S405, in the process of searching for a route, the changes in the status of vehicles and parking lots are monitored in real time, and the input of the road network model and the path planning algorithm are dynamically updated according to the new data. If a new vehicle enters or exits or the status of a parking space changes (such as a parking space on the originally planned route is suddenly occupied), the route is recalculated in time to ensure that the vehicle reaches the destination by the optimal route;

[0073] S406, outputting the calculated optimal path to the navigation system of the vehicle or the parking lot management system so that the vehicle can travel according to the planned path, and the vehicle is guided to travel along the optimal path by the vehicle's onboard navigation system.

[0074] It should be noted that: step S401 obtains the real-time location and destination information of the vehicle, as well as the status data of the parking lot through the vehicle-mounted GPS system to ensure that the path planning is based on the latest actual situation. Then, step S402 constructs a road network model of the parking lot and sets specific path planning constraints, such as one-way traffic, avoiding congested sections, and giving priority to parking spaces near the exit. The optimization goal is set to minimize the total driving time or distance of the vehicle, thereby improving the accuracy of path selection. Steps S403 and S404 use the Dijkstra algorithm to search for paths, calculate the total weight of the path based on the weight of the edge, and thus select the optimal path to ensure that the vehicle can reach the destination quickly. Step S405 can respond to changes in vehicle entry and exit or parking space status in a timely manner through real-time monitoring and dynamic updating of path planning, recalculate the optimal path, and ensure that the path is always optimal. Finally, step S406 outputs the optimal path to the vehicle navigation system to provide real-time driving guidance, which not only reduces the search time of the vehicle in the parking lot, but also effectively prevents traffic congestion, greatly improving the user's parking experience and the operation efficiency of the parking lot.

[0075] In this embodiment, each node represents a location in the parking lot, and each edge represents a road between two nodes.

[0076] In this embodiment, before obtaining the optimal path, it is necessary to collect the status information of each parking space in real time through ground sensors and obtain the type of vehicles entering the parking lot (such as small cars, SUVs, electric vehicles, etc.) and the needs of car owners (such as close to the exit, charging piles, special parking space requirements, etc.);

[0077] Based on the real-time monitoring of parking space status and vehicle information, the owner's needs and vehicle type are matched with the available parking spaces. Based on the matching of vehicle type and parking space type, the space utilization of the parking lot is optimized. For example, electric vehicles are given priority in being allocated to parking spaces close to charging piles.

[0078] In this embodiment, S4041, create a set Q, including all nodes in the parking lot road network model, select the node u closest to the starting point from the set Q, mark u as visited, and remove it from Q;

[0079] S4042, traverse all adjacent nodes v of node u, and for each adjacent node, calculate the total path distance d(u) from the starting point to node v, the formula is: d(v)=d(u)+w(u,v), where d(u) is the distance value of node u, and w(u,v) is the weight of edge (u,v);

[0080] S4043, if the calculated total distance d(v) of the new path is less than the current distance value of node v, then update the distance value and predecessor node of node v, that is, update d(v) to the new value, and record u as the predecessor node of v;

[0081] S4044, continue to select the next unvisited node closest to the starting point from the set Q, and repeat steps S4041 to S4043 until the set Q is empty or the target node v is marked as visited;

[0082] S4045, starting from the target end point, backtracking along the predecessor node step by step until backtracking to the starting point, forming an optimal path from the starting point to the end point, and finally, outputting all the nodes included in the optimal path in sequence as the recommended route for the vehicle.

[0083] It should be noted that: Step S4041 effectively reduces the search space and improves the execution efficiency of the algorithm by creating a node set Q and selecting the nearest node u; Steps S4042 and S4043 ensure the accuracy of selecting the shortest path and optimize the path calculation by traversing adjacent nodes and updating the total path distance; Step S4044 ensures full exploration of all possible paths by looping and iterating until the target node is visited; Finally, Step S4045 accurately constructs the optimal path by tracing back from the target node to the starting point, and outputs it as the recommended route for vehicle travel. This method can not only quickly respond to real-time changes and provide flexible path adjustments, but also significantly reduce the driving time of vehicles in the parking lot and the trouble of finding parking spaces, thereby improving the overall user experience and parking lot operation efficiency.

[0084] The present invention is not limited to the above-mentioned embodiments. Anyone should be aware that any structural changes made under the enlightenment of the present invention, and any technical solutions that are the same or similar to the present invention, fall within the protection scope of the present invention. The technology, shape, and structural parts not described in detail in the present invention are all well-known technologies.

Claims

1. A parking lot planning method based on real-time parking lot operation big data, characterized in that: The following steps are involved: S1, obtains vehicle entry and exit time and direction data, and combines it with operator mobile phone signaling data to calculate the starting and ending point information of regional vehicle flow; S2, obtaining the real-time flow data of the parking lot from the global parking lot management system and establishing a spatiotemporal state model of the parking lot based on the obtained real-time flow data; S3, using the spatiotemporal state model to analyze parking space utilization, parking space turnover rate and parking demand in different time periods, and predict future parking demand and vehicle flow trends; S4, based on the real-time location of the vehicle, the destination and the predicted parking lot status, the path planning is performed to optimize the optimal path for the vehicle to reach the destination.

2. A parking lot planning method based on real-time parking lot operation big data according to claim 1, characterized in that: The steps for calculating the starting point and end point information of regional vehicle flow in step S1 are: S101, install license plate recognition cameras on the road sections to capture the license plate numbers, entry and exit time and direction data of passing vehicles in real time and obtain mobile phone signaling data in the area. The mobile phone signaling data includes user location, movement trajectory and timestamp; S102, clean the vehicle data and mobile phone signaling data, remove noise and abnormal data and use the formula Match the correlation score R(V i , U j ), and according to the correlation score R(V i , U j ) determines the association between the vehicle and the corresponding mobile phone user, where: It is the difference between the vehicle identification time and the mobile phone user positioning time, indicating the time difference. is the Euclidean distance between the vehicle identification location and the mobile phone user's location, indicating the spatial difference. is the signal strength of the mobile phone user, α, β, γ are weight parameters used to adjust the importance of time difference, space difference and signal strength in matching, V i represents the i-th car, is the time when the vehicle is identified, is the coordinates of the location where the vehicle is identified, U j represents the jth mobile phone user, The time when the mobile phone user is located. The location coordinates of the mobile phone user. Signal strength for mobile phone users; S103, calculating the starting point and end point of the regional vehicle flow according to the cleaned and matched data, and generating an OD matrix OD ij , which reflects the travel mode and flow of vehicles in a specific time period, and its expression is: Where T represents the time range for calculating the OD matrix, N ij (t) represents the number of vehicles actually observed from the starting point i to the end point j within time t, w ij (t) is the weight function, which represents the importance of going from starting point i to end point j within time t.

3. A parking lot planning method based on real-time parking lot operation big data according to claim 1, characterized in that: In step S2, the spatiotemporal state model is a STAR-Grid model formed by a spatiotemporal autoregressive model and a grid model, wherein the steps of establishing the STAR-Grid model are: S201, collecting real-time data and historical data in the parking lot and cleaning and preprocessing the collected data; S202, converting the plan of the parking lot into grid units according to the preprocessed data, each grid representing a parking space, storing the number of parking spaces in the area and current status information, and discretizing the time axis of the grid into fixed time intervals, then recording the state changes of the parking spaces and the entry and exit of vehicles in each time interval; S203, converting the status record of each grid unit into time series data, describing the change of parking space status over time and constructing a time autoregressive model based on the spatiotemporal data, and then training the STAR-Grid model using the historical data obtained in step S1.

4. A parking lot planning method based on real-time parking lot operation big data according to claim 3, characterized in that: When combining the spatiotemporal autoregressive model and the grid model, first determine the size of each grid according to the layout of the parking lot and the size of the parking spaces. Then, use OpenCV to divide the parking lot plan into grids. Then, calculate the coordinates of the upper left corner and the lower right corner of each grid, where each grid unit includes four coordinate values, current status, number of parking spaces, and timestamp. Then, use a database table to store grid information, and then define spatial features for each grid unit and use geographic information system data to assist in extracting spatial features. Then extract the historical parking data of each grid to form a time series. Then, identify the order of the autoregressive term, the number of differences, and the order of the moving average term of the model through the ACF and PACF graphs, and construct an ARIMA model based on the order of the autoregressive term, the number of differences, and the order of the moving average term. Finally, use the ARIMA model to predict the future parking space status and output the future parking space occupancy rate and demand of each grid.

5. The parking lot planning method based on real-time parking lot operation big data according to claim 3 is characterized in that: When training the STAR-Grid model, it is necessary to obtain historical data from the parking management system and collect external data that affects parking demand, integrate the historical data and external data into a data set, and divide the data set into a training set and a validation set. The training set is used when training the STAR-Grid model, and the validation set is used to verify the performance of the STAR-Grid model after training is completed.

6. A parking lot planning method based on real-time parking lot operation big data according to claim 1, characterized in that: The steps to use the spatiotemporal state model to analyze parking space utilization, parking space turnover rate and parking demand in different time periods and predict future parking demand and vehicle flow trends are as follows: S301, inputting the parking space status data of the current parking lot and the latest external data into the trained STAR-Grid model, and the STAR-Grid model uses the current input data to predict the parking space status changes of each grid unit in the future time period; S302, using the data output by the model, calculate the parking space utilization rate and parking space turnover rate in each future time period and identify the peak and off-peak periods, predict the overall parking demand change trend of the parking lot, and finally analyze the flow trend of vehicles in and out of the parking lot based on the parking space status prediction results.

7. A parking lot planning method based on real-time parking lot operation big data according to claim 6, characterized in that: According to the real-time location, destination and predicted parking lot status of the vehicle, the route is planned and the steps to optimize the optimal route for the vehicle to reach the destination are as follows: S401, obtaining the real-time location data of the vehicle and the destination information of the vehicle and obtaining the parking lot status data through the vehicle-mounted GPS system; S402, constructing a road network model in the parking lot according to the plan of the parking lot, dividing the parking lot into a plurality of nodes and edges, and setting constraint conditions for path planning; S403, using the Dijkstra algorithm to perform path planning based on the road network model of the parking lot and the number of vehicles, then setting the real-time position of the vehicle as the starting point and the destination as the end point, and updating the edge attributes in the road network model based on the predicted parking lot status; S404, using the Dijkstra algorithm to search for the optimal path from the current position of the vehicle to the destination in the parking lot road network model. During the path planning process, the algorithm calculates the total weight of different paths according to the weight of each edge and selects the path with the smallest weight; S405, during the process of searching for a route, monitor the changes in the status of the vehicle and the parking lot in real time, and dynamically update the input of the road network model and the route planning algorithm according to the new data; S406: Output the calculated optimal path to the navigation system of the vehicle so that the vehicle can travel according to the planned path. The vehicle is guided by the on-board navigation system of the vehicle to travel along the optimal path.

8. A parking lot planning method based on real-time parking lot operation big data according to claim 7, characterized in that: Each node represents a location in the parking lot, and each edge represents a road between two nodes.

9. A parking lot planning method based on real-time parking lot operation big data according to claim 7, characterized in that: Before obtaining the optimal path, it is necessary to use ground sensors to collect the status information of each parking space in real time and obtain the type of vehicles entering the parking lot and the needs of car owners; Based on the real-time monitoring of parking space status and vehicle information, the owner's needs and vehicle types are matched with available parking spaces, and the space utilization of the parking lot is optimized based on the matching of vehicle types and parking space types.

10. A parking lot planning method based on real-time parking lot operation big data according to claim 7, characterized in that: The steps of using the Dijkstra algorithm in step S404 to search for the optimal path from the current position of the vehicle to the destination in the parking lot road network model are: S4041, create a set Q, including all nodes in the parking lot road network model, select the node u closest to the starting point from the set Q, mark u as visited, and remove it from Q; S4042, traverse all adjacent nodes v of node u, and for each adjacent node, calculate the total path distance d(u) from the starting point to node υ, the formula is: d(υ)=d(u)+w(u,υ), where d(u) is the distance value of node u, and w(u,υ) is the weight of edge (u,υ); S4043, if the calculated total distance d(v) of the new path is less than the current distance value of node v, then update the distance value and predecessor node of node v, that is, update d(v) to the new value, and record u as the predecessor node of v; S4044, continue to select the next unvisited node closest to the starting point from the set Q, and repeat steps S4041 to S4043 until the set Q is empty or the target node v is marked as visited; S4045, starting from the target end point, backtracking along the predecessor node step by step until backtracking to the starting point, forming an optimal path from the starting point to the end point, and finally, outputting all the nodes included in the optimal path in sequence as the recommended route for the vehicle.

Images