Multi-stage path searching method, device and equipment based on time-varying road conditions

Abstract

The invention discloses a multi-stage path searching method, device and equipment based on time-varying road conditions, and relates to the technical field of traffic network travel path searching. In the travel stage of driving, the idea of travel simulation is adopted, the real driving process is simulated from the departure moment along a candidate path sequence, and time consumption is accumulated section by section; in the process, firstly, dynamic time variation in driving path search is considered, and in a parking decision stage, a parking lot is dynamically associated to a specific node of a road network through UTM coordinate conversion and k-d tree spatial indexing, so that a complete search strategy from driving path search to parking decision is formed; the distance between the attachment point and the parking lot in the preset range of the attachment point is obtained, and the driving dynamic time consumption of the distance between the attachment point and the parking lot in the preset range of the attachment point is calculated, so that the parking decision is internalized into a path search variable in the process, and the dynamic time variation in the parking decision is considered; and finally, high-efficiency searching and matching of the multi-stage path are carried out.

Description

Technical Field

[0001] This invention relates to the field of travel route search technology for transportation networks, and in particular to a multi-stage route search method, apparatus, device, and medium based on time-varying road conditions. Background Technology

[0002] In recent years, with the improvement of the integration and intelligence of transportation systems, driving navigation services and route search as its core supporting technology have been widely used. As the core of navigation services, the optimization effect of route search directly affects travel efficiency and user experience.

[0003] At present, driving navigation routes often adopt a passive response strategy of "real-time traffic query + post-correction". This strategy plans the route based on the instantaneous traffic conditions at the departure time and passively corrects it according to the latest traffic conditions during the journey. When selecting parking lots, it first plans the shortest driving route to the vicinity of the destination, and then looks for available parking lots around the end of the route.

[0004] However, current methods for searching travel routes rely solely on the traffic conditions at the departure time as the sole basis for the entire trip planning, ignoring the evolution of road conditions over time. This makes it difficult to capture and respond to such dynamic changes, resulting in the planned "optimal route" often being based on incorrect or outdated premises. This easily leads to vehicles getting stuck in congested areas. Furthermore, treating route planning and parking lot selection as two separate processes artificially ignores the cost coupling and mutual constraints between different stages, ultimately making it difficult to accurately and efficiently search and match multi-stage routes in a dynamic and realistic urban traffic environment. Summary of the Invention

[0005] This invention provides a multi-stage path search method, apparatus, and device based on time-varying road conditions, which can solve the problems existing in the prior art.

[0006] This invention provides a multi-stage path search method based on time-varying road conditions, comprising the following steps: Obtain traffic network map data, parking data, and travel request data for trips to be made; During the travel phase, the travel time is obtained by simulating the journey: Based on traffic network map data, the travel time from the origin to any network node is obtained. Starting from the departure time of the travel request, the dynamic travel time of each road segment is accumulated sequentially along the path node sequence to obtain the dynamic travel time from the origin to any network node. The latitude and longitude of each parking lot in the parking lot data are converted into road network coordinates using UTM coordinate transformation. The nearest road network node is queried using kd-tree spatial index as the attachment point to obtain the distance from the attachment point to the parking lot within the preset range of the attachment point. Based on the real-time number of available parking spaces in each parking lot and the time taken to travel the distance from the attachment point to each parking lot, the parking search time of each parking lot is dynamically obtained using a nonlinear function. A path search algorithm is used to obtain multiple driving paths from the starting point of the trip to the attachment points of each parking lot, forming a candidate set of driving paths. For each driving path in the candidate set, the attachment point corresponding to the driving path is taken as the driving destination. The dynamic driving time from the starting point to any road network node, the time taken from the attachment point to the parking lot, and the parking space search time are accumulated. The driving path with the minimum time consumption is taken as the search path.

[0007] Preferably, obtaining the dynamic travel time from the origin of travel to any road network node includes: The relationship between real-time traffic flow and speed on a road segment is quantified using a piecewise function, expressed as follows: ; in: , All are coefficients, and α represents the congestion index. In this piecewise function, the direction of decreasing traffic speed indicates that the road segment is in a state of smooth traffic, slight congestion, moderate congestion, and severe congestion, respectively. For a continuously changing real-time throughput speed v ,use s Let the total length of the road segment be: ; ; For the passing speed v of the input and output road segments i v j Intersection turning impedance t ij The expression is created as follows: t ij = g ( v i , v j , ...); Combine road segment length with real-time throughput speed v The time taken to obtain the ratio, and the intersection turning impedance t ij , which is the dynamic time consumption from the starting point to any node in the road network.

[0008] Preferably, the process of obtaining the distance and time consumption from the attachment point to the parking lot within a preset range of the attachment point includes: The destination coordinates are used as the coordinate group to be queried. A circular search with a preset radius of r is performed in the pre-built parking lot kd tree to obtain the indexes of all parking lots that meet the conditions, forming a candidate parking lot set. The latitude and longitude of each parking lot in the candidate parking lot set are converted into road network coordinates using UTM coordinates. The nearest neighbor node corresponding to the coordinate is queried and output in the kd-tree of the road network node, and the node is marked as the attachment point of the corresponding parking lot. Add two key-value pairs to the external values ​​of the parking information table, which record the attachment point number and the route length from the attachment point to the parking lot, respectively. The route length is represented by the Manhattan distance between the two, and the time consumption of the segment is calculated based on the average speed of the road segment.

[0009] Preferably, the acquisition of parking search time for each parking lot includes: Set the vacancy rate δ of the total number of parking spaces and the parking search time t. p And design the parking space search time t p The quadratic nonlinear function of the vacancy rate δ is expressed as: ; in: Indicates coefficient; The penalty time t caused by the difficulty in finding parking spaces when the vacancy rate δ is low. p The parking availability of a parking lot is represented by a quadratic nonlinear function, which is used to dynamically calculate the parking search time for each parking lot.

[0010] Preferably, obtaining the driving route with the least time consumption includes: The starting point of the path search is set as the road network node s corresponding to the starting point of the trip, and the ending point is set as the road network node t corresponding to the attachment point of a certain parking lot. Based on the traffic road network map data, a path search is performed from node s to node t to obtain a single feasible path, which is added to the candidate path set. Based on the Yen algorithm, k-1 suboptimal paths from node s to node t are obtained and added to the candidate path set, finally obtaining a driving path candidate set containing k paths. For the k paths in the candidate path set, obtain their total time consumption based on dynamic traffic conditions, determine the time consumption of each of the k paths based on the total time consumption, and sort them in ascending order of priority. Return the paths sorted by priority from the k paths to the traveler.

[0011] Preferably, the total time consumption of the driving route is obtained as follows: C = C car,on + C car,of + C foot +t p ; in: C car,on This represents the dynamic time consumption from the starting point to any node in the road network; C car,of This indicates the time taken to travel the distance from the attachment point to the parking lot; C foot Indicates the time required to find a parking space in the parking lot; t p This indicates the time taken to walk.

[0012] This invention also provides a multi-stage path search device based on time-varying road conditions, comprising: The data module is used to acquire traffic network map data, parking data, and travel request data for travelers. The driving module is used to obtain driving time during the driving phase of a trip by simulating the journey: based on traffic network map data, it obtains the driving time from the starting point of the trip to any network node. Starting from the departure time of the trip request, it accumulates the dynamic travel time of each road segment along the path node sequence to obtain the dynamic driving time from the starting point of the trip to any network node. The parking module is used to convert the latitude and longitude of each parking lot in the parking lot data into road network coordinates using UTM coordinate transformation, and to use the kd-tree spatial index to query the nearest road network node as the attachment point to obtain the distance from the attachment point to the parking lot within the preset range of the attachment point; based on the real-time number of available parking spaces in each parking lot and the time taken to travel the distance from the attachment point to each parking lot, the module uses a nonlinear function to dynamically obtain the parking search time of each parking lot. The route selection module is used to obtain multiple driving routes from the starting point of the trip to the attachment points of each parking lot using a route search algorithm, forming a candidate set of driving routes. For each driving route in the candidate set, the attachment point corresponding to the driving route is taken as the driving destination. The dynamic driving time from the starting point to any road network node, the time taken from the attachment point to the parking lot, and the parking space search time are accumulated. The driving route with the minimum time consumption is taken as the search path.

[0013] This invention also provides an electronic device, including a memory and a processor; The memory is used to store computer programs; When the processor executes the computer program stored in the memory, it implements the steps of the multi-stage path search method based on time-varying traffic conditions as described above.

[0014] This invention also provides a computer-readable storage medium for storing a computer program, which, when executed by a processor, implements the steps of a multi-stage path search method based on time-varying traffic conditions as described above.

[0015] This invention provides a multi-stage path search method, apparatus, and device based on time-varying road conditions, which has the following advantages compared with the prior art: This invention employs a simulated travel approach during the driving phase. Starting from the departure time, it gradually accumulates travel time along a candidate path sequence through simulation, and queries the corresponding road segment impedance in real time at each predicted future time point. This process allows path cost calculation to proactively reflect upcoming traffic condition changes during the journey, fundamentally avoiding congestion caused by delayed road condition information updates. Furthermore, through UTM coordinate transformation and kd-tree spatial indexing technology, discrete parking lot locations are accurately and dynamically associated with specific attachment points in the continuous road network to calculate parking time consumption. This transforms parking lot selection from an additional step after path search is completed into a decision variable that is evaluated synchronously with the driving path and competes with it, achieving a leap from independent optimization of driving and parking to optimal optimization throughout the entire trip, ultimately searching for the optimal path in a dynamic and realistic urban traffic environment. Attached Figure Description

[0016] Figure 1 This is a schematic diagram of a road network map data management method based on a multi-stage path search method for time-varying road conditions provided in an embodiment of the present invention; Figure 2 This is a schematic diagram of the entire driving navigation process provided by an embodiment of the present invention, which is based on a multi-stage path search method for time-varying road conditions. Figure 3 A schematic diagram illustrating a method for calculating the time-varying impedance of a road segment using a multi-stage path search method based on time-varying road conditions, provided in an embodiment of the present invention. Figure 4 This is a schematic diagram of the continuous form of the parking location time function of a multi-stage path search method based on time-varying road conditions provided in an embodiment of the present invention. Figure 5 A schematic diagram illustrating a method for expressing the dynamic availability of parking lots based on a multi-stage path search method for time-varying road conditions, provided in an embodiment of the present invention; Figure 6 A schematic diagram of a parking lot access strategy based on a multi-stage path search method for time-varying traffic conditions, provided in an embodiment of the present invention; Figure 7 A schematic diagram of a road network path search method based on a multi-stage path search method with time-varying road conditions provided in an embodiment of the present invention; Figure 8 This is a schematic diagram of a path organization method for a multi-stage path search method based on time-varying traffic conditions provided in an embodiment of the present invention; Figure 9This is a schematic diagram of a complete multi-stage path search process for a multi-stage path search method based on time-varying road conditions, provided in an embodiment of the present invention. Detailed Implementation

[0017] To make the above-mentioned objects, features, and advantages of the present invention more apparent and understandable, specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings. Many specific details are set forth in the following description to provide a thorough understanding of the present invention. However, the present invention can be practiced in many other ways different from those described herein, and those skilled in the art can make similar modifications without departing from the spirit of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.

[0018] Currently, driving navigation route search still faces the following challenges: (1) In terms of road condition dynamics, the passive response strategy combination of searching for real-time road conditions and then correcting based on the latest road conditions in the later stage fails to capture the evolution of road conditions during the trip in a timely manner during the search stage, and is prone to getting caught in local optimization; (2) In terms of the integration of the entire travel process ("driving-parking-walking"), parking decisions and driving route search are separated from each other, making it difficult to achieve multi-stage global travel time optimization; It can be seen that how to consider time-varying road conditions, systematically improve the accuracy of route time cost estimation, enhance the ability of route search to discover potential better routes, and ultimately achieve time cost optimization of the entire driving navigation travel process has become a problem that needs to be solved in this field.

[0019] Therefore, a path search method is proposed that, compared with existing technologies, can better consider road condition evolution, improve the timeliness of driving time cost estimation, and integrate parking decisions and driving route search to achieve joint optimization of multi-stage driving path time costs under time-varying road conditions. Specifically, it includes: Step 1: Obtain the original traffic network map data, impedance information, parking lot data, and travel request data.

[0020] Step Two: First, perform dynamic traffic information representation: The driving section specifically proposes the formula for calculating the turning impedance function of road segments and intersections used in this invention. Utilizing the concept of "simulated journey", the dynamics of road conditions on the road segments and the complete driving path sequence are comprehensively considered to affect the passage time. Starting from the departure time t0, the time consumption of each road segment is accumulated according to the sequence of path nodes to obtain the arrival time t0 + Δt of each node, and the impedance value of the corresponding time period is queried in real time.

[0021] The parking section, following the approach of expressing the dynamic impedance of road segments in the driving section, represents the dynamic nature of "parking search time" in parking lots. Specifically, a quadratic function is used to characterize the relationship between parking lot vacancy rate and parking search time. First, parking lot location data is obtained and converted to road network coordinates using UTM coordinates. The nearest road network node is then used as the attachment point by querying a node kd-tree. The Manhattan distance from the attachment point to the parking lot is calculated to complete the connection of the parking lot to the road traffic network. A location index is established using the parking lot kd-tree, and a circular search with a preset radius is performed using the destination coordinates as the query point to obtain a set of candidate parking lots, achieving efficient parking lot search.

[0022] Step 3: Dynamic time consumption estimation also includes two parts: transit time consumption estimation and parking location time consumption estimation.

[0023] The specific steps of the dynamic on-the-go time estimation method are as follows: 1) Input the traveler's query time and candidate route. The former is used as the initial value of the current time, and the latter is composed of the node numbers passed through in sequence along the route, forming an array, including the starting point and the ending point; initialize the true total time to 0, and traverse the number groups formed by two adjacent points in the array, repeating step 2); 2) Use the number groups to obtain the corresponding time-varying impedance array in the time-varying impedance table; query the passage time of the road segment corresponding to the number group and add it to the true total time; 3) After the traversal is completed, output the true total time.

[0024] The parking space search time estimate (dynamic) will follow the online driving time estimate (dynamic) and offline driving time estimate (static) and precede the walking time estimate (static). The specific estimation method is similar to the dynamic driving time estimate, that is, the latest time is obtained by using the departure time and the current accumulated time, and the corresponding time cost is queried in the time-varying information table at this time as the parking space search time estimate for the current route.

[0025] Step 4: When performing path search, first obtain the coordinates of the starting point and the destination. Then, for the driving part, obtain multiple feasible paths as a driving path candidate set using a certain method (example: Yen algorithm). After that, perform multi-stage path integration, select feasible parking lots using a certain method (heuristic method), and integrate the "driving-parking-walking" complete path candidate set based on the driving path candidate set. Calculate the total time consumption using the path time consumption estimation strategy, and finally select the recommended path.

[0026] Specifically: As shown in Table 1, the present invention requires the following information from the original traffic network map: network topology, basic network attributes, and dynamic network attributes; in addition, it requires the UTM coordinate information of the node data and time-varying traffic flow data.

[0027] Table 1. Information Categories That Should Be Included in the Original Traffic Network Map This invention processes the information from the original traffic network map into data that can be directly used in subsequent steps. The specific method is as follows: (1) Construct the edge numbering attribute using the connection relationship of the nodes, i.e. (starting node number, ending node number), and add it to the basic attributes of the road network.

[0028] (2) The ratio of edge length and edge free flow velocity is used as the basic impedance value based on the road segment travel time and added to the basic attributes of the road network, while the edge length and edge free flow velocity attributes are removed.

[0029] This invention utilizes a structure to store basic information about nodes and edges, while using a hash table to manage the attributes of nodes and edges for subsequent steps; the management method is as follows: Figure 1 As shown.

[0030] This embodiment requires parking lot data to provide parking lot location information and time-varying parking space information, including latitude and longitude coordinates (which are later converted to UTM-51 format coordinates), total number of parking spaces, and vacancy rate for each time period.

[0031] As shown in Table 2, the present invention requires the following information from the travel request data: travel start time and travel origin and destination UTM-51 coordinates.

[0032] Table 2 Travel Request Attribute Information like Figure 2 As shown, the multi-stage driving navigation in this invention includes the following processes: 1) Driving process: The vehicle departs from the starting point and travels along a certain path on the road network to reach the vicinity of the destination; 2) Parking search process: The vehicle navigates within the parking lot to the vicinity of an available parking space; 3) Parking process: The vehicle begins parking operations near an available parking space; 4) Walking process: The traveler walks from the parking lot to the destination. The driving process, depending on differences in technical implementation, will subsequently be divided into "online driving process" and "offline driving process." Furthermore, since the absolute value and volatility of the time cost of the parking process are extremely small, this invention ignores it.

[0033] This invention argues that the multi-stage characteristics of driving navigation dictate that route search must consider the time cost coupling relationship of each stage. For example, route selection during the driving process directly affects the distance to the driving network and the time to find a parking space, while the dynamic availability of parking lots, in turn, restricts the optimization space of driving routes. Therefore, parking decisions need to be internalized as an endogenous variable of route search to achieve joint optimization of time costs across multiple stages.

[0034] This invention references a time-varying road network traffic flow dataset when estimating the dynamic time consumption of a road segment, and uses a piecewise function to quantify the relationship between real-time traffic flow and speed of the road segment, expressed as: .

[0035] in: , All are coefficients, also known as the congestion index α; in this piecewise function, the direction of decreasing traffic speed indicates that the road segment is in a state of smooth traffic, slight congestion, moderate congestion, and severe congestion, respectively.

[0036] For a continuously changing real-time passing speed v If used s Let represent the total length of the road segment, then: .

[0037] This invention takes into account that the predicted information collected is necessarily cross-sectional data with a certain time interval and is not a continuous variable. Therefore, the above formula needs to be approximated and simplified, and is expressed as: .

[0038] This invention takes into account the convenience of data storage. Since the stored time-varying impedance data inevitably has a certain time interval, it is necessary to classify continuous time points into different time periods, ensuring that the impedance within the same time period is consistent: for the center time t of a discrete time period... j Compared to other central time points, the distance t j All more recent moments are included in this time period; the calculation method is as follows: Figure 3 As shown.

[0039] This invention relates to the throughput speed v of input and output road segments. i v j Intersection turning impedance t ij Establish the following function expression: t ij = g ( v i , v j , ...).

[0040] And it also follows the same form of expression for the impedance of the road segment as described above.

[0041] This invention addresses the penalty time t caused by the difficulty in finding parking spaces when the proportion of available parking spaces to total parking spaces δ (i.e., vacancy rate) is low in the dynamic traffic information representation of the parking section. p(hereinafter referred to as "parking search time") is used to characterize the parking availability of a parking lot; among which, the relationship between δ and parking availability roughly follows the following rules: 1) From a certain threshold δ s Initially, as δ decreases, availability decreases; 2) When δ decreases to a certain extent, parking search time may increase significantly; 3) When δ approaches 0, availability decreases sharply.

[0042] This invention designs t p When considering functions of δ, a curve similar to a quadratic function is used to reflect the availability of parking spaces. This nonlinear function ensures that as δ approaches 0, the function value approaches a finite, relatively large value rather than infinity. The specific function formula is as follows: .

[0043] in: The coefficient is denoted by ; when the number of parking spaces in the parking lot is sufficient, this function can be considered a continuous function, and its graph is as follows. Figure 4 As shown in the figure (a is 1802 and s is 0.458, these values ​​are only used for range reference).

[0044] Referring to the above calculation formula and the previous method for expressing time-varying road condition information, to maintain consistency in information expression, this invention maintains a two-level nested hash table in the computer program when expressing the dynamic availability of parking lots. The foreign key is the parking lot number, and the foreign value contains three key-value pairs, representing the parking lot coordinates, the total number of parking spaces, and the time-varying number of available parking spaces, respectively. Specifically, as follows... Figure 5 As shown; the node coordinate format is the same as that of the road network, and the parking lot coordinates are based on the UTM coordinate system; during the path search process, after obtaining the total number of parking spaces and the number of available parking spaces at a specific location in the time-varying available parking space array, the parking search time when selecting the parking lot can be estimated.

[0045] This invention takes into account the off-network nature of parking lot coordinates and designs a method for integrating them into the road traffic network. After establishing a tree structure (hereinafter referred to as the node kd-tree) using a node table, all parking lots are traversed, and the following steps are repeated: 1) Using the parking lot coordinates as the coordinate group to be queried, find its nearest neighbors in the node kd-tree and return the indices of these nearest neighbors; 2) Query the nearest neighbor node number and output this node number, which is regarded as a necessary node for the vehicle to reach the corresponding parking lot (hereinafter referred to as "attachment point"); 3) Add two key-value pairs to the foreign values ​​of the parking lot information table, recording the attachment point number and the route length from the attachment point to the parking lot, respectively. This length is represented by the Manhattan distance (L1 distance) between the two, such as... Figure 6 As shown.

[0046] In the parking lot search phase of the path search, this invention does not only examine "parking lots near the destination," but also examines a certain number of parking lots near the destination. A circular search strategy is employed during the search. From the aforementioned parking lot information table, a key array (parking lot number array) and a parking lot coordinate array with consistent indices are separated. The latter is used to construct a kd-tree structure (hereinafter referred to as "parking lot kd-tree"). During the path search process, before running the graph search algorithm, the following steps are performed: 1) Using the destination coordinates as the coordinate group to be queried, perform a circular search with a preset radius of r in the parking lot kd-tree to obtain all parking lot indices that meet the conditions; 2) Based on the parking lot index array, query the corresponding parking lot number in the key array and convert the parking lot index array into a parking lot number array.

[0047] The specific steps of the dynamic on-the-go time estimation method of the present invention are as follows: 1) Input the traveler's query time and candidate routes. The former serves as the initial value of the current time, and the latter is represented by an array of node numbers passed sequentially along the route, in the form of a , where represents the start and end points, respectively. Initialize the actual total time consumption to 0, and iterate through the number groups formed by adjacent points in the array, repeating step 2).

[0048] 2) Use the number group to obtain the corresponding time-varying impedance array in the time-varying impedance table; query the passage time of the corresponding road segment of the number group and add it to the actual total time.

[0049] 3) After the traversal is complete, output the actual total time consumption; the parking space search time consumption estimate (dynamic) will follow the online driving time consumption (dynamic) and offline driving time consumption (static), and precede the walking time consumption (static). Its specific estimation method is similar to the dynamic driving time consumption estimation, that is, using the departure time and the current accumulated time consumption to obtain the latest time, and using this time to look up the corresponding time cost in the time-varying information table, as the parking space search time consumption for the current route.

[0050] Based on the above information, this invention proposes a road network path search method as follows: Figure 7As shown, the specific steps for querying the path are as follows: 1) Receive the starting point and ending point coordinates sent by the traveler, which are used as the starting point (hereinafter referred to as "node s") and ending point (hereinafter referred to as "node t") for the path search, respectively; 2) Perform a path search from node s to node t, obtain a single feasible path, and add it to the candidate path set; 3) Based on the principle of the Yen algorithm, obtain k-1 suboptimal paths from node s to node t (k is not less than 1), and add them to the candidate path set; 4) For the k paths in the candidate path set, calculate their actual total time consumption based on dynamic road conditions using the dynamic path time consumption calculation method designed in this invention, and determine the priority of each of the k paths based on this indicator; 5) Return the top n paths (n is not greater than k) in order of priority among the k paths to the traveler.

[0051] This invention ultimately integrates the three significantly different and multi-stage travel processes—driving, parking, and walking—in the context of driving navigation, and designs related heuristic methods to organize the entire path for driving navigation, such as... Figure 8 As shown; its general algorithm flow is as follows: 1) Obtain relevant information from the travel requests of drivers.

[0052] 2) Based on the destination, perform a parking lot search process, obtain a parking lot candidate set, pop up the parking lot number one by one, and enter the loop consisting of steps 2) to 6) below until a certain termination rule is met.

[0053] 3) Use the parking lot number to query the attachment point and perform a route search between the origin and the attachment point.

[0054] 4) If the route is not empty, calculate and record the off-grid driving time based on the route length from the above attachment point to the parking lot; otherwise, proceed to the next loop.

[0055] 5) Based on the departure time and the driving time mentioned above, obtain the time when the vehicle arrives at the parking lot, and then obtain the parking search time based on the availability of the parking lot at this time.

[0056] 6) Using the parking lot coordinates and the destination coordinates, estimate and record the walking time, integrate all the records of the loop, and output them to the result set.

[0057] 7) If the result set is not empty at this time, calculate the total travel time, output the first m records, and terminate the process; otherwise, terminate the process directly.

[0058] Based on the algorithm flow of the above route organization method, the total travel time is... C It should be the time consumed while driving online. C car,on Offline driving time consumption Ccar,of Parking search time C foot and walking time t p The sum is expressed as: C = C car,on + C car,of + C foot + t p .

[0059] The online driving time and parking space search time are obtained using the dynamic time estimation method described above. The parking space search time is based on the current time and the parking lot information table. Specifically, the operation is as follows: the parking lot number is used to query the time-varying available parking space array in the parking lot information table. The current time is divided by the smallest granularity of the time-varying information time interval to obtain the index of the current time in the time-varying array. Based on this, the latest available parking space number is obtained, and the parking space search time at the current time is obtained by combining the calculation formula. The offline driving time is the ratio of the route length from the attachment point to the parking lot to the preset offline driving speed. The walking time is the ratio of the L2 distance between the parking lot and the destination to the preset walking speed.

[0060] The complete multi-stage path search method of this invention can be abstractly represented as a module in a computer system, such as... Figure 9 As shown, the main operating mechanism is as follows: 1) The module receives the latest traffic data and traffic prediction data, and inputs them into the aforementioned road network path search method (hereinafter referred to as "Module A"); 2) The module receives and stores basic parking lot data and parking lot availability prediction data; 3) For each travel request, steps 4) to 11) are repeated; 4) The travel destination information is extracted, parking lot search is performed, and a candidate parking lot set is generated; 5) Before the preset termination rule is met, the target parking lot information is obtained multiple times, and for each target parking lot, the following steps 6) are repeated. 10) Step 6) Use the road connection point corresponding to the target parking lot as the destination of the online path search, and extract the origin information of the trip. After integrating it into a new trip request, input it into module A to obtain a single candidate online path; 7) Calculate the off-network driving time using the target parking lot information; 8) Estimate the parking space search time using the target parking lot information and parking lot availability prediction data; 9) Calculate the walking time using the target parking lot information and the destination information; 10) Organize the complete path and calculate the total time, and add the path to the candidate path set; 11) Prioritize the paths in the candidate path set, obtain m preferred paths and their corresponding path times, and output them.

[0061] The time complexity of the path lookup phase in this invention is O ( mkL ( E + V )) ,in m This represents the number of representative complete paths organized under preset rules. E For the number of nodes, V Let be the number of edges. L The representative length of the path, expressed in terms of the number of path nodes. k This represents the number of shortest paths to be found.

[0062] This invention also establishes an experimental study based on real urban road traffic networks and real travel survey data for a route search method for multi-stage driving navigation, verifying its effectiveness in practical application scenarios; the raw data included in the verification test are as follows: (1) Travel request data.

[0063] The travel request data was generated using travel survey data from Shanghai. Samples with the main travel characteristics of private cars were selected and randomly distributed into different sample sets, each with a duration of 1 hour. Finally, 5,000 samples were selected from each sample set through systematic sampling and filtered. The final data contains a total of 80,000 samples from 16 different time periods.

[0064] (2) Basic data of transportation network.

[0065] The traffic network map data used is Shanghai's road network data, which contains 42,144 road nodes and 100,504 road segments, including necessary road segment attributes and node UTM-51 coordinate information.

[0066] (3) Traffic network impedance data.

[0067] Traffic network impedance data were obtained by combining time-varying traffic flow data based on the MATSim traffic simulation model with the aforementioned dynamic time consumption estimation method for road segments and the intersection analysis method in the Road Capacity Manual (HCM).

[0068] (4) Parking lot data.

[0069] Parking data includes parking lot location data and time-varying parking space information.

[0070] The more precise latitude and longitude coordinates of each parking lot were obtained by web crawling in conjunction with the POI search service of Gaode Map API, and then converted into UTM-51 format coordinates and added to the list.

[0071] The latter is based on the time-varying parking pressure generated near the destination, calculated using data from the aforementioned traffic simulation model; the specific steps are as follows: 1) Set the average total number of parking spaces in public parking lots with reference to the total number of public parking spaces in Shanghai and travel survey data. N = 13, based on Shanghai's average vacancy rate of 0.446, the number of parking spaces to be controlled is n=6; 2) Record the destination POI coordinates at 10-minute intervals, allowing duplicates in the records (duplicates mean that the coordinates are used as the travel destination multiple times within this time interval); 3) For the destination POI coordinates of each time interval, mark the geographically nearest surrounding area. s (This test takes) s =5) There are 5 parking lots, and the number of markers can be accumulated (if a parking lot has multiple markers, it means that the parking lot has been considered as a candidate parking lot multiple times); 4) Integrate all marker data, adjust the outliers according to the empirical rule of 3σ (three-sigma rule of thumb), and standardize it to the range of [0, 1] to characterize the parking pressure λ of a parking lot in a certain time interval, and estimate the corresponding time-varying number of available parking spaces n0, expressed as: n0 = (1 − λ)n.

[0072] The hardware platform used for verification testing of the above data in this invention is a Mac computer equipped with an Apple M4 chip (10 cores and 10 threads), 24GB RAM, and the operating system is macOS Sequoia 15. The algorithm implementation uses Java 17, the Java development environment is IntelliJ IDEA, the source code is compiled into bytecode by the standard JDK 17 compiler, and executed on the corresponding JVM with the default JVM optimization settings.

[0073] The specific experimental steps of the above study include: 1) Establish a road network map based on the aforementioned real urban road traffic network containing necessary road segment attributes; 2) Preprocess the road network map by adding road traffic network map data containing time-varying impedance data and parking lot data containing time-varying availability information; 3) Set up a separate processing process for each travel request sample set; 4) In the processing process corresponding to each sample set, traverse the samples in the sample set and repeat the following steps; 5) Comprehensively consider the availability of multiple parking lots around the destination, search for and store parking lot information; 6) For the selected parking lot set, from a multi-stage perspective, comprehensively consider the entire process of "driving-parking-walking" and the impact of dynamic changes to estimate the time cost of each stage, and execute the complete travel route organization; 7) Calculate the time consumption of the organization results obtained in 6) and output it.

[0074] The cost analysis of the verification test was conducted. The average execution time was taken as the sample mean of the test cases, and the peak memory usage was taken as the maximum amount of memory used by the algorithm during parallel execution on a 10-core CPU. The final average execution time was 12458μs, and the peak memory usage was 9.30GB. It can be seen that after setting additional parking lot multi-target selection and parking lot availability dynamic evaluation mechanisms, the running efficiency of this invention is still within an acceptable range, and the path search strategy is considered to be effective.

[0075] This invention significantly improves the accuracy of path cost estimation by using dynamic impedance characterization and combining it with the concept of "simulated journey" to accumulate time consumption from the departure time and query the impedance of the corresponding time period in real time, effectively avoiding the local optimization problem caused by the lag in dynamic road condition processing. This invention overcomes the problem of the separation between parking decision and driving route search in existing research, internalizing parking decision into a path search variable, and improving the overall joint optimization capability through multi-objective strategy parking lot selection and multi-stage cost measurement. This invention has good usability when processing large-scale road networks and real-time data, and can be integrated into existing navigation systems to provide precise path optimization solutions for smart city travel, enhancing the adaptability and scalability of the method.

[0076] This invention applies the concept of "simulated journey" to driving time estimation. This method abandons the traditional mode of prediction based solely on instantaneous road conditions at the departure time or simple historical averages. Instead, it simulates the real driving process along the candidate path sequence from the departure time, accumulates the time consumption segment by segment, and queries the dynamic impedance value corresponding to the future time period in real time. This can proactively capture the evolution of road conditions during the journey. This invention uses UTM coordinate transformation and kd-tree spatial indexing technology to dynamically associate parking lots with specific nodes (attachment points) in the road network, and seamlessly embeds a dynamic parking space search time model that reflects real-time availability into the cost calculation process.

[0077] The embodiments described above are merely illustrative of several implementations of the present invention, and while the descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the invention patent. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of the present invention, and these all fall within the protection scope of the present invention. Therefore, the protection scope of this invention patent should be determined by the appended claims.

Claims

1. A multi-stage path search method based on time-varying road conditions, characterized in that, Includes the following steps: Obtain traffic network map data, parking data, and travel request data for trips to be made; During the travel phase, the travel time is obtained by simulating the journey: Based on traffic network map data, the travel time from the origin to any network node is obtained. Starting from the departure time of the travel request, the dynamic travel time of each road segment is accumulated sequentially along the path node sequence to obtain the dynamic travel time from the origin to any network node. The latitude and longitude of each parking lot in the parking lot data are converted into road network coordinates using UTM coordinate transformation. The nearest road network node is queried using kd-tree spatial index as the attachment point to obtain the distance from the attachment point to the parking lot within the preset range of the attachment point. Based on the real-time number of available parking spaces in each parking lot and the time taken to travel the distance from the attachment point to each parking lot, the parking search time of each parking lot is dynamically obtained using a nonlinear function. A path search algorithm is used to obtain multiple driving paths from the starting point of the trip to the attachment points of each parking lot, forming a candidate set of driving paths. For each driving path in the candidate set, the attachment point corresponding to the driving path is taken as the driving destination. The dynamic driving time from the starting point to any road network node, the time taken from the attachment point to the parking lot, and the parking space search time are accumulated. The driving path with the minimum time consumption is taken as the search path.

2. The multi-stage path search method based on time-varying road conditions according to claim 1, characterized in that, The process of obtaining the dynamic travel time from the origin of the journey to any road network node includes: The relationship between real-time traffic flow and speed on a road segment is quantified using a piecewise function, expressed as follows: ； in: , All are coefficients, and α represents the congestion index. In this piecewise function, the direction of decreasing traffic speed indicates that the road segment is in a state of smooth traffic, slight congestion, moderate congestion, and severe congestion, respectively. For a continuously changing real-time throughput speed v ,use s Let the total length of the road segment be: ； ； For the passing speed v of the input and output road segments i v j Intersection turning impedance t ij The expression is created as follows: t ij = g ( v i , v j , ...)； Combine road segment length with real-time throughput speed v The time taken to obtain the ratio, and the intersection turning impedance t ij , which is the dynamic time consumption from the starting point to any node in the road network.

3. The multi-stage path search method based on time-varying road conditions according to claim 1, characterized in that, The process of obtaining the distance and time consumption from the attachment point to the parking lot within a preset range of the attachment point includes: The destination coordinates are used as the coordinate group to be queried. A circular search with a preset radius of r is performed in the pre-built parking lot kd tree to obtain the indexes of all parking lots that meet the conditions, forming a candidate parking lot set. The latitude and longitude of each parking lot in the candidate parking lot set are converted into road network coordinates using UTM coordinates. The nearest neighbor node corresponding to the coordinate is queried and output in the kd-tree of the road network node, and the node is marked as the attachment point of the corresponding parking lot. Add two key-value pairs to the external values ​​of the parking information table, which record the attachment point number and the route length from the attachment point to the parking lot, respectively. The route length is represented by the Manhattan distance between the two, and the time consumption of the segment is calculated based on the average speed of the road segment.

4. The multi-stage path search method based on time-varying road conditions according to claim 1, characterized in that, The acquisition of parking space search time for each parking lot includes: Set the vacancy rate δ of the total number of parking spaces and the parking search time t. p And design the parking space search time t p The quadratic nonlinear function of the vacancy rate δ is expressed as: ； in: Indicates coefficient; The penalty time t caused by the difficulty in finding parking spaces when the vacancy rate δ is low. p The parking availability of a parking lot is represented by a quadratic nonlinear function, which dynamically calculates the parking search time for each parking lot.

5. The multi-stage path search method based on time-varying road conditions according to claim 1, characterized in that, The acquisition of the driving route with the least time consumption includes: The starting point of the path search is set as the road network node s corresponding to the starting point of the trip, and the ending point is set as the road network node t corresponding to the attachment point of a certain parking lot. Based on the traffic road network map data, a path search is performed from node s to node t to obtain a single feasible path, which is added to the candidate path set. Based on the Yen algorithm, k-1 suboptimal paths from node s to node t are obtained and added to the candidate path set, finally obtaining a driving path candidate set containing k paths. For the k paths in the candidate path set, obtain their total time consumption based on dynamic traffic conditions, determine the time consumption of each of the k paths based on the total time consumption, and sort them in ascending order of priority. Return the paths sorted by priority from the k paths to the traveler.

6. The multi-stage path search method based on time-varying road conditions according to claim 5, characterized in that, The total time consumption of the driving route is obtained as follows: C = C car,on + C car,of + C foot + t p ； in: C car,on This represents the dynamic time consumption from the starting point to any node in the road network; C car,of This indicates the time taken to travel the distance from the attachment point to the parking lot; C foot Indicates the time required to find a parking space in the parking lot; t p This indicates the time taken to walk.

7. A multi-stage path search device based on time-varying road conditions, characterized in that, include: The data module is used to acquire traffic network map data, parking data, and travel request data for travelers. The driving module is used to obtain driving time during the driving phase of a trip by simulating the journey: based on traffic network map data, it obtains the driving time from the starting point of the trip to any network node. Starting from the departure time of the trip request, it accumulates the dynamic travel time of each road segment along the path node sequence to obtain the dynamic driving time from the starting point of the trip to any network node. The parking module is used to convert the latitude and longitude of each parking lot in the parking lot data into road network coordinates using UTM coordinate transformation, and to use the kd-tree spatial index to query the nearest road network node as the attachment point to obtain the distance from the attachment point to the parking lot within the preset range of the attachment point; based on the real-time number of available parking spaces in each parking lot and the time taken to travel the distance from the attachment point to each parking lot, the module uses a nonlinear function to dynamically obtain the parking search time of each parking lot. The route selection module is used to obtain multiple driving routes from the starting point of the trip to the attachment points of each parking lot using a route search algorithm, forming a candidate set of driving routes. For each driving route in the candidate set, the attachment point corresponding to the driving route is taken as the driving destination. The dynamic driving time from the starting point to any road network node, the time taken from the attachment point to the parking lot, and the parking space search time are accumulated. The driving route with the minimum time consumption is taken as the search path.

8. An electronic device, characterized in that, include: Memory and processor; The memory is used to store computer programs; When the processor executes the computer program stored in the memory, it implements the steps of the multi-stage path search method based on time-varying traffic conditions as described in any one of claims 1 to 6.

9. A computer-readable storage medium, characterized in that, Used to store a computer program, which, when executed by a processor, implements the steps of a multi-stage path search method based on time-varying road conditions as described in any one of claims 1 to 6.

Images