Vehicle low-carbon path planning method and system based on real-time traffic carbon intensity atlas

Abstract

The invention provides a vehicle low-carbon path planning method and system based on a real-time traffic carbon intensity atlas, and relates to the technical field of intelligent traffic, and the method comprises the steps: obtaining the multi-dimensional traffic environment data of a vehicle at a to-be-planned road section, and constructing the traffic carbon intensity atlas of the to-be-planned road section; determining the vehicle type of the vehicle and the correction weight of the carbon emission corresponding to the vehicle type; constructing a road network directed graph according to the traffic carbon intensity map of the to-be-planned road section; and performing low-carbon path planning on the road network directed graph according to the traffic carbon intensity map and the correction weight to obtain an optimal driving path of the to-be-planned road section under the minimum carbon emission. The method is used for solving the problems that in the prior art, a vehicle navigation system cannot achieve low-carbon path guidance, and the error is large when carbon emission estimation is carried out.

Description

Technical Field

[0001] This invention relates to the field of intelligent transportation technology, specifically to a method and system for low-carbon route planning for vehicles based on real-time traffic carbon intensity maps. Background Technology

[0002] With the advancement of global "dual carbon" goals and the upgrading of urban transportation carbon emission control requirements, reducing carbon emissions during vehicle operation has become one of the core research directions in the field of intelligent transportation. However, current mainstream vehicle navigation systems are unable to meet the actual needs of low-carbon travel. Specific problems include: the core optimization dimensions of existing navigation systems are concentrated on "shortest time" or "shortest distance," with only a few systems offering an "economy mode." However, this mode only reduces driving costs by roughly estimating fuel consumption and does not take "lowest carbon emissions" as a direct and core optimization objective, thus failing to fundamentally guide low-carbon routes. Secondly, vehicle carbon emissions are affected by multiple dynamic factors such as real-time traffic flow status, traffic light phase sequence, road conditions, and road network geographical features. Existing technologies lack accurate quantitative models for these traffic environmental factors and rely solely on static "vehicle type-average carbon emissions" databases for estimation, resulting in large errors in carbon emission estimation and failing to reflect the carbon footprint in real-world driving scenarios. The carbon emission mechanisms of different vehicle types, such as gasoline vehicles, hybrid vehicles, and pure electric vehicles, differ significantly. For example, gasoline vehicles rely on fuel combustion efficiency, while pure electric vehicles rely on battery energy consumption and recycling efficiency. Furthermore, the displacement, curb weight, and powertrain parameters of the same vehicle type directly affect carbon emission levels. Existing systems lack differentiated carbon emission models based on individual vehicle parameters, and use a uniform estimation method to calculate carbon emissions for different vehicle types, further reducing the accuracy of low-carbon pathway planning. Summary of the Invention

[0003] In view of this, it is necessary to provide a vehicle low-carbon route planning method and system based on real-time traffic carbon intensity map, so as to solve the problem that the vehicle navigation system in the prior art cannot achieve low-carbon route guidance and has large errors when estimating carbon emissions.

[0004] To address the above problems, this invention provides a vehicle low-carbon route planning method based on real-time traffic carbon intensity maps, comprising the following steps: Acquire multidimensional traffic environment data of vehicles on the road segment to be planned, and construct a traffic carbon intensity map of the road segment to be planned based on the multidimensional traffic environment data; Determine the vehicle type and the corresponding carbon emission correction weight; Construct a directed road network graph based on the traffic carbon intensity map of the road segment to be planned; Based on the traffic carbon intensity map and the corrected weights, low-carbon path planning is performed on the directed graph of the road network to obtain the optimal driving path for the road segment to be planned with the minimum carbon emissions.

[0005] In some embodiments, constructing the traffic carbon intensity map of the road segment to be planned based on the multidimensional traffic environment data includes: The gradient boosting tree model is invoked to predict the multidimensional traffic environment data, and the weight coefficients for each carbon emission influencing factor are obtained. The spatiotemporal correlation features between the road network topology and traffic light phases corresponding to the road segment to be planned are extracted by calling a convolutional neural network model; Based on the spatiotemporal correlation features and the weighting coefficients, the real-time carbon intensity value of the road segment to be planned is calculated to form a traffic carbon intensity map.

[0006] In some embodiments, determining the correction weight for carbon emissions corresponding to the vehicle type includes: Construct a corresponding carbon emission calculation sub-model based on the vehicle type; The carbon emission correction weight is calculated based on the historical estimated carbon emissions of the vehicle along its historical driving routes and the historical actual carbon emissions calculated based on the carbon emission calculation sub-model.

[0007] In some embodiments, constructing a directed road network graph based on the traffic carbon intensity map of the road segment to be planned includes: The road intersections and road segments between the road intersections in the road segment to be planned are taken as nodes and edges, and the driving direction of the road segment is taken as the edge direction of the edge. The real-time carbon intensity value of each road segment is queried based on the traffic carbon intensity map of the road segment to be planned, and the real-time carbon intensity value is used as the edge weight of the edge. Based on the nodes, edges, edge directions, and edge weights, a directed graph of the road network is constructed.

[0008] In some embodiments, the step of performing low-carbon route planning on the directed graph of the road network to obtain the optimal driving route for the road segment to be planned with minimum carbon emissions includes: The starting point and ending point of the vehicle in the planned road segment are determined from the directed road network graph. Starting from the travel origin, Dijkstra's algorithm is invoked to perform shortest path planning in the directed graph of the road network, resulting in sub-road segments between the travel origin and the travel destination. The path weight is calculated for each sub-segment based on the traffic carbon intensity map and the corrected weight, and the path weights are accumulated. When the sub-segment reaches the destination, the optimal driving path for the segment to be planned with the minimum carbon emissions is determined based on the minimum cumulative result of the path weights.

[0009] In some embodiments, calculating the path weight for each sub-segment based on the traffic carbon intensity map and the corrected weight includes: The real-time carbon intensity value of each sub-road segment is retrieved from the traffic carbon intensity map; The real-time carbon intensity value is corrected based on the correction weight to obtain a personalized road segment carbon intensity value; The actual distance of the sub-segment is determined, and the path weight of the sub-segment is calculated based on the actual distance and the personalized segment carbon intensity value.

[0010] In some embodiments, the step of performing low-carbon route planning on the directed graph of the road network to obtain the optimal driving route for the road segment to be planned with minimum carbon emissions includes: The starting point and ending point of the vehicle in the road segment to be planned are determined from the directed graph of the road network, and a heuristic search algorithm is called to search the directed graph of the road network starting from the starting point. For any target node in the search process, the actual cumulative carbon emissions between the starting point of the journey and the target node are calculated based on the traffic carbon intensity map. Calculate the estimated minimum carbon emissions from the target node to the destination, and sum the actual cumulative carbon emissions with the estimated minimum carbon emissions to obtain the estimated value. The target node with the smallest estimated value is selected as the candidate node; When the candidate node is found to be the destination, the optimal driving route for the planned road segment with the minimum carbon emissions is determined based on the candidate node.

[0011] In some embodiments, calculating the estimated minimum carbon emissions from the target node to the travel destination includes: Determine the Euclidean geometric distance from the target node to the destination, and the global minimum carbon intensity value of the directed graph of the road network; The product of the Euclidean geometric distance and the global minimum carbon intensity value is used as the estimated minimum carbon emissions.

[0012] In some embodiments, the method further includes: After the vehicle travels along the optimal driving route, the actual total carbon emissions of the vehicle are obtained; Error samples are constructed based on the planned carbon emissions corresponding to the optimal driving route and the actual total carbon emissions. A gradient boosting tree model, a convolutional neural network model, and a carbon emission calculation sub-model are then trained based on the error samples.

[0013] This invention also provides a vehicle low-carbon route planning system based on real-time traffic carbon intensity maps, the system comprising the following modules: The carbon intensity calculation module is used to acquire multi-dimensional traffic environment data of vehicles on the road segment to be planned, and to construct a traffic carbon intensity map of the road segment to be planned based on the multi-dimensional traffic environment data. The personalized vehicle carbon model module is used to determine the vehicle type and the corresponding correction weight for the vehicle type. The road network construction module is used to construct a directed road network graph based on the traffic carbon intensity map of the road segment to be planned; The low-carbon route planning algorithm module is used to perform low-carbon route planning on the directed graph of the road network based on the traffic carbon intensity map and the modified weight, so as to obtain the optimal driving route of the road segment to be planned with the minimum carbon emissions.

[0014] The present invention also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and running on the processor, wherein the processor executes the computer program to implement the vehicle low-carbon route planning method based on real-time traffic carbon intensity map as described above.

[0015] The present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the vehicle low-carbon route planning method based on real-time traffic carbon intensity map as described above.

[0016] The beneficial effects of adopting the above implementation method are as follows: The vehicle low-carbon route planning method and system based on real-time traffic carbon intensity map provided by this invention, by integrating multi-dimensional traffic environment data of the road segment to be planned, constructs a real-time traffic carbon intensity map, upgrading the carbon emission quantification dimension from individual vehicles to the collaborative level of the road network system and individual vehicles, breaking through the existing static, single-dimensional carbon emission assessment framework. Furthermore, by designing correction weights for vehicle types to correct carbon emissions, personalized carbon emission predictions are adapted to various vehicle type parameters, improving the accuracy of carbon emission estimation and providing a reliable data foundation for low-carbon route planning. Based on this, a directed road network graph is constructed using the traffic carbon intensity map, and low-carbon route planning is performed with the minimum carbon emission as the planning objective, realizing low-carbon route guidance. Attached Figure Description

[0017] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0018] Figure 1 This is a flowchart illustrating the vehicle low-carbon route planning method based on real-time traffic carbon intensity maps provided by the present invention. Figure 2 This is a schematic diagram of the vehicle low-carbon route planning system based on real-time traffic carbon intensity map provided by the present invention. Figure 3 A schematic diagram of an embodiment of the electronic device provided by the present invention. Detailed Implementation

[0019] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of them. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the scope of protection of the present invention.

[0020] In the description of the embodiments of this application, unless otherwise stated, "a plurality of" means two or more.

[0021] In this embodiment of the invention, the terms "comprising" and "having" and any variations thereof are intended to cover non-exclusive inclusion, for example, a process, method, apparatus, product or device that includes a series of steps or modules is not necessarily limited to those steps or modules that are explicitly listed, but may include other steps or modules that are not explicitly listed or that are inherent to such process, method, product or device.

[0022] The naming or numbering of steps in the embodiments of the present invention does not mean that the steps in the method flow must be executed in the time / logical order indicated by the naming or numbering. The execution order of the named or numbered process steps can be changed according to the technical purpose to be achieved, as long as the same or similar technical effect can be achieved.

[0023] In this document, the term "embodiment" means that a particular feature, structure, or characteristic described in connection with an embodiment may be included in at least one embodiment of the invention. The appearance of this phrase in various places throughout the specification does not necessarily refer to the same embodiment, nor is it a mutually exclusive, independent, or alternative embodiment. It will be explicitly and implicitly understood by those skilled in the art that the embodiments described herein can be combined with other embodiments.

[0024] This invention provides a method and system for low-carbon route planning for vehicles based on real-time traffic carbon intensity maps, which will be described below in conjunction with the accompanying drawings.

[0025] The vehicle low-carbon route planning method based on real-time traffic carbon intensity map of the present invention can be applied to the scenario of automatic vehicle navigation. Its execution subject can be in the vehicle-to-everything (V2X) device, vehicle automatic diagnostic system or traffic monitoring system and other devices.

[0026] Figure 1 This is a flowchart illustrating the vehicle low-carbon route planning method based on real-time traffic carbon intensity maps provided in an embodiment of the present invention. Figure 1 As shown, the method is implemented by steps 101 to 104, which will be explained in detail below.

[0027] Step 101: Obtain multi-dimensional traffic environment data of vehicles on the road section to be planned, and construct a traffic carbon intensity map of the road section to be planned based on the multi-dimensional traffic environment data.

[0028] Here, multidimensional traffic environment data includes environmental data on multiple factors influencing carbon emissions. These factors include traffic flow, traffic lights, vehicles themselves, road conditions, and the road network. All of these factors can affect the carbon emissions from vehicle operation. Accordingly, multidimensional traffic environment data specifically includes: traffic flow data, traffic light data, vehicle data, and environmental and road network data. Traffic flow data includes the instantaneous average speed, vehicle density, and headway of each segment of the road network corresponding to the planned road segment. Traffic light data includes the phase sequence and timing scheme of traffic lights at each intersection within the planned road segment. Vehicle data includes Controller Area Network (CAN) data and On-Board Diagnostics (OBD) data. Environmental and road network data includes weather data (sunny / rainy), road network geographical features (dense / rich), and road surface conditions (congested / smooth).

[0029] The traffic carbon intensity map is formed based on the carbon emissions of each sub-segment in the planned road segment. By using real-time multi-dimensional traffic environment data, the real-time carbon intensity value of each corresponding sub-segment can be calculated, thereby determining the carbon emissions of the sub-segment and forming the traffic carbon intensity map.

[0030] In some embodiments, the traffic carbon intensity map of the road segment to be planned can be constructed based on multidimensional traffic environment data through the following steps 1011 to 1013, which are described in detail below.

[0031] Step 1011: Call the gradient boosting tree model to predict the multidimensional traffic environment data and obtain the weight coefficients for each carbon emission influencing factor.

[0032] Step 1012: Call the convolutional neural network model to extract the spatiotemporal correlation features of the road network topology and traffic light phases corresponding to the road segment to be planned; Step 1013: Calculate the real-time carbon intensity value of the road segment to be planned based on the spatiotemporal correlation characteristics and weight coefficients to form a traffic carbon intensity map.

[0033] The traffic carbon intensity map is constructed using a fusion model of Gradient Boosting Tree (GBDT) and Convolutional Neural Network (CNN). The GBDT model is pre-trained using historical carbon emission data. During training, it learns the weight coefficients of each carbon emission influencing factor, i.e., the percentage of influence of each factor on total carbon emissions. First, after predicting multi-dimensional traffic environment data using the GBDT model, the weight coefficients for each carbon emission influencing factor are obtained, such as traffic flow, traffic lights, vehicle components, road conditions, and road network, which are 0.2, 0.3, 0.4, and 0.1 respectively.

[0034] Secondly, a convolutional neural network model is used to extract the spatiotemporal correlation features between the road network topology and traffic light phases corresponding to the road segment to be planned. These spatiotemporal correlation features can correct for the carbon emission transfer effect between road segments, such as the transfer from high-carbon emission road segments to low-carbon emission road segments.

[0035] Finally, the fusion model calculates the real-time carbon intensity value of each road segment in the planned road segment using spatiotemporal correlation features and weighting coefficients, with the unit being gCO2 / km. Since the traffic environment changes in real time during vehicle movement, the calculated real-time carbon intensity value also changes in real time, and the resulting traffic carbon intensity map is also dynamically updated, generally at a frequency of once per minute.

[0036] In this embodiment of the invention, multi-dimensional traffic environment data of the road segment to be planned is incorporated into the low-carbon route planning to construct a real-time traffic carbon intensity map. This upgrades the carbon emission quantification dimension from individual vehicles to the collaborative level of the road network system and individual vehicles, breaking through the existing static, single-dimensional carbon emission assessment framework and filling the technical gap in dynamic road network carbon intensity modeling. In addition, it also provides a data foundation for subsequent low-carbon route planning.

[0037] Step 102: Determine the vehicle type and the corresponding carbon emission correction weight.

[0038] Because vehicles have different types, they are generally divided into gasoline vehicles, pure electric vehicles, and hybrid vehicles. Hybrid vehicles use a combination of gasoline and electricity, or alternate between the two, such as vehicles that use electricity at low speeds and gasoline at high speeds. The carbon emissions of different vehicle types are different, and there are errors between theoretical calculations and actual carbon emissions. Furthermore, the errors in the calculated carbon emissions also vary depending on the vehicle type. To address this difference, this invention proposes a correction weight to adjust the calculated carbon emissions for different vehicle types.

[0039] In some embodiments, determining the correction weight for carbon emissions corresponding to the vehicle type can be achieved in the following ways, as detailed below.

[0040] First, a corresponding carbon emission calculation sub-model is constructed based on the vehicle type. Then, the carbon emission correction weight is calculated based on the historical estimated carbon emissions of the vehicle on its historical driving routes and the historical actual carbon emissions calculated based on the carbon emission calculation sub-model.

[0041] The carbon emission calculation sub-model is used to calculate the carbon emissions of a corresponding type of vehicle during operation. Depending on the vehicle type, the carbon emission calculation sub-model includes a gasoline vehicle model, a pure electric vehicle model, and a hybrid vehicle model.

[0042] The gasoline vehicle model focuses on instantaneous fuel consumption, combined with engine speed, road slope, and wind speed. Carbon emissions are calculated using the formula "carbon emission = instantaneous fuel consumption × fuel carbon emission coefficient × slope correction coefficient × wind speed correction coefficient". The parameters of the formula can be calculated or verified based on specific vehicle parameters and driving data.

[0043] The pure electric vehicle model takes instantaneous power consumption as the core, and combines the remaining battery power (State of Charge, SOC), ambient temperature, and road resistance. It is calculated using the formula "carbon emissions = instantaneous power consumption × regional power carbon emission coefficient × temperature correction coefficient × road resistance correction coefficient". The regional power carbon emission coefficient is updated in real time according to the local power grid structure, and the other parameters can be calculated or verified based on specific vehicle parameters and driving data.

[0044] The hybrid vehicle model dynamically switches between a fuel vehicle model and a pure electric vehicle model based on the vehicle's power switching logic (such as using electricity at low speeds and fuel at high speeds). The corresponding carbon emissions are then calculated cumulatively based on the model, and the energy consumption loss coefficient during the power switching process is added to correct the cumulative calculation results.

[0045] The correction weight is determined by combining the vehicle's historical data. Based on the vehicle's historical driving path, the corresponding carbon emission calculation sub-model is invoked to calculate the historical actual carbon emissions. Then, based on the vehicle's historical estimated carbon emissions along the historical driving path, the ratio of the estimate to the actual emissions is calculated. This ratio is the correction weight for the vehicle's carbon emissions for that vehicle type. Specifically, the correction weight is a vehicle-specific correction coefficient used to eliminate the error between the estimated and actual carbon emissions. Different vehicle types have different correction weights.

[0046] In this embodiment of the invention, a carbon emission calculation sub-model is constructed based on the vehicle type, thereby calculating the correction weight of carbon emissions and correcting the carbon emission estimate. This eliminates the impact of differences in carbon emission mechanisms among different types of vehicles on carbon emission levels, providing a reliable data foundation for low-carbon path planning.

[0047] Step 103: Construct a directed road network graph based on the traffic carbon intensity map of the road segment to be planned.

[0048] Here, route planning requires constructing a corresponding directed graph of the road network based on the road network of the road segment to be planned. This allows for route planning with low carbon emissions as the planning objective on the directed graph. Specifically, road intersections and road segments between intersections in the road segment to be planned can be abstracted as nodes and edges. The road segment length or the real-time carbon intensity value calculated in the above embodiment can be used as the edge weight to form a directed graph of the road network.

[0049] In some embodiments, constructing a directed road network graph based on the traffic carbon intensity map of the road segment to be planned can be achieved in the following ways, as detailed below.

[0050] First, the road intersections and the road segments between them in the road segment to be planned are designated as nodes and edges, respectively, with the travel direction of the road segment as the direction of the edge. The road segment to be planned generally includes a starting point and a ending point, and there may be multiple paths between these two points. These paths include road intersections and the road segments between them. Let the road intersections be designated as nodes, and the set of nodes be denoted as V. Then, the road segments between the road intersections are the edges connecting the nodes, denoted as e(i, j), representing the road segment from node i to node j, and the set of edges is denoted as E.

[0051] Furthermore, the real-time carbon intensity value of each road segment is queried from the traffic carbon intensity map of the road segment to be planned, and this real-time carbon intensity value is used as the edge weight. Since the real-time carbon intensity value of each road segment in the road segment to be planned has already been calculated when constructing the traffic carbon intensity map, this embodiment of the invention assigns the real-time carbon intensity value corresponding to the road segment as the edge weight by querying the traffic carbon intensity map. The edge weight is denoted as C(i, j), representing the real-time carbon intensity value of the road segment from node i to node j. Each road segment has a travel direction, which can be used as the edge direction.

[0052] Finally, a directed graph of the road network is constructed based on nodes, edges, edge directions, and edge weights. Here, edge directions and edge weights are assigned to each edge, thus abstracting the road network of the road segment to be planned into a directed graph, denoted as G=(V,E).

[0053] In this embodiment of the invention, when constructing a directed graph of a road network, in order to facilitate path planning with low carbon as the goal, the real-time carbon intensity value of each road segment to be planned is used as the edge weight. This links path planning with the dynamic traffic environment and the carbon emissions of the vehicle itself, rather than the static distance and time in traditional technologies. This solves the technical problem that environmental protection goals cannot be directly optimized when planning vehicle navigation paths.

[0054] Step 104: Based on the traffic carbon intensity map and the corrected weights, perform low-carbon path planning on the directed graph of the road network to obtain the optimal driving path for the road segment to be planned with the minimum carbon emissions.

[0055] After constructing the directed graph of the road network, the theoretical carbon emissions of vehicles after each planning step can be calculated in real time based on the traffic carbon intensity map and the corrected weights during low-carbon route planning. After calculating the theoretical carbon emissions through the traffic carbon intensity map, the theoretical carbon emissions are corrected using the corrected weights, and the corrected carbon emissions are accumulated in each planning step. Finally, under the condition of achieving the minimum carbon emissions, the optimal travel path between the starting point and the ending point of the planned road segment is determined.

[0056] In this embodiment of the invention, the path planning problem is still abstracted as finding the shortest path in a weighted directed graph. The key difference is that this embodiment sets the edge weight of each edge in the graph to a real-time carbon intensity value, rather than "distance" or "time." This real-time carbon intensity value is dynamically calculated and updated in real time; that is, the traffic carbon intensity map is also updated in real time during path planning. This low-carbon path planning can be implemented using path planning algorithms, such as Dijkstra's algorithm or heuristic search algorithms.

[0057] In some embodiments, low-carbon path planning is performed on the directed graph of the road network to obtain the optimal driving path of the road segment to be planned with the minimum carbon emissions. This can be achieved through the following steps 1041A to 1044A, which are described in detail below.

[0058] Step 1041A: Determine the starting point and ending point of the vehicle's journey on the planned road segment from the directed road network graph.

[0059] Here, low-carbon route planning is implemented using an improved Dijkstra algorithm. During initialization, Dijkstra's algorithm first determines the starting and ending points of the vehicle's journey on the planned route segment from the directed graph of the road network, denoted as S and D respectively. It maintains two node attributes for each vertex v in the directed graph. The first attribute, denoted as minCarbon[v], represents the currently known minimum cumulative carbon emissions from the starting point S to the fixed point v. Initially, minCarbon[S] = 0, and other vertices are set to infinity (∞). The second attribute, denoted as precursor[v], represents the previous vertex visited when obtaining minCarbon[v], used for backtracking. Furthermore, a priority queue is created as a min-heap to store vertices v and minCarbon[v] during the planning process. At the start of planning, the starting point S and minCarbon[S] = 0 are enqueued.

[0060] Step 1042A: Starting from the starting point of the journey, Dijkstra's algorithm is called to perform shortest path planning in the directed graph of the road network to obtain the sub-segment between the starting point and the ending point of the journey.

[0061] Step 1043A: Calculate the path weight for each sub-segment based on the traffic carbon intensity map and the corrected weight, and accumulate the path weights.

[0062] Step 1044A: When the sub-segment reaches the destination, determine the optimal driving path for the planned segment with the minimum carbon emissions based on the minimum cumulative result of the path weights.

[0063] Here, when the Dijkstra algorithm starts low-carbon path planning after initialization, it starts searching in the road network directed graph from the starting point S of travel, traversing the neighbor nodes u of the starting point S of travel and the subsequent neighbor nodes v corresponding to u, forming various sub-sections, that is, e(u, v). When forming the sub-sections, the traffic carbon intensity map is queried in real time to determine the real-time carbon intensity value c(u, v) of each sub-section, and then the real-time carbon intensity value obtained by query is corrected based on the correction weight, so as to form the path weight corresponding to the sub-section. In each path planning, the goal is to minimize the path weight, and the calculated path weights are accumulated to update minCarbon[v] until, in a certain search, the neighbor node v is searched as the end point D of travel, the search ends, and minCarbon[v] is returned.

[0064] Specifically, in the improved Dijkstra algorithm, the starting point S of travel and the end point D of travel are input into the algorithm, and two attributes, namely minCarbon[v] and predecessor[v], are maintained for each vertex v in the road network directed graph. Then, a priority queue (min heap) is created, and the starting point S is queued with its minCarbon[S]=0, and the search is carried out through loop iteration.

[0065] The loop condition here is that the priority queue is not empty. In each loop search, the vertex u with the current minimum minCarbon is taken out and removed from the priority queue. If it is judged that u is the end point D of travel, the loop is jumped out, and the final path is generated by backtracking through the record of predecessor, that is, the optimal travel path with the minimum carbon emission.

[0066] If it is judged that u is not the end point D of travel, continue to search and traverse all neighbor vertices v of vertex u, that is, continue to search for the edge e(u, v). For each edge e(u, v) in the search process, the current carbon intensity value is obtained through querying the traffic carbon intensity map and the correction calculation through the correction weight as the path weight, denoted as C(u, v). In the cumulative calculation, the minimum cumulative result is denoted as alternative, where alternative = minCarbon[u] + C(u, v). This represents the cumulative carbon emission of a new path searched through u to reach v. In path decision-making, if alternative < minCarbon[v], then update minCarbon[v]=alternative and update predecessor[v]=u for backtracking, and thus a round of loop search ends. In the next round of iterative training search, vertex v and the updated minCarbon[v] value are put into the priority queue until the searched vertex v is the end point D of travel, and the loop search ends.

[0067] After the loop search is completed, starting from the destination D, the predecessor nodes are traced back sequentially according to the predecessor array until the starting point S, thus generating a complete vertex sequence P = {S, ..., D} as the final path, which is the optimal driving path with the minimum carbon emissions.

[0068] In real-world scenarios, the final path P can be mapped back to the actual road where the vehicle is located, serving as a candidate solution for low-carbon navigation. The corresponding carbon emission minCarbon[D] can be displayed to the vehicle user, while the carbon emission of traditional navigation solutions can be compared to enhance the user's perception.

[0069] In this embodiment of the invention, based on a directed graph of a road network, an improved Dijkstra algorithm is called to perform path search, establishing a path planning system with "lowest carbon emissions" as the direct objective, overcoming the shortcomings of the traditional path planning framework centered on "time / distance".

[0070] Furthermore, in some embodiments, the path weight for each sub-segment is calculated based on the traffic carbon intensity map and the adjusted weights. This can be achieved in the following ways, which are explained in detail below.

[0071] First, the real-time carbon intensity value of each sub-segment is retrieved from the traffic carbon intensity map. Since the traffic carbon intensity map is calculated based on real-time multidimensional traffic environment data and is constantly updated, when the Dijkstra algorithm searches for a node that forms a sub-segment e(u, v) in each iteration, the carbon intensity value I(u, v) of the corresponding sub-segment can be retrieved by querying the traffic carbon intensity map.

[0072] Then, the real-time carbon intensity value is corrected based on the correction weight to obtain a personalized road segment carbon intensity value. The correction weight is determined by the vehicle type and denoted as K_vehicle. For fuel vehicles, K_vehicle is generally greater than 1, while for pure electric vehicles, K_vehicle≈0 is taken under the background of green electricity. The personalized road segment carbon intensity obtained by correction calculation is denoted as I'(i,j) = I(i,j)*K_vehicle.

[0073] Furthermore, the actual distance of the sub-segment is determined, and the path weight of the sub-segment is calculated based on the actual distance and the personalized carbon intensity value of the sub-segment. Here, the actual distance of the sub-segment e(u,v) can be obtained, denoted as L(i,j), and then multiplied with the personalized carbon intensity of the sub-segment to finally obtain the path weight of the sub-segment, denoted as C(u,v) = I'(i,j) * L(i,j).

[0074] In this embodiment of the invention, during the route planning process, the aforementioned traffic carbon intensity map and the corrected weights are used to calculate the carbon intensity value, thereby forming the route weights of the corresponding sub-segments, which effectively improves the accuracy of low-carbon route planning.

[0075] In some embodiments, low-carbon path planning is performed on the directed graph of the road network to obtain the optimal driving path of the road segment to be planned with the minimum carbon emissions. This can also be achieved through the following steps 1041B to 1044B, which are described in detail below.

[0076] Step 1041B: Determine the starting point and ending point of the vehicle's journey on the road segment to be planned from the directed graph of the road network, and start from the starting point to call the heuristic search algorithm to search in the directed graph of the road network.

[0077] Here, we first determine the starting point and ending point of the vehicle's journey on the planned road segment from the directed graph of the road network, denoted as S and D respectively. The heuristic search algorithm can be the A* algorithm, which guides the search direction through a heuristic function and is particularly suitable for large-scale urban road networks. The core idea is to use an evaluation function f(n) to determine the nodes to be expanded in the next iteration. Specifically, the evaluation function f(n) is the sum of the cost function g(n) and the heuristic function h(n). Here, we define a cost function g(n) and a heuristic function h(n), where n is the target node, representing any search node between the starting point S and the ending point D in the heuristic search algorithm. The cost function g(n) represents the actual cumulative carbon emissions from the starting point S to the target node n, which is obtained by the cumulative path weight C(i,j) of each road segment actually traversed from the starting point S to the target node n. The heuristic function h(n) is the estimated minimum carbon emissions from the target node n to the destination D. This value is acceptable, meaning it cannot overestimate the actual minimum carbon cost from the target node n to the destination D.

[0078] In low-carbon route planning, starting from the starting point of the journey, a heuristic search algorithm is invoked to search the directed graph of the road network to obtain multiple target nodes.

[0079] Step 1042B: For any target node in the search process, calculate the actual cumulative carbon emissions between the starting point and the target node based on the traffic carbon intensity map.

[0080] Here, the heuristic search algorithm searches for a target node in each iteration. For any target node n found, it calculates the actual cumulative carbon emissions from the starting point S to the target node n based on the traffic carbon intensity map, which is the calculation of the cost function g(n). Specifically, it first counts the sub-segments traversed between the starting point S and the target node n, then calculates the path weights of each sub-segment and accumulates them. The calculation method can be referred to step 1043A above, and will not be repeated here.

[0081] Step 1043B: Calculate the estimated minimum carbon emissions from the target node to the driving end point, and sum the actual cumulative carbon emissions with the estimated minimum carbon emissions to obtain the estimated value.

[0082] When determining the target node n, it is also necessary to calculate the estimated minimum carbon emissions from the target node n to the destination D, which is to calculate the heuristic function h(n). Then, the actual cumulative carbon emissions calculated by the cost function g(n) are summed with the estimated minimum carbon emissions h(n) to obtain the evaluation function f(n), which is used as the estimated value of the target node n.

[0083] In some embodiments, the estimated minimum carbon emissions from the target node to the destination can be calculated in the following ways, as detailed below.

[0084] First, determine the Euclidean geometric distance from the target node to the destination and the global minimum carbon intensity value of the directed graph of the road network. Then, use the product of the Euclidean geometric distance and the global minimum carbon intensity value as the estimated minimum carbon emissions.

[0085] Here, the Euclidean distance from the target node n to the destination D is calculated, denoted as L_air(n, D). Then, the global minimum carbon intensity value is obtained. This global minimum carbon intensity value is the lowest carbon intensity value I_min in the entire road network of the road segment to be planned (e.g., the lowest carbon intensity of an expressway during off-peak hours, or the optimal carbon intensity for an electric vehicle). Finally, the product of the Euclidean distance and the global minimum carbon intensity value is used as the estimated minimum carbon emission, denoted as h(n), expressed as h(n) = L_air(n, D) × I_min. Here, the lowest carbon intensity value I_min is not a value obtained through complex real-time calculations, but rather a pre-calculated theoretical lower limit based on historical data and ideal scenarios. Its purpose is to provide a reliable and acceptable benchmark for the heuristic function h(n) of the A* algorithm.

[0086] In some embodiments, the global minimum carbon intensity value I_min in the heuristic function h(n) is a constant pre-calculated through the following steps: First, driving segments that meet ideal driving conditions are selected from historical traffic databases or standard vehicle operating condition tests. Ideal driving conditions include: smooth and constant speed, gentle driving, zero gradient, and no extreme weather. Next, based on the energy consumption data of the driving segments, the carbon intensity value of each driving segment is calculated using the corresponding carbon emission conversion coefficient according to the vehicle's energy type. Finally, all calculated carbon intensity values ​​are statistically analyzed, and the minimum value or the 5th percentile is taken as the global minimum carbon intensity value I_min. Here, the corresponding global minimum carbon intensity value can be calculated and stored separately for different vehicle types, which can be denoted as I_min_type. I_min is determined as a fixed constant (or a typed constant lookup table), pre-stored in a configuration file, and can be called when the path planning algorithm is invoked to calculate the heuristic function h(n). The carbon emission conversion coefficient is related to the vehicle type. For gasoline vehicles, fuel consumption (L / 100km) is converted into carbon intensity using a fixed carbon emission coefficient (e.g., 2.3kgCO2 / L gasoline). Specifically, carbon intensity (gCO2 / km) = fuel consumption (L / 100km) × carbon emission coefficient (gCO2 / L) / 100. For electric vehicles, electricity consumption (kWh / 100km) is converted into carbon intensity by multiplying the average or marginal carbon emission factor of the current power grid (e.g., 0.5kgCO2 / kWh). Carbon intensity (gCO2 / km) = electricity consumption (kWh / 100km) × power grid carbon factor (gCO2 / kWh) / 100.

[0087] In other embodiments, the estimated minimum carbon emissions from the target node to the destination can be calculated by pre-calculating a road network with a fixed minimum carbon intensity as the weight. On this road network, Dijkstra's algorithm is used to calculate the "ideal" minimum carbon cost from the target node n to the destination D, which is then used as the estimated minimum carbon emissions h(n). The specific calculation steps are as follows: (1) Constructing a fixed-weight carbon cost map: Based on historical ideal driving data, each road segment in the actual road network is assigned a fixed carbon intensity value per unit distance, and then the total carbon cost of each road segment is calculated as a fixed weight to form a static carbon cost map G_fixed; (2) Calculate the theoretical minimum carbon cost: On the carbon cost graph G_fixed, calculate the shortest path cost from the target node n to the destination D, and use this cost as the heuristic function h(n).

[0088] The shortest path cost can be calculated and stored in advance through preprocessing, or calculated in real time when a path planning request is received. The specific steps are as follows: Step 1: Construct a fixed-weight carbon cost graph and define the ideal carbon intensity: Assign a fixed carbon intensity value I_ideal(i,j) to each road segment e(i,j) in the road network. I_ideal(i,j) is determined by taking the lowest observed carbon intensity of the road segment under historical ideal conditions (smooth traffic, constant speed, good weather), or the theoretical carbon intensity value of the road segment at the economic speed, or by assigning a standardized minimum carbon intensity to similar roads based on road grade and design speed. The fixed edge weights are calculated by calculating the fixed carbon cost weight for each edge e(i,j): W_fixed(i,j) = I_ideal(i,j) * L(i,j), where L(i,j) is the road segment length. Further, generate a fixed-weight graph using the original road network topology, but replacing the weights of all edges with W_fixed(i,j), resulting in a static, idealized carbon cost graph G_fixed.

[0089] The second step is preprocessing and heuristic value calculation. First, preprocessing (offline calculation) is performed. Since the graph G_fixed is static, the shortest path cost from all nodes to common destination locations (such as city centers, transportation hubs, and popular destinations) can be pre-calculated. This can be achieved by running Dijkstra's algorithm backwards from each important destination, storing the results as a lookup table. Then, h(n) is queried online. When planning a path for destination D, if D is one of the preprocessed destinations, h(n) is directly retrieved from the table as the shortest carbon cost (n→D) from target node n to destination D. If D is a new destination (i.e., a non-preprocessed destination), the reverse Dijkstra's algorithm is run on the G_fixed graph starting from D, calculating h(n) for all target nodes in real time.

[0090] In the main loop of the heuristic algorithm, for each extended target node n, its heuristic function h(n) is no longer a simple geometric estimate, but the shortest path carbon cost from the target node n to the destination D on the G_fixed graph.

[0091] In this embodiment of the invention, the estimated minimum carbon emissions from the target node to the destination are calculated using Euclidean geometric distance and the global minimum carbon intensity value of the directed graph of the road network. This is used as a heuristic function for heuristic path search, rather than the distance or time commonly used in the traditional A* algorithm. This ensures that the heuristic algorithm always searches in the direction of "lowest total carbon emission potential".

[0092] Step 1044B: Select the target node with the smallest estimated value as the candidate node.

[0093] Step 1045B: When a candidate node is found to be the destination, determine the optimal driving path for the planned road segment with the minimum carbon emissions based on the candidate node.

[0094] Here, after calculating the estimated value of the target node, the target node with the smallest estimated value is selected as a candidate node, ending the current iteration. In the next iteration, the neighboring nodes of the candidate node are searched again as new target nodes, and so on. This process continues until a candidate node is found as the destination, indicating that the search goal has been achieved. Based on the determined candidate nodes, the corresponding road segment can be determined, thus forming the final path, which serves as the optimal driving path for the planned road segment with the minimum carbon emissions.

[0095] Specifically, in the heuristic search algorithm, the starting point S and the ending point D are input into the algorithm. For each vertex v in the directed graph of the road network, two attributes are maintained, denoted as gScore[v] and fScore[v]. gScore[v] represents the minimum actual cumulative carbon emissions from the starting point S to the target node v. During algorithm initialization, gScore[S] = 0, and all other nodes are set to ∞. fScore[v] represents the estimated value for the target node v, specifically fScore[v] = gScore[v] + h(v), where h(v) is the heuristic function, representing the estimated minimum carbon emissions. During algorithm initialization, fScore[S] = h(S), and all other nodes are set to ∞. Then, a priority queue is created, storing fScore values ​​in ascending order. Initially, the starting point S and gScore[S] = 0 are added to the queue.

[0096] The loop condition here remains that the priority queue is not empty. During each loop search, the vertex u with the smallest current gScore is taken from the priority queue and removed. If it is determined that u is the destination D, the loop is exited, and the final path, i.e., the optimal driving path with the minimum carbon emissions, is generated by backtracking.

[0097] If it is determined that u is not the driving end point D, continue to search and traverse all neighbor vertices v of vertex u, that is, continue to search for the edge e(u, v). For each edge e(u, v) in the search process, obtain the current carbon intensity value as the path weight, denoted as C(u, v), by querying the traffic carbon intensity map and through the corrected calculation of the corrected weight. When calculating the cumulative cost function, denote tentative_gScore = gScore[u] + C(u, v). This represents the actual cumulative carbon emissions of a new path searched through u to v. When making a path decision, if tentative_gScore < gScore[v], then update gScore[v] = tentative_gScore, fScore[v] = gScore[v] + h(v), and record the predecessor node of v as u for backtracking. Thus, one round of cyclic search ends. In the next round of iterative training search, if v is not in the priority queue, add v and the updated gScore[v] to the priority queue, put the vertex v and the updated gScore[v] into the priority queue until the searched vertex v is the driving end point D, and the cyclic search ends.

[0098] After the cyclic search ends, starting from the driving end point D, perform backtracking according to the recorded predecessor nodes until the driving starting point S, so as to generate a complete vertex sequence P = {S,..., D} as the final path, that is, the optimal driving path with the minimum carbon emissions.

[0099] In the actual scenario, the final path P can be mapped back to the actual road where the vehicle is located as a candidate solution for the vehicle's low-carbon navigation driving, and display the corresponding carbon emissions gScore[D] to the vehicle user. At the same time, compare the carbon emissions of the traditional navigation solution to enhance the vehicle user's perception.

[0100] In the embodiment of the present invention, in the heuristic search algorithm, the search direction is guided by designing a heuristic function. The heuristic function is designed as the estimated carbon emissions from the current target node to the driving end point, combined with the real-time carbon intensity map of the end area, to reduce the search for invalid paths, which can significantly improve the search efficiency of the algorithm and is especially applicable to large-scale urban road networks.

[0101] In step 104 of the above embodiment, low-carbon route planning on the directed graph of the road network is performed according to a low-carbon priority mode, with the minimum total carbon emissions as the sole objective. This yields the optimal driving path for the planned road segment with the minimum carbon emissions. This path prioritizes road segments with low carbon intensity and is suitable for users with extremely high low-carbon requirements, such as corporate fleet carbon emission assessment scenarios. However, it has some limitations and cannot meet the needs of various vehicle users. Based on this, the embodiments of the present invention also provide two other route planning modes: a carbon-time balance mode and a carbon-time-distance balance mode, which will be described in detail below.

[0102] First, both the carbon-time equilibrium mode and the carbon-time distance equilibrium mode are implemented using the improved Dijkstra algorithm or heuristic algorithm mentioned above. The difference lies in that the carbon-time equilibrium mode does not solely aim to minimize total carbon emissions, but rather balances carbon emissions and travel time. After calculating the path weight for each sub-segment based on the traffic carbon intensity map and corrected weights, a time weight is introduced for recalculation. The path weight of each sub-segment is denoted as C(i,j), and a preset time weight T(i,j) is introduced. Based on the vehicle user's emphasis on low carbon emissions and travel efficiency, a carbon path weight a and a time path weight b are designed, both set to 0.5 when considering both. Thus, the final path weight of the sub-segment is expressed as a·C(i,j) + b·T(i,j), which is then used in the cumulative calculation to determine the optimal travel path for the planned segment with the minimum carbon emissions. Therefore, by designing the carbon-time equilibrium mode, a two-dimensional optimization is achieved, balancing low carbon emissions and travel efficiency, making it suitable for daily commuters.

[0103] The carbon-time-distance balancing model, after calculating the path weight for each sub-segment based on the traffic carbon intensity map and adjusted weights, recalculates by introducing time and distance weights. The path weight of each sub-segment is denoted as C(i,j). Preset time weights T(i,j) and distance weights D(i,j) are introduced. Based on vehicle users' priorities regarding low carbon emissions, travel efficiency, and travel costs, carbon path weights a, time path weights b, and distance path weights c are designed, taking one-third of each when all three are considered. Therefore, the final path weight of the sub-segment is expressed as a·C(i,j) + b·T(i,j) + c·D(i,j), which is then used in the cumulative calculation to determine the optimal travel path for the planned road segment with the minimum carbon emissions. Thus, by designing the carbon-time-distance balancing model, three-dimensional optimization can be achieved, balancing low carbon emissions, travel efficiency, and travel costs, making it suitable for users sensitive to travel costs.

[0104] In practical applications, when vehicle users input their starting point S, destination, and route mode preferences (e.g., low-carbon priority mode, carbon-time balance mode, or carbon-time-distance balance mode) via in-vehicle terminal or mobile app, the method of this invention and the in-vehicle equipment will return the planned route, estimated total carbon emissions, estimated time, and estimated energy consumption in real time. During navigation, if the carbon intensity of the road ahead changes significantly due to factors such as congestion or traffic light changes, the route will be recalculated in real time and an update notification will be pushed. Users can choose whether to switch to the optimal driving route.

[0105] In some embodiments, once the optimal driving route for the planned road segment with the minimum carbon emissions is determined, vehicle users can choose the optimal driving route to drive, and after the vehicle drives along the optimal driving route, the vehicle's actual total carbon emissions are obtained.

[0106] Here, after the vehicle travels along the optimal driving path, data such as instantaneous fuel or electricity consumption, driving trajectory, and dwell time during the actual driving process can be obtained through the OBD interface, vehicle network, or other on-board devices to calculate the vehicle's actual total carbon emissions.

[0107] Then, error samples are constructed based on the planned carbon emissions corresponding to the optimal driving route and the actual total carbon emissions. Based on the error samples, a gradient boosting tree model, a convolutional neural network model, and a carbon emission calculation sub-model are trained.

[0108] Here, the planned carbon emissions minCarbon[D] or gScore[D] corresponding to the optimal driving path in the above embodiments are compared with the actual total carbon emissions. The difference between the two is compared, and the error rate is calculated. These error data are then used as training samples to retrain the gradient boosting tree model, the convolutional neural network model, and the carbon emission calculation sub-model. The weight coefficients of various carbon emission influencing factors are adjusted, and the weights are optimized. After each low-carbon route planning, error data is collected as training samples to retrain the model, improving the accuracy of the next low-carbon route planning.

[0109] The following describes the retraining process of the Gradient Boosting Tree (GBDT) model. GBDT is an ensemble model that corrects previous errors by iteratively training a series of decision trees. New data (real carbon emission intensity values) is collected periodically (weekly / monthly), and a new model is retrained on the full dataset, then used to replace the online model. An online evolution mechanism for the GBDT model based on real vehicle feedback data is established, enabling the carbon intensity prediction model to maintain high accuracy over time and with changes in traffic patterns.

[0110] Data preparation includes: Features (X): Same as the initial training, including specific indicators such as temperature, wind speed, vehicle type, vehicle weight, average vehicle speed, speed variance, congestion index, traffic light waiting time, and road gradient. New label (y): Uses actual carbon emissions per unit distance (gCO2 / km), calculated from the total carbon emissions measured by OBD / driving distance.

[0111] The loss function uses mean squared error (MSE) or root mean squared error (MRSE) as the loss function Loss, expressed as Loss = Mean( ), The values ​​represent the predictions of the GBDT model. In each iteration, the learning objective of the new decision tree is to fit the negative gradient predicted by the current model, that is, to approximate the residual y - y_pred. The retraining process involves merging the newly collected (X, y) data with the historical training data and loading the current production GBDT model as the pre-trained model. Using the merged dataset, further training is performed on the pre-trained model. A small learning rate (e.g., 0.05-0.1) is set to allow the model to "fine-tune" rather than "forget". The performance of the new model is evaluated on an independent validation set to ensure that MSE / RMSE decreases until convergence.

[0112] The following describes the retraining process of a Convolutional Neural Network (CNN) model. Here, carbon intensity prediction is modeled as a spatiotemporal image regression problem, and the feature extraction capabilities of CNNs are utilized to achieve automatic calibration of model parameters through closed-loop feedback. Offline batch retraining is employed. Data preparation includes input (X): constructing a spatiotemporal tensor. For example, an H×W×C image, where H and W represent the height and width of the map grid, respectively, and C represents the number of channels, each channel representing a feature (such as average speed channel, flow rate channel, traffic light status channel, etc.). Labels (y): H×W grid cells, where the value of each cell is the true average carbon intensity (gCO2 / km) of that region.

[0113] The loss function can be the mean squared error (MSE), the most direct choice, penalizing large errors. Alternatively, Huber loss can be used, which is less sensitive to outliers than MSE and more robust to noisy training data. In this embodiment, the specific loss function used is Loss = Mean(L(y - y_pred)), where L is either MSE or the Huber loss function. The retraining process mainly involves transfer learning or fine-tuning, loading the structure and weights of the current production CNN model. The earlier layers are frozen (these layers may learn general features, such as basic spatial patterns), and only the later fully connected layers or the last few convolutional layers are fine-tuned. This prevents overfitting on small datasets. The newly constructed (X, y) dataset is used for training with a small learning rate. Through backpropagation, the error signal is passed forward layer by layer from the output layer, fine-tuning the network weights to make the network's mapping of carbon intensity more accurate. The model is deployed after validating its performance.

[0114] The following describes the retraining process of the carbon emission calculation sub-model, which is used to calculate the carbon emissions of corresponding vehicle types during operation. Depending on the vehicle type, the carbon emission calculation sub-model includes models for gasoline vehicles, pure electric vehicles, and hybrid vehicles. Based on the calculation formulas for the aforementioned different models, the retraining of the carbon emission calculation sub-model mainly involves calibrating the key parameters in the model. Taking the gasoline vehicle model as an example, carbon emissions = instantaneous fuel consumption × fuel carbon emission coefficient × slope correction coefficient × wind speed correction coefficient.

[0115] Data preparation involves collecting real-time or historical data during vehicle operation, including but not limited to instantaneous fuel consumption, gradient data, wind speed data, and corresponding actual carbon emission data. This data constitutes the training dataset for parameter calibration. Initial values ​​are set for key parameters in the model (such as fuel carbon emission coefficient, gradient correction coefficient, and wind speed correction coefficient). The default fuel carbon emission coefficient published by relevant regulations is adopted.

[0116] Based on vehicle design parameters and historical experience, initial estimates for slope correction and wind speed correction factors are set (e.g., the initial value can be set to 1, indicating no correction). The preprocessed model input data (instantaneous fuel consumption, slope, wind speed) and the current parameter set (fuel carbon emission coefficient Kfuel, slope correction coefficient Kgrade, wind speed correction coefficient Kwind) are substituted into the fuel vehicle carbon emission calculation model for forward calculation, yielding the predicted carbon emission sequence Cpred.

[0117] The formula for calculating the predicted value sequence is Cpred(t) = E(t) × Kfuel × Kgrade(t) × Kwind(t), where E(t) represents the instantaneous fuel consumption at time t. When constructing the loss function, the predicted carbon emissions calculated by the model are compared with the actual carbon emissions data, and the error value is calculated. The error can be calculated using mean squared error (MSE), mean absolute error (MAE), or other suitable loss functions.

[0118] The retraining process is as follows: First, based on the error value, key parameters are iteratively adjusted using optimization algorithms to minimize the error. Optimization algorithms can employ gradient descent-type algorithms (such as stochastic gradient descent, Adam optimizer) or nonlinear least squares methods (such as the Levenberg-Marquardt algorithm). Specifically, parameter values ​​are updated by calculating the gradient of the error with respect to the parameters or by fitting data, making the model output closer to the actual values. Taking gradient descent as an example: using backpropagation, the partial derivatives (i.e., gradients) of the loss function with respect to each parameter to be calibrated (Kfuel, Kgrade, Kwind) are calculated. This indicates the direction and magnitude of adjustment for each parameter to reduce the loss function. Based on the calculated gradients, the parameters are updated according to the following rule: θnew = θold η· L(θold).

[0119] Where θ represents any parameter to be calibrated, and η is the learning rate (a preset positive number that controls the step size of each update). L is the gradient of the loss function with respect to that parameter.

[0120] When the optimization algorithm meets the preset convergence conditions (e.g., the change in the loss function value is less than a threshold or the maximum number of iterations is reached), the iteration process terminates. The adjusted parameter values ​​are then updated in the carbon emission calculation sub-model, replacing the original parameters. The updated model is used for subsequent carbon emission calculations, ensuring that the model adapts to changes in vehicle driving conditions.

[0121] To ensure the continued effectiveness of the model, a retraining trigger mechanism can be configured, which can be triggered periodically, such as automatically performing retraining once after accumulating a certain mileage or at fixed time intervals (such as monthly).

[0122] It can also be triggered by events, such as when a significant change in vehicle performance is detected, or when the vehicle is driven for a long time in a specific new environment (such as a plateau or windy area), the retraining process can be actively triggered.

[0123] In this embodiment of the invention, a feedback mechanism for low-carbon path planning is designed. The actual carbon emissions are fed back to the model to achieve continuous optimization of the model. As the number of times the vehicle is used by users increases, the carbon emission data accumulated in different scenarios becomes richer, which continuously improves the generalization ability of the model and forms a positive cycle of "data-model-planning-feedback-data".

[0124] This invention integrates multi-dimensional traffic environment data of the planned road segments to construct a real-time traffic carbon intensity map, upgrading the carbon emission quantification dimension from individual vehicles to a collaborative level between the road network system and individual vehicles, thus breaking through the existing static, single-dimensional carbon emission assessment framework. Furthermore, by designing correction weights for vehicle types to adjust carbon emission amounts, personalized carbon emission predictions are adapted to various vehicle type parameters, improving the accuracy of carbon emission estimation and providing a reliable data foundation for low-carbon route planning. Based on this, a directed road network graph is constructed using the traffic carbon intensity map, and low-carbon route planning is performed with the minimum carbon emission as the planning objective, achieving low-carbon route guidance.

[0125] This invention also provides a vehicle low-carbon route planning system based on real-time traffic carbon intensity maps. For example... Figure 2 As shown, the vehicle low-carbon route planning system based on real-time traffic carbon intensity map specifically includes a carbon intensity calculation module 201, a personalized vehicle carbon model module 202, a road network construction module 203, and a low-carbon route planning algorithm module 204.

[0126] Specifically, the carbon intensity calculation module 201 is used to acquire multi-dimensional traffic environment data of vehicles on the road segment to be planned, and construct a traffic carbon intensity map of the road segment to be planned based on the multi-dimensional traffic environment data; the personalized vehicle carbon model module 202 is used to determine the vehicle type of the vehicle and determine the correction weight corresponding to the vehicle type; the road network construction module 203 is used to construct a directed road network graph based on the traffic carbon intensity map of the road segment to be planned; and the low-carbon path planning algorithm module 204 is used to perform low-carbon path planning on the directed road network graph based on the traffic carbon intensity map and the correction weight, and obtain the optimal driving path of the road segment to be planned with the minimum carbon emissions.

[0127] In some embodiments, the vehicle low-carbon route planning system based on real-time traffic carbon intensity maps further includes a closed-loop and feedback module 205. Specifically, the closed-loop and feedback module 205 is used to obtain the actual total carbon emissions of the vehicle after the vehicle travels along the optimal driving route; construct error samples based on the planned carbon emissions corresponding to the optimal driving route and the actual total carbon emissions; and train a gradient boosting tree model, a convolutional neural network model, and a carbon emission calculation sub-model based on the error samples.

[0128] The vehicle low-carbon route planning system based on real-time traffic carbon intensity map provided in the above embodiments can realize the technical solutions described in the above embodiments of the vehicle low-carbon route planning method based on real-time traffic carbon intensity map. The specific implementation principles of each module or unit can be referred to the corresponding content in the above embodiments of the vehicle low-carbon route planning method based on real-time traffic carbon intensity map, and their technical effects can also be referred to accordingly, which will not be repeated here.

[0129] like Figure 3 As shown, the present invention also provides an electronic device 300. The electronic device 300 includes a processor 301, a memory 302, and a display 303. Figure 3 Only some components of the electronic device 300 are shown, but it should be understood that it is not required to implement all of the components shown, and more or fewer components may be implemented instead.

[0130] In some embodiments, memory 302 may be an internal storage unit of electronic device 300, such as a hard disk or memory of electronic device 300. In other embodiments, memory 302 may also be an external storage device of electronic device 300, such as a plug-in hard disk, smart media card (SMC), secure digital (SD) card, flash card, etc. equipped on electronic device 300.

[0131] Furthermore, the memory 302 may include both internal storage units of the electronic device 300 and external storage devices. The memory 302 is used to store application software and various types of data installed on the electronic device 300.

[0132] In some embodiments, processor 301 may be a central processing unit (CPU), microprocessor, or other data processing chip for running a vehicle low-carbon route planning program stored in memory 302, such as the vehicle low-carbon route planning method based on real-time traffic carbon intensity map in this invention.

[0133] In some embodiments, display 303 may be an LED display, a liquid crystal display, a touch-sensitive liquid crystal display, or an OLED (Organic Light-Emitting Diode) touchscreen. Display 303 is used to display information from electronic device 300 and to display a visual user interface. Components 301-303 of electronic device 300 communicate with each other via a system bus.

[0134] In some embodiments of the present invention, when the processor 301 executes the vehicle low-carbon route planning program in the memory 302, the following steps can be implemented: acquiring multi-dimensional traffic environment data of the vehicle on the road segment to be planned, constructing a traffic carbon intensity map of the road segment to be planned based on the multi-dimensional traffic environment data; determining the vehicle type of the vehicle, and determining the correction weight of the carbon emission corresponding to the vehicle type; constructing a directed road network graph based on the traffic carbon intensity map of the road segment to be planned; and performing low-carbon route planning on the directed road network graph based on the traffic carbon intensity map and the correction weight to obtain the optimal driving path of the road segment to be planned with the minimum carbon emission.

[0135] It should be understood that when the processor 301 executes program code in the memory 302 or processes data, in addition to the functions mentioned above, it can also perform other functions, as can be found in the description of the corresponding method embodiments above.

[0136] Furthermore, the embodiments of the present invention do not specifically limit the type of electronic device 300 mentioned. Electronic device 300 can be a mobile phone, tablet computer, personal digital assistant (PDA), wearable device, laptop computer, or other portable electronic device. Exemplary embodiments of portable electronic devices include, but are not limited to, portable electronic devices running iOS, Android, Microsoft, or other operating systems. The aforementioned portable electronic device can also be other portable electronic devices, such as a laptop computer with a touch-sensitive surface (e.g., a touch panel). It should also be understood that in some other embodiments of the present invention, electronic device 300 may not be a portable electronic device, but rather a desktop computer with a touch-sensitive surface (e.g., a touch panel).

[0137] In another aspect, the present invention also provides a non-transitory computer-readable storage medium storing a computer program thereon. When executed by a processor, the computer program implements the vehicle low-carbon path planning method based on real-time traffic carbon intensity map provided by the above methods. The method includes: acquiring multi-dimensional traffic environment data of a vehicle on a road segment to be planned; constructing a traffic carbon intensity map of the road segment to be planned based on the multi-dimensional traffic environment data; determining the vehicle type of the vehicle and determining the correction weight for carbon emissions corresponding to the vehicle type; constructing a directed road network graph based on the traffic carbon intensity map of the road segment to be planned; and performing low-carbon path planning on the directed road network graph based on the traffic carbon intensity map and the correction weight to obtain the optimal driving path of the road segment to be planned with minimum carbon emissions.

[0138] Those skilled in the art will understand that all or part of the processes of the methods described in the above embodiments can be implemented by a computer program instructing related hardware, and the program can be stored in a computer-readable storage medium. The computer-readable storage medium may be a disk, optical disk, read-only memory, or random access memory, etc.

[0139] The above provides a detailed description of the vehicle low-carbon route planning method based on real-time traffic carbon intensity maps provided by this invention. Specific examples have been used to illustrate the principles and implementation methods of this invention. The descriptions of the above embodiments are only for the purpose of helping to understand the method and core ideas of this invention. At the same time, for those skilled in the art, there will be changes in the specific implementation methods and application scope based on the ideas of this invention. Therefore, the content of this specification should not be construed as a limitation of this invention.

Claims

1. A method for low-carbon route planning for vehicles based on real-time traffic carbon intensity maps, characterized in that, include: Acquire multidimensional traffic environment data of vehicles on the road segment to be planned, and construct a traffic carbon intensity map of the road segment to be planned based on the multidimensional traffic environment data; Determine the vehicle type and the corresponding carbon emission correction weight; Construct a directed road network graph based on the traffic carbon intensity map of the road segment to be planned; Based on the traffic carbon intensity map and the corrected weights, low-carbon path planning is performed on the directed graph of the road network to obtain the optimal driving path for the road segment to be planned with the minimum carbon emissions.

2. The vehicle low-carbon route planning method based on real-time traffic carbon intensity map according to claim 1, characterized in that, The process of constructing the traffic carbon intensity map of the road segment to be planned based on multidimensional traffic environment data includes: The gradient boosting tree model is invoked to predict the multidimensional traffic environment data, and the weight coefficients for each carbon emission influencing factor are obtained. The spatiotemporal correlation features between the road network topology and traffic light phases corresponding to the road segment to be planned are extracted by calling a convolutional neural network model; Based on the spatiotemporal correlation features and the weighting coefficients, the real-time carbon intensity value of the road segment to be planned is calculated to form a traffic carbon intensity map.

3. The vehicle low-carbon route planning method based on real-time traffic carbon intensity map according to claim 1, characterized in that, The correction weight for determining the carbon emissions corresponding to the vehicle type includes: Construct a corresponding carbon emission calculation sub-model based on the vehicle type; The carbon emission correction weight is calculated based on the historical estimated carbon emissions of the vehicle along its historical driving routes and the historical actual carbon emissions calculated based on the carbon emission calculation sub-model.

4. The vehicle low-carbon route planning method based on real-time traffic carbon intensity map according to claim 1, characterized in that, The construction of a directed road network graph based on the traffic carbon intensity map of the road segment to be planned includes: The road intersections and road segments between the road intersections in the road segment to be planned are taken as nodes and edges, and the driving direction of the road segment is taken as the edge direction of the edge. The real-time carbon intensity value of each road segment is queried based on the traffic carbon intensity map of the road segment to be planned, and the real-time carbon intensity value is used as the edge weight of the edge. Based on the nodes, edges, edge directions, and edge weights, a directed graph of the road network is constructed.

5. The vehicle low-carbon route planning method based on real-time traffic carbon intensity map according to claim 4, characterized in that, The process of performing low-carbon route planning on the directed graph of the road network to obtain the optimal driving route for the road segment to be planned with the minimum carbon emissions includes: The starting point and ending point of the vehicle in the planned road segment are determined from the directed road network graph. Starting from the travel origin, Dijkstra's algorithm is invoked to perform shortest path planning in the directed graph of the road network, resulting in sub-road segments between the travel origin and the travel destination. The path weight is calculated for each sub-segment based on the traffic carbon intensity map and the corrected weight, and the path weights are accumulated. When the sub-segment reaches the destination, the optimal driving path for the segment to be planned with the minimum carbon emissions is determined based on the minimum cumulative result of the path weights.

6. The vehicle low-carbon route planning method based on real-time traffic carbon intensity map according to claim 5, characterized in that, The calculation of path weights for each sub-segment based on the traffic carbon intensity map and the corrected weights includes: The real-time carbon intensity value of each sub-road segment is retrieved from the traffic carbon intensity map; The real-time carbon intensity value is corrected based on the correction weight to obtain a personalized road segment carbon intensity value; The actual distance of the sub-segment is determined, and the path weight of the sub-segment is calculated based on the actual distance and the personalized segment carbon intensity value.

7. The vehicle low-carbon route planning method based on real-time traffic carbon intensity map according to claim 1, characterized in that, The process of performing low-carbon route planning on the directed graph of the road network to obtain the optimal driving route for the road segment to be planned with the minimum carbon emissions includes: The starting point and ending point of the vehicle in the road segment to be planned are determined from the directed graph of the road network, and a heuristic search algorithm is called to search the directed graph of the road network starting from the starting point. For any target node in the search process, the actual cumulative carbon emissions between the starting point of the journey and the target node are calculated based on the traffic carbon intensity map. Calculate the estimated minimum carbon emissions from the target node to the destination, and sum the actual cumulative carbon emissions with the estimated minimum carbon emissions to obtain the estimated value. The target node with the smallest estimated value is selected as the candidate node; When the candidate node is found to be the destination, the optimal driving route for the planned road segment with the minimum carbon emissions is determined based on the candidate node.

8. The vehicle low-carbon route planning method based on real-time traffic carbon intensity map according to claim 7, characterized in that, The calculation of the estimated minimum carbon emissions from the target node to the travel destination includes: Determine the Euclidean geometric distance from the target node to the destination, and the global minimum carbon intensity value of the directed graph of the road network; The product of the Euclidean geometric distance and the global minimum carbon intensity value is used as the estimated minimum carbon emissions.

9. The vehicle low-carbon route planning method based on real-time traffic carbon intensity map according to claim 1, characterized in that, The method further includes: After the vehicle travels along the optimal driving route, the actual total carbon emissions of the vehicle are obtained; Error samples are constructed based on the planned carbon emissions corresponding to the optimal driving route and the actual total carbon emissions. A gradient boosting tree model, a convolutional neural network model, and a carbon emission calculation sub-model are then trained based on the error samples.

10. A vehicle low-carbon route planning system based on real-time traffic carbon intensity maps, characterized in that, include: The carbon intensity calculation module is used to acquire multi-dimensional traffic environment data of vehicles on the road segment to be planned, and to construct a traffic carbon intensity map of the road segment to be planned based on the multi-dimensional traffic environment data. The personalized vehicle carbon model module is used to determine the vehicle type and the corresponding correction weight for the vehicle type. The road network construction module is used to construct a directed road network graph based on the traffic carbon intensity map of the road segment to be planned; The low-carbon route planning algorithm module is used to perform low-carbon route planning on the directed graph of the road network based on the traffic carbon intensity map and the modified weight, so as to obtain the optimal driving route of the road segment to be planned with the minimum carbon emissions.

Images