A new energy vehicle dynamic charging navigation method and system based on a hybrid algorithm

Abstract

The application provides a new energy vehicle dynamic charging navigation method and system based on a hybrid algorithm, relates to the technical field of charging path planning, and responds to a navigation request of a target new energy vehicle to perform the following steps: acquiring sensing information; based on the sensing information, dynamically deciding an optimal charging navigation path of the new energy vehicle through a MSDA-DQN hybrid algorithm of fusing a mirror symmetry Dijkstra algorithm and deep reinforcement learning, the charging navigation path comprising a wireless charging lane and a corresponding optimal executable path, the executable path comprising a path from a current vehicle position to an entrance of the wireless charging lane and a path from an exit of the wireless charging lane to a terminal position; and performing driving and charging navigation on the new energy vehicle according to the optimal charging navigation path; the application designs a hybrid algorithm to calculate the optimal charging navigation path in real time and realize collaborative optimization of path planning and charging strategies in a complex dynamic environment.

Description

Technical Field

[0001] This invention relates to the field of charging route planning technology, specifically to a dynamic charging navigation method and system for new energy vehicles based on a hybrid algorithm. Background Technology

[0002] With increasing global awareness of environmental protection and a deepening understanding of sustainable development, electric vehicles (EVs) are gradually becoming an ideal alternative to traditional internal combustion engine vehicles due to their advantages such as clean energy drive, lower travel and maintenance costs, and environmental friendliness. However, the widespread application of EVs has also brought new challenges. On the one hand, the construction of charging infrastructure lags behind the growth in the number of EVs, leading to increasingly prominent charging difficulties. On the other hand, the concentrated charging of EVs during peak electricity consumption periods may cause excessive load on the power grid, thereby affecting its stability.

[0003] Currently, the main charging methods include plug-in charging, battery swapping, vehicle-to-vehicle charging, mobile charging, and wireless charging. Traditional plug-in charging is limited by insufficient physical connection interfaces, resulting in complex operation, poor flexibility, and, more importantly, long charging times, hindering the widespread adoption of new energy vehicles. Vehicle-to-vehicle charging requires time for electric vehicles to pair and find suitable charging locations, consuming significant time and energy. While battery swapping offers rapid charging, issues such as battery compatibility across different vehicle brands and battery ownership remain to be addressed.

[0004] Against this backdrop, dynamic wireless charging technology has emerged. By embedding charging facilities into roads, electric vehicles can replenish their energy while driving, providing a revolutionary solution to range anxiety and battery dependence. The core value of dynamic wireless charging systems lies in their unique operating mode. Instead of traditional fixed charging stations, they embed charging facilities into specific road segments within the road network, achieving simultaneous energy transfer and vehicle movement without additional vehicle dwell time. This characteristic necessitates that charging navigation systems simultaneously optimize route selection and charging strategies, a fundamental difference from traditional charging station navigation. Further complicating matters, the varying risk preferences exhibited by different car owners in their charging decisions can significantly impact system operation—conservative users may avoid congested wireless charging lanes (WCLs), while adventurous users may actively choose congested road segments to extend charging time. This diversity of individual behavior not only affects charging efficiency but also has a cascading impact on grid stability through dynamic changes in load distribution.

[0005] Existing technologies have the following problems when solving this type of real-time navigation problem:

[0006] (1) Challenges of dynamic information fusion and path coupling optimization: Traditional navigation is difficult to integrate the real-time status of wireless charging lanes and electricity price fluctuations, and cannot uniformly optimize the three-segment coupled path of “starting point - charging entrance - charging exit - destination”, resulting in a disconnect between driving costs and charging strategies.

[0007] (2) Challenges of personalized modeling and real-time computation efficiency: Existing methods lack a mechanism to map user risk preferences and other personalized parameters into an interpretable cost model. At the same time, directly solving related optimization problems is too computationally complex, making it difficult to achieve navigation decisions that balance personalization and real-time performance in seconds.

[0008] Therefore, existing dynamic charging navigation technologies for new energy vehicles still have shortcomings in terms of robustness, accuracy, and efficiency. Summary of the Invention

[0009] To address the aforementioned issues, this invention proposes a dynamic charging navigation method and system for new energy vehicles based on a hybrid algorithm. It designs an MSDA-DQN hybrid algorithm that integrates Mirror-Symmetrical Dijkstra (MSDA) and Deep Q Network (DQN) to calculate the optimal charging navigation path in real time, achieving coordinated optimization of path planning and charging strategy in complex dynamic environments.

[0010] According to some embodiments, the present invention adopts the following technical solution:

[0011] A dynamic charging navigation method for new energy vehicles based on a hybrid algorithm performs the following steps in response to a navigation request from a target new energy vehicle:

[0012] Acquire perceived information, including prerequisite information reflecting user preferences and basic task requirements, and real-time information reflecting the instantaneous state of the vehicle and the environment;

[0013] Based on perception information, the optimal charging navigation path for new energy vehicles is dynamically determined by fusing the mirror-symmetric Dijkstra algorithm with the MSDA-DQN hybrid algorithm of deep reinforcement learning. The charging navigation path includes a wireless charging lane and the corresponding optimal executable path. The executable path includes the path from the current vehicle location to the entrance of the wireless charging lane and the path from the exit of the wireless charging lane to the destination location.

[0014] Based on the optimal charging navigation route, the system provides driving and charging navigation for new energy vehicles.

[0015] The MSDA-DQN hybrid algorithm employs a two-layer decision-making process. The lower layer utilizes the mirror-symmetric Dijkstra algorithm to search for the optimal executable path for each candidate wireless charging lane in the road network, with the goal of minimizing travel cost. The upper layer utilizes deep reinforcement learning, combined with perception information, to select the wireless charging lane with the lowest sum of travel and charging costs that meets safety constraints from the candidate wireless charging lanes. The selected wireless charging lane and its corresponding optimal executable path are then combined to form the optimal charging navigation path.

[0016] According to some embodiments, the present invention adopts the following technical solution:

[0017] A dynamic charging navigation system for new energy vehicles based on a hybrid algorithm includes:

[0018] The perception module is configured to acquire perception information, including prerequisite information reflecting user preferences and basic task requirements, and real-time information reflecting the instantaneous state of the vehicle and the environment.

[0019] The decision-making module is configured to: based on perception information, dynamically decide the optimal charging navigation path for new energy vehicles by fusing the mirror-symmetric Dijkstra algorithm and the MSDA-DQN hybrid algorithm of deep reinforcement learning. The charging navigation path includes a wireless charging lane and the corresponding optimal executable path. The executable path includes the path from the current vehicle location to the entrance of the wireless charging lane and the path from the exit of the wireless charging lane to the destination location.

[0020] The execution module is configured to: provide driving and charging navigation for new energy vehicles based on the optimal charging navigation path;

[0021] The MSDA-DQN hybrid algorithm employs a two-layer decision-making process. The lower layer utilizes the mirror-symmetric Dijkstra algorithm to search for the optimal executable path for each candidate wireless charging lane in the road network, with the goal of minimizing travel cost. The upper layer utilizes deep reinforcement learning, combined with perception information, to select the wireless charging lane with the lowest sum of travel and charging costs that meets safety constraints from the candidate wireless charging lanes. The selected wireless charging lane and its corresponding optimal executable path are then combined to form the optimal charging navigation path.

[0022] According to some embodiments, the present invention adopts the following technical solution:

[0023] A computer program product includes a computer program that, when executed by a processor, implements the aforementioned dynamic charging navigation method for new energy vehicles based on a hybrid algorithm.

[0024] According to some embodiments, the present invention adopts the following technical solution:

[0025] A non-transitory computer-readable storage medium is provided for storing computer instructions, which, when executed by a processor, implement the aforementioned dynamic charging navigation method for new energy vehicles based on a hybrid algorithm.

[0026] According to some embodiments, the present invention adopts the following technical solution:

[0027] An electronic device includes a processor, a memory, and a computer program; wherein the processor is connected to the memory, the computer program is stored in the memory, and when the electronic device is running, the processor executes the computer program stored in the memory to enable the electronic device to implement the aforementioned dynamic charging navigation method for new energy vehicles based on a hybrid algorithm.

[0028] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0029] (1) Optimization of route and charging to reduce overall travel costs: By integrating route planning and wireless charging strategies into a single optimization objective, this invention can minimize overall travel costs (energy consumption + time loss + charging costs) while ensuring the vehicle reaches its destination smoothly. Simulation experiments based on real traffic data show that this method can significantly reduce the total travel cost of the vehicle compared to traditional strategies that only plan routes or only select charging stations. At the same time, the vehicle can complete charging while driving, reducing the waiting time caused by static charging and greatly improving travel efficiency.

[0030] (2) The two-layer algorithm is efficient and reliable, meeting the requirements of real-time decision-making at the second level: This invention innovatively adopts the MSDA algorithm for optimal executable path search, decomposing the complex navigation problem into two symmetric shortest path solutions, reducing the computational complexity to Levels, among which, For road segments, For wireless charging lane sets, this method is more than 100 times faster than traditional exhaustive search or planning algorithms. This feature ensures that the system can calculate candidate driving costs in milliseconds on complex road networks, laying the foundation for real-time decision-making at higher levels. Combined with the decision generalization capabilities of deep reinforcement learning, the entire navigation algorithm achieves second-level response in road networks with hundreds of nodes, significantly improving real-time performance and scalability.

[0031] (3) Strong real-time adaptive capability and personalized and intelligent decision-making: The method of this invention can respond sensitively to real-time dynamics such as electricity prices and traffic congestion, as well as user risk preferences. Simulation tests show that the model will automatically adjust its strategy under different urban charging infrastructure layouts: for example, it prioritizes short-distance multiple charging in areas with concentrated charging lanes, and tends to charge in a long distance in a single charge in a sparse ring network of charging lanes, demonstrating adaptability to infrastructure layout; users with different risk preferences will also receive personalized navigation schemes in the same scenario, more reasonably balancing travel time and charging gains; this intelligent decision-making capability ensures flexible and reasonable route planning, improving user satisfaction and grid load balancing benefits. Attached Figure Description

[0032] The accompanying drawings, which form part of this invention, are used to provide a further understanding of the invention. The illustrative embodiments of the invention and their descriptions are used to explain the invention and do not constitute an improper limitation of the invention.

[0033] Figure 1 This is a framework diagram of Example 1. Detailed Implementation

[0034] The present invention will be further described below with reference to the accompanying drawings and embodiments.

[0035] It should be noted that the following detailed descriptions are exemplary and intended to provide further illustration of the invention. Unless otherwise specified, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.

[0036] It should be noted that the terminology used herein is for the purpose of describing particular embodiments only and is not intended to limit the scope of exemplary embodiments according to the invention. As used herein, the singular form is intended to include the plural form as well, unless the context clearly indicates otherwise. Furthermore, it should be understood that when the terms "comprising" and / or "including" are used in this specification, they indicate the presence of features, steps, operations, devices, components, and / or combinations thereof.

[0037] Example 1

[0038] One embodiment of the present invention provides a dynamic charging navigation method for new energy vehicles based on a hybrid algorithm, which, in response to a navigation request from a target new energy vehicle, performs the following steps:

[0039] Step S1: Acquire perception information, including prerequisite information reflecting user preferences and basic task requirements, and real-time information reflecting the instantaneous state of the vehicle and the environment;

[0040] Step S2: Based on the perception information, the optimal charging navigation path for the new energy vehicle is dynamically determined by fusing the mirror-symmetric Dijkstra algorithm with the MSDA-DQN hybrid algorithm of deep reinforcement learning. The charging navigation path includes the wireless charging lane and the corresponding optimal executable path. The executable path includes the path from the current vehicle location to the entrance of the wireless charging lane and the path from the exit of the wireless charging lane to the destination location.

[0041] Step S3: Navigate the new energy vehicle for driving and charging according to the optimal charging navigation route;

[0042] The MSDA-DQN hybrid algorithm employs a two-layer decision-making process. The lower layer utilizes the mirror-symmetric Dijkstra algorithm to search for the optimal executable path for each candidate wireless charging lane in the road network, with the goal of minimizing travel cost. The upper layer utilizes deep reinforcement learning, combined with perception information, to select the wireless charging lane with the lowest sum of travel and charging costs that meets safety constraints from the candidate wireless charging lanes. The selected wireless charging lane and its corresponding optimal executable path are then combined to form the optimal charging navigation path.

[0043] This embodiment addresses a dynamic wireless charging navigation scenario in urban road networks where vehicles replenish power while driving, aiming to solve the following key technical challenges:

[0044] (1) Challenges of dynamic information fusion: Traditional route planning usually finds the shortest or lowest energy consumption path based on a static road network. It cannot integrate real-time information such as the availability of wireless charging lanes (WCL), congestion status, and electricity price fluctuations into navigation decisions, resulting in a disconnect between route planning and charging strategies.

[0045] (2) Path coupling optimization challenge: Existing navigation methods mostly focus on the selection of a single static charging station, making it difficult to take into account the three-segment coupled path of "starting point - WCL entrance - WCL exit - destination". In the dynamic wireless charging scenario, the vehicle's driving cost (energy consumption cost and time loss) is closely related to the charging cost, forming a complex comprehensive travel cost optimization problem. How to unify and optimize the comprehensive travel cost of the driving segment and the charging segment is a challenge.

[0046] (3) Challenges in Personalized Decision Modeling: Different users have different risk preferences regarding congestion delays and charging benefits. Conservative users may tend to avoid congested charging lanes, while aggressive users may be willing to stay longer on congested roads in order to obtain more energy. Currently, there is a lack of a mechanism to map users' prior personality parameters (such as risk preference coefficient, time value coefficient, minimum energy safety threshold, etc.) into an interpretable cost model, thereby reflecting user personality differences in navigation decisions.

[0047] (4) Real-time computational efficiency challenge: Achieving second-level real-time navigation decision-making in urban road networks containing hundreds of nodes is extremely difficult. Directly solving the above mixed integer nonlinear programming (MINLP) problem, which includes path selection and charging behavior, has extremely high computational complexity, belonging to the NP-hard problem and is difficult to implement in engineering. An effective algorithm is needed to reduce computational complexity while ensuring that the result is close to the optimal level and meeting the real-time requirements.

[0048] To address the above issues, this embodiment proposes a dynamic wireless charging navigation method for new energy vehicles based on a "Mirror-Symmetrical Dijkstra (MSDA) algorithm and Deep Q Network (DQN)" architecture. This method integrates the advantages of classic path algorithms and deep reinforcement learning, enabling collaborative optimization of path planning and charging strategies in complex dynamic environments. The specific implementation process is described below.

[0049] First, let's explain "responding to the navigation request of the target new energy vehicle":

[0050] This embodiment adopts an event-driven navigation request mechanism, rather than real-time automatic charging navigation. Specifically, when the vehicle's current remaining battery power SoC drops below a safety threshold or the vehicle approaches a certain WCL while driving, the driver is asked on the in-vehicle display or voice interface whether to proceed with charging navigation. Only after the driver confirms and agrees is a navigation request issued, and the new energy vehicle dynamic wireless charging navigation method of this embodiment is executed.

[0051] The dynamic wireless charging navigation method for new energy vehicles proposed in this embodiment, such as Figure 1 As shown, the overall architecture is divided into three parts: perception layer, decision layer and execution layer, to achieve closed-loop optimization control of vehicle route selection and dynamic charging process.

[0052] I. Perception Layer: Acquisition of prerequisite and real-time information

[0053] The top layer is the perception layer, which acquires and integrates various types of information required for navigation decisions: vehicles and users provide prerequisite information reflecting user preferences and basic task requirements, while intelligent transportation systems and power grid platforms provide real-time information reflecting the instantaneous state of vehicles and the environment.

[0054] Prerequisite information consists of fixed parameters set by the user or system at the initialization of each navigation session and remains unchanged throughout the navigation process, including: destination. Value coefficient per user per unit of time User risk preference coefficient for congestion (Values ​​range [-1, 1], -1 indicates extreme avoidance of congestion, 1 indicates pursuing charging benefits regardless of congestion) and the minimum safe threshold of vehicle battery power, etc.; prerequisite information reflects user preferences and basic task requirements, providing prior constraints in navigation decision-making.

[0055] Real-time information refers to environmental and vehicle status parameters that are updated over time during navigation. It reflects the instantaneous state of the vehicle's environment, including its current location. Current moment The vehicle's current remaining battery power. Real-time traffic speed of vehicles on adjacent road sections Current charging electricity price for candidate WCL road sections (Dynamically adjusted according to grid load) and the congestion coefficient on WCL Etc., among which, congestion coefficient It is calculated based on the deviation of the actual vehicle speed on that road section from the normal speed, and is expressed by the formula:

[0056]

[0057] in, For the current moment The actual speed of vehicles on the WCL section This represents the expected vehicle speed under normal conditions for this WCL section. The larger the value, the more congested the traffic.

[0058] In the perception layer, the aforementioned prerequisite information is used to construct the road network topology and travel constraints with WCL, and real-time information is used to update the time-varying weights and vehicle states of each road segment, thereby forming a dynamic charging road network model and state vector for use by the lower-level MSDA and the upper-level DQN, serving as the basis for constructing optimization decisions.

[0059] Before the decision-making level can perform calculations, a Deterministic Dynamic Charging Routing Model (DDCRM) needs to be established for the urban road network with wireless charging roads.

[0060] Define an urban electrified transportation network as a directed graph. ,in For traffic nodes (such as intersections, charging entrances, etc.). This is a set of road segments.

[0061] Ordinary road sections are formed by edge collection This indicates that each edge represents a road segment. Related road segment length Real-time traffic speed Attributes such as...

[0062] Wireless charging lanes (WCLs) are special road segments that utilize edge sets. This means that each WCL corresponds to one entry node. To the export node Oriented edge .

[0063] The Wireless Charging Lane (WCL) has two key features:

[0064] Firstly, the wireless charging module embedded in the road surface provides a constant unit charging power through a segmented activation mechanism. This allows vehicles to continuously receive electrical energy while traveling on this section of road.

[0065] Secondly, dedicated short range communication (DSRC) or 5G units deployed along the roadside can enable real-time data interaction between vehicles and infrastructure, such as vehicles reporting their current remaining battery SoC, obtaining electricity prices and traffic conditions (i.e., real-time traffic speed) for the road segment ahead.

[0066] It should be noted that the level of congestion on the WCL (Wideline Charge Center) directly affects the charging effect. The more vehicles in congestion, the slower the traffic speed, the longer the vehicles stay on the WCL, and the higher the amount of electricity that can be obtained per unit distance. However, excessive congestion can also cause time delays. Therefore, this embodiment introduces a congestion coefficient. To measure The degree to which the current vehicle speed deviates from the normal speed range is used to calculate subsequent charging costs, thereby achieving a trade-off between charging benefits and delays under different congestion conditions.

[0067] II. Decision-making level

[0068] The middle layer is the decision-making layer, which uses the MSDA-DQN two-layer algorithm to calculate the optimal charging path in real time. The lower layer MSDA calculates the candidate WCL set based on the road network and real-time data, and outputs the optimal driving cost, charging cost and executable path of each candidate. The upper layer DQN, based on the output of the lower layer MSDA, combines the perception state, evaluates the Q value and selects the optimal charging lane (outputs the decision action, i.e. which WCL to select).

[0069] Here, Q represents the total cost, i.e., the cost of the vehicle choosing a particular wireless charging lane. The corresponding overall travel cost includes two parts: the cost of travel during the journey. and charging costs incurred during the charging process This embodiment takes minimizing the overall travel cost as the route optimization objective, that is, seeking:

[0070]

[0071] in, Includes all optional WCLs, For the case where no WCL is selected and the destination is reached directly, it should be noted that the decision involves discrete path selection (such as whether to pass through a certain WCL) and continuous charging calculation, which is a complex mixed integer optimization problem. Direct solution is extremely difficult, and it is not feasible to obtain results in seconds on a road network with hundreds of nodes.

[0072] Therefore, this embodiment employs a two-layer optimization approach to decompose the aforementioned problem: the original problem is divided into two sub-problems: a lower-layer path feature extraction and an upper-layer path decision optimization. The lower layer uses fixed decisions to quickly calculate the optimal executable path (consisting of an executable entry path and an executable exit path) and cost features (including driving cost and charging cost) for each candidate WCL. The upper layer, based on the lower layer's output, selects the optimal WCL decision within the action space, thereby significantly reducing the solution complexity. The MSDA algorithm is precisely the tool used to efficiently solve the lower-layer problem. It extracts the structured path features corresponding to each WCL from the high-dimensional road network, achieving dimensionality reduction and speed improvement: since only two shortest path searches are performed each time, the time complexity of MSDA is approximately... In actual road networks with hundreds of nodes and hundreds of edges, it can effectively reduce computation time, making real-time decision-making at the upper level possible.

[0073] 1. Lower layer of the decision-making level: Path feature extraction method based on MSDA

[0074] To address the efficiency and interpretability issues of path planning, this embodiment incorporates an MSDA algorithm module at the lower layer of the decision layer as a path feature extractor. This module outputs path-related features for each candidate Wireless Charging Lane (WCL), including cost features such as travel cost and charging cost, as well as executable entrance and exit paths. MSDA utilizes the symmetric search principle of the classic Dijkstra's shortest path algorithm to quickly evaluate the travel cost of each candidate WCL and extract low-dimensional path features for upper-layer decision-making. Its core idea is to decompose the three-segment path through a WCL into two symmetric shortest path subproblems, including (a) from the current vehicle position... Shortest path to the WCL entry point (b) The shortest path from the WCL exit to the destination. This algorithm module is designed to efficiently calculate the optimal driving cost for traversing this WCL.

[0075] (1) Path cost modeling:

[0076] Before performing path search, a weight needs to be defined for each edge in the road network to quantify the path cost. Considering the energy and time costs of electric vehicle travel, this embodiment defines arbitrary road segments. Instant path cost weight It is the sum of two parts: the electricity cost converted from energy consumption and the travel time cost, as shown in the following formula:

[0077]

[0078] in, Energy consumption per unit distance of vehicle (kWh / km) The price per unit of electricity is (RMB / kWh, using the current basic electricity price of the power grid). For the edge The length of the road section, therefore This indicates the energy cost of a vehicle traveling on that road segment (i.e., the cost converted from electricity consumption). The second item... middle, The user's time value coefficient (yuan / hour). The edge weight represents the time required to traverse the road segment, and its product is the delay cost incurred by the travel time. This is how edge weights are defined. The road segment at the current moment is considered The energy and time costs are considered, which helps navigation algorithms to consider both energy conservation and timeliness. This path cost is used for both ordinary road sections and wireless charging vehicle road sections (WCL sections also consume a certain amount of energy and time during travel).

[0079] (2) Mirror-symmetric path search:

[0080] Based on the path costs mentioned above, MSDA performs two shortest path searches for each candidate WCL:

[0081] 1) Forward search (current vehicle location) From the WCL entry point, obtain the executable entry path: Starting from the vehicle's current node (location). Start by running Dijkstra's algorithm to find the WCL entry node that reaches the target. shortest path distance The distance here is actually obtained from the cumulative path cost, that is... The minimum path cost, where, Given the set of all directed edges contained in the shortest path from the previous position to the WCL entry node, the algorithm returns the optimal executable entry path. and the corresponding path cost .

[0082] 2) Reverse search (WCL exit to endpoint) to obtain the executable exit path: [The node is then...] Set as the endpoint node (Vehicle's final destination), from the exit node of the candidate WCL. Run Dijkstra's algorithm in reverse to find the origin from... To the finish line shortest path distance In a directed graph, the path from the endpoint to the destination can be achieved by inverting or equivalent edge operations. The reverse search yields the optimal executable exit path. and the corresponding path cost .

[0083] Through the above two steps, corresponding to a specific wireless charging lane The cost of a vehicle traveling through this WCL can be expressed as the sum of the costs of the three optimal paths:

[0084]

[0085] in, For entry from WCL To WCL Exit The cumulative path cost.

[0086] Here, if the vehicle chooses to use the wireless charging lane To drive, you need to first move from the current vehicle position to... Entrance , and then from exit Heading to the destination The MSDA algorithm calculates the optimal path cost for each of the three segments and sums them to obtain the optimal path cost through the charging route. Minimum operating cost .

[0087] At the same time, MSDA will also output the lowest operating cost. The corresponding actual path sequence, i.e., the execution entry path. and executable export path ,in, Indicates the current vehicle position To the wireless charging lane Entrance The shortest path node sequence, Indicates from the wireless charging lane exit To the finish line The shortest path sequence is obtained, and this path information will be used by the execution layer to guide vehicles to travel along the optimal route.

[0088] (3) Charging cost calculation:

[0089] Excluding driving costs In addition, the amount of electricity the vehicle receives while driving on the WCL and the associated costs are also important considerations in the decision-making process. This embodiment will include charging costs. As a separate component, it refers to the cost incurred by a vehicle for dynamic charging on selected WCL road sections. The charging cost is only related to the WCL road sections themselves and is independent of ordinary road sections. Its model definition is as follows:

[0090] 1) Charging capacity When the vehicle enters the selected Then, it will be charged at a constant power. The amount of electricity a vehicle can obtain while driving and charging on this road section. It depends on the length of the road segment, the charging power, and the acceptable margin of the vehicle's battery.

[0091] To ensure safety, the charging capacity... It is subject to two constraints: first, the maximum battery capacity minus the remaining chargeable space of the SoC; and second, the time taken by the vehicle to traverse the road segment multiplied by the charging power. The maximum energy that can be provided can be expressed by the formula:

[0092]

[0093] in, The length of the WCL road segment. This represents the real-time traffic speed of this road segment. Full charge capacity, The predicted battery level upon entering the WCL is determined by the current battery level. Subtract the "from current position" calculated along the lower-level MSDA The estimated energy consumption is obtained by taking the shortest path to WCL entrance m. This minimum value operation ensures that the energy consumption does not exceed the available power of the road segment or the remaining battery capacity.

[0094] 2) Cost Calculation and Risk Preference Adjustment: The vehicle receives the above-mentioned charging electricity. The fee must be paid according to the current electricity price. for Given the current unit electricity price, the basic charging cost is: Furthermore, this embodiment innovatively introduces a congestion coefficient. (The value is non-negative; the more severe the congestion, the higher the value.) (larger) and user congestion risk preference coefficient The charging cost is adjusted to reflect different users' subjective weighting of congestion delays. The formula for calculating the charging cost is as follows:

[0095]

[0096] When traffic is smooth, the congestion factor is... Approaching 0, the above formula indicates that the charging cost is approximately equal to the electricity cost; when congestion occurs (congestion coefficient) ), correction factor This will have an impact on risk-averse users ( This factor, being greater than 1, is equivalent to increasing the equivalent charging cost under congested conditions, thus penalizing inefficient charging behavior caused by congestion; for risk-seeking users ( The factor, ranging from 0 to 1, is equivalent to reducing the unit cost under congestion (or giving more positive weight to charging during congestion), encouraging vehicles to charge more under acceptable congestion conditions. This is achieved through the user's congestion risk preference coefficient. The personalized settings will reflect different driving styles in the charging cost calculation formula, making the optimization decision more in line with user preferences.

[0097] (4) Interface definition between lower layer output and upper layer:

[0098] To facilitate communication with upper-level decision-making and execution layers, MSDA at every moment... The calculation results (i.e. the extracted paths) are organized into a unified data packet containing three types of information: optional, evaluable, and drivable.

[0099] The first "optional" is the candidate WCL set. It is the set of currently feasible WCLs obtained under connectivity and energy safety constraints, only if "from Can reach the entrance "+" in "Charging without exceeding the limit" + "Leave" The WCL will only enter the process when all conditions are met, including "can continue to be driven". .

[0100] Secondly, "evaluable" refers to the set of cost characteristics of candidate WCLs. For each candidate in the set The system provides structured feature records, with the core field being the optimal travel cost via this WCL. With charging costs It includes intermediate parameters such as congestion coefficient and real-time electricity price, which are used by the upper layer to select candidate WCLs.

[0101] Thirdly, "drivable" refers to drivable information, and "drivable" is a set of drivable paths. Including the executable entry path and executable export paths Once the upper level selects the wireless charging lane The execution layer can directly call Drive into the selected WCL without repeating the shortest path calculation.

[0102] Therefore, the output of the lower layer is used as the action space and evaluation features in the upper layer, and the selected WCL corresponds to... It is used as an immediately executable entry path to form a closed-loop link.

[0103] 2. Upper Decision Layer: Path Selection Reinforcement Learning Strategy Based on DQN

[0104] To ensure consistency across layers, this section continues to use the candidate WCL set and candidate WCL cost feature set provided by MSDA in the previous section before proceeding to the next layer decision. Specifically, the candidate WCL set output by MSDA... and candidate WCL cost feature set Together with real-time vehicle-environment information, it is assembled into the state input of DQN. .

[0105] Based on this, DQN no longer blindly selects across the entire graph, but only selects from the candidate WCL set in MSDA. Internal action selection, for each candidate Evaluation value function ,Pick The road segment with the highest value is selected as the optimal charging road segment. Based on this connection, the upper layer of the decision-making layer adopts a deep reinforcement learning method, designing DQN as a path selection strategy optimizer. Its task is to intelligently select an optimal WCL from all "optional and evaluated" WCLs, under the premise of integrating vehicle state and environmental dynamics, so as to minimize the overall travel cost and ensure reachability. DQN learns an approximate state-action value function through interaction with the environment. It enables adaptive navigation under complex and uncertain traffic and electricity price conditions.

[0106] (1) Markov Decision Process (MDP) Modeling:

[0107] The dynamic charging navigation process is abstracted as a Markov decision process. In this context, at discrete decision-making moments, the agent (navigation strategy) selects an action based on the current state, receives a reward, and transitions to a new state until it reaches its destination and terminates. Its elements are defined as follows:

[0108] 1) State Space State vector Includes real-time information on vehicles and the environment, user prior preferences, and candidate cost features extracted by MSDA:

[0109]

[0110] in, This is the current vehicle location node. This is the current battery level. For decision-making moments; This represents the real-time speed collection of surrounding road segments, reflecting the local traffic conditions; and These are the user's prior parameters regarding the value of time and the risk of congestion, respectively. and The candidate WCL set and candidate cost features extracted by the MSDA module are used to evaluate the cost-effectiveness of each WCL for DQN.

[0111] 2) Action Space :action Defined as either selecting a WCL from the candidate WCL set for charging or selecting an empty set. (This means skipping all WCLs and going directly to the finish line):

[0112]

[0113] when At that time, it indicates the choice to go via Perform dynamic charging; when When the signal is active, it indicates that no additional power is needed to proceed directly to the destination; in actual navigation, the decision to trigger the signal can be determined based on the SoC safety threshold or the distance to the entrance.

[0114] 3) State transition After the action is performed, the vehicle enters the next state. .if The vehicle is in accordance with the specifications provided by MSDA. Drive to the entrance ,exist Charge at a constant power and drive away to the exit. New state The update rules for each component are as follows:

[0115] Location: Updated to (Or directly update to the next node, for) Then proceed along the shortest path towards the destination.

[0116] Battery level: ,in To pass And the driving energy consumption of WCL road sections, The amount of electricity obtained during WCL recharging (as defined in the MSDA module) (Transformed from)

[0117] time: The time increment depends on the time taken for driving and charging.

[0118] Other statuses: neighboring traffic conditions, real-time electricity prices, and congestion levels. It is updated in real time by the perception layer.

[0119] When the vehicle reaches its destination, the transfer is terminated.

[0120] 4) Reward Function To guide the DQN strategy to minimize the overall travel cost, this embodiment designs a cost-related reward function. Since the goal is to minimize cost, the reward can be a negative of the cost. Let's assume the time step... If you choose an action Corresponding to the candidate wireless charging lane WCL For driving, the single-step reward is defined as:

[0121]

[0122] in, For the current navigation scheme via candidate WCL The cost of driving For the corresponding charging cost, This is the additional reward constant added upon reaching the finish line, i.e., the finish line bonus. This is an additional penalty constant added when a SoC fails due to insecurity or other reasons, i.e., a security penalty, used in rewards. This indicates that a large negative reward is given in one go.

[0123] For ease of understanding, the meanings and calculation methods of each term in the above formula are explained below:

[0124] Operating cost item Reflected in the current navigation scheme, from the current location via candidate WCL The travel cost incurred in reaching the destination. Based on the aforementioned road network modeling, assume that the complete travel path corresponding to this scheme consists of several road segments. Composition, with the unit driving cost weight of each road segment as follows: Then the driving cost can be expressed as .in, To get from the current location via WCL The set of road segments along the path to the destination. The calculation is performed according to the aforementioned formula, based on road length, real-time vehicle speed, energy consumption model, etc. In actual execution, when the vehicle is at a certain time step... Through the road section At that time, the reward increment generated by driving in this step can be set to... Accumulate all throughout the entire navigation round That is, the negative value of the above-mentioned driving cost item is obtained. , and the reward function Each item corresponds to one another.

[0125] Charging cost item Reflects the vehicle in the selected WCL The electricity cost incurred during one dynamic charging process. Based on the previous discussion of charging volume... And from the definitions of unit electricity price and congestion coefficient, we can obtain ,in, For WCL At any moment The unit electricity price This is the congestion coefficient for that road segment (reflecting that the more severe the congestion, the higher the cost). To pass WCL this time The amount of charging available at any given time is calculated using the formula described above. In practice, this cost can be proportionally allocated as a single-step cost increment over several time steps when the vehicle is in the WCL (Wide Limit Chain). And make the rewards for this step include Then the sum of all charging-related rewards during the entire navigation round equals , and in the reward function The items must remain consistent.

[0126] Finish line reward With security penalties This is used to introduce task-level preference constraints in addition to cost items. To be a fixed positive number, it is only triggered at the final step when the final destination is successfully reached in the current navigation round, corresponding to the reward of that step. One line. This encourages intelligent agents to successfully complete the entire travel task while meeting energy constraints. This is the SoC safety constraint penalty coefficient, triggered when the current SoC determines that it is impossible to reach any charging location or destination from the current node, or when the SoC falls below the safety lower limit, causing the round to fail and terminate. It corresponds to the penalty coefficient in the reward function. One line. By Setting it to a positive number much greater than the normal driving cost can create a strong penalty during the learning process, forcing the strategy to avoid choosing infeasible paths that would lead to battery depletion.

[0127] In summary, when the agent learns to accumulate rewards... The most effective strategy is to minimize overall travel costs while meeting SoC security constraints. As small as possible.

[0128] (2) DQN architecture training:

[0129] Considering the large and complex state and action spaces, this embodiment uses DQN to approximate the value function. DQN uses a parameterized... A deep neural network representing the state-action value function. The input is the state. The output is the Q-value estimate for each action. The policy adopts... - Greedy method: using probability Choose random actions to explore new paths, using probability. Select the action with the highest current Q-value and utilize experience. Gradually increase the... The value is decayed from an initial value of 1.0 to a smaller value (0.1) to balance exploration and exploitation.

[0130] To ensure stable training, DQN employs experience replay and a target network mechanism. Specifically, when the agent interacts with the environment, it replays the experienced samples... Store in the replay cache pool At fixed time intervals, from A small batch of samples is randomly selected for training. The mean squared error loss of the Q-network is calculated, and gradient descent is performed to update the parameters. Specifically:

[0131] First, for each sample, based on the current target network parameters... Calculate the target Q value ,in, As a discount factor, The target network parameters are copied from the online network parameters from several steps ago. It is the next state of the sample.

[0132] Then, use the currently online Q network. Calculate the estimated Q-value by minimizing the loss function. To update network parameters ( (This refers to the batch size).

[0133] After every few training steps, the online network parameters are softly updated or periodically copied and synchronized to the target network to ensure stable convergence of training.

[0134] Through repeated iterations, the above process will gradually bring the Q-network closer to the true value function. After training convergence, for any given state... Both DQN and DQN can output the various actions. Value estimation, selection during navigation decision-making The action with the highest value This means obtaining the optimal wireless charging decision under the current state. Through this deep reinforcement learning method, the system can learn the optimal path selection strategy hidden in complex environments, overcoming the shortcomings of traditional algorithms in adapting to environmental uncertainties.

[0135] (3) Strategy deployment and online adaptation:

[0136] The trained DQN model can be deployed in the cloud or in a roadside unit (RSU) for online navigation services. When a vehicle sends a navigation request, real-time vehicle status and environmental data are acquired as state inputs, and the DQN model calculates and outputs the optimal action (i.e., the selected WCL).

[0137] Considering the constantly changing real-world environment, the DQN model can be periodically trained or fine-tuned using newly collected data, ensuring its adaptability to new road conditions and driving behaviors over long-term operation. Since the DQN decision time is only one forward network computation (milliseconds), and the MSDA algorithm's cost calculation for each candidate path is also in milliseconds, the total decision-making time of the two-layer algorithm in this embodiment is within seconds, fully meeting the real-time requirements of dynamic navigation.

[0138] (4) Upper-layer output and interface to the execution layer:

[0139] The DQN reinforcement learning module at time... Output optimal action (or (This indicates that the current round will not enter the WCL). To achieve execution with a single issuance, the decision-making layer packages the action and the result of MSDA into a navigation instruction and issues it to the execution layer. The navigation instruction includes: the executable entry path. (like This is the shortest path to the destination. Selecting a charging segment and charging strategy parameters (such as) (SoC security upper and lower limits and interrupt conditions), and can include a reference path after the exit. As a prompt; finally, the replanning trigger condition (reaching) is given simultaneously. (or SoC threshold) so that the execution layer can immediately return to the loop when the environment changes or a stage is completed.

[0140] III. Execution Level: Decision Execution and Status Feedback Mechanism

[0141] The bottom layer is the execution layer, where electric vehicles follow the decision-making route (including the current vehicle position). Shortest path to WCL entrance + WCL(m,n) - Shortest path from WCL exit to destination It travels and charges on the WCL, while simultaneously feeding back its status to the decision-making level through vehicle-to-infrastructure communication, forming a closed-loop control.

[0142] Specifically, this embodiment utilizes a mechanism where a vehicle agent navigates according to navigation instructions and provides timely status feedback, thereby achieving a closed-loop integration of planning and execution. The entire execution and feedback process is as follows:

[0143] (1) Decision-making and path execution:

[0144] Whenever the decision-making level determines the current optimal wireless charging lane Then, the corresponding atomic navigation command is sent to the vehicle agent, and the command is executed in three stages:

[0145] Arrival at the entrance: Vehicles follow the shortest path sequence output by MSDA. (The entry path can be executed) from the current vehicle location Drive to the selected WCL entrance. .

[0146] Dynamic charging: the vehicle along Drive and charge at a constant power Dynamic charging yields an amount of electricity determined by the length of the road segment, the actual vehicle speed, and the remaining battery capacity.

[0147] Exit: After charging is complete, exit from the exit. Drive out and take the shortest path from the exit to the destination. Proceed to the destination. If the DQN decision is... In this case, the vehicle will not enter any WCL (Widelined Logic Container Registry), but will instead travel directly from its current location along the shortest path on a regular road. Heading to the destination.

[0148] (2) Status feedback and continuous optimization:

[0149] During the process, the vehicle continuously reports its status via the vehicle-to-everything (V2X) network, including its current location, remaining battery power, driving time, driving speed, and the amount of electricity acquired during charging. This information is used to calculate the actual driving cost and charging cost, and to generate rewards that are fed back to DQN for policy updates.

[0150] After each navigation execution, record the experience quadruple. :in It is the state before execution. It is WCL that was selected. It is a reward in the form of a negative number representing the overall travel cost. This is the new state after execution. The accumulated data in the experience replay pool is used for DQN training, and the strategy is continuously optimized through the target Q-value formula.

[0151] (3) Closed-loop control and multi-step decision-making:

[0152] Once the vehicle completes the current navigation instructions and reaches the exit... (Or the battery level drops to a threshold or the distance to the WCL is less than a threshold), a new set of candidate wireless charging lanes is recalculated using MSDA. and characteristics The system then proceeds to the next round of decision-making in the DQN reinforcement learning module. This iterative mechanism ensures that the navigation scheme can be continuously optimized in a dynamic environment, taking into account path cost, charging efficiency, and grid load balance.

[0153] Example 2

[0154] One embodiment of the present invention provides a dynamic charging navigation system for new energy vehicles based on a hybrid algorithm, comprising:

[0155] The perception module is configured to acquire perception information, including prerequisite information reflecting user preferences and basic task requirements, and real-time information reflecting the instantaneous state of the vehicle and the environment.

[0156] The decision-making module is configured to: based on perception information, dynamically decide the optimal charging navigation path for new energy vehicles by fusing the mirror-symmetric Dijkstra algorithm and the MSDA-DQN hybrid algorithm of deep reinforcement learning. The charging navigation path includes a wireless charging lane and the corresponding optimal executable path. The executable path includes the path from the current vehicle location to the entrance of the wireless charging lane and the path from the exit of the wireless charging lane to the destination location.

[0157] The execution module is configured to: provide driving and charging navigation for new energy vehicles based on the optimal charging navigation path;

[0158] The MSDA-DQN hybrid algorithm employs a two-layer decision-making process. The lower layer utilizes the mirror-symmetric Dijkstra algorithm to search for the optimal executable path for each candidate wireless charging lane in the road network, with the goal of minimizing travel cost. The upper layer utilizes deep reinforcement learning, combined with perception information, to select the wireless charging lane with the lowest sum of travel and charging costs that meets safety constraints from the candidate wireless charging lanes. The selected wireless charging lane and its corresponding optimal executable path are then combined to form the optimal charging navigation path.

[0159] Example 3

[0160] One embodiment of the present invention provides a computer program product, including a computer program that, when executed by a processor, implements the aforementioned dynamic charging navigation method for new energy vehicles based on a hybrid algorithm.

[0161] Example 4

[0162] In one embodiment of the present invention, a non-transitory computer-readable storage medium is provided for storing computer instructions. When the computer instructions are executed by a processor, they implement the aforementioned dynamic charging navigation method for new energy vehicles based on a hybrid algorithm.

[0163] Example 5

[0164] One embodiment of the present invention provides an electronic device, including: a processor, a memory, and a computer program; wherein, the processor is connected to the memory, and the computer program is stored in the memory. When the electronic device is running, the processor executes the computer program stored in the memory, so that the electronic device executes the aforementioned dynamic charging navigation method for new energy vehicles based on a hybrid algorithm.

[0165] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0166] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0167] While the specific embodiments of the present invention have been described above in conjunction with the accompanying drawings, this is not intended to limit the scope of protection of the present invention. Those skilled in the art should understand that various modifications or variations that can be made by those skilled in the art without creative effort based on the technical solutions of the present invention are still within the scope of protection of the present invention.

Claims

1. A hybrid algorithm-based dynamic charging navigation method for new energy vehicles, characterized in that, In response to a navigation request of a target new energy vehicle, the following steps are performed: acquiring perception information, including prerequisite information reflecting user preferences and task basic requirements and real-time information reflecting instantaneous state of the vehicle and the environment; based on the perception information, dynamically deciding an optimal charging navigation path of the new energy vehicle by fusing a mirror-symmetry Dijkstra algorithm and a MSDA-DQN hybrid algorithm of deep reinforcement learning, the charging navigation path including a wireless charging lane and a corresponding optimal executable path, the executable path including a path from a current vehicle location to an entrance of the wireless charging lane and a path from an exit of the wireless charging lane to a terminal location; the optimal executable path search is to perform twice shortest path search on each candidate wireless charging lane, the twice shortest path forming the optimal executable path, and the twice shortest path search including: starting from the current vehicle location, running the Dijkstra algorithm to find a shortest path to the entrance of the wireless charging lane; starting from the exit of the wireless charging lane, running the Dijkstra algorithm in reverse to find a shortest path to the terminal location; according to the optimal charging navigation path, performing driving and charging navigation on the new energy vehicle; wherein the MSDA-DQN hybrid algorithm adopts upper and lower layer decision-making, the lower layer uses the mirror-symmetry Dijkstra algorithm to perform optimal executable path search on each candidate wireless charging lane in the road network with the optimization target of minimum driving cost; the upper layer uses deep reinforcement learning to select, from the candidate wireless charging lanes, a wireless charging lane with the lowest sum of driving cost and charging cost and meeting safety constraints, and to form the optimal charging navigation path by combining the selected wireless charging lane and the corresponding optimal executable path; The travel cost is the path cost of the vehicle passing through the charging navigation path, and is the sum of the instantaneous path cost weights of all road segments in the charging navigation path quantifying; wherein the real-time path cost weight of the road segment includes an energy consumption converted power cost and a driving time cost, which is expressed by a formula as: wherein, represents the cost of electricity for the vehicle to pass through the road segment , is the electric energy consumption per unit distance for the vehicle, is the unit price of electric energy, is the length of the road segment ; represents the travel time cost of the road segment , is the user time value coefficient, represents the time required to pass through the road segment , is the real-time traffic speed; The charging cost refers to the expense incurred by a vehicle during dynamic charging in a wireless charging lane. It is calculated based on the amount of electricity charged, the unit price of electricity, the user's congestion risk preference coefficient, and the congestion coefficient. The calculation is expressed by the formula: wherein, is a charging amount, is a wireless charging lane is a unit price, is a user congestion risk preference coefficient, is a congestion coefficient; Charging power subject to two constraints: one is the available charging space left by the upper limit of the battery capacity minus the current remaining power SoC, and the other is the maximum energy that can be provided by the vehicle through the road segment multiplied by the charging power , expressed in the formula as: wherein, is the WCL road segment length, is the real-time traffic speed of the road segment, is the full capacity, is the predicted state of charge upon entering the WCL from the current state of charge is the minimum of the predicted consumption along the shortest path from the current location to the WCL entrance m calculated along the underlying MSDA, ensuring neither the road segment provided power nor the battery remaining capacity is exceeded.

2. The new energy vehicle dynamic charging navigation method based on a hybrid algorithm according to claim 1, wherein, the prerequisite information includes a start point and a terminal location, an initial power at the time of departure, a user value coefficient per unit time, a user risk preference coefficient for congestion, and a minimum safety SoC threshold of the vehicle battery; the real-time information includes a current vehicle location, a remaining power of the current vehicle, a real-time traffic speed of a vehicle adjacent road segment, a current charging price of the wireless charging lane, and a congestion coefficient on the wireless charging lane.

3. The new energy vehicle dynamic charging navigation method based on a hybrid algorithm according to claim 1, characterized in that, The selection of the wireless charging lane from candidate wireless charging lanes that minimizes the sum of driving and charging costs while satisfying safety constraints involves defining a Markov decision model for the dynamic charging navigation process, including a state space. Action space reward function State transition Based on Markov decision models, deep reinforcement learning is used to evaluate the current state of each candidate wireless charging lane. Value, take The candidate wireless charging lane with the highest value is selected as the optimal wireless charging lane. The state space The states in the state space include vehicle and environment instantaneous states, user prior preferences, candidate wireless charging lanes set and driving cost and charging cost of each candidate wireless charging lane. Action space The action in the action space is defined as selecting one candidate wireless charging lane from the candidate wireless charging lane set for charging or skipping all the wireless charging lanes to go directly to the destination. Reward function includes a driving phase penalty, a charging completion reward, and a power safety constraint penalty; State transition After performing the action, the vehicle enters the next state .

4. A new energy vehicle dynamic charging navigation system based on a hybrid algorithm, characterized in that, A new energy vehicle dynamic charging navigation method based on a hybrid algorithm is adopted, as claimed in any one of claims 1-3, including: a perception module configured to acquire perception information, including prerequisite information reflecting user preferences and task basic requirements and real-time information reflecting instantaneous state of the vehicle and the environment; a decision module configured to, based on the perception information, dynamically decide an optimal charging navigation path of the new energy vehicle by fusing a mirror-symmetry Dijkstra algorithm and a MSDA-DQN hybrid algorithm of deep reinforcement learning, the charging navigation path including a wireless charging lane and a corresponding optimal executable path, the executable path including a path from a current vehicle location to an entrance of the wireless charging lane and a path from an exit of the wireless charging lane to a terminal location; The execution module is configured to perform driving and charging navigation on the new energy vehicle according to the optimal charging navigation path. The MSDA-DQN hybrid algorithm adopts upper and lower layer decision-making, the lower layer uses the mirror symmetry Dijkstra algorithm to perform optimal executable path search on each candidate wireless charging lane in the road network with the minimum driving cost as the optimization target; the upper layer uses deep reinforcement learning to select a wireless charging lane with the lowest sum of driving cost and charging cost and meeting the safety constraint from the candidate wireless charging lanes in combination with the perception information, and the selected wireless charging lane and the corresponding optimal executable path are combined to form the optimal charging navigation path.

5. A computer program product comprising a computer program, characterized in that, The computer program is executed by the processor to implement the new energy vehicle dynamic charging navigation method based on the hybrid algorithm according to any one of claims 1-3.

6. A non-transitory computer-readable storage medium, comprising: The non-transitory computer readable storage medium is used to store computer instructions, and the computer instructions are executed by the processor to implement the new energy vehicle dynamic charging navigation method based on the hybrid algorithm according to any one of claims 1-3.

7. An electronic device, comprising: It comprises: A processor, a memory and a computer program; wherein the processor is connected with the memory, the computer program is stored in the memory, and when the electronic device is running, the processor executes the computer program stored in the memory to enable the electronic device to implement the new energy vehicle dynamic charging navigation method based on the hybrid algorithm according to any one of claims 1-3.

Images