A tcpn-based electric vehicle charging path planning method and system

Abstract

This invention belongs to the field of path planning technology, specifically relating to a method and system for electric vehicle charging path planning based on TCPN. This method maps data such as the travel time and maximum capacity of each road segment in the traffic network, the number of idle charging piles, charging power, queuing time, and total charging service cost in charging stations, as well as grid load factor, charging carbon emissions, and electricity price fluctuation coefficients in the power grid, into locations, transitions, and time constraints in a time-constrained Petri net. A structural controller is added to control the model. The optimal charging path is obtained by solving all feasible charging paths in the model with the comprehensive goal of minimizing time, cost, and carbon emissions. This shortens the charging path planning solution time, establishes a more refined electric vehicle charging path model, and the planned paths better meet users' real-time response needs, achieving synergistic optimization of charging efficiency and grid stability.

Description

Technical Field

[0001] This invention belongs to the field of path planning technology, specifically relating to a method and system for electric vehicle charging path planning based on TCPN. Background Technology

[0002] With the accelerated global energy transition, the number of electric vehicles (EVs) has increased significantly. The large-scale application of EVs not only helps alleviate the energy crisis and environmental pollution, but is also an important pathway to achieving carbon neutrality in the transportation sector. However, with the rapid increase in EV ownership, the operational efficiency and service quality of charging infrastructure face severe challenges. The randomness of vehicle arrivals, the variability of charging times, and the fluctuations in grid load make the allocation and scheduling of charging resources extremely complex, thus necessitating research on real-time optimization decisions for charging routes.

[0003] Traditional path planning algorithms (such as) Traditional methods, such as Dijkstra's algorithm, only consider static road network topology and employ single-objective optimization, neglecting dynamic constraints like user time windows, leading to planning results that deviate from actual needs. Traditional modeling methods are inefficient, and solving integer linear programming (ILP) problems in large-scale road networks is time-consuming and cannot meet real-time response requirements. With technological advancements, Chinese patent document CN119737970B proposes a new energy vehicle charging guidance method and system based on controlled Petri nets. This method constructs an extended road network topology map based on surrounding map information, charging station capacity information, and traffic conditions of the new energy vehicle's current location. Electrical attributes are added to charging station nodes to construct a Petri net model containing charging station information. A structured controller with constrained transition emission is added to the Petri net model to obtain a controlled Petri net model. A depth-first search method and the ILP method are used to perform distance, road condition, and cost optimization searches on the feasible space of the controlled Petri net model to obtain the optimal transition sequence. The optimal transition sequence is then selected from the optimal transition sequence; the optimal transition sequence is the optimal path for the new energy vehicle from the starting marker, through the optimal set of charging station markers, to the ending marker. This method can improve the efficiency of path planning to some extent, but it only considers road information, traffic information and charging station information to build a Petri net model, resulting in low modeling accuracy and thus poor path planning performance, which cannot meet the dynamic balance of traffic, charging stations, power grid and user needs. Summary of the Invention

[0004] The purpose of this invention is to provide a method and system for electric vehicle charging path planning based on TCPN, in order to solve the problem of poor planning results in existing path planning methods.

[0005] This invention provides an electric vehicle charging path planning method based on TCPN to solve the above-mentioned technical problems. The method includes: acquiring traffic data, charging station data, power grid data, vehicle status data, and user demand data; establishing an electric vehicle charging path planning model based on TCPN; the model's database includes road segment database, charging station database, and power grid database; transitions include road segment traffic transitions and charging transitions; the model is controlled by a structural controller, which dynamically constrains the charging path according to the enabling conditions of the transitions; traffic data includes road segment travel time and maximum road segment capacity; charging station data includes the number of idle charging piles, charging power, queuing time, and total charging service cost; power grid data includes power grid load factor, charging carbon emissions, and electricity price fluctuation coefficient; vehicle status data includes the vehicle's current state of charge, target state of charge, average driving speed, and remaining driving range; and determining the optimal electric vehicle charging path from all feasible charging paths in the model with the objective of minimizing the sum of time, cost, and carbon emissions.

[0006] Furthermore, the enabling conditions for road segment traffic change are: the road segment in the road segment database is in a free-flowing state; the vehicle has already arrived at the starting point of the road segment; the time when the vehicle arrives at the starting point of the road segment meets the time constraint of road segment traffic change; and the real-time traffic flow of the road segment is less than or equal to the maximum capacity of the road segment. The time constraint of road segment traffic change is determined based on the basic travel time of the road segment, the congestion coefficient, the road grade coefficient, and the time fluctuation coefficient. The larger the real-time traffic flow, the larger the congestion coefficient; the higher the road traffic efficiency, the smaller the road grade coefficient; and the longer the travel time caused by random factors, the larger the fluctuation coefficient.

[0007] Furthermore, the enabling conditions for charging transition are: there are a number of idle charging piles in the charging station warehouse, the current battery level of the vehicle is less than or equal to the charging trigger threshold set by the user, the charging time required meets the time constraint of charging transition, and the real-time load of the grid node associated with the charging station is less than or equal to the maximum carrying capacity. The lower limit of the time constraint of charging transition is determined based on the charging time taken to charge the vehicle from its current remaining battery level to the user's desired charging target battery level under the current charging power, and the upper limit of the time constraint of charging transition is determined based on the maximum charging time acceptable to the user.

[0008] Furthermore, the upper limit of the time constraint for charging transitions is:

[0009] in, The maximum charging time acceptable to the user. This indicates the current remaining battery power of the vehicle.

[0010] Furthermore, the structure controller is also used to perform at least one of the following constraint controls: grid overload constraint, charging station congestion constraint, and emergency demand constraint. The grid overload constraint refers to the structure controller reducing the charging power of associated charging stations and extending the charging time constraint when it detects that the grid load rate is greater than a set grid load rate. The charging station congestion constraint refers to the structure controller lowering the charging transition enable priority of a charging station when it detects that the queuing time at a charging station is greater than a set queuing time. The emergency demand constraint refers to the structure controller extending the upper limit of the charging time constraint and prioritizing the nearest charging station when it detects that the user's travel urgency is high.

[0011] Furthermore, the objective of minimizing the sum of time, cost, and carbon emissions refers to minimizing the weighted sum of the time objective, cost objective, and carbon emission objective; the time objective is determined based on road segment travel time, charging time, and queuing time; the cost objective is determined based on the total cost of charging services and road segment travel costs; and the carbon emission objective is determined based on charging carbon emissions and driving carbon emissions.

[0012] Furthermore, the process for determining the weights of the time target, cost target, and carbon emission target is as follows: vehicle status data is used as the query, and traffic data, charging station data, power grid data, and user demand data are used as the key-value pairs. These four modal data and vehicle status data are input into the attention mechanism to obtain the weights of the four modal data. The weights of the time target, cost target, and carbon emission target are determined based on the weights of each modal data.

[0013] Furthermore, the weights of the time target are determined based on the weights of traffic data and charging station data, the weights of the cost target are determined based on the weights of charging station data and user demand, and the weights of the carbon emission target are determined based on the weights of power grid data and user demand.

[0014] Furthermore, the process for determining the weights of the time target, cost target, and carbon emission target is as follows: Vehicle status data is used as the query, and five modal data types—traffic data, charging station data, power grid data, user demand data, and user behavior perception data—are used as the key-value pairs. These five modal data types and vehicle status data are input into the attention mechanism to obtain the weights of the five modal data types. The weights of the time target, cost target, and carbon emission target are determined based on the weights of each modal data type. User behavior perception data includes high-frequency travel density, historical charging trigger thresholds, and average charging amount.

[0015] To address the aforementioned technical problems, this invention also provides an electric vehicle charging path planning system based on TCPN, including a processor. The processor is used to output charging paths according to a TCPN-based electric vehicle charging path planning method. The method includes: acquiring traffic data, charging station data, power grid data, vehicle status data, and user demand data; establishing an electric vehicle charging path planning model based on TCPN; the database of the electric vehicle charging path planning model includes road segment databases, charging station databases, and power grid databases; transitions include road segment traffic transitions and charging transitions; the electric vehicle charging path planning model is controlled by a structural controller, which dynamically constrains the charging path according to the enabling conditions of the transitions; traffic data includes road segment travel time and maximum road segment capacity; charging station data includes the number of idle charging piles, charging power, queuing time, and total charging service cost; power grid data includes power grid load factor, charging carbon emissions, and electricity price fluctuation coefficient; vehicle status data includes the vehicle's current state of charge, target state of charge, average driving speed, and remaining driving range; and determining the optimal electric vehicle charging path from all feasible charging paths in the electric vehicle charging path planning model with the objective of minimizing the sum of time, cost, and carbon emissions.

[0016] Furthermore, the enabling conditions for road segment traffic change are: the road segment in the road segment database is in a free-flowing state; the vehicle has already arrived at the starting point of the road segment; the time when the vehicle arrives at the starting point of the road segment meets the time constraint of road segment traffic change; and the real-time traffic flow of the road segment is less than or equal to the maximum capacity of the road segment. The time constraint of road segment traffic change is determined based on the basic travel time of the road segment, the congestion coefficient, the road grade coefficient, and the time fluctuation coefficient. The larger the real-time traffic flow, the larger the congestion coefficient; the higher the road traffic efficiency, the smaller the road grade coefficient; and the longer the travel time caused by random factors, the larger the fluctuation coefficient.

[0017] Furthermore, the enabling conditions for charging transition are: there are a number of idle charging piles in the charging station warehouse, the current battery level of the vehicle is less than or equal to the charging trigger threshold set by the user, the charging time required meets the time constraint of charging transition, and the real-time load of the grid node associated with the charging station is less than or equal to the maximum carrying capacity. The lower limit of the time constraint of charging transition is determined based on the charging time taken to charge the vehicle from its current remaining battery level to the user's desired charging target battery level under the current charging power, and the upper limit of the time constraint of charging transition is determined based on the maximum charging time acceptable to the user.

[0018] Furthermore, the upper limit of the time constraint for charging transitions is:

[0019] in, The maximum charging time acceptable to the user. This indicates the current remaining battery power of the vehicle.

[0020] Furthermore, the structure controller is also used to perform at least one of the following constraint controls: grid overload constraint, charging station congestion constraint, and emergency demand constraint. The grid overload constraint refers to the structure controller reducing the charging power of associated charging stations and extending the charging time constraint when it detects that the grid load rate is greater than a set grid load rate. The charging station congestion constraint refers to the structure controller lowering the charging transition enable priority of a charging station when it detects that the queuing time at a charging station is greater than a set queuing time. The emergency demand constraint refers to the structure controller extending the upper limit of the charging time constraint and prioritizing the nearest charging station when it detects that the user's travel urgency is high.

[0021] Furthermore, the objective of minimizing the sum of time, cost, and carbon emissions refers to minimizing the weighted sum of the time objective, cost objective, and carbon emission objective; the time objective is determined based on road segment travel time, charging time, and queuing time; the cost objective is determined based on the total cost of charging services and road segment travel costs; and the carbon emission objective is determined based on charging carbon emissions and driving carbon emissions.

[0022] Furthermore, the process for determining the weights of the time target, cost target, and carbon emission target is as follows: vehicle status data is used as the query, and traffic data, charging station data, power grid data, and user demand data are used as the key-value pairs. These four modal data and vehicle status data are input into the attention mechanism to obtain the weights of the four modal data. The weights of the time target, cost target, and carbon emission target are determined based on the weights of each modal data.

[0023] Furthermore, the weights of the time target are determined based on the weights of traffic data and charging station data, the weights of the cost target are determined based on the weights of charging station data and user demand, and the weights of the carbon emission target are determined based on the weights of power grid data and user demand.

[0024] Furthermore, the process for determining the weights of the time target, cost target, and carbon emission target is as follows: Vehicle status data is used as the query, and five modal data types—traffic data, charging station data, power grid data, user demand data, and user behavior perception data—are used as the key-value pairs. These five modal data types and vehicle status data are input into the attention mechanism to obtain the weights of the five modal data types. The weights of the time target, cost target, and carbon emission target are determined based on the weights of each modal data type. User behavior perception data includes high-frequency travel density, historical charging trigger thresholds, and average charging amount.

[0025] The beneficial effects of the above technical solution are as follows: This invention is an improved invention that maps data such as the travel time and maximum capacity of each road segment in the traffic network, the number of idle charging piles, charging power, queuing time, and total charging service cost in charging stations, and the grid load rate, charging carbon emissions, and electricity price fluctuation coefficient in the power grid into locations, transitions, and time constraints in a time-constrained Petri net (TCPN). A structural controller is added to control the model, resulting in a dynamically constrained TCPN electric vehicle charging path planning model. The optimal charging path is obtained by solving all feasible charging paths in the model with the goal of minimizing time, cost, and carbon emissions. The capacity of road segments, charging stations, and grid locations is adaptively adjusted according to real-time traffic flow, number of idle charging piles, and grid load, shortening the charging path planning solution time. The established electric vehicle charging path model is more refined, and the planned path is more in line with the real-time response needs of users. Furthermore, it achieves a dynamic balance between traffic network status, grid load, user demand, and charging facilities, realizing the synergistic optimization of charging efficiency and grid stability. Attached Figure Description

[0026] Figure 1 This is a flowchart of the electric vehicle charging path planning based on TCPN of the present invention;

[0027] Figure 2 This is a diagram of the charging process structure expressed by TCPN modeling in this invention;

[0028] Figure 3 This is a flowchart of the multimodal fusion process of the present invention;

[0029] Figure 4 This is a flowchart of the multimodal fusion process that incorporates user behavior perception in this invention. Detailed Implementation

[0030] To make the objectives, technical solutions, and advantages of the present invention clearer, the specific embodiments of the present invention will be further described below with reference to the accompanying drawings.

[0031] This invention integrates parameters such as traffic network status, power grid status, user demand, and charging facilities to establish a multimodal data fusion framework based on a 6-tuple time-constrained Petri net (6-Tuple TCPN) to realize dynamic planning of electric vehicle charging paths, enabling real-time optimization decisions for charging paths.

[0032] Method Implementation

[0033] This invention provides a TCPN-based electric vehicle charging path planning method, the implementation principle of which is as follows: Figure 1 As shown below, a detailed explanation will follow.

[0034] S1. Acquire traffic data, charging station data, power grid data, vehicle status data, and user demand data, and establish an electric vehicle charging path planning model based on TCPN.

[0035] like Figure 2 As shown, the elements of electric vehicle charging path planning are abstracted into a six-tuple time-constrained Petri net (6-Tuple TCPN), using... This involves mapping traffic networks, charging stations, power grids, vehicle status, and user demand to data sources, transitions, and time constraints. A dynamic structure controller is then incorporated to dynamically constrain charging paths through transition enablement conditions. Traffic data includes road segment travel time and maximum road segment capacity; charging station data includes the number of idle charging piles, charging power, queuing time, and total charging service cost; power grid data includes grid load factor, charging carbon emissions, and electricity price fluctuation coefficients; vehicle status data includes current state of charge, target state of charge, average driving speed, and remaining driving range; and user demand data includes travel urgency, cost tolerance threshold, carbon emission preference value, and time tolerance threshold.

[0036] 1. The set of places, P.

[0037] The depot set P includes four types of depots: traffic nodes, charging station nodes, power grid nodes, and vehicles, which are also known as road segment depots. Charging station warehouse Power grid storage : and vehicle garage .

[0038] 1) Road section depot .

[0039] Each road section of the warehouse The attributes representing the road segment (ij) from node i to j include:

[0040] ①Dynamic passage time : Represents the road segment database that takes into account real-time traffic flow changes. The passage time, ,in This refers to the basic travel time of a road segment when there is no congestion. Let be the congestion coefficient at time t. ≥1 indicates that the higher the real-time traffic flow, the greater the congestion coefficient.

[0041] ② Road grade coefficient Distinguish between highways (roads with high traffic efficiency, which can...) =0.8), urban arterial roads, secondary arterial roads and branch roads, etc., are used to correct the travel time. The higher the road traffic efficiency, the smaller the road grade coefficient.

[0042] ③ Weather influencing factors Rain, snow, strong winds and other weather conditions can increase travel time, and the more severe the weather, the greater the impact of weather factors. ,in The influence coefficient, Let t represent the weather level at time t, for example, 0 for sunny, 1 for light rain / light breeze, 2 for heavy rain / strong wind, and 3 for torrential rain / typhoon.

[0043] ④ Time fluctuation coefficient The longer the travel time caused by random factors, the greater the fluctuation coefficient, and the larger the allowable range of travel time fluctuation. The time fluctuation coefficient increases the travel time fluctuation caused by random factors such as traffic lights or temporary lane closures.

[0044] Status update rules: The data is updated periodically based on the traffic cloud platform. The data is updated periodically based on weather station data.

[0045] 2) Charging station storage .

[0046] Each charging station warehouse The attributes include:

[0047] ① Number of available charging stations Number of available charging stations at time t.

[0048] ② Charging power The charging power of the charging piles in charging station k at time t is affected by the grid load.

[0049] ②Queueing time Calculated based on real-time charging order data, ,in Let t be the number of vehicles queuing at the charging station. This represents the average charging time.

[0050] ④ Total cost of charging service Users at charging station locations The total cost required to complete a single charge is the core calculation basis for the "cost target" in path planning.

[0051] in, The total electricity consumption measured at the charging station; To improve the charging efficiency of charging stations; The actual amount of electricity charged to the user's vehicle; The basic charging electricity price indicates the charging station The base price per unit of electricity set at time t for different charging types; This is the peak-valley electricity price coefficient, which is a coefficient that dynamically adjusts the basic charging electricity price in conjunction with the local power grid's peak-valley time-of-use electricity price policy. It is used to reflect the impact of differences in power grid supply costs at different times on charging costs. This is a service surcharge, representing the charging station. A fixed fee or percentage-based fee is charged for providing additional services to users.

[0052] ⑤ Service reliability Calculated based on historical fault data, .

[0053] ⑥ Geographical coordinates It is used to calculate the distance between road segments, determine the location of vehicles, and connect road segments with charging stations. It is the foundation for building the entire road network topology.

[0054] ⑦ Distribution of charging pile types This parameter is used to distinguish between the number of DC fast chargers and AC slow chargers, and is used when calculating charging time. If the vehicle does not support fast charging, or the user prefers slow charging (which may be better for the battery or cheaper), then when selecting a charging route, it is necessary to match suitable charging stations based on the distribution of charging station types and use the corresponding charging power to calculate the charging time.

[0055] Service reliability, geographic location coordinates, and charging pile type distribution serve as the foundational static data for the model, primarily utilized in the selection, calculation, and constraint phases of route planning. For example, service reliability can be used as a soft constraint or cost item in the route optimization objective.

[0056] 3) Power grid storage .

[0057] Each power grid depot Corresponding charging station locations The power grid partition m it belongs to has the following attributes:

[0058] ① Real-time grid load factor ,in Let be the real-time load of grid zone m at time t. Let m be the maximum load capacity of the power grid zone.

[0059] ② Renewable energy share This is used for the dynamic calculation of carbon emission targets. Let be the amount of renewable energy generation, such as solar and wind power, within grid zone m at time t. Let m be the total power generation of grid zone m at time t.

[0060] ③ Electricity price fluctuation coefficient In conjunction with the formulation of time-of-use electricity pricing policies, Let t be the multiple of the electricity price relative to the benchmark electricity price.

[0061] ④ Carbon emissions from charging At time t, the corresponding charging station location of the power grid depot The carbon emission intensity value of charging.

[0062] in, It is the carbon emission intensity of the basic energy source (such as thermal power) in this power grid zone. It is the carbon emission intensity of renewable energy (typically 0 or very low).

[0063] 4) Vehicle storage area .

[0064] Vehicle storage Electric vehicles participating in path planning In the Petri net model, vehicles exist as tokens carrying data colors. Essentially, it is the data structure definition of this token, which is used to centrally store all the dynamic states and static attributes of the vehicle for use in transition enable judgment, time constraint calculation, attention mechanism query, and objective function calculation.

[0065] The attributes of each vehicle parking space include:

[0066] Vehicle_SN: Unique vehicle identifier;

[0067] Battery capacity (kWh);

[0068] : User's desired target state of charge (%) for the vehicle; User-defined charging trigger threshold (%), when At that time, the charging demand is triggered; : The maximum charging time (in hours) acceptable to the user; Vehicle's current state of charge (%) : Location identifier of the vehicle's current location. This attribute is dynamically updated and points to the location of the road segment where the vehicle token is currently located. or charging station storage ;

[0069] Average driving speed (km / h); Remaining driving range (km); : Travel urgency; C_max: Cost tolerance threshold (yuan);

[0070] λ_carbon_pref: Carbon emission preference.

[0071] 2. The transition set T and the time constraint function τ of the transition.

[0072] The change set includes changes in road traffic conditions. and the evolution of charging .

[0073] 1) Changes in road traffic conditions Traffic changes on each road section Corresponding road section depot Time constraints on passage behavior The time constraints for changes in road segment traffic are determined based on the basic traffic time of the road segment, congestion coefficient, road grade coefficient, and time fluctuation coefficient:

[0074] 2) Evolution of charging technology Every charging transition Corresponding charging station locations Time constraints on charging behavior The lower limit of the time constraint for charging transition is determined based on the charging time required to charge the vehicle from its current remaining charge to the user's desired charge level at the current charging power, while the upper limit is determined based on the maximum charging time acceptable to the user.

[0075] in, This refers to the vehicle's current state of charge. B represents the target state of charge of the vehicle, and B represents the battery capacity. The maximum charging time acceptable to the user (time tolerance threshold), when At lower speeds, users are more tolerant of charging time. At higher levels, tolerance decreases.

[0076] 3. Initial Identification .

[0077] Defined in vector form, ,in:

[0078] This is the initial state vector of the road segment. Indicates road segment Initially unobstructed; 0 indicates initial closure (e.g., during construction). Let be the initial state vector of the charging station. (Initial number of idle stakes); Let be the initial state vector of the power grid. (Initial grid load factor).

[0079] 4. Flow relation F (arc set)

[0080] Represented using an adjacency matrix, F is ( Matrix, where:

[0081] This indicates that there is a flow relationship from place p to transition t (where p is the preceding place of t); This indicates a flow relationship from transition t to place p (where p is the subsequent place of t); the remaining elements =0.

[0082] For example, road section storage Traffic changes (from node 1 to 2) The flow relationship is (Assuming that the road section will be reached after passage) (The starting point).

[0083] 5. Structural controller C.

[0084] The electric vehicle charging path planning model is controlled by a structural controller.

[0085] 1) The structural controller is used to dynamically constrain the charging path based on the enabling conditions of the transition. For any transition t( The basic enabling condition is to meet the triple constraints, namely, the state constraints, time constraints, and resource constraints of the preceding warehouse must all be met.

[0086] ①Pre-existing warehouse status constraints.

[0087] The triggering of transition t requires all its preceding libraries to be present. (Referring to the place that is pointed to by the flow relation F) is in a valid state as a prerequisite, and the requirements for the preceding place state are different for different types of transitions.

[0088] In response to changes in road traffic conditions Corresponding road section depot (The preceding depot is the road section depot) The location of the vehicle is currently identified by the database. The following must be met: Status indicators of road section depots , indicating road segment Currently unobstructed; status indicator This indicates that the vehicle has now reached the section of road. The starting point has the necessary conditions for passage.

[0089] In response to the changes in charging Corresponding charging station locations The front-end depot is the charging station depot. With vehicle status database It must meet the following requirements: charging station storage location status indicator This indicates a charging station. There are available charging stations. The value is equal to the current number of idle stakes. Vehicle status database status indicator satisfy: , The charging trigger threshold set by the user indicates that the vehicle's current battery level has reached the point where charging is required.

[0090] ②Time constraints.

[0091] The triggering of transition t must meet its time constraint interval. That is, the current moment Falling Within the upper and lower limits, the time constraint calculation logic and enabling requirements for different types of transitions are as follows:

[0092] In response to changes in road traffic conditions : That is, the section of road the vehicle arrives at. The starting point must be within the currently passable time range of that road segment; the model no longer needs to set a static "vehicle location database" for each node. Conversely, because vehicle tokens move within the Petri net, when the vehicle token is in the warehouse... When the time is right, it means the vehicle is in the section of road. Driving upwards. When the token passes through a transition. Then, it will be moved to the next depot (such as...) This indicates that the vehicle has arrived at a new road segment. If the delay is due to traffic congestion... The extension will cause the structural controller C to update in real time. Ensure that passage is permitted only during permitted times. trigger.

[0093] In response to the changes in charging This means that the theoretical minimum charging time has been met, and the current time has not exceeded the user's maximum tolerable charging time.

[0094] ③ Resource constraints.

[0095] change The triggering of this event requires ensuring that the relevant resources (transportation resources, charging resources, and power grid resources) are in a sustainable state. The specific constraints are as follows:

[0096] In response to changes in road traffic conditions : Must meet the requirements of the road section Real-time traffic flow , For road section The maximum traffic capacity is determined to avoid traffic safety risks caused by exceeding the flow limit. The structural controller C determines this by acquiring real-time traffic monitoring data. Whether it meets the standard, if Then disable This guides vehicles to choose alternative routes.

[0097] In response to the changes in charging : Must meet charging station storage requirements Associated power grid warehouse Real-time load To prevent charging activities from causing grid overload, the structural controller C obtains data through its connection to the grid monitoring platform. ,like Then it is allowed. Triggered; if Then temporarily disable This continues until the grid load returns to a safe range. Meanwhile, Must meet charging station storage requirements Number of idle piles Constraints.

[0098] In other words, the enabling conditions for road segment traffic change are: the road segment in the road segment database is in a free-flowing state, the vehicle has already arrived at the starting point of the road segment, the time when the vehicle arrives at the starting point of the road segment meets the time constraint of road segment traffic change, and the real-time traffic flow of the road segment is less than or equal to the maximum capacity of the road segment. The time constraint of road segment traffic change is determined based on the basic travel time of the road segment, the congestion coefficient, the road grade coefficient, and the time fluctuation coefficient. The larger the real-time traffic flow, the larger the congestion coefficient; the higher the road traffic efficiency, the smaller the road grade coefficient; and the longer the travel time caused by random factors, the larger the fluctuation coefficient.

[0099] The enabling conditions for charging transition are: there are a number of idle charging piles in the charging station warehouse, the current battery level of the vehicle is less than or equal to the charging trigger threshold set by the user, the charging time required meets the time constraint of charging transition, and the real-time load of the grid node associated with the charging station is less than or equal to the maximum carrying load. The time constraint of charging transition is determined based on the charging time taken to charge the vehicle from its current remaining battery level to the user's desired charging target battery level under the current charging power, and the maximum charging time acceptable to the user.

[0100] 2) Dynamic triggering logic.

[0101] The structural controller C performs at least one constraint control among power grid overload constraint, charging station congestion constraint and emergency demand constraint through a three-step process of "state monitoring - condition judgment - constraint adjustment".

[0102] ① Grid Overload Constraint: When the structural controller detects that the grid load rate is higher than the set grid load rate, it reduces the charging power of the associated charging station, extends the charging time constraint, and sends a notification to the user stating "The current grid load is high, and the charging time may be extended." For example, when it detects... When the value is greater than 0.9, the structural controller C triggers the charging power limit:

[0103] Reduce charging power :

[0104] Simultaneously extend the upper limit of the charging transition time constraint:

[0105] ② Charging station congestion constraint: When the structure controller detects that the queuing time of a charging station is greater than the set queuing time, it will reduce the enabling priority of the charging station's charging transition.

[0106] For example, when monitored The structural controller C will The enabling priority is reduced, prioritizing the transition of charging stations with more idle charging piles and shorter queuing times.

[0107] ③ Emergency Demand Constraint: When the structure controller detects that the user's travel urgency is high, it extends the upper limit of the charging time constraint and prioritizes the nearest charging station.

[0108] For example, users set an "emergency travel" mode (such as...) (If the current range is less than 20% and the destination distance is greater than the current range), the structure controller C relaxes the upper limit of the time constraint and prioritizes the nearest charging station for transition.

[0109] S2. Determine the optimal electric vehicle charging path from all feasible charging paths in the electric vehicle charging path planning model with the goal of minimizing the sum of time, cost, and carbon emissions.

[0110] The objective of minimizing the sum of time, cost, and carbon emissions refers to minimizing the weighted sum of time, cost, and carbon emission objectives. The time objective is determined based on road segment travel time, charging time, and queuing time. The cost objective is determined based on the total cost of charging services and road segment travel costs. The carbon emission objective is determined based on charging carbon emissions and driving carbon emissions. An attention-weighted multimodal fusion mechanism is used to generate an optimization objective function.

[0111] Multi-objective optimization function for:

[0112] in, The weights for time targets, cost targets, and carbon emission targets are respectively. Let the time objective function be... The objective function is cost. The objective function is for carbon emissions.

[0113] The time objective is determined based on road segment travel time, charging time, and queuing time. The time objective function minimizes the total time consumption of "road segment travel time + charging time + queuing time", and the formula is:

[0114] in, This is a set of road segments and charging stations within the planned route. The higher the weight of travel time and queuing time in the fusion features, the better. The higher the priority in the overall objective function, the better.

[0115] The cost target is determined based on the total cost of charging services and the road segment toll cost, minimizing "total cost of charging services + road segment toll cost", as shown in the formula:

[0116] in, The toll for road segment ij. This represents the unit electricity consumption cost of the vehicle. When the charging cost has a high weight in the fused features, With the optimization priority increased, the algorithm will prioritize selecting low-cost charging stations.

[0117] Carbon emission targets are determined based on charging carbon emissions and driving carbon emissions, minimizing the sum of "charging carbon emissions + driving carbon emissions" and linking the charging carbon emission intensity in the fusion characteristics. The formula for the proportion of renewable energy in the power grid is:

[0118] in, Carbon emissions per unit mileage traveled by a vehicle. This refers to vehicle speed. The higher the weight of carbon emission intensity in the fusion features, the better. The higher the optimization priority, the more likely the algorithm will select charging stations associated with low-carbon power grids.

[0119] 1. Determine the weights of time targets, cost targets, and carbon emission targets.

[0120] like Figure 3 As shown, in one implementation, vehicle status data is used as the query, and traffic data, charging station data, power grid data, and user demand data—four modalities—are used as key-value pairs. These four modalities and vehicle status data are then input into an attention mechanism to obtain the weights of the four modalities. The attention mechanism is as follows:

[0121]

[0122] Where Q represents the vehicle state vector query, and the multimodal data consists of key-value pairs (K, V). Let Variance be the variance.

[0123] The following is a detailed introduction.

[0124] 1) Data standardization.

[0125] In the process of multimodal data fusion, due to significant differences in the dimensions and numerical ranges of the various modal data (e.g., travel time is measured in "minutes", SOC in "%", and cost in "yuan"), it is necessary to first standardize all data to map them to a unified dimension [0, 1] to eliminate dimensional interference. Let a certain modal data item be x, and its standardization formula is:

[0126] in, This is the historical minimum value of this data item. This represents the historical maximum value for that data item. For example, road segment travel time. The value range is [5, 60 minutes], when at a certain moment =At 25 minutes, after standardization

[0127] .

[0128] 2) Similarity calculation and dynamic weight generation.

[0129] Vehicle Status Query Vector ;in This is the standardized value of the current state of charge; The target state of charge is standardized. This is a standardized value for the average driving speed; This is a standardized value for the remaining driving range.

[0130] Traffic data key vector Value vector ;in, Standardized / real-time value of basic travel time for road segment; / Standardized value / real-time value of congestion coefficient; / The standard value of the road grade coefficient / the actual value; This represents the standardized value of the maximum traffic capacity of the road segment divided by the actual value.

[0131] Charging station data key vector Value vector ;in, Standardized value / real-time value of the number of idle piles; / This refers to the standardized / real-time value of the charging power. Standardized / real-time value for queue waiting time; / This represents the standardized / real-time value of the total charging cost.

[0132] Power grid data key vector Value vector ;in, Standardized / real-time value of grid load factor; Standardized / real-time value of carbon emission intensity for charging; This represents the standardized / real-time value of the electricity price fluctuation coefficient.

[0133] User Requirement Key Vector Value vector ;in, / Standardized value / actual value for travel urgency; / The standardized value of the cost tolerance threshold / the actual value; Standardized value / actual value for carbon emission preference; The time tolerance threshold is the standardized value / actual value.

[0134] Using the vehicle state query vector Q as a baseline, the correlation between Q and each modal key vector is calculated using the cosine similarity formula, quantifying the impact of different modal data on the current vehicle charging path planning. The similarity formula is:

[0135] in, For modal identification, Represents the vector dot product. This represents the L2 norm. Similarity results. The closer the value is to 1, the higher the degree of matching between the modal data and the current vehicle state, and the greater the decision value for charging route planning.

[0136] Preferably, to avoid an imbalance in weight allocation due to absolute differences in similarity calculation results, a Softmax function is introduced to normalize the similarity of each modality and generate initial weights. :

[0137] Where λ is a temperature coefficient (value 2.0), used to amplify similarity differences and enhance weight discrimination. The initial weights must satisfy... This ensures that the sum of the contributions of each modality's data is 1.

[0138] Preferably, the initial weights are dynamically adjusted based on the preference coefficients in the user's needs to form the final weights. It meets customers' typical needs such as "emergency travel mode", "low-carbon mode" and "economic mode".

[0139] 3) Determine the weights of time targets, cost targets, and carbon emission targets based on the weights of each modal data.

[0140] The value vectors of each modality are assigned according to their final weights. Perform weighted summation to generate a unified multimodal fusion feature vector. The formula is:

[0141] in, As for traffic data weighting, Weighting of charging station data, For power grid data weights, Weighting based on user demand.

[0142] The attention mechanism outputs the weights of the four modalities. and multimodal fusion feature vector The multimodal fusion feature vector is used as a path score or a state value function. The multimodal fusion feature vector output from the attention mechanism and the weights of the four modalities are input into a neural network (such as a fully connected layer), outputting the target weights. , and .

[0143] Specifically, the weights for the time objective are determined based on traffic data weights and charging station data weights; the weights for the cost objective are determined based on charging station data weights and user demand weights; and the weights for the carbon emission objective are determined based on power grid data weights and user demand weights. The weights of the time objective, cost objective, and carbon emission objective are dynamically adjusted by the attention score. + + =1, ensuring that the priority of each objective is highly matched with user needs and vehicle status.

[0144] For example: , , 0.4, 0.6, 0.3, and 0.7 are the hyperparameters of the attention mechanism + neural network model architecture, obtained through model training, thus avoiding manual setting of hyperparameters and improving the interpretability of the output results.

[0145] 2. Solve the multi-objective optimization function to determine the optimal electric vehicle charging path among all feasible charging paths in the electric vehicle charging path planning model.

[0146] The optimization algorithm used during solution can be a hybrid GA-PSO algorithm, with particle encoding as transition sequences and fitness function as a multi-objective optimization function. The GA-PSO hybrid algorithm consists of five stages: initialization → GA evolution → PSO refinement → convergence determination → result output. Each stage is deeply integrated with the TCPN model and multimodal fusion mechanism, which will be explained in detail below.

[0147] Phase 1: Population initialization (compatible with GA chromosomes and PSO particles).

[0148] TCPN-based storage topology and The connection relationships are used to generate all feasible charging paths through a "random walk" method, and the transitions of each path are transformed into particle position vectors. Simultaneously, as the chromosome of GA, t represents time:

[0149] PSO particle velocity vector Initialize to random small values To avoid path jumps caused by excessive initial velocity; incorporate multimodal fusion. With constraint penalties, calculate the penalty for each particle. Record the best position in the particle's history (Initially self) ) and the population's historical best position (Initially) The largest particle.

[0150] in: for The theoretical maximum value; k is the constraint violation penalty coefficient; To restrict violations of the signage.

[0151] Phase 2: GA Evolution Operation (Global Search for Diverse Paths).

[0152] For the initial population, perform selection, crossover, and mutation operations of GA to generate a new generation population, and calculate the selection probability of each particle. N particles are selected according to probability to enter the cross-pool. The higher the particle's height, the greater the probability of it being selected, thus preserving high-quality paths;

[0153] Particles in the randomized crossover pool are paired and their gene segments at the crossover points are exchanged to generate offspring chromosomes. The mutated offspring population is used as a temporary population after GA evolution, and its... ,renew .

[0154] Phase 3: PSO Refinement Operation (Locally Converged Optimization Solution).

[0155] Using the population evolved from GA as the initial particles for PSO, the optimal path is refined through a velocity-position update mechanism to accelerate convergence:

[0156] Speed ​​update: Combining the particle's own historical best Best in history Update particle velocity:

[0157] in, Inertia factor and As a learning factor, and It is a random number. For the total number of particles in this group, For time;

[0158] Position Update: Adjust particle position based on updated velocity (path encoding):

[0159] in, To ensure that the gene position is an integer (e.g., library ID), rounding is used.

[0160] calculate of ,like of Then update If there are in the population Greater than The particles then update .

[0161] Phase 4: Collaborative Iteration and Convergence Determination.

[0162] Collaborative Iteration: Repeat the cycle of "GA Evolution (Phase 2) → PSO Refinement (Phase 3)" and update based on the multimodal fusion mechanism after a certain number of iterations. The weights;

[0163] Convergence criterion: The optimal fitness value of a population over a certain number of generations. Stop iterating when the rate of change is small or after the number of iterations reaches the limit.

[0164] Phase 5: Optimal path output and TCPN mapping.

[0165] After iteration, Pareto-level 1 particles (paths) are extracted from the population to form the Pareto optimal front. The integer encoding of the optimal particles is converted into actual path descriptions. The "segment-charging station" sequence in the path is mapped to a TCPN transition sequence, ensuring that the transitions satisfy the time constraint τ and the enabling conditions of the structural controller C. The path in the optimal front and its associated path are then displayed to the user. Sub-target values ​​are selected by the user based on their needs, allowing them to choose the final path.

[0166] The six-tuple time-constrained Petri net is a dynamic storage capacity mechanism. The capacity of road segments, charging stations, and grid storage facilities is adaptively adjusted according to real-time traffic flow, number of idle charging piles, and grid load, thereby resolving resource competition among multiple vehicles or during transitions. The GA-PSO hybrid algorithm combines the global search capability of the genetic algorithm (GA) with the local convergence advantage of particle swarm optimization (PSO), thus shortening the solution time for charging route planning and significantly improving solution efficiency.

[0167] As a preferred implementation, a user behavior-aware charging function is integrated. That is, based on the existing framework, a new user historical behavior modality is added. By sensing the user's daily driving routes and charging behavior, a "historical-real-time" dual-dimensional demand prediction model is constructed to automatically calculate charging time and required energy, providing a more personalized decision-making basis for multimodal fusion. Historical (e.g., the last 30 days) driving route data, historical charging behavior data, and correlations between energy consumption per unit mileage and destination mileage in different scenarios are automatically collected through the in-vehicle system, charging app, and navigation records to form a user historical behavior dataset. Historical driving route data includes frequently traveled routes, high-frequency road segments, and destination types. Data is updated daily at midnight. Historical charging behavior data includes historical charging records, charging trigger thresholds, and single charging amounts.

[0168] based on Based on real-time vehicle status, a dual-drive prediction model of "rules + statistics" is constructed to predict users' charging time and charging capacity.

[0169] Charging time prediction: If the real-time location is less than or equal to a set distance (e.g., 5km) from a historically frequent destination, and the current time falls within a historical charging period, the predicted charging time is the historical average charging period. If the real-time traffic congestion coefficient ρ is higher than a set value (e.g., 20%) than the same period in history, the charging time will be postponed to a set time after arrival at the destination (e.g., 30 minutes) to avoid missing charging due to travel delays. If a user is detected heading to a new destination, the predicted charging time is the middle of the stay, based on the destination type and the current time, and referencing the average stay duration of similar destinations.

[0170] Charging power prediction: Based on the distance from the current location to the predicted destination. Energy consumption under the same road conditions as in the past Calculate the energy consumption of the remaining journey. Based on the user's historical target SOC, the required charging amount is estimated. If a high-frequency trip is predicted shortly after charging (e.g., within 12 hours), an additional redundancy (e.g., 10%) of power will be added; if the real-time weather is low temperature or heavy rain, energy consumption will be adjusted based on historical data. Increase charging capacity and make corresponding corrections. .

[0171] Based on charging demand forecasts and real-time vehicle SOC status, a multi-threshold trigger + personalized push SMS reminder mechanism is designed to ensure users receive timely charging advice and avoid range anxiety. The system personalizes SMS push notifications and reminders based on four core pieces of information: charging necessity, recommended time, recommended charging station, and required battery capacity, and according to user historical preferences. The SMS messages include three feedback options: "Confirm Charging," "Remind Later," and "No Charging Required." After the user replies with the corresponding command, the system records the feedback and adjusts subsequent reminder strategies accordingly.

[0172] The newly added user behavior perception and intelligent reminder module is integrated into the existing multimodal fusion framework in the following ways:

[0173] like Figure 4 As shown, user behavior perception data (i.e., user historical behavior data) ) as the sixth type of modal data New user historical behavior key vector With Value Vector .in, / This represents the standardized / real-time value of high-frequency stroke density. Standardize the historical charging trigger threshold value / real-time value. / This represents the standardized / real-time value of the average charge amount.

[0174] Vehicle status data is used as the query, and traffic data, charging station data, power grid data, user demand data, and user behavior perception data are used as the key-value pairs. These five modal data and vehicle status data are input into the attention mechanism to obtain the weights of the five modal data. The weights of time target, cost target, and carbon emission target are determined based on the weights of each modal data.

[0175] If the user selects "Confirm Charging" after the smart reminder is triggered, the multimodal fusion mechanism automatically increases the Key vector similarity Sim(Q, K) of the recommended charging stations. char The GA-PSO algorithm is guided to prioritize including the charging station in the optimal path; if the user selects "no charging required", the charging change T for the current time period is reduced. charge The enabling priority is adjusted to avoid forced charging recommendations during path planning.

[0176] System Implementation

[0177] The present invention provides an electric vehicle charging path planning system based on TCPN, comprising a processor, the processor being configured to output a charging path according to a TCPN-based electric vehicle charging path planning method, the method comprising:

[0178] This system acquires traffic data, charging station data, power grid data, vehicle status data, and user demand data to establish an electric vehicle charging path planning model based on TCPN. The model's database includes road segment databases, charging station databases, and power grid databases. Changes include road segment traffic changes and charging changes. The model is controlled by a structural controller, which dynamically constrains the charging path based on the enabling conditions of the changes. Traffic data includes road segment travel time and maximum road segment capacity. Charging station data includes the number of idle charging piles, charging power, queue waiting time, and total charging service cost. Power grid data includes grid load factor, charging carbon emissions, and electricity price fluctuation coefficient. Vehicle status data includes the vehicle's current state of charge, target state of charge, average driving speed, and remaining driving range.

[0179] The enabling conditions for road segment traffic change are: the road segment status in the road segment database is unobstructed, the vehicle has currently arrived at the starting point of the road segment, the time when the vehicle arrives at the starting point of the road segment meets the time constraint of road segment traffic change, and the real-time traffic flow of the road segment is less than or equal to the maximum capacity of the road segment. The time constraint of road segment traffic change is determined based on the basic travel time of the road segment, the congestion coefficient, the road grade coefficient, and the time fluctuation coefficient. The larger the real-time traffic flow, the larger the congestion coefficient; the higher the road traffic efficiency, the smaller the road grade coefficient; and the longer the travel time caused by random factors, the larger the fluctuation coefficient.

[0180] The enabling conditions for charging transition are: there are a number of idle charging piles in the charging station warehouse, the current battery level of the vehicle is less than or equal to the charging trigger threshold set by the user, the charging time required meets the time constraint of charging transition, and the real-time load of the grid node associated with the charging station is less than or equal to the maximum carrying load. The time constraint of charging transition is determined based on the charging time taken to charge the vehicle from its current remaining battery level to the user's desired charging target battery level under the current charging power, and the maximum charging time acceptable to the user.

[0181] The optimal electric vehicle charging path is determined from all feasible charging paths in the electric vehicle charging path planning model with the objective of minimizing the sum of time, cost, and carbon emissions. The specific implementation process of this method has been described in detail in the method implementation description, and will not be repeated here.

[0182] This invention integrates data from traffic, charging, power grid, vehicle status, and user modalities to establish a TCPN-based electric vehicle charging path planning model. Optimization aims to minimize the sum of time, cost, and carbon emissions, overcoming the limitations of traditional single-objective optimization. It achieves synergistic optimization of charging efficiency and power grid stability, integrating structural controllers with dynamic time delay constraints to solve multimodal resource competition problems. The modeling incorporates multi-dimensional time constraints, including absolute time constraints (such as the maximum acceptable charging time for users, fixed time periods for time-of-use pricing, the latest arrival time of users at their destination, and fixed operating hours of charging stations), relative time constraints (such as the impact of road congestion coefficients, battery SOC, power grid load, and charging power on real-time state time), and probabilistic time distributions (such as sudden road congestion, charging station queuing time, charging pile power fluctuations, and vehicle speed fluctuations). This enables refined modeling of the entire charging system process, including vehicle arrival, resource allocation, charging execution, and fault handling.

Claims

1. A method for electric vehicle charging path planning based on TCPN, characterized in that, include: The system acquires traffic data, charging station data, power grid data, vehicle status data, and user demand data, and establishes an electric vehicle charging path planning model based on TCPN. The system includes road segment locations, charging station locations, and power grid locations. The system also includes road segment traffic changes and charging changes. The electric vehicle charging path planning model is controlled by a structural controller, which dynamically constrains the charging path based on the enabling conditions of the changes. Traffic data includes road segment travel time and maximum road segment capacity; charging station data includes the number of available charging piles, charging power, queuing time and total charging service cost; power grid data includes power grid load factor, charging carbon emissions and electricity price fluctuation coefficient; vehicle status data includes the vehicle's current state of charge, vehicle's target state of charge, average driving speed and remaining driving range. The optimal electric vehicle charging path is determined from all feasible charging paths in the electric vehicle charging path planning model with the goal of minimizing the sum of time, cost, and carbon emissions.

2. The electric vehicle charging path planning method based on TCPN according to claim 1, characterized in that, The enabling conditions for road segment traffic change are: the road segment status in the road segment database is unobstructed, the vehicle has now arrived at the starting point of the road segment, the time when the vehicle arrives at the starting point of the road segment meets the time constraint of road segment traffic change, and the real-time traffic flow of the road segment is less than or equal to the maximum traffic capacity of the road segment. The time constraints on road segment traffic changes are determined based on the basic travel time of the road segment, the congestion coefficient, the road grade coefficient, and the time fluctuation coefficient. The larger the real-time traffic flow, the larger the congestion coefficient; the higher the road traffic efficiency, the smaller the road grade coefficient; and the longer the travel time caused by random factors, the larger the fluctuation coefficient.

3. The electric vehicle charging path planning method based on TCPN according to claim 1, characterized in that, The enabling conditions for charging transition are: there are available charging piles in the charging station warehouse, the vehicle's current battery level is less than or equal to the user-set charging trigger threshold, the charging time meets the time constraint of charging transition, and the real-time load of the grid node associated with the charging station is less than or equal to the maximum carrying capacity. The lower limit of the time constraint of charging transition is determined based on the charging time required to charge the vehicle from its current remaining battery level to the user's desired charging target battery level under the current charging power, and the upper limit of the time constraint of charging transition is determined based on the maximum charging time acceptable to the user.

4. The electric vehicle charging path planning method based on TCPN according to claim 3, characterized in that, The upper limit of the time constraint for charging transition is: ， in, The maximum charging time acceptable to the user. This indicates the current remaining battery power of the vehicle.

5. The electric vehicle charging path planning method based on TCPN according to claim 1, characterized in that, The structure controller is also used to perform at least one of the following constraint controls: grid overload constraint, charging station congestion constraint, and emergency demand constraint. Grid overload constraint refers to the structure controller reducing the charging power of associated charging stations and extending the charging time constraint when it detects that the grid load rate is greater than a set grid load rate. Charging station congestion constraint refers to the structure controller lowering the charging transition enable priority of a charging station when it detects that the queuing time at a charging station is greater than a set queuing time. Emergency demand constraint refers to the structure controller extending the upper limit of the charging time constraint and prioritizing the nearest charging station when it detects that the user's travel urgency is high.

6. The electric vehicle charging path planning method based on TCPN according to claim 1, characterized in that, The objective of minimizing the sum of time, cost, and carbon emissions refers to minimizing the weighted sum of time, cost, and carbon emission targets. The time target is determined based on road travel time, charging time, and queuing time; the cost target is determined based on the total cost of charging services and road travel costs; and the carbon emission target is determined based on charging carbon emissions and driving carbon emissions.

7. The electric vehicle charging path planning method based on TCPN according to claim 6, characterized in that, The process of determining the weights of time objectives, cost objectives, and carbon emission objectives is as follows: vehicle status data is used as the query, and traffic data, charging station data, power grid data, and user demand data are used as the key-value pairs. These four types of data and vehicle status data are input into the attention mechanism to obtain the weights of the four types of data. The weights of time objectives, cost objectives, and carbon emission objectives are determined based on the weights of each type of data.

8. The electric vehicle charging path planning method based on TCPN according to claim 7, characterized in that, The weights for time targets are determined based on traffic data weights and charging station data weights; the weights for cost targets are determined based on charging station data weights and user demand weights; and the weights for carbon emission targets are determined based on power grid data weights and user demand weights.

9. The electric vehicle charging path planning method based on TCPN according to claim 6, characterized in that, The process for determining the weights of time, cost, and carbon emission targets is as follows: Vehicle status data is used as the query, and five modal data types—traffic data, charging station data, power grid data, user demand data, and user behavior perception data—are used as the key-value pairs. These five modal data types and vehicle status data are input into an attention mechanism to obtain the weights of the five modal data types. The weights of the time, cost, and carbon emission targets are determined based on the weights of each modal data type. User behavior perception data includes high-frequency trip density, historical charging trigger thresholds, and average charging amount.

10. A TCPN-based electric vehicle charging path planning system, comprising a processor, characterized in that, The processor is used to output a charging path according to the TCPN-based electric vehicle charging path planning method as described in any one of claims 1-9.

Images