Differential privacy-based electric vehicle track synthesis method

Abstract

The invention relates to an electric vehicle track synthesis method based on differential privacy, and the method is characterized in that the method comprises the steps: obtaining a real electric vehicle track data set of an urban geographic space, carrying out the meshing of the urban geographic space according to regions, and obtaining a divided geographic grid; according to a geographic grid and a road network of an urban geographic space, track movement mode extraction is carried out on a real electric vehicle track to obtain a geographic grid transfer matrix and a region transfer matrix which meet differential privacy constraints; according to the charging pile set in the urban geographic space, modeling is carried out on the consumption of the vehicle driving electric quantity, and an electric quantity state updating rule and a charging guiding strategy of the vehicle are obtained; based on Markov sampling, combined with the region transfer matrix, the grid cell transfer matrix, the electric quantity state updating rule and the charging guide strategy, a synthetic track is generated, the legality of the synthetic track is verified, and an electric vehicle synthetic track data set is obtained; according to the method, the privacy leakage risk is effectively avoided, electric vehicle trajectory data with high behavior fidelity can be generated, and the method is suitable for various privacy protection data analysis and intelligent travel decision-making scenes and has good engineering practicability and popularization prospects.

Description

Technical Field

[0001] The present invention belongs to the field of privacy data protection, and in particular relates to an electric vehicle trajectory synthesis method based on differential privacy. Background Art

[0002] As a key vehicle for clean energy transportation, electric vehicles are experiencing rapid growth in both ownership and usage. With the advancement of intelligent connected technologies, the massive amounts of trajectory data generated by electric vehicles are becoming increasingly crucial for key tasks such as urban traffic management, energy scheduling optimization, infrastructure planning, and user behavior modeling. In particular, high-quality trajectory data can effectively improve model prediction accuracy and system efficiency in applications such as charging station site selection, grid load forecasting, and green travel recommendations.

[0003] However, electric vehicle trajectory data often carries sensitive personal information about users, such as where they live, their daily travel patterns, and frequently visited charging stations. It strongly identifies their identities and spatial behavior preferences. If this data is directly disclosed or used without processing, it is vulnerable to privacy threats such as re-identification attacks and trajectory inference attacks, posing serious challenges to user privacy protection and data sharing compliance. Therefore, a trajectory data synthesis mechanism that protects individual privacy while preserving the behavioral characteristics of electric vehicle trajectories is urgently needed to support various research and industrial practices involving electric vehicle trajectory analysis.

[0004] At the same time, electric vehicles exhibit distinct trajectory characteristics distinct from those of traditional fuel-powered vehicles. These are primarily manifested in limited range, a strong dependence on the distribution of charging stations for route accessibility, and the prevalence of charging activities within travel routes. Traditional vehicle trajectory generation models typically focus solely on spatial mobility patterns and fail to capture the unique characteristics of electric vehicles in terms of range, battery life, and recharging behavior. Therefore, directly applying general trajectory synthesis strategies to electric vehicle scenarios often results in trajectories that fail to visit any charging points, unreachable routes, and energy depletion, severely impacting the practical value and usability of the generated data. Summary of the Invention

[0005] In order to solve the problems existing in the background technology, the present invention provides an electric vehicle trajectory synthesis method based on differential privacy, comprising:

[0006] S1: Obtain a real electric vehicle trajectory dataset in the urban geographic space, and grid the urban geographic space by region to obtain a completed geographic grid;

[0007] S2: Extract the trajectory movement pattern of real electric vehicle trajectories based on the geographic grid and road network of the urban geographic space to obtain the geographic grid transfer matrix and regional transfer matrix that meet the differential privacy constraints;

[0008] S3: Model the vehicle's power consumption based on the collection of charging piles within the city's geographic space to obtain the vehicle's power status update rules and charging guidance strategy;

[0009] S4: Based on Markov sampling, the synthetic trajectory is generated and its legitimacy is verified by combining the regional transfer matrix, grid unit transfer matrix, power state update rule and charging guidance strategy to obtain the electric vehicle synthetic trajectory dataset.

[0010] Another aspect of the present invention provides an electric vehicle trajectory synthesis system based on differential privacy, the system comprising a memory and a processor; the memory is used to store an application; the processor is used to run the application and execute the electric vehicle trajectory synthesis method based on differential privacy.

[0011] Another aspect of the present invention provides a computer storage medium, on which a remote monitoring program is stored. When the remote monitoring program is executed by a processor, the electric vehicle trajectory synthesis method based on differential privacy is implemented.

[0012] The present invention has at least the following beneficial effects

[0013] This method, based on abstract modeling of real electric vehicle trajectory data, introduces "range limitation" and "charging preference guidance" mechanisms while maintaining the statistical characteristics of existing mobility patterns. By constructing a semantically guided sampling strategy, it dynamically controls charging point access and energy accessibility during trajectory synthesis. Compared to traditional trajectory generation methods, this method not only effectively avoids the risk of privacy leakage but also generates electric vehicle trajectory data with high behavioral fidelity, making it suitable for a variety of privacy-preserving data analysis and intelligent travel decision-making scenarios. It has excellent engineering practicality and widespread application prospects. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] Figure 1 Schematic diagram of the method of the present invention. DETAILED DESCRIPTION

[0015] The following describes the embodiments of the present invention by means of specific examples, and those skilled in the art can easily understand other advantages and effects of the present invention from the contents disclosed in this specification. The present invention can also be implemented or applied through other different specific embodiments, and the details in this specification can also be modified or changed in various ways based on different viewpoints and applications without departing from the spirit of the present invention. It should be noted that the illustrations provided in the following embodiments are only schematic illustrations of the basic concept of the present invention, and the following embodiments and features in the embodiments can be combined with each other without conflict.

[0016] See also Figure 1The present invention provides an electric vehicle trajectory synthesis method based on differential privacy, comprising:

[0017] S1: Obtain a real electric vehicle trajectory dataset in the urban geographic space, and grid the urban geographic space by region to obtain a completed geographic grid;

[0018] Preferably, said gridding the urban geographic space by region comprises:

[0019] The urban geographic space is divided into multiple non-overlapping areas, each area is gridded, and the grids that cannot be reached by vehicles are removed. Each area unit A i The grid division rules are as follows:

[0020]

[0021] S i =g i ×a 2

[0022] Among them, L(A i ) represents area A i The total length of all roads in i ) represents area A i The area of ​​​​the β∈[0~1] represents the weight parameter; S u Indicates area A i Each grid size; a 2 Represents the area of ​​the base grid.

[0023] In this embodiment, the trajectory of an electric vehicle (EV) can be represented in geographic space as a sequence of locations recorded in chronological order, including charging events and sensitive location access information. Therefore, the trajectory is formally represented as: Tr = {p1, p2, ..., p n}, where p i =(x i ,y i ,t i ), (x i ,y i ) represents geographic coordinates (such as longitude and latitude), t i Indicates the recorded position point p i The acquisition time is n, and n is the number of sampling positions in the trajectory. We use represents the trajectory dataset, Indicates the number of trajectories in the trajectory dataset.

[0024] Road network model: A road network is a spatial topological structure consisting of roads and intersections in the real world. It can be formally modeled as a directed graph G = (V, E), where V represents a set of nodes and each node v∈V corresponds to a geographical location in the road network, such as an intersection, junction, or road endpoint. Represents a set of edges, each edge (v i ,v j )∈E represents the slave node v i To node v j There is one directional road segment.

[0025] Given a trajectory dataset Its geographic space is covered by the geographic network G = (V, E), so a trajectory can be represented by the geographic network, and each trajectory point represents a grid in the geographic network. Based on this, the geographic space can be divided into a grid of non-overlapping units, and each trajectory point l i Can be mapped to a grid cell In fact, the density of the road network is positively correlated with the probability of a vehicle trajectory visiting an urban area. Since the cell grid density in different areas is different, the entire urban geographic space can be divided into non-overlapping areas. A i Represents an area unit.

[0026] In this embodiment, step S1 obtains a real electric vehicle trajectory dataset in the urban geographic space, providing raw data support based on actual behavior for subsequent trajectory movement pattern extraction, power modeling, and synthetic trajectory generation. At the same time, the urban geographic space is gridded by region, converting the continuous geographic space into discrete, computable grid units. By removing grids that vehicles cannot reach and adjusting the grid size based on the total length, area, and weight parameters of the roads in the region, the grid division is made more consistent with the actual road network distribution and vehicle accessibility. This not only achieves spatial structuring to facilitate quantitative analysis of vehicle movement patterns, but also provides the basic spatial unit for the construction of the geographic grid transfer matrix in step S2 and the trajectory generation based on Markov sampling in step S4, becoming a key bridge connecting the raw trajectory data and synthetic trajectory generation.

[0027] S2: Extract the trajectory movement pattern of real electric vehicle trajectories based on the geographic grid and road network of the urban geographic space to obtain the geographic grid transfer matrix and regional transfer matrix that meet the differential privacy constraints;

[0028] This method aims to synthesize trajectory data that captures the behavioral characteristics of electric vehicles under a differential privacy protection mechanism. This addresses the problem that existing trajectory synthesis methods fail to consider range constraints and charging behavior, resulting in a lack of semantic rationality in the synthesized data. Based on real electric vehicle trajectories, this method retains the statistical movement patterns of the original trajectories and introduces a unique range constraint modeling and charging behavior guidance mechanism for electric vehicles. This allows the synthesized trajectories to retain the same travel logic and behavioral structure as real trajectories while preserving privacy.

[0029] Preferably, the extraction of the trajectory movement pattern of the real electric vehicle trajectory includes:

[0030] S21: Mapping the trajectory points in the real electric vehicle trajectory to the geographic grid, counting the number of transfers from each geographic grid to other geographic grids in the real electric vehicle trajectory, and normalizing the transfer numbers to construct the initial geographic grid transfer matrix;

[0031]

[0032] Among them, M(C i ,C j ) represents the element in the initial geographic grid transfer matrix M, N(Tr τ ,C i ,C j ) represents the trajectory Tr τ From C i to C j The normalized count of transfers; m-1 represents the trajectory Tr τ Total number of transfers on C; #transfer times C i →C j in Tr τ Represents the trajectory Tr τ From C i to C j The number of transfers.

[0033] S22: Add differential privacy-based Laplace noise to each element in the initial geographic grid transfer matrix, and perform post-processing and normalization in sequence to obtain the geographic grid transfer matrix;

[0034] M(C i ,C j )=M(C i ,C j )+Lap(1 / ∈1)

[0035] Where Lap(1 / ∈1) represents the Laplace noise distribution and ∈1 represents the differential privacy parameter.

[0036] The post-processing includes setting the probability between geographic grid pairs other than those that simultaneously meet the following three conditions to 0:

[0037] Condition 1: Two geographic grids are adjacent;

[0038] Condition 2: The two geographic grids contain the same roads or adjacent roads in the road network;

[0039] Condition 3: In the road network G, the transfer direction between two geographical grids satisfies the legal direction of travel of the road.

[0040] In this embodiment, step S21 maps the trajectory points in the real electric vehicle trajectory to geographic grids, counts the number of transitions between grids, and normalizes them to construct an initial geographic grid transfer matrix. This effectively transforms the scattered original trajectory data into a quantitative basis reflecting the movement patterns of vehicles between grids, extracts the spatial transition patterns of the real trajectory, and provides the original statistical basis for subsequent privacy protection processing. Step S22 adds Laplace noise based on differential privacy to the initial matrix, and obtains the final geographic grid transfer matrix through post-processing (such as retaining only the transition probabilities of adjacent grids that conform to the road network and travel direction) and normalization. This not only prevents the risk of privacy leakage through the differential privacy mechanism, but also ensures the rationality of the transfer matrix in accordance with the actual traffic scenario. It provides quantitative support for the calculation of grid transition probabilities in the subsequent synthetic trajectory, which combines privacy security with behavioral authenticity.

[0041] S23: Extract the start and end regions of each real electric vehicle trajectory, construct region pairs, count the number of identical start and end region pairs, and perform normalization to construct the initial region transfer matrix;

[0042] S24: Add differential privacy-based Laplace noise to each element of the initial regional transfer matrix and normalize it to obtain the regional transfer matrix;

[0043]

[0044] in, represents the elements in the regional transfer matrix; U(A i ,A j ) indicates that the start and end areas after normalization are located at A i and A j The number of trajectories; Lap(1 / ∈2) represents the Laplace noise distribution, ∈2 represents the differential privacy parameter; Represents a collection of regions.

[0045] In this embodiment, step S23 constructs an initial regional transfer matrix by extracting the start and end regions of each real electric vehicle trajectory, constructing region pairs, counting the number of the same start and end region pairs, and performing normalization processing, effectively extracting the start and end movement patterns of real trajectories at the regional level, and converting the scattered trajectory data into a quantitative initial basis reflecting the transfer rules between regions; step S24 adds Laplace noise based on differential privacy to each element on the basis of the initial regional transfer matrix and performs normalization to obtain the regional transfer matrix, which not only prevents the risk of privacy leakage through the differential privacy mechanism, but also ensures the rationality of the regional transfer matrix, enabling it to retain the movement statistical characteristics between regions in the real trajectory, providing quantitative support with both privacy security and behavioral authenticity for the start and end region sampling when synthesizing trajectories later.

[0046] S3: Model the power consumption of the vehicle during driving according to the charging pile set in the urban geospatial area, and obtain the power state update rule and charging guidance strategy of the vehicle;

[0047] Preferably, the power state update rule of the vehicle includes:

[0048] Define that when the vehicle starts, the initial energy of the vehicle is E1, and E1~Uniform(E min ,E max ) follows a uniform distribution; the vehicle updates the power every step during trajectory sampling, and the update rule is as follows:

[0049] Define the geographical grid where the vehicle is currently located as C i ; if there is a charging pile s that satisfies d(C i ,s)<d0, then E i =min(E i-1 -B+ΔE,E max ), otherwise, E i =E i-1 -B, where Uniform represents the uniform distribution function; E[[ID=3l]] min represents the minimum power threshold of the vehicle; E max represents the maximum power threshold of the vehicle; E i represents the power of the vehicle in the geographical grid C i ; d(C i ,s) represents the Euclidean distance between the geographical grid C i and the charging pile s; d0 represents the set distance threshold; ΔE represents the charging amount; B represents the power consumed by the vehicle for each movement of a geographical grid.

[0050] Preferably, the charging guidance strategy of the vehicle includes:

[0051] Define the geographical grid where the vehicle is currently located as C i [[ID=SO]], if the current power E of the vehiclei When it is less than the set threshold value θ0, it enters the charging guidance stage; the charging guidance factor Defined as:

[0052]

[0053] in, Belongs to geographic grid C i adjacent grids, exp represents the exponential function, and r represents the guide radius parameter; Represents a geographic grid The nearest charging station near The Euclidean distance of Represents a geographic grid The Euclidean distance to any charging station s; exp represents the exponential function.

[0054] In this embodiment, step S3 models the vehicle's power consumption based on the collection of charging piles within the city's geographic space, and obtains the vehicle's power state update rules and charging guidance strategies. This process specifically captures the core characteristics that distinguish electric vehicles from traditional fuel vehicles: endurance limitations and charging behavior requirements. Among them, the power state update rules dynamically adjust the vehicle's power level in combination with the distribution of charging piles (for example, the power can be replenished when close to the charging pile, otherwise it continues to consume), accurately simulating the energy changes of the vehicle in different geographical grids; the charging guidance strategy guides the vehicle to the grid near the charging pile through the guidance factor when the power is too low, effectively reflecting the electric vehicle's energy replenishment logic. This not only makes up for the defect of the traditional trajectory model that does not consider endurance and charging factors, but also provides the subsequent synthetic trajectory (step S4) with energy constraints and behavior guidance based on the actual behavior laws of electric vehicles, ensuring that the synthetic trajectory can truly reflect the travel characteristics of electric vehicles under power changes and charging needs, and improving the semantic rationality and practical usability of the synthetic data.

[0055] S4: Based on Markov sampling, the synthetic trajectory is generated and its legitimacy is verified by combining the regional transfer matrix, grid unit transfer matrix, power state update rule and charging guidance strategy to obtain the electric vehicle synthetic trajectory dataset.

[0056] Preferably, step S4 includes:

[0057] S41: Initialize synthetic trajectory dataset

[0058] S42: Define the number of synthetic tracks as represents a dataset of real electric vehicle trajectories;

[0059] S43: For each synthesized trajectory Tr sampled, initialize the power of the vehicle and sample the starting and ending regions (A s , A e ) from the regional transfer matrix, where A s represents the starting region and A e represents the ending region; randomly sample the starting geographic grid C s and the ending geographic grid C e of the synthesized trajectory from the starting region A s and the ending region A e ;

[0060] S44: Sample the length l of each synthesized trajectory from , where represents the trajectory length distribution between the starting geographic grid C s and the ending geographic grid C e ;

[0061] Preferably, the trajectory length distribution between the starting geographic grid and the ending geographic grid includes:

[0062]

[0063] where, Μ l (C s , C e ) represents the transition probability of passing through l steps from the starting geographic grid C s to the ending geographic grid C e ; l max represents the maximum path length from the starting geographic grid C s to the ending geographic grid C e ; l mun represents the minimum path length from the starting geographic grid C s to the ending geographic grid C e .

[0064] In this embodiment, both l max and l min are obtained based on the statistics of the real electric vehicle trajectory dataset.

[0065] S45: Starting from the starting geographic grid of the synthesized trajectory, for the current geographic grid where the vehicle is located as C i , if there is a charging pile s that satisfies d(C i , s) < d0, then E i = min(E i-1 - B + ΔE, E max ), otherwise, E i = E i-1 - B;

[0066] S46: Starting from the starting point of the synthetic trajectory, the current geographic grid is C i , if the current vehicle power E i When it is less than the set threshold θ0, it enters the charging guidance stage; then the geographic grid C i Transfer to its adjacent geographic grid Probability as follows:

[0067]

[0068] Otherwise, the geographic grid C i Transfer to its adjacent geographic grid The probability is as follows:

[0069]

[0070] Among them, M l-i+1 (C i ,C e ) indicates that the i to C e The probability of transition through l-i+1; Represents the geographic grid C in the geographic grid transfer matrix i Transfer to geographic grid The transition probability of

[0071] S47: According to geographic grid C i Transfer to its adjacent geographic grid The probability of getting from the geographic grid C i The next geographic grid is sampled from the adjacent geographic grid until l sampling times to obtain the synthetic trajectory Tr;

[0072] S48: Repeat the above sampling process Synthetic tracks.

[0073] Preferably, performing a validity check on each sampled synthetic trajectory includes:

[0074] Suppose a synthetic trajectory Tr={C1,C2,…,C i ,…,C L}, where C i Represents the i-th grid unit in the synthetic trajectory Tr; the charging point set contained in the synthetic trajectory Tr is defined as i k represents the position index of the kth charging point in the synthetic trajectory Tr, K represents the number of charging points included in the synthetic trajectory Tr; if there is C i If the distance to any charging pile is less than d0, then C iAs charging points; every two adjacent charging points will form a trajectory segment If there is a trajectory segment The length is greater than the vehicle's maximum cruising capacity R max , then the synthetic trajectory is illegal and needs to be resampled to generate a synthetic trajectory.

[0075] In this embodiment, step S4 is based on Markov sampling, integrating the regional transfer matrix, grid unit transfer matrix, power state update rules, and charging guidance strategy. By initializing the synthetic trajectory dataset, determining the number of trajectories, sampling the start and end regions and grids, and setting the trajectory length, the grid transition probability is dynamically calculated based on power changes (a charging guidance factor is introduced to adjust the transition direction when the battery is low). Trajectories are gradually sampled and verified for legitimacy (for example, whether a trajectory segment exceeds the vehicle's maximum range), ultimately generating a synthetic trajectory dataset for electric vehicles. This step fully utilizes the mobility patterns and unique electric vehicle behavior rules extracted in the previous step. Markov sampling ensures that the statistical characteristics of the synthetic trajectory are consistent with the actual trajectory. Legacy verification also avoids unreasonable battery depletion in the trajectory. The resulting dataset effectively protects user privacy while also having high behavioral fidelity and practical usability, supporting scenarios such as privacy-preserving data analysis and intelligent travel decision-making.

[0076] Another aspect of the present invention provides an electric vehicle trajectory synthesis system based on differential privacy, the system comprising a memory and a processor; the memory is used to store an application; the processor is used to run the application and execute the electric vehicle trajectory synthesis method based on differential privacy.

[0077] Another aspect of the present invention provides a computer storage medium, on which a remote monitoring program is stored. When the remote monitoring program is executed by a processor, the electric vehicle trajectory synthesis method based on differential privacy is implemented.

[0078] Those skilled in the art will understand that all or part of the processes in the above-mentioned embodiment methods can be implemented by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, storage, database or other media used in the embodiments provided in this application may include non-volatile and / or volatile memory. Non-volatile memory may include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory may include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in many forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), Synchronous Link DRAM (SLDRAM), Rambus Direct 10RAM (RDRAM), Direct Memory Bus Dynamic RAM (DRDRAM), and Rambus Dynamic RAM (RDRAM), etc.

[0079] In summary, this invention, based on abstract modeling of real electric vehicle trajectory data, introduces "range limitation" and "charging preference guidance" mechanisms while maintaining the statistical characteristics of existing mobility patterns. By constructing a semantically guided sampling strategy, it achieves dynamic control of charging point access and energy accessibility during trajectory synthesis. Compared to traditional trajectory generation methods, this invention not only effectively avoids the risk of privacy leakage but also generates electric vehicle trajectory data with high behavioral fidelity, making it suitable for a variety of privacy-preserving data analysis and intelligent travel decision-making scenarios, with excellent engineering practicality and widespread application prospects.

[0080] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not limiting. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention can be modified or replaced by equivalents without departing from the purpose and scope of the technical solutions, which should all be included in the scope of the claims of the present invention.

Claims

1. A method for synthesizing electric vehicle trajectories based on differential privacy, characterized in that: include: S1; obtain a real electric vehicle trajectory dataset in the urban geographic space, and grid the urban geographic space by region to obtain a completed geographic grid; S2: Extract the trajectory movement pattern of real electric vehicle trajectories based on the geographic grid and road network of the urban geographic space to obtain the geographic grid transfer matrix and regional transfer matrix that meet the differential privacy constraints; S3: Model the vehicle's power consumption based on the collection of charging piles within the city's geographic space to obtain the vehicle's power status update rules and charging guidance strategy; S4: Based on Markov sampling, the synthetic trajectory is generated and its legitimacy is verified by combining the regional transfer matrix, grid unit transfer matrix, power state update rule and charging guidance strategy to obtain the electric vehicle synthetic trajectory dataset.

2. The electric vehicle trajectory synthesis method based on differential privacy according to claim 1 is characterized in that: The gridding of the urban geographic space by region includes: The urban geographic space is divided into multiple non-overlapping areas, each area is gridded, and the grids that cannot be reached by vehicles are removed. Each area unit A i The grid division rules are as follows: S i =g i ×a 2 Among them, L(A i ) represents area A i The total length of all roads in i ) represents area A i The area of ​​​​the β∈[0~1] represents the weight parameter; S i Indicates area A i Each grid size; a 2 Represents the area of ​​the base grid.

3. The electric vehicle trajectory synthesis method based on differential privacy according to claim 1 is characterized in that: The extraction of the trajectory movement pattern of the real electric vehicle trajectory includes: S21: Mapping the trajectory points in the real electric vehicle trajectory to the geographic grid, counting the number of transfers from each geographic grid to other geographic grids in the real electric vehicle trajectory, and normalizing the transfer numbers to construct the initial geographic grid transfer matrix; S22: Add differential privacy-based Laplace noise to each element in the initial geographic grid transfer matrix, and perform post-processing and normalization in sequence to obtain the geographic grid transfer matrix; S23: Extract the start and end regions of each real electric vehicle trajectory, construct region pairs, count the number of identical start and end region pairs, and perform normalization to construct the initial region transfer matrix; S24: Add differential privacy-based Laplace noise to each element in the initial regional transfer matrix and normalize it to obtain the regional transfer matrix.

4. The electric vehicle trajectory synthesis method based on differential privacy according to claim 1 is characterized in that: The vehicle's power status update rules include: When the vehicle starts, the initial energy of the vehicle is defined as E1, E1 ~ Uniform (E min ,E max ) obeys a uniform distribution; the vehicle updates its battery power at each step in trajectory sampling, and the update rules are as follows: Define the geographical grid where the vehicle is currently located as C i ; If there exists a charging pile s such that d(C i , s) < d0, then E i = min(E i-1 - B + ΔE, E max ), otherwise, E i = E i-1 - B, where Uniform represents the uniform distribution function; E min represents the minimum power threshold of the vehicle; E max represents the maximum power threshold of the vehicle; E i represents the power of the vehicle in the geographical grid C i ; d(C i , s) represents the Euclidean distance between the geographical grid C i and the charging pile s; d0 represents the set distance threshold; ΔE represents the charging amount; B represents the power consumed by the vehicle for each movement of one geographical grid.

5. The electric vehicle trajectory synthesis method based on differential privacy according to claim 4 is characterized in that: The vehicle charging guidance strategy includes: Define the geographic grid where the vehicle is currently located as C i , if the current vehicle power E i When it is less than the set threshold value θ0, it enters the charging guidance stage; the charging guidance factor Defined as: in, Belongs to geographic grid C i adjacent grids, exp represents the exponential function, and r represents the guide radius parameter; Represents a geographic grid The nearest charging station near The Euclidean distance of Represents a geographic grid The Euclidean distance to any charging station s; exp represents the exponential function.

6. The electric vehicle trajectory synthesis method based on differential privacy according to claim 5 is characterized in that: The step S4 comprises: S41: Initialize synthetic trajectory dataset S42: Define the number of synthetic tracks as represents a dataset of real electric vehicle trajectories; S43: For each synthetic trajectory Tr, initialize the vehicle's power and sample the start and end areas (A s ,A e ), where A s Indicates the starting area, A e Indicates the end area; from the starting area A s and termination area A e The starting point geographic grid C of the randomly sampled synthetic trajectory s and the end point geographic grid C e ; S44: From The length l of each synthetic trajectory is sampled in Indicates the starting point geographic grid C s and the end point geographic grid C e The distribution of trajectory lengths between S45: Starting from the starting geographical grid of the synthetic trajectory, for the geographical grid where the vehicle is currently located as C i , if there exists a charging pile s such that d(C i , s) < d0, then E i = min(E i-1 - B + ΔE, E max ), otherwise, E i = E i-1 - B; S46: Starting from the starting point of the synthetic trajectory, the current geographic grid is C i , if the current vehicle power E i When it is less than the set threshold θ0, it enters the charging guidance stage; then the geographic grid C i Transfer to its adjacent geographic grid Probability as follows: Otherwise, the geographic grid C i Transfer to its adjacent geographic grid The probability is as follows: Among them, M l-i+1 (C i ,C e ) indicates that the i to C e The probability of transition through l-i+1; Represents the geographic grid C in the geographic grid transfer matrix i Transfer to geographic grid The transition probability of S47: According to geographic grid C i Transfer to its adjacent geographic grid The probability of getting from the geographic grid C i The next geographic grid is sampled from the adjacent geographic grid until l sampling times to obtain the synthetic trajectory Tr; S48: Repeat the above sampling process Synthetic tracks.

7. The electric vehicle trajectory synthesis method based on differential privacy according to claim 6 is characterized in that: The legitimacy check for each sampled synthetic trajectory includes: Suppose a synthetic trajectory Tr={C1,C2,…,C i ,…,C L }, where C i Represents the i-th grid unit in the synthetic trajectory Tr; the charging point set contained in the synthetic trajectory Tr is defined as i k represents the position index of the kth charging point in the synthetic trajectory Tr, K represents the number of charging points included in the synthetic trajectory Tr; if there is C i If the distance to any charging pile is less than d0, then C i As charging points; every two adjacent charging points will form a trajectory segment If there is a trajectory segment The length is greater than the vehicle's maximum cruising capacity R max , then the synthetic trajectory is illegal and needs to be resampled to generate a synthetic trajectory.

8. The electric vehicle trajectory synthesis method based on differential privacy according to claim 6 is characterized in that: The track length distribution between the starting point geographic grid and the end point geographic grid includes: Among them, M l (C s ,C e ) represents the geographic grid C from the starting point s To the destination geographic grid C e The transition probability after l steps; max Represents the geographic grid C from the starting point s To the destination geographic grid C e The maximum path length; l mun Represents the geographic grid C from the starting point s To the destination geographic grid C e The minimum path length.

9. An electric vehicle trajectory synthesis system based on differential privacy, characterized in that: The system includes a memory and a processor; the memory is used to store an application; the processor is used to run the application and execute the electric vehicle trajectory synthesis method based on differential privacy as described in any one of claims 1 to 8.

10. A computer storage medium, characterized in that The computer storage medium stores a remote monitoring program, and when the remote monitoring program is executed by the processor, an electric vehicle trajectory synthesis method based on differential privacy according to any one of claims 1 to 8 is implemented.

Images