AI-based electric vehicle positioning data auxiliary correction method and system

Abstract

The invention relates to the technical field of data processing, in particular to an AI-based electric vehicle positioning data auxiliary correction method and system. The method comprises the following steps: acquiring a positioning data sequence containing multiple dimensions at each historical moment in the historical travel of the electric vehicle; determining a movement coefficient of the positioning data sequence at each dimension at each historical moment; obtaining the categories of a plurality of electric vehicle driving conditions, and determining the confidence degree of each category and each vehicle-mounted sensor; determining a correction value of each vehicle-mounted sensor at the current moment in the travel; and based on the correction value, utilizing an AI algorithm to realize prediction of the positioning data at the current moment. According to the method, the positioning data movement coefficient is calculated, the driving condition is clustered, and the current sensor value is corrected in combination with the sensor confidence coefficient, so that track offset caused by vehicle-mounted sensor error accumulation in satellite signal shielding scenes such as urban canyons and tunnels is effectively inhibited, and positioning is closer to a real path.

Description

Technical Field

[0001] The present invention relates to the field of data processing technology, and in particular to an AI-based auxiliary correction method and system for electric vehicle positioning data. Background Art

[0002] As electric vehicles (new energy vehicles) continue to become more intelligent, precise positioning has become a key technology to ensure navigation accuracy, path planning rationality, and the proper functioning of safety-assisted driving features. Currently, electric vehicle positioning primarily relies on data fusion solutions combining satellite positioning systems (e.g., GPS, Beidou), and onboard sensors (e.g., inertial measurement units (IMUs) and wheel speed sensors). This solution provides highly accurate positioning results in open environments.

[0003] However, in scenarios such as urban canyons (streets lined with tall buildings), tunnels, and underground parking lots, satellite signals are susceptible to obstruction or interference, leading to signal weakening or even loss. In these situations, existing technologies typically employ dead reckoning (DR). This involves using on-board sensor data (such as acceleration and angular velocity) to estimate the trajectory based on the positioning point before the satellite signal was blocked. However, this approach also has significant drawbacks: over time, sensor measurement errors accumulate, causing significant trajectory drift in positioning results, and even leading to navigation failures and erroneous autonomous driving decisions. Furthermore, the accumulation characteristics of sensor errors vary significantly in different driving scenarios (such as sudden acceleration, cornering, and constant speed driving). Traditional single dead reckoning models are difficult to adapt to complex and changing driving conditions, further exacerbating positioning errors. Therefore, how to effectively suppress the accumulation of on-board sensor errors and achieve high-precision electric vehicle positioning in scenarios where satellite signals are lost has become a pressing issue in the field of electric vehicle positioning. Summary of the Invention

[0004] In order to solve the problem that the existing electric vehicle positioning technology mainly relies on the data fusion of satellite positioning and on-board sensors, but in scenarios where satellite signals are blocked, such as urban canyons and tunnels, the positioning of the electric vehicle is the positioning point before the satellite signal is blocked, and the trajectory is calculated using the on-board sensor data, resulting in error accumulation in the electric vehicle positioning, thereby causing trajectory deviation or navigation failure, the present invention provides an AI-based electric vehicle positioning data auxiliary correction method and system.

[0005] In a first aspect, the present invention provides an AI-based auxiliary correction method for electric vehicle positioning data, which adopts the following technical solutions: An AI-based electric vehicle positioning data auxiliary correction method includes: obtaining a positioning data sequence containing multiple dimensions at each historical moment in the historical travel of the electric vehicle; sliding the sliding window on the positioning data sequence with a preset length as the length of the sliding window, and for any historical moment, when the center of the sliding window is located at the moment, determining the movement coefficient of the positioning data sequence in each dimension at the moment based on the standard deviation of the positioning data in each dimension of all historical moments in the sliding window and the positioning data in each dimension of the historical moments at both ends of the sliding window; clustering each historical moment based on the movement coefficient to obtain multiple categories of electric vehicle driving conditions, and clustering the historical moments based on all the positioning data in each category. The confidence level of each category and each on-board sensor is determined based on the standard deviation of the values of each on-board sensor of the electric vehicle at historical moments, as well as the standard deviation of the moving coefficient of the positioning data sequence in each dimension at all historical moments within each category. The reference category at the current moment and the correction direction at the current moment are obtained based on the values of each on-board sensor at all historical moments in this trip. The values of each on-board sensor at the current moment are corrected according to the reference category, the confidence level of each on-board sensor, and the correction direction to obtain the corrected values of each on-board sensor at the current moment. Based on the corrected values, an AI algorithm is used to predict the positioning data at the current moment.

[0006] The beneficial effects are: by utilizing the statistical characteristics of historical positioning data and on-board sensor data (such as standard deviation, movement coefficient, etc.), combined with AI algorithms to assist in correcting the current positioning, it is effectively reduced the error accumulation problem caused by relying solely on on-board sensors for calculation after satellite signals are blocked; using sliding window analysis and clustering methods to realize the confidence assessment of on-board sensor data under different driving conditions of electric vehicles, making the correction more targeted, thereby improving positioning accuracy and navigation stability in complex environments (such as urban canyons, tunnels, etc.); through AI-assisted methods, it can dynamically learn and adapt to various driving conditions and environmental changes, enhance the system's perception of vehicle status, improve the robustness of the positioning algorithm, and reduce the risk of positioning failure or obvious trajectory deviation; based on the correction value at the current moment, it quickly predicts positioning data, provides more accurate location information for the electric vehicle navigation system, assists in path planning and driving decisions, and improves user driving experience and safety.

[0007] Furthermore, the dimensions include longitude, latitude and altitude.

[0008] Furthermore, the mobility coefficient satisfies: Where, For the The historical moment in The moving coefficient of the positioning data sequence of dimensions, The center of the sliding window is located at All historical moments in the sliding window are in the The standard deviation of the positioning data in each dimension, For all historical moments The standard deviation of the positioning data in each dimension, The center of the sliding window is located at The location data of the historical moment at one end of the sliding window at a historical moment, The center of the sliding window is located at The location data of the historical moment at the other end of the sliding window at the historical moment, is the absolute value symbol.

[0009] The beneficial effects are as follows: the movement coefficient effectively depicts the position change trend of each historical moment in different dimensions by combining the local standard deviation and the global standard deviation within the sliding window, and the difference in positioning data at both ends, and can reflect the dynamic fluctuation characteristics of the data; using the difference between the two ends of the positioning data as a weight, a higher movement coefficient is given to moments with significant position changes, which helps to identify possible mutation points or abnormal states in the trajectory and improve the accuracy of subsequent clustering and classification.

[0010] Furthermore, the clustering adopts DBSCAN clustering.

[0011] Furthermore, the confidence level satisfies: Where, For the Category and Confidence of vehicle-mounted sensors, For the All historical moments in the category The standard deviation of the values of the vehicle-mounted sensor data, is the number of types of vehicle-mounted sensors, For the All historical moments in the category The standard deviation of the moving coefficient of the positioning data sequence in the dimension, is the number of dimensions, is the number of categories, is the maximum value function, is the natural exponential function.

[0012] The beneficial effects are as follows: the confidence index combines the degree of dispersion (standard deviation) of the on-board sensor data and the fluctuation range of the positioning data characteristics (standard deviation of the moving coefficient), and is weighted by an exponential function. It can quantify the reliability between each category (driving condition, such as straight driving, left turn, etc.) and each sensor, and comprehensively reflect the performance of the sensor in that category; the maximum function is used to reflect the impact of the most significant fluctuation dimension in the positioning data on the confidence, which helps to identify those situations where sensor data is most mismatched with positioning changes, thereby giving reasonable weight to the sensor data; by continuously evaluating and updating the confidence, the system can identify and adapt to changes in sensor performance and complex environmental interference, enhance the positioning algorithm's resistance to abnormal data, and improve the overall positioning accuracy and stability.

[0013] Furthermore, obtaining the reference category at the current moment and the correction direction at the current moment includes: constructing a sequence at the current moment based on the value of each on-board sensor at the current moment in this trip, constructing a sequence for each category based on the average of the values of each sensor at all historical moments in each category, and recording the category with the largest cosine similarity between the sequence at the current moment and the sequence of each category as the reference category at the current moment; and determining the correction direction of each on-board sensor at the current moment based on the size of the value of each on-board sensor at the current moment and the reference category.

[0014] The beneficial effects are: by calculating the cosine similarity between the on-board sensor data at the current moment and the mean sequence of each category, it is possible to accurately match the current driving state of the electric vehicle and achieve effective classification and attribution of positioning data in complex driving environments; by comparing the sensor data at the current moment with similar historical moments within the reference category, the offset trend of each sensor data can be scientifically determined, guiding the correction direction and reducing the impact of sensor noise or anomalies on positioning accuracy.

[0015] Furthermore, the correction direction satisfies: Where, For the current moment Correction direction of vehicle-mounted sensors, For the current moment The value of the vehicle-mounted sensor, For all historical moments in the reference category The mean of the values of the vehicle-mounted sensors.

[0016] Furthermore, the correction value satisfies: Where, For the current moment Correction values for vehicle-mounted sensors, The reference category at the current moment is Correction direction of vehicle-mounted sensors, The reference category at the current moment and Confidence of vehicle-mounted sensors, For the current moment The value of the vehicle-mounted sensor, The moment before the current moment The value of the vehicle-mounted sensor, is the natural exponential function, is the absolute value symbol.

[0017] The beneficial effects are: the correction value is based on the ratio difference between the sensor data at the current moment and the previous moment, which can capture the mutation trend of the sensor data, and combined with the exponential decay weight, a smooth transition in the correction process is achieved to avoid positioning jumps caused by mutation data; the correction amplitude is adjusted by the exponential function of the confidence level. When the confidence level is high, the correction amplitude is automatically reduced to avoid over-correction. When the confidence level is low, the correction strength is increased, which effectively responds to sensor data anomalies and improves the flexibility and accuracy of the correction; the correction direction plays a directional role, ensuring that the correction value is adjusted in a reasonable direction, effectively reducing the positioning deviation caused by sensor errors, and enhancing the physical rationality and practicality of the correction results; by dynamically weighing historical data and current observations, combined with confidence and correction direction, intelligent correction of sensor data is achieved, significantly reducing the risk of positioning error accumulation, and ensuring navigation continuity and accuracy in complex environments.

[0018] Furthermore, the use of the AI algorithm to predict the positioning data at the current moment includes: the AI algorithm adopts the LSTM algorithm, takes the positioning data of the previous moment and the correction value of each sensor at the current moment as the input of the LSTM algorithm, obtains the predicted value of the positioning data at the current moment, and completes the AI-based auxiliary correction of the electric vehicle positioning data.

[0019] In a second aspect, the present invention provides an AI-based electric vehicle positioning data auxiliary correction system, which adopts the following technical solutions: An AI-based electric vehicle positioning data auxiliary correction system includes: a processor and a memory, wherein the memory stores computer program instructions, and when the computer program instructions are executed by the processor, the above-mentioned AI-based electric vehicle positioning data auxiliary correction method is implemented.

[0020] By adopting the above technical solution, the above-mentioned AI-based electric vehicle positioning data auxiliary correction method is generated into a computer program and stored in the memory to be loaded and executed by the processor, so that a terminal device is made based on the memory and the processor for easy use.

[0021] The present invention has the following technical effects: By calculating the movement coefficient of positioning data and clustering driving conditions, and combining the sensor confidence to correct the current sensor value, the trajectory deviation caused by the accumulation of on-board sensor errors in satellite signal obstruction scenarios such as urban canyons and tunnels can be effectively suppressed, making the positioning closer to the actual path; integrating multi-dimensional data, using AI algorithms to establish a prediction model, and dynamically adjusting the sensor correction strategy according to different driving scenarios, it can automatically identify scenes such as stillness and turning, avoiding the lack of adaptability of traditional methods and ensuring stable and reliable positioning; real-time calibration of sensor errors to prevent "jump points" or route planning failures in navigation, and continue to provide accurate navigation guidance even during signal obstruction; at the same time, it can adapt to new environments and reduce maintenance costs through iterative optimization through accumulated data. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] Figure 1 This is a flow chart of a method for assisting correction of electric vehicle positioning data based on AI in an embodiment of the present invention. DETAILED DESCRIPTION

[0023] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of them. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative work shall fall within the scope of protection of the present invention.

[0024] The embodiment of the present invention discloses an AI-based electric vehicle positioning data auxiliary correction method, referring to Figure 1 , including steps S1 to S5: S1: Obtain a positioning data sequence containing multiple dimensions at each historical moment in the historical travel of the electric vehicle.

[0025] It should be noted that new energy vehicles are equipped with a variety of sensors, such as Global Navigation Satellite System (GNSS) receivers, which can receive satellite signals to obtain real-time location information such as the vehicle's longitude, latitude, and altitude. Inertial Measurement Units (IMUs) can measure the vehicle's acceleration, angular velocity, and other data to infer the vehicle's motion state and position changes. Wheel speed sensors calculate the vehicle's distance and speed by monitoring wheel speed. These sensors work together to continuously collect data while the vehicle is in motion, forming a positioning data sequence containing multiple dimensions. The collected positioning data is transmitted to a cloud server or the vehicle manufacturer's data center for storage via the vehicle's communication module, such as the in-vehicle infotainment system (IVI) or telematics unit (T-Box), using mobile communication networks (such as 4G, 5G) or other wireless communication technologies. During the current trip, accurate historical positioning data for multiple segments closest to the starting location of the current trip is obtained for subsequent analysis.

[0026] Specifically, the dimensions include longitude, latitude, and altitude.

[0027] S2: Determine the movement coefficient of the positioning data sequence in each dimension at each historical moment.

[0028] It should be noted that while the electric vehicle's positioning is relatively accurate when there is no satellite signal obstruction, there will be some deviation in the positioning after the satellite signal is blocked. Therefore, to measure the degree of deviation in the electric vehicle's positioning after the loss of satellite signal support, this step obtains the mobility coefficient based on the historical positioning data sequence.

[0029] The sliding window is set to a preset length and is made to slide on the positioning data sequence. For any historical moment, when the center of the sliding window is located at that moment, the moving coefficient of the positioning data sequence in each dimension at that moment is determined based on the standard deviation of the positioning data in each dimension of all historical moments within the sliding window and the positioning data in each dimension of the historical moments at both ends of the sliding window.

[0030] Implementers can set the length of the sliding window according to specific implementation conditions, for example, 11. For the first 10 moments and the last 10 moments in each historical journey, these moments are used as auxiliary reference moments and are not calculated.

[0031] Specifically, the mobility coefficient satisfies: ; Where, For the The historical moment in The moving coefficient of the positioning data sequence of dimensions, The center of the sliding window is located at All historical moments in the sliding window are in the The standard deviation of the positioning data in each dimension, For all historical moments The standard deviation of the positioning data in each dimension, The center of the sliding window is located at The location data of the historical moment at one end of the sliding window at a historical moment, The center of the sliding window is located at The location data of the historical moment at the other end of the sliding window at the historical moment, is the absolute value symbol.

[0032] in, The larger the value, the more discrete the positioning data distribution in the sliding window is compared with the overall positioning data distribution, and the greater the movement of the car at the corresponding moment of the sliding window. The larger the positioning mobility coefficient of each dimension; The smaller the value, the more concentrated the positioning data distribution in the sliding window is compared with the overall positioning data distribution, and the smaller the movement of the car at the corresponding moment of the sliding window. The smaller the positioning movement coefficient of each dimension is, the smaller the positioning movement coefficient of each dimension is. The larger the value, the higher the electric vehicle is in the first The larger the movement span in the first dimension, the The larger the positioning mobility coefficient of each dimension; The smaller it is, the more electric vehicles are in the first The smaller the moving span in the first dimension, the The smaller the positioning movement coefficient of each dimension is, the smaller the positioning movement coefficient of each dimension is.

[0033] S3: Obtain multiple categories of electric vehicle driving conditions and determine the confidence level of each category and each onboard sensor.

[0034] It should be noted that a vehicle may experience various driving conditions, such as going straight, turning left, and turning right. The vehicle's movement varies under different driving conditions. To adapt the electric vehicle's positioning data correction to different driving conditions, this step clusters historical moments based on the positioning movement coefficients. Based on the onboard sensor data corresponding to each category of historical moments, the confidence level of each category and the corresponding onboard sensor data is obtained.

[0035] Based on the mobility coefficient, each historical moment is clustered to obtain multiple categories of electric vehicle driving conditions. The confidence of each category and each on-board sensor is determined based on the standard deviation of the values of each on-board sensor of the electric vehicle at all historical moments in each category, and the standard deviation of the mobility coefficient of the positioning data sequence in each dimension at all historical moments in each category.

[0036] Specifically, the clustering adopts DBSCAN clustering.

[0037] In another embodiment, the clustering adopts mean shift clustering.

[0038] Specifically, the confidence level satisfies: ; Where, For the Category and Confidence of vehicle-mounted sensors, For the All historical moments in the category The standard deviation of the values of the vehicle-mounted sensor data, is the number of types of vehicle-mounted sensors, For the All historical moments in the category The standard deviation of the moving coefficient of the positioning data sequence in the dimension, is the number of dimensions, is the number of categories, is the maximum value function, is the natural exponential function.

[0039] in, Indicates the The corresponding sensor data of all historical moments in the category The relative size of the discrete degree of the vehicle sensor data. The larger the value, the more Under the driving conditions corresponding to the historical moments in the categories, The greater the fluctuation of the vehicle-mounted sensor data itself is relative to other sensors, the higher the instability and noise level is. Category and The lower the confidence of the vehicle-mounted sensor (vehicle-mounted sensor data), the less likely it is to cause over-compensation or correction deviation of the positioning data caused by its fluctuation; the smaller the value, the lower the confidence of the vehicle-mounted sensor (vehicle-mounted sensor data). The vehicle sensor data The more stable the driving state of each category, the lower the noise level, and the better the reading consistency, the higher the confidence level of the on-board sensor in this state, and the more its output can be relied upon for positioning correction. Indicates the The maximum relative proportion of the discrete degree of the moving coefficient of the category in all dimensions. The larger the value, the The more complex the driving conditions corresponding to the historical moments in a category are (for example, when driving straight, sometimes the vehicle accelerates first and then decelerates, sometimes decelerates first and then accelerates, and sometimes the vehicle moves at a constant speed), the lower the confidence level of the sensor data under this condition should be, so as to avoid blindly trusting the vehicle sensor data and causing deviation in the positioning data. The smaller the value, the more complex the driving conditions are. The simpler the driving condition corresponding to a historical moment in a category (such as the need to slow down when turning), the higher the confidence level of the sensor data under this condition should be.

[0040] S4: Determine the correction value of each vehicle-mounted sensor at the current moment in this trip.

[0041] It should be noted that due to bumps during driving or interference from the electric vehicle's internal circuits, the data collected by the onboard sensors inevitably contains some inaccurate or noisy data. Acquiring the electric vehicle's positioning data based on this data will lead to deviations in the electric vehicle's positioning data. Therefore, to reduce the deviations in the electric vehicle's positioning data, this step performs auxiliary corrections to the electric vehicle's positioning based on the confidence level of each category and each onboard sensor, as well as the data from each onboard sensor.

[0042] Based on the values of each vehicle sensor at the current time in this trip and all historical times within each category, the reference category of the current time and the correction direction of the current time are obtained.

[0043] Specifically, obtaining the reference category at the current moment and the correction direction at the current moment includes: constructing a sequence at the current moment based on the values of each onboard sensor at the current moment in the current trip, constructing a sequence for each category based on the average of the values of each sensor at all historical moments within each category, and recording the category with the largest cosine similarity between the sequence at the current moment and the sequence of each category as the reference category at the current moment; According to the magnitude of the value of each vehicle-mounted sensor at the current moment and the value of the reference category at each vehicle-mounted sensor, the correction direction of each vehicle-mounted sensor at the current moment is determined.

[0044] Specifically, the correction direction satisfies: ; Where, For the current moment Correction direction of vehicle-mounted sensors, For the current moment The value of the vehicle-mounted sensor, For all historical moments in the reference category The mean of the values of the vehicle-mounted sensors.

[0045] in, This means that the value of this type of vehicle-mounted sensor at the current moment is higher than the average level of the reference category. The more likely it is to be overvalued, the more it should be revised in a negative direction to be closer to the historical normal level; When the value of the vehicle sensor of this type at the current moment is lower than the average level of the reference category, The more likely it is to be undervalued, the more it should be corrected in the positive direction to get closer to the historical normal level. It is 1 or -1, and the implementer can also determine it according to the actual situation , for example, 0.9 or -0.9, etc.

[0046] According to the reference category and the confidence of each onboard sensor, and the correction direction, the value of each onboard sensor at the current moment is corrected to obtain the corrected value of each onboard sensor at the current moment.

[0047] Specifically, the correction value satisfies: ; Where, For the current moment Correction values for vehicle-mounted sensors, The reference category at the current moment is Correction direction of vehicle-mounted sensors, The reference category at the current moment and Confidence of vehicle-mounted sensors, For the current moment The value of the vehicle-mounted sensor, The moment before the current moment The value of the vehicle-mounted sensor, is the natural exponential function, is the absolute value symbol.

[0048] in, Indicates the adjacent time The difference between the values of the on-board sensors is that the driving of electric vehicles is continuous, so the sensor data at adjacent moments are usually similar. The larger it is, the more the sensor data at the current moment should be corrected. The smaller it is, the less need there is to correct the sensor data at the current moment. The larger the The more reliable the data from the vehicle-mounted sensors, the less need there is for Make corrections; The smaller the The less reliable the data from the vehicle-mounted sensors, the more important it is to Make corrections.

[0049] S5: Based on the correction value, use the AI algorithm to predict the positioning data at the current moment.

[0050] Specifically, the use of AI algorithms to predict the current positioning data includes: The AI algorithm adopts the LSTM algorithm, takes the positioning data of the previous moment and the correction value of each sensor at the current moment as the input of the LSTM algorithm, obtains the predicted value of the positioning data at the current moment, and completes the AI-based auxiliary correction of electric vehicle positioning data.

[0051] An embodiment of the present invention also discloses an AI-based electric vehicle positioning data auxiliary correction system, including a processor and a memory, wherein the memory stores computer program instructions. When the computer program instructions are executed by the processor, an AI-based electric vehicle positioning data auxiliary correction method according to the present invention is implemented.

[0052] The above system also includes other components well known to those skilled in the art, such as a communication bus and a communication interface. The configuration and functions of these components are known in the art and will not be described in detail here.

[0053] The above are all preferred embodiments of the present invention, and are not intended to limit the scope of protection of the present invention. Therefore, any equivalent changes made based on the structure, shape, and principle of the present invention should be included in the scope of protection of the present invention.

Claims

1. An AI-based electric vehicle positioning data auxiliary correction method, characterized in that: include: Obtaining a sequence of positioning data containing multiple dimensions at each historical moment in the electric vehicle's historical travel; A sliding window is set to a preset length and is slid across the positioning data sequence. For any historical moment, when the center of the sliding window is at that moment, the shift coefficient of the positioning data sequence in each dimension at that moment is determined based on the standard deviation of the positioning data in each dimension for all historical moments within the sliding window, as well as the positioning data in each dimension at the historical moments at both ends of the sliding window. Based on the mobility coefficient, clustering each historical moment to obtain multiple categories of electric vehicle driving conditions, and determining the confidence level of each category and each onboard sensor based on the standard deviation of the values of each onboard sensor of the electric vehicle at all historical moments in each category, and the standard deviation of the mobility coefficient of the positioning data sequence in each dimension at all historical moments in each category; Based on the values of each vehicle sensor at the current moment in this trip and all historical moments within each category, obtain the reference category at the current moment and the correction direction at the current moment; Correcting the value of each onboard sensor at the current moment according to the reference category and the confidence of each onboard sensor, and the correction direction, to obtain a corrected value of each onboard sensor at the current moment; Based on the correction value, the AI algorithm is used to predict the positioning data at the current moment.

2. The AI-based electric vehicle positioning data auxiliary correction method according to claim 1, characterized in that: The dimensions include longitude, latitude, and altitude.

3. The AI-based electric vehicle positioning data auxiliary correction method according to claim 1, characterized in that: The mobility coefficient satisfies: ； Where, For the The historical moment in The moving coefficient of the positioning data sequence of dimensions, The center of the sliding window is located at All historical moments in the sliding window are in the The standard deviation of the positioning data in each dimension, For all historical moments The standard deviation of the positioning data in each dimension, The center of the sliding window is located at The location data of the historical moment at one end of the sliding window at a historical moment, The center of the sliding window is located at The location data of the historical moment at the other end of the sliding window at the historical moment, is the absolute value symbol.

4. The AI-based electric vehicle positioning data auxiliary correction method according to claim 1, characterized in that: The clustering adopts DBSCAN clustering.

5. The AI-based electric vehicle positioning data auxiliary correction method according to claim 1, characterized in that: The confidence level satisfies: ； Where, For the Category and Confidence of vehicle-mounted sensors, For the All historical moments in the category The standard deviation of the values of the vehicle-mounted sensor data, is the number of types of vehicle-mounted sensors, For the All historical moments in the category The standard deviation of the moving coefficient of the positioning data sequence in the dimension, is the number of dimensions, is the number of categories, is the maximum value function, is the natural exponential function.

6. The AI-based electric vehicle positioning data auxiliary correction method according to claim 1, characterized in that: The obtaining of the reference category at the current moment and the correction direction at the current moment includes: Based on the values of each onboard sensor at the current moment of this trip, a sequence for the current moment is constructed. Based on the mean of the values of each sensor at all historical moments within each category, a sequence for each category is constructed. The category with the largest cosine similarity between the current moment's sequence and the sequence of each category is recorded as the reference category for the current moment. According to the magnitude of the value of each vehicle-mounted sensor at the current moment and the value of the reference category at each vehicle-mounted sensor, the correction direction of each vehicle-mounted sensor at the current moment is determined.

7. The AI-based electric vehicle positioning data auxiliary correction method according to claim 6, characterized in that: The correction direction satisfies: ； Where, For the current moment Correction direction of vehicle-mounted sensors, For the current moment The value of the vehicle-mounted sensor, For all historical moments in the reference category The mean of the values of the vehicle-mounted sensors.

8. The AI-based electric vehicle positioning data auxiliary correction method according to claim 7, characterized in that: The correction value satisfies: ； Where, For the current moment Correction values for vehicle-mounted sensors, The reference category at the current moment is Correction direction of vehicle-mounted sensors, The reference category at the current moment and Confidence of vehicle-mounted sensors, For the current moment The value of the vehicle-mounted sensor, The moment before the current moment The value of the vehicle-mounted sensor, is the natural exponential function, is the absolute value symbol.

9. The AI-based electric vehicle positioning data auxiliary correction method according to claim 1, characterized in that: The use of AI algorithms to predict the current positioning data includes: The AI algorithm adopts the LSTM algorithm, takes the positioning data of the previous moment and the correction value of each sensor at the current moment as the input of the LSTM algorithm, obtains the predicted value of the positioning data at the current moment, and completes the AI-based auxiliary correction of electric vehicle positioning data.

10. An AI-based electric vehicle positioning data auxiliary correction system, characterized in that: include: A processor and a memory, wherein the memory stores computer program instructions, and when the computer program instructions are executed by the processor, an AI-based electric vehicle positioning data auxiliary correction method according to any one of claims 1 to 9 is implemented.

Images