Method and system for predicting driving mileage of vehicle based on real-time data and electronic equipment

Abstract

The invention relates to the field of electric vehicle mileage prediction, in particular to a method and system for predicting the vehicle driving mileage based on real-time data and electronic equipment. The system comprises a battery charging and discharging period identification module, a data extraction and cleaning module, a track segment identification and distance calculation module, an energy consumption behavior modeling and mileage coefficient calculation module, a sample screening and aggregation module and a remaining mileage prediction output module. According to the method, the battery working state, the GPS track and the historical charging and discharging behavior are used as basic data input, multi-layer data processing, track analysis, energy consumption modeling and prediction output are used as intermediate processing paths, and high-precision prediction of the mileage of the electric vehicle under different batteries, different environments and different working conditions can be achieved.

Description

Technical Field

[0001] The present invention relates to the field of electric vehicle mileage prediction, and in particular to a method, system and electronic equipment for predicting vehicle mileage based on real-time data. Background Art

[0002] With the rapid development of the electric vehicle market, accurately predicting the remaining range of electric vehicles has become an important research topic. Traditional range prediction methods are mostly based on static battery data (such as the initial charge and battery health status). However, these methods fail to fully consider dynamic environmental factors and real-time changes in battery status. Battery performance (such as charge and discharge efficiency and temperature sensitivity) changes with time, frequency of use, and environmental changes, and these changes cannot be effectively reflected in static data models. During the use of electric vehicles, the range of electric vehicles is affected by many factors, including but not limited to the battery charge, battery charge and discharge performance, temperature, driving behavior, road conditions, and vehicle load. Therefore, when predicting the range of electric vehicles, static models rely on a single data source (such as the initial charge and SOC of the battery), ignoring the influence of these dynamic factors, and often resulting in large errors.

[0003] Furthermore, battery performance varies significantly across brands, models, and usage conditions. Many existing range prediction methods lack personalized adjustments for different battery models, resulting in widely varying accuracy for the same prediction model across different battery types. While some methods attempt to adjust for battery state of health (SOH), these adjustments are typically based on simplified battery models and ignore the nonlinear variations and complexity of battery performance. Summary of the Invention

[0004] The present invention aims to provide a method and system for predicting vehicle mileage based on real-time data. By collecting real-time battery data (such as voltage, current, SOC, and energy change) and external environmental information such as GPS, combined with a dynamic adjustment mechanism, the system comprehensively analyzes various factors affecting mileage prediction, thereby accurately predicting the remaining mileage of an electric vehicle. The present invention also provides an electronic device for executing this method.

[0005] To achieve the above object, the present invention adopts the following technical solutions:

[0006] A method for predicting vehicle mileage based on real-time data comprises the following steps: S1: identifying battery charge and discharge cycles, identifying complete charge and discharge cycles existing in the battery's historical operation process;

[0007] S2: Data extraction and cleaning: extract voltage, current, SOC, temperature, and GPS information from each cycle and clean the data;

[0008] S3: Track segment identification and distance calculation, segmenting the GPS data and extracting representative vehicle track segments;

[0009] S4: Energy consumption behavior modeling and mileage coefficient calculation. By modeling, the driving distance corresponding to unit power consumption is calculated for each trajectory segment to obtain the mileage coefficient.

[0010] S5: Sample screening and aggregation to eliminate abnormal data;

[0011] S6: Dynamic correction, adjusting the mileage coefficient or final prediction value based on battery and vehicle load information;

[0012] S7: Output of remaining mileage prediction.

[0013] Furthermore, in step S1 , the discharge cycle identification is based on the following multiple conditions: the SOC starting value is greater than or equal to 95%, the ending value is less than or equal to 10%; and the current direction remains consistent throughout the entire cycle.

[0014] Furthermore, step S2 specifically includes, S21: data standardization, unifying the sampling frequency of all fields to 1 Hz, and performing linear interpolation synchronization through timestamp alignment;

[0015] S22: Field completion and anomaly elimination. For intermittently missing fields, the fields are filled by interpolating the previous and next average values. If the duration of continuous missing data exceeds the set threshold, the entire sub-segment is discarded.

[0016] S23: GPS drift correction, through spatial continuity and speed rationality verification, detects and eliminates coordinate anomalies caused by positioning drift;

[0017] S24: Units are standardized, with electricity measured in ampere-hours, distance measured in kilometers, and time measured in seconds, ensuring consistency of all physical quantities for easy subsequent calculations.

[0018] Furthermore, in step S3, the trajectory segments are divided based on two main criteria:

[0019] Time discontinuity: If the time interval between adjacent GPS sampling points is greater than the set threshold, they are considered as two different sections;

[0020] Spatial mutation: If the geographical distance between two points exceeds a reasonable range, they are determined to be trajectory segment cutting points.

[0021] Furthermore, in step S4, the coefficient based on the change in power and the coefficient based on the change in SOC are calculated simultaneously, and by comparing and analyzing the difference between the two, a deviation rate index is further obtained to determine the data stability.

[0022] Furthermore, in step S5, the screening strategy mainly includes: box analysis, median and interquartile range analysis of the coefficient sample set, and elimination of data points falling outside the outlier interval; extreme behavior elimination, excluding energy consumption abnormal fluctuation segments, GPS drift segments, and short-term static segments.

[0023] Furthermore, in step S5, the retained samples after screening are weighted according to the length of the trajectory segment to finally obtain the global average coefficient.

[0024] Furthermore, in step S6, the correction factors include: temperature correction factor, rate correction factor, model confidence, and adjustment weights are set according to the battery cell temperature range. The average rate is estimated in real time and dynamically adjusted through an exponential decay function.

[0025] A system for predicting vehicle mileage based on real-time data includes a battery charge and discharge cycle identification module for identifying complete charge and discharge cycles existing in the battery's historical operation process;

[0026] Data extraction and cleaning module, used to extract voltage, current, SOC, temperature and GPS information from each cycle and clean the data;

[0027] The track segment identification and distance calculation module is used to segment the GPS data and extract representative vehicle track segments;

[0028] Energy consumption behavior modeling and mileage coefficient calculation module, used to model and calculate the driving distance corresponding to unit power consumption for each trajectory segment to obtain the mileage coefficient;

[0029] Sample screening and aggregation module, used to eliminate abnormal data;

[0030] Dynamic correction module, which adjusts the mileage coefficient or final prediction value based on battery and vehicle load information;

[0031] The remaining mileage prediction output module predicts the maximum distance the vehicle can travel in the future based on the corrected mileage coefficient and the current battery remaining energy status.

[0032] An electronic device includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the computer program is executed by the processor, it implements any of the above-mentioned methods for predicting vehicle mileage based on real-time data.

[0033] Beneficial effects of the present invention:

[0034] The present invention uses battery operating status, GPS trajectory and historical charging and discharging behavior as basic data input, and uses multi-layer data processing, trajectory analysis, energy consumption modeling and prediction output as intermediate processing paths to ultimately achieve high-precision prediction of the vehicle's remaining driving capacity.

[0035] The present invention provides a mileage coefficient prediction method based on the fusion analysis of multi-dimensional battery operating parameters and geographical driving trajectories, which can achieve high-precision prediction of the mileage of electric vehicles under different batteries, different environments, and different operating conditions. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] Figure 1 It is a logical step diagram of the method for predicting vehicle mileage based on real-time data of the present invention;

[0037] Figure 2 It is a logic principle diagram of the method for predicting vehicle mileage based on real-time data of the present invention. DETAILED DESCRIPTION

[0038] The technical solutions in the embodiments of the present invention will be described clearly and completely below with reference to the accompanying drawings in the embodiments of the present invention.

[0039] Example:

[0040] like Figure 1-Figure 2 As shown in the figure, a system for predicting vehicle mileage based on real-time data consists of seven main functional modules, each responsible for different tasks from data access to model inference. These modules include: battery charge and discharge cycle identification module, real-time operation data extraction and cleaning module, trajectory segment identification and distance calculation module, energy consumption behavior modeling and mileage coefficient calculation module, sample screening and aggregation module, dynamic correction and adaptability module, and remaining range prediction output module.

[0041] 1.Battery charge and discharge cycle identification module:

[0042] In the process of predicting the actual mileage capability of electric vehicles, accurately identifying the complete discharge behavior cycle of the battery is a prerequisite for building an energy consumption model. The function of the battery charge and discharge cycle identification module is to identify the complete charge and discharge cycles that existed in the historical operation process based on the battery number. Based on the historical operation records of the battery (including current, voltage, SOC time series, etc.), it automatically divides and extracts complete discharge cycle segments with strong representativeness and continuous data. This module not only extracts the start and end time of the discharge cycle, but also verifies the integrity of the cycle, time span, SOC change amplitude and other factors, eliminates abnormal cycles such as charging interruption and energy rebound, and ensures that the data used for analysis is representative and consistent.

[0043] The system will prioritize identifying the time period that meets the following conditions as a valid period:

[0044] (1) The SOC starting value is greater than or equal to 95%, and the ending value is less than or equal to 10%, indicating that the discharge process is sufficient;

[0045] (2) There is no reverse charging behavior during the discharge process (i.e., no continuous negative current recording);

[0046] (3) The time span of the entire process is greater than a preset threshold (e.g., 30 minutes) to ensure the physical integrity of the cycle;

[0047] (4) The time intervals of all data sampling points are continuous, and there is no large-scale data missing.

[0048] For some cycles with "rebound charging" or "charging interruption" characteristics, such as a sudden increase in the SOC value or a reversal of the current in the middle of the cycle, they will be eliminated through the differential mutation detection algorithm to ensure the purity and representativeness of the subsequent modeling data, significantly improving the stability of the mileage coefficient prediction.

[0049] Discharge cycle identification is based on the following multiple conditions:

[0050] The SOC starting value is greater than or equal to 95%, and the ending value is less than or equal to 10%:

[0051] SOC start ≥95%, SOC end ≤10%

[0052] The current direction remains consistent throughout the cycle, i.e. continuous discharge behavior:

[0053]

[0054] In addition, a sliding window is used to perform continuity verification on the data integrity within the cycle, eliminating cycles containing interrupted charging and rebound behavior, ensuring that the interval data used for modeling has physical continuity and statistical representativeness.

[0055] 2. Run the data extraction and cleaning module in real time:

[0056] Real-time vehicle data during driving typically comes from two sources: battery operating status, such as current, voltage, SOC, and temperature, recorded by the onboard battery management system (BMS); and geographic location and trajectory coordinates provided by the GPS module. Considering that raw data may have issues such as inconsistent formats, discontinuous time periods, GPS drift, and missing fields, this module performs structured cleaning of the raw data by setting field specifications, standardizing time granularity, filling missing values, and filtering out duplicate points, resulting in a clean, parseable multidimensional dataset.

[0057] The function of this module is to structurally integrate the above multi-source data and perform standardized cleaning to ensure that the data is time-aligned, field-complete, and error-controlled. It mainly includes the following technical steps:

[0058] (1) Data standardization: Unify the sampling frequency of all fields to 1 Hz, and perform linear interpolation synchronization through timestamp alignment to solve the problem of inconsistent time bases among multiple modules;

[0059] (2) Field completion and anomaly elimination: For intermittently missing fields (such as temperature or SOC value), the fields are completed by interpolation of the previous and next average values; if the continuous missing duration exceeds the set threshold (such as 10 seconds), the entire sub-segment is discarded;

[0060] (3) GPS drift correction: By checking spatial continuity and speed rationality, coordinate anomalies caused by positioning drift are detected and eliminated to prevent them from affecting trajectory segment identification and subsequent energy consumption distribution modeling;

[0061] (4) Unit standardization: Unify the unit of electricity into ampere-hour (Ah), distance into kilometers (km), and time into seconds (s) to ensure the consistency of all physical quantities and facilitate subsequent calculations.

[0062] Through the above standardization processing, the system can eventually form a data set with clear structure, controlled errors, and time-space matching, which can be directly used by the energy consumption analysis and trajectory modeling modules.

[0063] Data alignment:

[0064] Use a unified sampling frequency (usually 1 Hz) and use linear interpolation to time-align signals from different clock sources:

[0065]

[0066] Right now:

[0067] x(t): refers to the value of a physical quantity (such as current, voltage, SOC, temperature, etc.) at time t;

[0068] t: refers to the timestamp (seconds);

[0069] t k : The timestamp to which the interpolation is to be aligned (for example, to a time point of a certain second at a 1Hz frequency);

[0070] t1, t2: two adjacent timestamps in the original data;

[0071] K: the number of the new sampling point;

[0072] Data cleaning:

[0073] Missing or drifting data is repaired or removed. GPS trajectory anomalies are identified using a "spatial jump + speed smoothing" strategy. If the distance between two consecutive points at any moment exceeds 222m, or the calculated speed does not meet the physical limits of the vehicle, they will be removed.

[0074] Finally, a unified, synchronized and complete time series data set [I(t), V(t), SOC(t), T(t), (lat, lon)(t)] is formed for subsequent modeling. The previous data represent the current, voltage, remaining power, temperature and coordinate information corresponding to a certain time t.

[0075] 3. Trajectory segment recognition and distance calculation module:

[0076] Considering the fact that electric vehicles often experience brief stops, data interruptions, and atypical driving, this module uses a rational segmentation strategy to divide the entire driving process into multiple trajectory segments with independent physical meanings. Each segment is considered an independent energy consumption analysis unit. The total driving distance, running time, and speed distribution corresponding to each trajectory segment are calculated, providing spatial information support for energy consumption modeling.

[0077] The division of trajectory segments is mainly based on two criteria:

[0078] Temporal discontinuity: If the time interval between adjacent GPS sampling points is greater than the set threshold (e.g. 120 seconds), they are considered as two different segments;

[0079] Spatial mutation: If the geographical distance between two points exceeds a reasonable range (e.g., 222 meters, which is far beyond the distance a vehicle can instantly travel), it is determined to be a trajectory segment cutting point.

[0080] For each identified trajectory segment, the system calculates statistical characteristics such as total distance traveled, duration, average speed, and speed distribution. Specifically, distance calculation uses the Haversine formula in a spherical coordinate system, which more accurately estimates the vehicle's path length under various terrain conditions and avoids the error accumulation problem associated with planar approximations.

[0081] The system uses the following rules to divide paragraphs:

[0082] If the time interval between two consecutive sampling points is greater than 120 seconds, or the spatial jump is greater than 222 meters, it is considered a trajectory segment breakpoint:

[0083] Δt k >120s or s k >222m

[0084] The driving distance in each segment is calculated using the Haversine spherical distance formula:

[0085]

[0086] The total length of the trajectory segment is:

[0087]

[0088]

[0089] 4. Energy consumption behavior modeling and mileage coefficient calculation module:

[0090] After completing the trajectory segment extraction and distance calculation, the system will use each trajectory segment as a basic unit to establish a mathematical mapping relationship between power consumption and actual driving distance, thereby deriving the core indicator of "mileage coefficient". This coefficient is the basic parameter for prediction and represents the distance that can be traveled per unit energy consumption. The system adopts a dual-channel modeling method, that is, it simultaneously calculates the coefficient based on the change in power and the coefficient based on the change in SOC, and by comparing and analyzing the difference between the two, it further derives a deviation rate indicator to determine the stability of the data. The modeling results will be output as a set of K value samples for each segment, which will be used to construct a global prediction benchmark later. The following are the specific implementation steps:

[0091] The energy mileage coefficient is constructed based on the charge change obtained by current integration.

[0092] The SOC mileage coefficient is constructed based on the SOC change value and the energy consumption estimated from the battery nominal capacity.

[0093] The mileage factor, defined as the average distance traveled per unit of energy consumption, can be calculated in two ways:

[0094] Calculation of power coefficient:

[0095] For each trajectory segment, calculate the power consumption within the segment:

[0096]

[0097] From this we can get:

[0098] (Unit: km / Ah)

[0099] SOC coefficient calculation:

[0100] The SOC change and the nominal capacity Q nom Estimated power consumption:

[0101]

[0102] Data stability determination:

[0103] If the deviation between the two exceeds 20%, the segment is eliminated:

[0104]

[0105] Through dual-channel modeling and consistency judgment mechanism, it is ensured that all sections involved in the modeling have good physical rationality and measurement stability.

[0106] 5. Sample screening and aggregation module:

[0107] Given that real-world data may contain noise behaviors such as GPS loss, voltage distortion, and SOC jumps that could affect model stability, this module uses statistical methods to thoroughly screen and cleanse the mileage coefficient samples generated through preliminary modeling. Using quantile analysis, outlier removal, and extreme deviation removal, the effective value interval in the sample set is controlled within the median ±IQR range, ensuring the representativeness of the final calculated average K value. Furthermore, the module incorporates multiple built-in anomaly rejection strategies, including those for segments with excessively high machine variability and unreasonable energy consumption, to enhance the robustness of model predictions.

[0108] Screening strategies mainly include:

[0109] Box analysis: Perform median and interquartile range (IQR) analysis on the coefficient sample set and remove data points that fall outside the outlier interval;

[0110] Extreme behavior elimination: excludes low-reliability samples such as abnormal energy consumption fluctuations, GPS drift, and short-term static segments;

[0111] Outlier removal:

[0112] The boxplot method is used for anomaly detection. The median M and interquartile range IQR are defined, and the samples that meet the following conditions are retained:

[0113] M-1.5·IQR≤K Q,j ≤M+1.5·IQR

[0114] Aggregate calculation:

[0115] The retained samples are weighted according to the length of the trajectory segment, and the global average coefficient is finally obtained:

[0116]

[0117] This value represents the average energy consumption efficiency under current battery and environmental conditions and is an important basis for calculating the remaining mileage.

[0118] 6. Dynamic correction and adaptability module:

[0119] Since different battery models, temperature environments, and driving conditions have a significant impact on vehicle energy consumption, this module provides an adaptive correction mechanism for each prediction based on contextual environmental information, ensuring that the prediction results maintain a high level of accuracy in a variety of scenarios, greatly enhancing the system's adaptability to complex scenarios.

[0120] Specific correction factors include but are not limited to:

[0121] Temperature correction factor: Set adjustment weights according to the battery cell temperature range (low temperature, high temperature, normal temperature) to compensate for the trend of increasing energy consumption at low temperatures;

[0122] Rate correction factor: real-time estimation of average rate (current / capacity) and dynamic adjustment through exponential decay function to simulate the energy decay effect under high current;

[0123] Model confidence adjustment: When the system detects a large deviation between the current environment and the historical modeling environment (such as a temperature change of more than 10°C), it will automatically enlarge the predicted deviation range to remind users to pay attention to driving risks.

[0124] Temperature correction factor, used to compensate for the effect of low or high temperatures on energy efficiency:

[0125]

[0126] 7. Remaining mileage prediction output module:

[0127] The system uses the corrected global mileage coefficient as a basis and predicts the maximum distance that can be traveled in the future based on the current battery remaining energy status.

[0128] Based on SOC prediction:

[0129]

[0130] K corr Indicates the corrected global mileage coefficient after environmental correction (such as temperature multiplier correction);

[0131] corr: abbreviation of corrected;

[0132] rem: remaining meaning;

[0133] The system has voltage and current integration capabilities and can predict the following based on the remaining energy:

[0134]

[0135] At the same time, the system also outputs confidence intervals and data quality assessment indicators for reference by the navigation path planning system or battery management system.

[0136] The present invention also provides an electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor. When the computer program is executed by the processor, the method for predicting vehicle mileage based on real-time data as described above is implemented.

[0137] Experimental example:

[0138] Based on the complete data collected during the daily operation of an urban distribution electric logistics vehicle equipped with a 48V / 100Ah ternary lithium battery for 5 days, the mileage coefficient modeling and remaining mileage prediction process were completed by the method of the present invention, and good practical application results were achieved.

[0139] Implementation Scenario and Data Overview Test Vehicle: Light electric logistics vehicle, battery rated capacity 48V×100Ah=4.8kWh;

[0140] Sampling period: March 15 to March 20, 2025, a total of 5 days;

[0141] Operating environment: Urban roads + park roads, with an average daily delivery distance of approximately 40 km;

[0142] Data collection:

[0143] Complete discharge cycle: 18;

[0144] Number of valid trajectory segments: 149 (after excluding drift / interruption segments);

[0145] Average distance of a single segment: 2.6km, maximum segment length: 8.1km;

[0146] Operating temperature range: 5℃~28℃;

[0147] Forecasting process description:

[0148] Step (1): Cycle identification

[0149] The system automatically identified and extracted 18 discharge cycles, with typical cycle lengths ranging from 45 minutes to 1.8 hours, and an average SOC reduction of 82% per cycle. All cycles passed validation after continuity checks and current direction verification.

[0150] Step (2): Data extraction and cleaning

[0151] Voltage, current, SOC, temperature, and GPS information were extracted from each cycle, processing over 250,000 data points in total. During the data cleaning process, 11 GPS drift points due to network latency were removed, 73 missing temperature fields were added, and complete time series data was generated after unified timestamps.

[0152] Step (3): Trajectory segment identification and distance analysis

[0153] By detecting time intervals and spatial jump points, the system automatically divides continuous data into 162 trajectory segments, and retains 149 segments after eliminating signal interruption or parking segments.

[0154] The shortest driving distance is 0.4km (short-distance U-turn), and the longest section is 8.1km (long-distance delivery section). The average speed ranges from 28 to 52km / h, and some sections require waiting at traffic lights.

[0155] Step (4): Energy consumption modeling results

[0156] During the modeling process, the system calculated the distance traveled per unit of energy consumption for each trajectory segment. After removing signal fluctuations, it ultimately retained 138 valid energy consumption samples. The sample range was concentrated between 4.9 and 6.1 km / Ah, with a stable data distribution and low skewness.

[0157] Step (5): Sample aggregation and correction

[0158] After sample screening and weighted calculation, the average mileage coefficient under the current vehicle operating conditions is finally obtained as: 5.47km / Ah.

[0159] Considering the temperature in the area where the vehicle is located is 7–10°C, the system automatically applies a temperature correction factor, and the coefficient is attenuated to 0.965; at the same time, based on the vehicle's average rate of 0.62°C, the system applies a rate correction attenuation of 0.912, and the final corrected coefficient is: 4.81km / Ah.

[0160] Step (6): Remaining mileage prediction output

[0161] Before a delivery, the vehicle's SOC was 76%, corresponding to a remaining charge of approximately 76 Ah (3.65 kWh). The system predicted a remaining driving range of 30.4 km, but the actual distance traveled was 29.1 km, with a prediction error of 4.3%.

[0162] Comparative analysis results:

[0163]

[0164] The present invention is not limited to the above-mentioned optimal implementation mode. Anyone can derive other forms of products under the inspiration of the present invention. However, no matter what changes are made in the shape or structure, any technical solution that is the same or similar to that of the present application falls within the scope of protection of the present invention.

Claims

1. A method for predicting vehicle mileage based on real-time data, characterized by: The following steps are included: S1: Battery charge and discharge cycle identification, identifying the complete charge and discharge cycles that exist in the battery's historical operation process; S2: Data extraction and cleaning: extract voltage, current, SOC, temperature, and GPS information from each cycle and clean the data; S3: Track segment identification and distance calculation, segmenting the GPS data and extracting representative vehicle track segments; S4: Energy consumption behavior modeling and mileage coefficient calculation. By modeling, the driving distance corresponding to unit power consumption is calculated for each trajectory segment to obtain the mileage coefficient. S5: Sample screening and aggregation to eliminate abnormal data; S6: Dynamic correction, adjusting the mileage coefficient or final prediction value based on battery and vehicle load information; S7: Output of remaining mileage prediction.

2. The method for predicting vehicle mileage based on real-time data according to claim 1, characterized in that: In step S1 , the discharge cycle identification is based on the following multiple conditions: the SOC starting value is greater than or equal to 95%, the ending value is less than or equal to 10%; and the current direction remains consistent throughout the entire cycle.

3. The method for predicting vehicle mileage based on real-time data according to claim 1, characterized in that: Step S2 specifically includes, S21: data standardization, unifying the sampling frequency of all fields to 1 Hz, and performing linear interpolation synchronization through timestamp alignment; S22: Field completion and anomaly elimination. For intermittently missing fields, the fields are filled by interpolating the previous and next average values. If the duration of continuous missing data exceeds the set threshold, the entire sub-segment is discarded. S23: GPS drift correction, through spatial continuity and speed rationality verification, detects and eliminates coordinate anomalies caused by positioning drift; S24: Units are standardized, with electricity measured in ampere-hours, distance measured in kilometers, and time measured in seconds, ensuring consistency of all physical quantities for easy subsequent calculations.

4. The method for predicting vehicle mileage based on real-time data according to claim 1, characterized in that: In step S3, the trajectory segments are divided based on two main criteria: Time discontinuity: If the time interval between adjacent GPS sampling points is greater than the set threshold, they are considered as two different sections; Spatial mutation: If the geographical distance between two points exceeds a reasonable range, they are determined to be trajectory segment cutting points.

5. The method for predicting vehicle mileage based on real-time data according to claim 1, characterized in that: In step S4, the coefficient based on the change in power and the coefficient based on the change in SOC are calculated simultaneously, and by comparing and analyzing the difference between the two, a deviation rate index is further obtained to determine the data stability.

6. The method for predicting vehicle mileage based on real-time data according to claim 1, characterized in that: In step S5, the screening strategy mainly includes: box analysis, median and interquartile range analysis of the coefficient sample set, and elimination of data points falling outside the outlier range; extreme behavior elimination, excluding abnormal energy consumption fluctuation segments, GPS drift segments, and short-term static segments.

7. The method for predicting vehicle mileage based on real-time data according to claim 6, characterized in that: In step S5, the retained samples after screening are weighted according to the length of the trajectory segment to finally obtain the global average coefficient.

8. The method for predicting vehicle mileage based on real-time data according to claim 7, characterized in that: In step S6, the correction factors include: temperature correction factor and rate correction factor. The adjustment weights are set according to the battery cell temperature range, and the average rate is estimated in real time and dynamically adjusted through an exponential decay function.

9. A system for predicting vehicle mileage based on real-time data, characterized by: It includes a battery charge and discharge cycle identification module for identifying the complete charge and discharge cycles that have existed in the battery's historical operation process; Data extraction and cleaning module, used to extract voltage, current, SOC, temperature and GPS information from each cycle and clean the data; The track segment identification and distance calculation module is used to segment the GPS data and extract representative vehicle track segments; Energy consumption behavior modeling and mileage coefficient calculation module, used to model and calculate the driving distance corresponding to unit power consumption for each trajectory segment to obtain the mileage coefficient; Sample screening and aggregation module, used to eliminate abnormal data; Dynamic correction module, which adjusts the mileage coefficient or final prediction value based on battery and vehicle load information; The remaining mileage prediction output module predicts the maximum distance the vehicle can travel in the future based on the corrected mileage coefficient and the current battery remaining energy status.

10. Electronic equipment, characterized in that: The method comprises a memory, a processor, and a computer program stored in the memory and executable on the processor. When the computer program is executed by the processor, the method for predicting vehicle mileage based on real-time data as claimed in any one of claims 1 to 8 is implemented.

Images