New energy automobile regional energy consumption adaptability evaluation method

Abstract

According to the method provided by the invention, regional energy consumption adaptability evaluation is carried out based on real vehicle operation data, and three steps of S100 data preprocessing, S200 regional secondary fragment feature database construction and S300 regional energy consumption adaptability evaluation are carried out in sequence. And comprehensively evaluating the regional energy consumption adaptability of the target vehicle model in three dimensions of regional energy consumption forward fluctuation adaptability and cross-regional energy consumption forward fluctuation adaptability. According to the method, the statistical energy consumption value is calculated by using real vehicle operation data, and data distribution balance and weighting are performed for multiple working conditions, so that the evaluation is closer to reality, and the universality of different working conditions is good; meanwhile, data of different areas are independently processed and counted, the pertinence of the evaluated areas is high, and the reference value is higher; and finally, introducing an evaluation index of the energy consumption forward fluctuation amplitude, and raising the dimension of energy consumption evaluation from a traditional single point to an interval.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of new energy vehicles, and particularly relates to a new energy vehicle regional energy consumption adaptability evaluation method. BACKGROUND

[0002] New energy vehicles are gradually becoming the focus of the automobile industry and market due to their low carbon emissions, high economy, strong power performance and other advantages. The energy consumption of a vehicle is closely related to the energy charging cost, driving range and carbon emission level, and as an important performance indicator, it is always closely watched by users and manufacturers of new energy vehicle products. Since a vehicle usually travels in one or several fixed areas, and different areas often have different natural climates and traffic conditions, there are obvious differences in vehicle energy consumption. Therefore, the evaluation of regional energy consumption is of great significance. From the user's perspective, the energy consumption evaluation of different vehicle models in a specific area helps users make a reasonable choice when purchasing a vehicle. From the manufacturer's perspective, the energy consumption evaluation of a specific vehicle model in different areas helps the manufacturer find the deficiencies of the vehicle design in terms of economy and reasonably carry out regional differentiation and promotion. At present, the commonly used new energy vehicle energy consumption evaluation method is mainly based on the NEDC, CLTC and other cycle test to obtain the unit mileage energy consumption of the vehicle, such as GB / T 18386 Electric Vehicle Energy Consumption and Driving Range Test Method. Since the working condition conditions of the cycle test (such as vehicle speed, air temperature, kinetic energy recovery intensity, battery health, etc.) are fixed in one or a few cases, this method has many shortcomings. First, the single working condition makes it difficult to widely reflect the energy consumption under complex and variable real-world conditions. Second, its process is not related to the region, so its accuracy and reference value in different regions are uneven, and it is not region-specific. Finally, this method only evaluates the level of vehicle energy consumption, but ignores the stability of energy consumption under different conditions, and fails to represent the fluctuation range of energy consumption under certain conditions (such as low temperature), which is closely related to the driving experience including range anxiety. Through energy consumption statistics based on real vehicle operation data, data segment division based on regional changes, and the introduction of energy consumption positive fluctuation adaptability index, the new energy vehicle regional energy consumption adaptability evaluation method of the application improves the above shortcomings and provides a more accurate, reasonable and widely used new energy vehicle energy consumption evaluation method. SUMMARY

[0003] The new energy vehicle regional energy consumption adaptability evaluation method of the application improves the above shortcomings and provides a more accurate, reasonable and widely used new energy vehicle energy consumption evaluation method through energy consumption statistics based on real vehicle operation data, data segment division based on regional changes, and the introduction of energy consumption positive fluctuation adaptability index.

[0004] The application discloses a new energy automobile regional energy consumption adaptability evaluation method.

[0005] S100, preprocessing original message data of vehicle operation data, cleaning abnormal data and filling missing data;

[0006] S200, based on administrative division, cutting the pretreated vehicle operation data into regional secondary segments, and constructing a regional secondary segment feature database corresponding to a single regional secondary segment of a single line;

[0007] S300, based on the working condition characteristics, layering and sample balancing of unit mileage energy consumption characteristic data in the regional secondary segment feature database, and calculating regional energy consumption adaptability evaluation indexes considering the working condition layer weight;

[0008] In step 300, the following steps are further included:

[0009] S301, distributing the unit mileage energy consumption characteristic data to different layers according to the regional independent working condition to which the unit mileage energy consumption characteristic data belongs, so that different predetermined weights are applied to the data of different working conditions in the energy consumption adaptability evaluation, and then using an oversampling method to balance the sample amount of each layer, so as to avoid random interference of regional independent mixed factors on the regional energy consumption adaptability evaluation;

[0010] S302, calculating the weighted average value of the unit mileage energy consumption characteristic based on the different working condition layer weight as the regional energy consumption average adaptability evaluation index, measuring the comprehensive energy consumption level of the target vehicle model in the fixed region;

[0011] S303, calculating the weighted generalized positive standard deviation of the unit mileage energy consumption characteristic relative to the regional energy consumption average adaptability based on the different working condition layer weight as the regional energy consumption positive fluctuation adaptability evaluation index, measuring the positive fluctuation amplitude of the comprehensive energy consumption of the target vehicle model in the fixed region;

[0012] S304, calculating the overall regional energy consumption average, and based on the value, calculating the generalized positive standard deviation of the regional energy consumption average adaptability, measuring the energy consumption positive fluctuation amplitude of the target vehicle model when the region changes.

[0013] More further, in step 100, the following steps are further included:

[0014] S101, by decoding, transforming, segmenting, sorting operations, the original message of the vehicle operation data is converted into a structured data table corresponding to a single frame, a single column corresponding to a single field, and the time sequence and data value are corrected;

[0015] S102, the driving state of the data frame is identified, and the driving state data frame is cut into trip segments for subsequent further cutting and feature extraction;

[0016] S103, the abnormal, invalid value and time repeated frame of each field are cleaned to ensure the accuracy of subsequent processing and statistics;

[0017] S104, the missing values in the data are filled by using regression interpolation and time series interpolation method, so as to ensure the normal processing and accurate calculation of subsequent processing and statistics as much as possible, and avoid forced deletion of a large number of data frames due to programmed error.

[0018] Further, in step 200, the following steps are further included:

[0019] S201, based on the longitude and latitude fields of the data frame, the data frame area label is identified by using the electronic map platform reverse geocoding interface, which supports further cutting of secondary segments and feature extraction in subsequent steps;

[0020] S202, the trip segment is further cut into regional secondary segments based on the change of regional label and combined with the regional mutation tolerance mechanism;

[0021] S203, the unit mileage energy consumption, regional label and working condition feature are extracted from each regional secondary segment, and the regional secondary segment feature database corresponding to a single regional secondary segment of a single line is constructed, which provides statistical samples and hierarchical basis for regional energy consumption adaptability evaluation;

[0022] The working condition features include start time, end time, start cumulative mileage and end cumulative mileage;

[0023] S204, the secondary segment feature data cleaning step performs abnormal cleaning on part of the feature data, and removes the secondary segment corresponding row with extreme, abnormal and invalid feature value.

[0024] Further, in step 302, the regional energy consumption average adaptability of the target vehicle model is calculated:

[0025]

[0026] Among them, RECA mean is the regional energy consumption average adaptability; is each component of the working condition joint hierarchical vector c s of the sample s in all regional secondary segments in a certain region, that is, the selected working condition feature value after discretization of the sample; is the weight corresponding to the working condition layer to which the sample s belongs in the weight matrix W; ECR sLet be the energy consumption characteristic per unit mileage of sample s; sum(W) and size(W) are the sum of elements and the number of elements in W, respectively; N is the total number of secondary segments in a certain region.

[0027] Furthermore, in step 303, the regional energy consumption positive fluctuation adaptability of the target vehicle model is calculated:

[0028]

[0029] Among them, RECA pf To adapt to positive fluctuations in regional energy consumption; The energy consumption per unit mileage within a certain region is greater than RECA. mean The joint hierarchical vector c of the working conditions of sample p in the secondary fragment of the region p The components of the sample are the values ​​of the selected working condition features after discretization.

[0030] The weight of the working condition layer to which sample p belongs in the weight matrix W; ECR p N represents the energy consumption characteristics per unit mileage for sample p; + The energy consumption per unit mileage within a certain region is greater than RECA. mean The total number of secondary segments in the region.

[0031] Furthermore, in step 304, a weighted average of the energy consumption characteristics per unit mileage across the entire region is calculated.

[0032]

[0033] in, This is a weighted average of energy consumption characteristics per unit mileage across the entire region; v (r) N (r) Here, r represents the region weight and the number of secondary segments in region r, respectively; R is the total number of regions. Let W be the weight of the working condition layer to which the secondary segment i of region r belongs, i.e., the joint working condition layer weight matrix W of region r. (r) Elements; ECR i (r) The energy consumption characteristics per unit mileage of the secondary segment i of region r;

[0034] Calculate the cross-regional energy consumption positive fluctuation adaptability of the target vehicle model:

[0035]

[0036] RECA crpf For cross-regional energy consumption positive fluctuation adaptability; v (pr) The regional average energy consumption adaptability is greater than The weight of region pr in all regions; For the regional average energy consumption adaptability of region pr; R + The regional average energy consumption adaptability is greater than The total number of regions.

[0037] Furthermore, in step 101, the original message includes a unique identifier, data acquisition time, positioning status, longitude, latitude, vehicle status, charging status, SOC, total voltage, total current, cumulative mileage, and vehicle speed.

[0038] Furthermore, in step 104, for the total voltage and SOC fields where the data fluctuation frequency is much lower than the data acquisition frequency, a time-series interpolation method is used to fill in their missing values, specifically linear interpolation:

[0039]

[0040] In the formula, v h ,v1,v2,...,v i ,...,v n-1 ,v n ,v t For a series of consecutive frames, the values ​​in the target interpolation field are excluding v. h With v t All values ​​outside the range are missing. For missing value v i The corresponding interpolation value; t h ,t1,t2,...,t i ,...,t n-1 ,t n ,t t Let t be the data acquisition time for each of the consecutive frames in the string, and t t -t h <thr。;

[0041] For the cumulative mileage field, first traverse the frames to be interpolated in chronological order, and then perform interpolation using a velocity-time numerical integration method based on the trapezoidal formula:

[0042]

[0043] Among them, mileage, mileage -1 These represent the cumulative mileage for the current frame and the previous frame, respectively; t,t -1 These represent the data acquisition times for the current frame and the previous frame, respectively. -1 The unit is h; speed, speed -1 These represent the vehicle speeds in the current frame and the previous frame, respectively. If tt -1 Greater than the predetermined threshold Δt, or mileage -1If a value is missing, the missing value for that frame will not be filled. After the traversal is complete, the frames that are still missing after accumulating mileage are traversed in reverse chronological order, and the missing values ​​are interpolated according to the above formula.

[0044] Furthermore, in step 203, the method for calculating the net battery energy consumption per unit mileage of the regional secondary segment is as follows:

[0045]

[0046] Among them, ECR is the net energy consumption of the battery per unit mileage; ΔSOC and Δmileage are the SOC reduction of the end frame of the secondary segment relative to the start frame and the cumulative mileage increase; rating_energy is the nominal energy of the power battery of the target vehicle.

[0047] The beneficial effects achieved by this invention are:

[0048] This invention enables energy consumption adaptability evaluation results to accurately reflect the impact of natural environment and traffic characteristics in a specific region by independently statistically analyzing and processing energy consumption data by region. This improves the geographical adaptability of the evaluation and avoids the masking effect of common energy consumption statistics across all regions on regional differences. It helps users choose vehicle models according to their own region and also facilitates manufacturers to optimize regional layout and differentiated design, thus enhancing the reference value of the evaluation results.

[0049] This invention performs stratification and sample balancing of energy consumption data based on region-independent operating condition characteristics, which effectively avoids random interference from region-independent confounding factors on energy consumption adaptability evaluation indicators, enhances the stability of indicator calculation results on sample data with different distributions, and makes the indicators between different vehicle models and regions more comparable.

[0050] This invention introduces an evaluation index for the positive fluctuation range of energy consumption, reflecting the average increase in energy consumption of a specific vehicle model based on mean adaptability. Energy consumption evaluation is upgraded from a traditional single-point to a range-based approach, providing important reference for the redundant design and use of vehicle electrical energy. Specifically, regional energy consumption positive fluctuation adaptability measures the average increase in energy consumption of a specific vehicle model within a specific region under predetermined operating conditions, helping manufacturers and users establish reasonable expectations for the economic degradation of target models in that region. Cross-regional energy consumption positive fluctuation adaptability measures the average increase in energy consumption of a target model when the region changes, helping to reflect the geographical adaptability of the vehicle model from an energy consumption perspective and promoting related design improvements. Attached Figure Description

[0051] Figure 1 This is a diagram showing the overall steps of the method proposed in this invention;

[0052] Figure 2 This is a flowchart of the S100 data preprocessing steps in the method proposed in this invention;

[0053] Figure 3 This is a flowchart illustrating the steps involved in constructing the secondary fragment feature database of the S200 region in the method proposed in this invention.

[0054] Figure 4 This is a flowchart of the secondary segmentation of region S202 in the method proposed in this invention;

[0055] Figure 5 This is a flowchart of the steps for evaluating the energy consumption adaptability of the S300 region in the method proposed in this invention. Detailed Implementation

[0056] The present invention will be further described below with reference to specific embodiments, and the advantages and features of the present invention will become clearer as a result. However, these embodiments are merely exemplary and do not constitute any limitation on the scope of the present invention. Those skilled in the art should understand that modifications or substitutions can be made to the details and form of the technical solutions of the present invention without departing from the spirit and scope of the present invention, but all such modifications and substitutions fall within the protection scope of the present invention.

[0057] To more realistically evaluate the energy consumption of new energy vehicles while also being regionally specific, this invention proposes a method for evaluating the regional energy consumption adaptability of new energy vehicles. Regional energy consumption adaptability refers to the degree to which a particular vehicle model adapts its energy consumption to a specific region and its adaptability to changes in the region. A lower average energy consumption with minimal fluctuations when operating conditions / regions change is considered optimal.

[0058] like Figure 1 The specific implementation process of the method proposed in this invention can be divided into three main steps:

[0059] S100 data preprocessing,

[0060] Construction of the S200 region secondary fragment feature database

[0061] S300 regional energy consumption adaptability evaluation;

[0062] A complete process has been established to calculate the regional energy consumption adaptability index of the target model from massive amounts of raw vehicle operation data.

[0063] S100 preprocesses the raw message data of vehicle operation data, cleans up abnormal data and fills in missing data.

[0064] like Figure 2 As shown, this step includes:

[0065] S101 data structuring;

[0066] S102 travel segment segmentation;

[0067] S103 Data Abnormal Cleaning;

[0068] S104 missing value filling.

[0069] S101, through a series of decoding, transformation, segmentation and sorting operations, transforms the original message of vehicle operation data into a structured data table with a single row corresponding to a single frame and a single column corresponding to a single field, and corrects the timing and data values.

[0070] Taking vehicle operation data collected in accordance with GB / T 32960 (hereinafter referred to as the national standard) as an example, the first step is to parse the hexadecimal raw message into detailed message according to the relevant protocol, and initially form structured data in tabular form.

[0071] Due to the needs of transmission and storage processes, most national standard data fields have been scaled and offset. Therefore, the second step is to filter the required fields and restore the scaled and offset data of the selected fields based on their definitions to facilitate subsequent processing. The filtered fields and their definitions are shown in Table 1. In other embodiments, it is feasible to select more fields to enrich the working condition characteristics of subsequent steps.

[0072] Table 1 shows the required data fields and their definitions.

[0073]

[0074]

[0075] The third step is to separate the data of different vehicles based on their unique identification codes, forming multiple structured data tables that correspond one-to-one with each vehicle.

[0076] The fourth step involves arranging the data frames in ascending order of time based on the data acquisition time field within each table, thus completing the S101 data structuring step, since the original order of the data frames may contain timing errors.

[0077] S102 identifies whether the data frame is in motion or not, and segments the motion data frame into trip segments for extraction and retention, so as to facilitate further segmentation and feature extraction in the future.

[0078] The first step identifies whether a data frame is in a driving state based on the vehicle status and charging status fields. According to the field definitions (see Table 1), a vehicle status of 1 and a charging status of 2 or 3 indicate that the vehicle is in a driving state, and the identification result is recorded in a newly added field. Since the start and end points of a vehicle's journey typically correspond to the transition edges of time-series frame-by-frame data from non-driving to driving and from driving to non-driving states, this invention uses driving state transitions as one of the conditions for segmenting journey segments. Furthermore, during the journey, prolonged data transmission interruptions may occur due to equipment failure or poor signal quality. If the interruption period is included in the segmented segments, the long-term data loss may prevent or reduce the accuracy of statistical calculations of energy consumption and operating condition characteristics. Therefore, this invention also introduces the inter-frame time interval as another condition for segmenting journey segments.

[0079] Combining the two segmentation conditions mentioned above, the second step involves segmenting the travel segments for each vehicle's data frame. All data frames are traversed. If a frame is in a driving state and the previous frame is in a non-driving state (or the previous frame does not exist), that frame is marked as the starting frame of a new travel segment. If the time interval between two consecutive driving state frames exceeds a preset threshold δ (e.g., 300s), the second frame is marked as the starting frame of a new travel segment. The driving state data frames between two adjacent new travel segment starting frames (excluding the second new travel segment starting frame) constitute one travel segment. These travel segments are numbered sequentially and recorded in a new field. Finally, non-driving state data frames are deleted, completing S102, the travel segment segmentation.

[0080] S103 cleans up abnormal, invalid values ​​and duplicate time frames in each field to ensure the accuracy of subsequent processing and statistics.

[0081] The first step involves identifying outliers in each field using a fixed threshold method based on information such as the specifications and parameters of the target vehicle's components. Data exceeding the defined reasonable value range is replaced with missing values ​​for subsequent steps. The reasonable value ranges for each field are shown in Table 2.

[0082] Table 2 Examples of Reasonable Value Ranges for Each Field

[0083]

[0084] According to the field definitions in Table 1, there are some invalid data in each field. Some of them are represented by special values ​​(such as the maximum value under the specified data type), and some are marked in the status field (such as invalid values ​​of longitude and latitude fields are marked in the positioning status field).

[0085] The first step has replaced some invalid data with missing values, and the second step is to replace the remaining invalid data with missing values.

[0086] Finally, the third step processes frames with repeated data acquisition times (hereinafter referred to as time-repeated frames). For a series of time-repeated frames, check whether they have data similarity in each field—that is, the discrete fields have the same value, and the range of the continuous fields is within a predetermined threshold. Delete all frames except the first frame, and replace the field values ​​of the first frame that do not have data similarity with missing values. The range thresholds for continuous fields in time-repeated frames are shown in Table 3.

[0087] Table 3. Examples of range thresholds for continuous fields of time-repeating frames.

[0088]

[0089] S104 uses regression imputation and time-series imputation methods to fill in missing values ​​in the data, thereby ensuring the normal progress and accurate calculation of subsequent processing and statistics as much as possible, and avoiding the forced deletion of a large number of data frames due to errors in the programmed processing.

[0090] This invention selects different types of interpolation methods to fill missing values ​​for different fields based on their data characteristics. In a preferred embodiment, for the fields of total current and vehicle speed, whose data fluctuation frequency is much higher than the data acquisition frequency, a regression interpolation method is used to fill their missing values, specifically using the LightGBM model. The data of the other fields in the frame where the missing value is located and the data of the adjacent p frames before and after the field where the missing value is located (the value is adjusted according to the data acquisition frequency, for example, p=2 when the acquisition frequency is 0.1Hz) are used as the model input features. The data frames in the dataset where the target interpolation field is not empty are selected to construct the training set. The SequentialFeatureSelector of sklearn is used to perform stepwise regression feature selection, and optuna is used to carry out Bayesian hyperparameter optimization based on TPE. Finally, the model training is completed and the model prediction is used to fill the missing values ​​of the corresponding fields. For the fields of total voltage and SOC, whose data fluctuation frequency is much lower than the data acquisition frequency, a time-series interpolation method is used to fill their missing values, specifically using linear interpolation. The interpolation algorithm is shown in Equation (1).

[0091]

[0092] In the formula, v h ,v1,v2,...,v i ,...,v n-1 ,v n ,v t For a series of consecutive frames, the values ​​in the target interpolation field are excluding v. h With v t All values ​​outside the range are missing. For missing value v i The corresponding interpolation value; t h ,t1,t2,...,ti ,...,t n-1 ,t n ,t t are the data acquisition times for each of the consecutive frames in the string, and t t -t h <thr (for example, thr = 120 s).

[0093] For the cumulative mileage field, traverse the frames to be interpolated in chronological order, and use the velocity-time numerical integration method based on the trapezoidal formula for interpolation, as shown in Equation (2).

[0094]

[0095] where mileage, mileage -1 are the cumulative mileage (km) of this frame and the previous frame respectively; t, t -1 are the data acquisition times of this frame and the previous frame respectively, and the unit of t - t -1 is h; speed, speed -1 are the vehicle speeds (km / h) of this frame and the previous frame respectively. If t - t -1 is greater than the predetermined threshold Δt (for example, Δt = 15 s), or mileage -1 is a missing value, then the missing value of this frame is not filled. After the traversal is completed, then traverse the frames with missing cumulative mileage in reverse chronological order and perform interpolation according to Equation (2).

[0096] After filling the missing values of the cumulative mileage, perform interpolation and filling on the longitude and latitude fields. Use a linear interpolation algorithm similar to that shown in Equation (1), but replace the data acquisition time in the equation with the cumulative mileage.

[0097] After the above steps, there may still be some missing values in the data, which are often due to the fact that a continuous string of data frames where the missing values are located contains the start and end frames of a trip segment or the duration of data loss is too long. In this case, it is difficult to accurately fill the data missing values, so the trip segment data still containing missing values is deleted at the end of the missing value filling step in S104.

[0098] S200 Based on the administrative division, the preprocessed vehicle operation data is sliced into regional sub-segments, and a regional sub-segment feature database with a single row corresponding to a single regional sub-segment is constructed.[[ID=4a]]

[0099] such as Figure 3 shown, this step includes:

[0100] S201 Region identification;

[0101] S202 Regional sub-segment slicing;

[0102] S203 Regional sub-segment feature extraction; It should be noted that in the above translation, "mileage" is used as a general term for cumulative mileage. You can adjust it according to the specific context if there are more accurate terms in the original language. Also, for the special tags like i , etc., they are kept as they are according to the requirements. And for the "thr" and "Δt" which are likely specific notations in the original text, they are directly translated and the values in brackets are also translated as they are for the sake of maintaining the original information.

[0103] S204 secondary fragment feature data cleaning.

[0104] First, S201 uses the longitude and latitude fields of the data frame to identify the region label of the data frame using the reverse geocoding interface of the electronic map platform, which supports the subsequent steps of further segmenting secondary segments and feature extraction.

[0105] Administrative boundaries are often set based on geographical elements (mountains, rivers, etc.). A single administrative region typically has relatively uniform natural conditions, while the natural conditions of different administrative regions often differ significantly. Furthermore, road traffic conditions also vary across administrative boundaries due to economic and social factors. Therefore, this invention uses the geographical boundaries of provincial, prefectural, and county-level administrative divisions as the regional segmentation standard to make regional energy consumption adaptability assessments more accurate and better targeted to specific regions. Simultaneously, this regional division standard also provides a more intuitive reference. Specifically, regional identification is based on the reverse geocoding interface provided by the electronic map platform, which typically takes a location's latitude and longitude coordinates as input and returns information about the region where that location is located. In a preferred embodiment, the data frames of each travel segment are traversed using a step size h (adjusted according to the data acquisition frequency, e.g., h=3 when the data acquisition frequency is 0.1Hz). The longitude and latitude fields of the data frames are converted to the geographic coordinate system specified by the reverse geocoding interface. Then, the data is input into the reverse geocoding interface of an electronic map platform to obtain the 6-digit administrative division code (adcode) of the region specified by the Ministry of Civil Affairs for that frame. This code is recorded as the region label in a newly added field. If region identification fails, an empty string is recorded. Data frames that have undergone region identification are hereinafter referred to as identification frames. For data frames that have not undergone region identification, it is assumed that their region label is consistent with the region label of the most recent identification frame.

[0106] S202 uses regional label changes as the segmentation condition and combines it with a regional mutation tolerance mechanism to further segment the travel segment into regional sub-segments.

[0107] The purpose of this step is to segment vehicle operation data according to its location, laying the foundation for subsequent extraction and statistical analysis of energy consumption and operating characteristics for specific regions, thus making the regional energy consumption adaptability evaluation more region-specific. Due to the potential non-convexity and enclave phenomena at regional boundaries, vehicles may briefly enter other regions while driving within a certain region and then quickly return to their original region. Strictly segmenting secondary regional segments at the moment of regional change could lead to a large number of extremely short secondary regional segments. Such excessively short segments are difficult to accurately calculate energy consumption and are significantly affected by random factors. Therefore, a mutation tolerance mechanism is introduced in the segmentation of secondary regional segments to ignore brief regional changes. In a preferred embodiment, the specific process of S202 regional secondary segmentation is as follows: Figure 4As shown. For a certain journey segment, the first frame is marked as the starting frame of the new secondary segment. Then, the identification frames are traversed to find the first frame in which the region is successfully identified. The region label of this frame is taken as the region where the current secondary segment is located. The identification frames are traversed again. If the region identification of the current frame fails, the region label is considered to be consistent with the previous one. If the region label is inconsistent with the region where the current secondary segment is located, it is checked whether the region label is restored in the subsequent mt (e.g., mt = 500m). If it is restored, the region abrupt change is "tolerated" and no secondary segment is split. If it cannot be restored, the frame is marked as the starting frame of the new secondary segment, and the region where the current secondary segment is located is updated to the region label of this frame. After the traversal is completed, the data between the starting frame of the new secondary segment and the starting frame of the next new secondary segment (or the last frame of the journey segment if it does not exist) (excluding the starting frame of the next new secondary segment) is taken as a region secondary segment, and the duration of each region secondary segment is checked. If the duration of a region secondary segment exceeds a predetermined threshold fdt (e.g., fdt = ...), the traversal continues.

[0108] If the data frame is within 7200s, then the appropriate data frame is marked as the starting frame of the new secondary segment, thus minimizing and uniformly interrupting the secondary segments in that region, so that the duration of the newly formed secondary segment is exactly less than fdt. Finally, the secondary segment sequence number is assigned to each secondary segment data frame according to the time sequence and recorded in the newly added field.

[0109] S203 extracts unit mileage energy consumption, regional labels, and operating condition characteristics (such as start time, end time, start cumulative mileage, end cumulative mileage, etc.) from each regional sub-segment to construct a regional sub-segment feature database with each row corresponding to a single regional sub-segment, providing statistical samples and hierarchical basis for regional energy consumption adaptability evaluation.

[0110] In a preferred embodiment, the net battery energy consumption per unit mileage, regional label, monthly and cumulative mileage characteristics of the regional secondary segment are extracted (the listed characteristics are only examples, and the energy consumption characteristics and operating condition characteristics may vary depending on the data used and the requirements of energy consumption adaptability evaluation under specific conditions). The method for calculating the net battery energy consumption per unit mileage of a certain regional secondary segment is shown in Equation (3).

[0111]

[0112] Among them, ECR is the net energy consumption of the battery per unit mileage (Wh / km); ΔSOC and Δmileage are the SOC reduction (1%) of the end frame of the secondary segment relative to the start frame and the cumulative mileage increase (km); rating_energy is the nominal energy of the power battery of the target vehicle (kWh).

[0113] Based on the proposed regional secondary segmentation method, the regional label feature is determined by the regional label of the first frame in the regional secondary segment where the region is successfully identified. The cumulative mileage and month features are determined by the cumulative mileage field and data acquisition time field of the first frame in the regional secondary segment. By traversing all regional secondary segments of the target vehicle data and extracting features using the above algorithm, a data table can be obtained where each row corresponds to a single regional secondary segment and each column corresponds to a single feature, thus initially constructing a regional secondary segment feature database.

[0114] S204 performs anomaly cleaning on some fields of the feature data, removing rows corresponding to secondary fragments with extreme, abnormal, or invalid feature values.

[0115] S103 data anomaly cleaning only identifies outliers independently for each field in each frame, failing to address anomalies in multi-field numerical relationships or multi-frame numerical temporal changes, such as excessive mileage changes within a unit time or a significant increase in SOC during driving. These anomalies are often reflected in the extracted energy consumption and operating condition features. Therefore, this step identifies and processes outliers in the feature data to prevent them from affecting the accuracy of the energy consumption adaptability evaluation. Furthermore, due to reasons such as failed region identification, features extracted in previous steps may contain invalid values; this step deletes the corresponding rows of secondary segments containing invalid values. In a preferred embodiment, the quantile truncation method is used to identify extreme values, i.e., outliers, of the net battery energy consumption per unit mileage. The quantile thresholds are set as q... min ,q max (e.g., take q) min =0.3%, q max =99.7%), calculate the corresponding quantiles of all net energy consumption data per unit mileage, and delete quantiles less than q. min or greater than q max The data rows were processed to remove, to some extent, secondary segments from areas exhibiting abnormal fluctuations in SOC / cumulative mileage or operating under extreme conditions. Subsequently, rows with empty string labels were removed. These secondary segment regions failed to be identified throughout the entire trip and could not be used for subsequent regional energy consumption adaptability evaluation.

[0116] S300 performs stratification and sample balancing on the unit mileage energy consumption characteristic data in the regional secondary segment feature database based on operating condition characteristics, and calculates the regional energy consumption adaptability evaluation index considering the weight of the operating condition layer.

[0117] like Figure 5 As shown, this step includes:

[0118] S301 Working Condition Stratification and Quantity Balance;

[0119] S302 Regional Energy Consumption Average Adaptability Evaluation;

[0120] S303 Regional Energy Consumption Positive Fluctuation Adaptability Evaluation;

[0121] S304 Cross-regional energy consumption positive fluctuation adaptability evaluation.

[0122] S301, the energy consumption characteristic data per unit mileage is allocated to different layers according to the region-independent operating conditions, so that different predetermined weights are applied to the data of different operating conditions during the energy consumption adaptability evaluation. Then, the oversampling method is used to balance the sample size of each layer, so as to avoid the random interference of region-independent confounding factors in the regional energy consumption adaptability evaluation as much as possible.

[0123] This step significantly reduces the impact of data sampling randomness on the calculated evaluation indicators, making the indicators for each region more stable and more comparable, and allowing the operating condition weights to be effectively applied to the evaluation. To define the operating condition stratification, the operating condition features of interest are first selected, preferably those unrelated or weakly related to the region. This ensures that after the stratified sample size is balanced, the energy consumption adaptability evaluation indicators can reflect regional differences and avoid comparability due to confounding biases caused by regionally unrelated operating condition distribution differences. In a preferred embodiment, the operating condition features of interest are month and cumulative mileage. Next, the operating condition features are discretized. For discrete features (e.g., month), maintaining the original classification or performing category fusion simplification is feasible. For continuous features (e.g., cumulative mileage), a certain range should be selected and binned. For example, the month feature retains its original classification 1, 2, ..., 11, 12; the cumulative mileage feature is discretized according to a binning rule of evenly dividing 0-300,000 km into 6 bins, and assigned numbers 1-6 based on the specific value (data that does not belong to the specified range and cannot obtain discretization labels is deleted). Based on the selected and discretized operating condition characteristics, their entire combination is defined as a joint operating condition layer. The unit mileage energy consumption characteristic data of each region is allocated to the layer according to the corresponding operating condition characteristics, and a layer label is assigned to each layer. In the preferred embodiment proposed in this step, the discretized combination of month and cumulative mileage forms 72 layers, such as 1_1 (corresponding to January and cumulative mileage belonging to [0km, 50000km)) and 1_2 (corresponding to January and cumulative mileage belonging to [50000km, 100000km)). The unit mileage energy consumption characteristic data of all regional sub-segments obtains the layer label based on the month and cumulative mileage field values ​​of their corresponding sub-segments, and is recorded in a newly added field. Finally, sample size balancing is performed for each operating condition layer. To minimize the loss of usable data, sample size balancing is accomplished using oversampling methods such as Bootstrap and SMOTE. In a preferred embodiment, the SMOTE method is used in this step. For the secondary segment feature data of a certain region, the dataset used by SMOTE is initially constructed by selecting data from four fields: net energy consumption per unit mileage, month, cumulative mileage, and operating condition label, where the operating condition label is the category label. Before officially starting oversampling, discrete feature encoding and data standardization are carried out to support the inter-sample distance calculation steps included in SMOTE. For the data of the month field, it is encoded according to equations (4) and (5) to transform the month field into a continuous field, and the original month field is deleted in the SMOTE input dataset.

[0124] month_sin=sin(πmonth / 6) (4)

[0125] month_cos=cos(πmonth / 6) (5)

[0126] Where month_sin and month_cos are the encoded results of the month field data; month is the original month field data.

[0127] For the net energy consumption per unit mileage, the new field obtained by encoding the two months' fields, and the cumulative mileage field, the data are standardized according to formula (6).

[0128]

[0129] Where z represents the standardized data; x represents the original data; and μ and σ represent the mean and standard deviation of all data in the field to which x belongs, respectively.

[0130] The dataset after discrete feature encoding and data standardization is input into the SMOTE algorithm implemented by the imblearn package in Python to balance the number of samples in each working condition layer. At this point, the number of samples in each working condition layer should be consistent, and all samples should show a uniform distribution across the selected working condition features. While ensuring the consistency of working condition stratification in each region, the aforementioned working condition stratification and quantity balancing process is performed on the data in each region to complete the S301 working condition stratification and quantity balancing.

[0131] This invention proposes three evaluation indicators for the regional energy consumption adaptability of target vehicle models from different dimensions: regional average energy consumption adaptability, regional positive energy consumption fluctuation adaptability, and cross-regional positive energy consumption fluctuation adaptability.

[0132] S302 calculates the weighted average of the unit mileage energy consumption characteristics based on the weight of different operating conditions, and uses it as an evaluation index for the regional energy consumption mean adaptability to measure the overall energy consumption level of the target vehicle under various operating conditions in a fixed area.

[0133] For a certain region, first determine its joint hierarchical weight matrix W for operating conditions, and its elements... The weights are corresponding to the operating conditions. W is usually determined based on the travel needs of vehicle users in a specific region under various operating conditions, and is obtained through methods such as questionnaires and expert discussions. This invention does not specifically specify the method for determining W. It should be noted that W may be different in different regions. Subsequently, for a certain region, the regional average energy consumption adaptability of the target vehicle model is calculated according to equation (7).

[0134]

[0135] Among them, RECA mean To adapt to the regional average energy consumption; Let c be the joint hierarchical vector of working conditions for samples s in all secondary segments of a certain region. s The components of the sample are the values ​​of the selected working condition features after discretization. The weight of the working condition layer to which sample s belongs in the weight matrix W; ECR s Let be the energy consumption characteristic per unit mileage of sample s; sum(W) and size(W) are the sum of elements and the number of elements in W, respectively; N is the total number of secondary segments in a certain region.

[0136] Regional average energy consumption adaptive RECA mean This indicator, based on a weighted and comprehensive consideration of various operating conditions, measures the energy consumption of a target vehicle in a specific region. From a mean perspective, it evaluates the energy consumption adaptability of the target vehicle to a particular region. It compares the RECA (Regenerative Calories and Capacities) of multiple (similar) vehicles in a given region. mean It can effectively reflect the advantages and disadvantages of energy consumption of each vehicle model, and has regional specificity and versatility under changing operating conditions.

[0137] S303 calculates the weighted generalized positive standard deviation of the unit mileage energy consumption characteristics based on different operating condition layer weights relative to the regional energy consumption mean, as an evaluation index for the positive fluctuation of regional energy consumption, to measure the positive fluctuation range of the comprehensive energy consumption of the target vehicle under various operating conditions in a fixed region.

[0138] Given the significant fluctuations in vehicle energy consumption under real-world driving conditions, a single energy consumption evaluation inevitably lacks practical reference value. Therefore, evaluating energy consumption fluctuations is crucial. Under certain operating conditions, increased energy consumption can lead to a significantly shorter driving range than expected, potentially causing substantial losses for vehicle users. Conversely, negative fluctuations in vehicle energy consumption only result in small savings in electricity costs for vehicle users, with minimal impact on the driving experience. Therefore, the evaluation of energy consumption fluctuations should focus primarily on positive fluctuations.

[0139] For a certain region, the regional energy consumption positive fluctuation adaptability of the target vehicle model is calculated according to formula (8).

[0140]

[0141] Among them, RECA pf To adapt to positive fluctuations in regional energy consumption; The energy consumption per unit mileage within a certain region is greater than RECA. mean The joint hierarchical vector c of the working conditions of sample p in the secondary fragment of the region p The components of the sample are the values ​​of the selected working condition features after discretization. The weight of the working condition layer to which sample p belongs in the weight matrix W (consistent with S302); ECR p N represents the energy consumption characteristics per unit mileage for sample p; + The energy consumption per unit mileage within a certain region is greater than RECA. mean The total number of secondary segments in the region.

[0142] Regional Energy Consumption Positive Fluctuation Adaptive RECA pf From the perspective of generalized positive standard deviation, this indicator evaluates the energy consumption adaptability of a target vehicle model to a specific region. It comprehensively considers various operating conditions and indicates a reference value for the increase in energy consumption of a specific vehicle model relative to the regional average energy consumption. This elevates energy consumption evaluation from a traditional scalar to a range-based approach, helping manufacturers and vehicle users to reasonably understand the potential deterioration in the fuel economy of the target vehicle model.

[0143] The S304 cross-regional energy consumption positive fluctuation adaptability evaluation steps calculate the average energy consumption of the entire region, and based on this value, calculate the generalized positive standard deviation of the energy consumption average adaptability evaluation of each region, measuring the positive fluctuation range of the target vehicle's energy consumption when the region changes.

[0144] The aforementioned regional energy consumption mean adaptability and regional energy consumption positive fluctuation adaptability indicators are both calculated under fixed regional conditions, failing to reflect the target vehicle's energy consumption performance in variable regions. For most vehicle users, their specific location renders regional energy consumption evaluations lacking reference value. However, for vehicle manufacturers with a wide sales range, regional energy consumption evaluation is crucial for assessing the energy consumption adaptability of vehicle products under regional changes and highlighting economic shortcomings. Referring to the aforementioned two energy consumption adaptability evaluation indicators, evaluations can still be conducted from the perspectives of mean and generalized positive standard deviation under variable regional conditions. Considering the regional energy consumption mean adaptability (RECA) under fixed regional conditions... mean Because of its finer granularity, it has higher accuracy compared to the average energy consumption of the entire region. Therefore, the energy consumption adaptability evaluation in a variable region only focuses on the positive fluctuation range of energy consumption across regions.

[0145] First, calculate the weighted average value of the energy consumption characteristics per unit mileage of the entire region according to formula (9).

[0146]

[0147] in, This is a weighted average of energy consumption characteristics per unit mileage across the entire region; v (r) N (r) These represent the regional weight and the number of secondary segments in region r (after working condition stratification and quantity balancing); r is the total number of regions. Let W be the weight of the working condition layer to which the secondary segment i of region r belongs, i.e., the joint working condition layer weight matrix W of region r. (r) Elements; ECR i (r) Let v be the energy consumption characteristic per unit mileage of the secondary segment i of region r. It is worth noting that this invention does not specifically specify the region weight v. (r) The method for determining this is as follows. In a preferred embodiment, it is proportional to the per capita vehicle ownership in the region.

[0148] Next, the cross-regional energy consumption positive fluctuation adaptability of the target model is calculated according to Equation (10).

[0149]

[0150] Among them, RECA crpf For cross-regional energy consumption positive fluctuation adaptability; v (pr) The regional average energy consumption adaptability is greater than The weight of region pr in all regions; For the regional average energy consumption adaptability of region pr; R + The regional average energy consumption adaptability is greater than The total number of regions.

[0151] It is worth noting that the cross-regional energy consumption positive fluctuation adaptive algorithm proposed in this invention directly calculates the weighted generalized positive standard deviation of the regional energy consumption mean adaptability of each region, rather than calculating the weighted average squared deviation of all positive fluctuation regions' secondary segments. This avoids including energy consumption fluctuations within the region in the evaluation index, and only reflects the positive energy consumption fluctuations caused by regional changes. Furthermore, compared to the calculation of between-group variance in traditional ANOVA, the proposed algorithm replaces the sample size weight with a custom regional weight, eliminating the influence of the sample size in each region on the index. Cross-regional energy consumption positive fluctuation adaptive RECA crpf This helps vehicle manufacturers measure the positive fluctuation range of energy consumption of target models when regional changes occur, thereby revealing the economic deficiencies of battery thermal management and HVAC systems.

[0152] The three energy consumption adaptability evaluation indicators proposed in this invention are: Regional Average Energy Consumption Adaptability (RECA) and Regional Average Energy Consumption. mean Regional energy consumption positive fluctuation adaptive RECA pf and cross-regional energy consumption positive fluctuation adaptive RECA crpf Together, they form a regional energy consumption adaptability evaluation system for the target vehicle model, addressing energy consumption adaptability in specific regions and under regional changes. Among them, the Regional Average Energy Consumption Adaptability (RECA) is... mean RECA adaptability to positive fluctuations in regional energy consumption pf It can be used to compare the energy consumption adaptability of different models in a specific region (of the same class), while the cross-regional energy consumption positive fluctuation adaptability RECA crpf This is used to compare the energy consumption adaptability of different models (of the same class) when the region changes.

[0153] The above are merely specific steps of the present invention and do not constitute any limitation on the scope of protection of the present invention; all technical solutions formed by equivalent transformation or equivalent substitution fall within the scope of protection of the present invention; the parts of the present invention not described in detail are common knowledge to those skilled in the art.

Claims

1. A method for evaluating the regional energy consumption adaptability of new energy vehicles, characterized in that, The method for evaluating the regional energy consumption adaptability of new energy vehicles includes the following steps: S100 preprocesses the raw message data of vehicle operation data, cleans up abnormal data and fills in missing data. S200 divides the preprocessed vehicle operation data into regional sub-segments based on administrative divisions and constructs a regional sub-segment feature database with each row corresponding to a single regional sub-segment. S300 performs stratification and sample balancing on the unit mileage energy consumption characteristic data in the regional secondary segment characteristic database based on operating condition characteristics, and calculates the regional energy consumption adaptability evaluation index considering the weight of the operating condition layer. Step 300 also includes the following steps: S301, the energy consumption characteristic data per unit mileage is allocated to different layers according to the region-independent operating conditions, so that different predetermined weights are applied to the data of different operating conditions during the energy consumption adaptability evaluation. Then, the oversampling method is used to balance the sample size of each layer, so as to avoid the random interference of region-independent confounding factors in the regional energy consumption adaptability evaluation as much as possible. S302, calculate the weighted average of the unit mileage energy consumption characteristics based on the weight of different working condition layers, as the regional energy consumption mean adaptability evaluation index, to measure the level of comprehensive energy consumption of the target vehicle under various working conditions in a fixed area. S303 calculates the weighted generalized positive standard deviation of the unit mileage energy consumption characteristics based on the weight of different working condition layers relative to the regional energy consumption mean, as an evaluation index of regional energy consumption positive fluctuation adaptability, and measures the positive fluctuation range of the comprehensive energy consumption of the target vehicle under various working conditions in a fixed region. S304 calculates the average energy consumption of the entire region, and based on this value, calculates the generalized positive standard deviation of the adaptability evaluation of the average energy consumption of each region, measuring the positive fluctuation range of the target vehicle's energy consumption when the region changes.

2. The method for evaluating the regional energy consumption adaptability of new energy vehicles according to claim 1, characterized in that, Step 100 also includes the following steps: S101, through decoding, transformation, segmentation and sorting operations, transforms the original message of vehicle operation data into a structured data table with a single row corresponding to a single frame and a single column corresponding to a single field, and corrects the timing and data values. S102, the driving status of the data frame is identified, and the driving status data frame is segmented into trip segments for extraction and retention, so as to facilitate further segmentation and feature extraction in the future; S103 cleaned up abnormal, invalid values ​​and duplicate time frames in each field to ensure the accuracy of subsequent processing and statistics. S104 uses regression imputation and time-series imputation methods to fill in missing values ​​in the data, thereby ensuring the normal progress and accurate calculation of subsequent processing and statistics as much as possible, and avoiding the forced deletion of a large number of data frames due to errors in the programmed processing.

3. The method for evaluating the regional energy consumption adaptability of new energy vehicles according to claim 1, characterized in that, Step 200 also includes the following steps: S201, based on the longitude and latitude fields of the data frame, uses the reverse geocoding interface of the electronic map platform to identify the region label of the data frame, supporting subsequent steps to further segment secondary segments and extract features; S202, using regional label changes as the segmentation condition and combining a regional mutation tolerance mechanism, further segments the travel segment into regional secondary segments; S203 extracts unit mileage energy consumption, regional labels and operating condition features from each regional sub-segment, and constructs a regional sub-segment feature database with a single row corresponding to a single regional sub-segment, providing statistical samples and hierarchical basis for regional energy consumption adaptability evaluation; Operating condition characteristics include start time, end time, start cumulative mileage, and end cumulative mileage; S204, the secondary fragment feature data cleaning step performs anomaly cleaning on some fields of the feature data, removing the corresponding rows of secondary fragments with extreme, abnormal, or invalid feature values.

4. The method for evaluating the regional energy consumption adaptability of new energy vehicles according to claim 1, characterized in that, In step 302, the regional average energy consumption adaptability of the target vehicle model is calculated: Among them, RECA mean To adapt to the regional average energy consumption; Let c be the joint hierarchical vector of working conditions for samples s in all secondary segments of a certain region. s The components of the sample are the values ​​of the selected working condition features after discretization. The weight of the working condition layer to which sample s belongs in the weight matrix W; ECR s Let be the energy consumption characteristic per unit mileage of sample s; sum(W) and size(W) are the sum of elements and the number of elements in W, respectively; N is the total number of secondary segments in a certain region.

5. The method for evaluating the regional energy consumption adaptability of new energy vehicles according to claim 1, characterized in that, In step 303, the regional energy consumption positive fluctuation adaptability of the target vehicle model is calculated: Among them, RECA pf To adapt to positive fluctuations in regional energy consumption; The energy consumption per unit mileage within a certain region is greater than RECA. mean The joint hierarchical vector c of the working conditions of sample p in the secondary fragment of the region p The components of the sample are the values ​​of the selected working condition features after discretization. The weight of the working condition layer to which sample p belongs in the weight matrix W; ECR o N represents the energy consumption characteristics per unit mileage for sample p; + The energy consumption per unit mileage within a certain region is greater than RECA. mean The total number of secondary segments in the region.

6. The method for evaluating the regional energy consumption adaptability of new energy vehicles according to claim 1, characterized in that, In step 304, the weighted average value of the energy consumption characteristics per unit mileage of the entire region is calculated. in, This is a weighted average of energy consumption characteristics per unit mileage across the entire region; v (r) N (r) Here, r represents the region weight and the number of secondary segments in region r, respectively; R is the total number of regions. Let W be the weight of the working condition layer to which the secondary segment i of region r belongs, i.e., the joint working condition layer weight matrix W of region r. (r) Elements; ECR i (r) The energy consumption characteristics per unit mileage of the secondary segment i of region r; Calculate the cross-regional energy consumption positive fluctuation adaptability of the target vehicle model: RECA crpf For cross-regional energy consumption positive fluctuation adaptability; v (pr) The regional average energy consumption adaptability is greater than The weight of region pr in all regions; For the regional average energy consumption adaptability of region pr; R + The regional average energy consumption adaptability is greater than The total number of regions.

7. The method for evaluating the regional energy consumption adaptability of new energy vehicles according to claim 2, characterized in that, In step 101, the original message includes a unique identifier, data acquisition time, positioning status, longitude, latitude, vehicle status, charging status, SOC, total voltage, total current, cumulative mileage, and vehicle speed.

8. The method for evaluating the regional energy consumption adaptability of new energy vehicles according to claim 2, characterized in that, In step 104, for the total voltage and SOC fields whose data fluctuation frequency is much lower than the data acquisition frequency, a time-series interpolation method is used to fill in their missing values, specifically linear interpolation: In the formula, v h ,v1,v2,...,v i ,...,v n-1 ,v n ,v t For a series of consecutive frames, the values ​​in the target interpolation field are excluding v. h With v t All values ​​outside the range are missing. For missing value v i The corresponding interpolation value; t h ,t1,t2,...,t i ,...,t n-1 ,t n ,t t Let t be the data acquisition time for each of the consecutive frames in the string, and t t -t h <thr。; For the cumulative mileage field, first traverse the frames to be interpolated in chronological order, and then perform interpolation using a velocity-time numerical integration method based on the trapezoidal formula: Among them, mileage, mileage -1 These represent the cumulative mileage for the current frame and the previous frame, respectively; t,t -1 These represent the data acquisition times for the current frame and the previous frame, respectively. -1 The unit is h; speed, speed -1 These represent the vehicle speeds in the current frame and the previous frame, respectively. If tt -1 Greater than the predetermined threshold Δt, or mileage -1 If a value is missing, the missing value for that frame will not be filled. After the traversal is complete, the frames that are still missing after accumulating mileage are traversed in reverse chronological order, and the missing values ​​are interpolated according to the above formula.

9. The method for evaluating the regional energy consumption adaptability of new energy vehicles according to claim 3, characterized in that, In step 203, the method for calculating the net battery energy consumption per unit mileage of the regional secondary segment is as follows: Among them, ECR is the net energy consumption of the battery per unit mileage; ΔSOC and Δmileage are the SOC reduction of the end frame of the secondary segment relative to the start frame and the cumulative mileage increase; rating_energy is the nominal energy of the power battery of the target vehicle.

Images