Prediction method of remaining driving range of pure electric vehicles

Abstract

The present invention discloses a method for predicting the remaining driving range of a pure electric vehicle, comprising collecting and processing raw data of the vehicle over a preset time span; training a joint prediction model based on the raw data, and obtaining a predicted battery health state through the joint prediction model; and training a mileage prediction model based on the joint prediction model and the raw data; obtaining a change in the state of charge during a charging segment, and calculating the actual energy consumed by the battery based on the change in the state of charge and the predicted battery health state; obtaining the current ambient temperature, and obtaining a predicted non-driving energy demand based on the raw data and the current ambient temperature, and calculating the driving energy demand; obtaining a prediction segment based on the driving energy demand, and inputting the operating condition information, average current, and remaining charge change of the prediction segment into the mileage prediction model to obtain the remaining driving range. The method for predicting the remaining driving range of a pure electric vehicle of the present invention can predict the remaining driving range by combining low-temperature environments and battery degradation conditions to improve prediction accuracy.

Description

Technical Field

[0001] The present invention relates to the technical field of vehicle endurance prediction, and in particular to a method for predicting the remaining endurance of a pure electric vehicle. Background Art

[0002] This section merely provides background information related to the present disclosure and is not necessarily prior art.

[0003] With the continuous advancement of battery technology, the number of electric vehicles in low-temperature areas has maintained a rapid growth trend. However, in the existing technology, the remaining driving range of electric vehicles is predicted to be too high because the effects of low temperature and battery degradation are not taken into account, resulting in low accuracy in the prediction of the remaining driving range of electric vehicles in low-temperature environments. Summary of the Invention

[0004] The present invention aims to at least address the problem of low accuracy in predicting the remaining range of electric vehicles in low-temperature environments. This objective is achieved through the following technical solutions:

[0005] The present invention proposes a method for predicting the remaining driving range of a pure electric vehicle, comprising:

[0006] Collecting and processing raw data of the vehicle for a preset time span, wherein the raw data is used to characterize the vehicle status and battery status, and the preset time span is greater than or equal to 1 year;

[0007] extracting a plurality of pure driving kinematics segments from the raw data, wherein the vehicle only has driving motion in each of the pure driving kinematics segments;

[0008] Obtaining historical prediction features affecting the vehicle's driving range based on each of the pure driving kinematic segments, and obtaining the vehicle's current charging segment and current prediction features affecting the vehicle's driving range;

[0009] Screening out the pure driving kinematics segments whose similarity between the historical prediction feature and the current prediction feature is greater than or equal to the target similarity, and forming a template library;

[0010] Training a joint prediction model based on the raw data, wherein the joint prediction model is used to predict the remaining battery power and the battery health status;

[0011] Inputting the characteristic parameters of the current charging segment into the joint prediction model to obtain a predicted battery health state;

[0012] A mileage prediction model is trained based on the combined prediction model and the original data, where the mileage prediction model is used to predict the remaining battery range;

[0013] Obtaining a change in the state of charge during the charging section, and calculating actual energy consumption of the battery based on the change in the state of charge and the predicted battery health status;

[0014] Acquiring a current ambient temperature, and obtaining a predicted non-driving demand energy based on the raw data and the current ambient temperature;

[0015] Calculating the required energy for driving according to the actual energy consumption of the battery and the predicted non-driving required energy;

[0016] Obtaining a prediction segment from the template library according to the required driving energy, wherein the prediction segment is one or more pure driving kinematic segments in the template library, and a total energy consumption generated by the prediction segment is less than or equal to the required driving energy;

[0017] Obtaining operating condition information, average current, and remaining charge change of the predicted segment, wherein the operating condition information is used to characterize the type of vehicle operating condition and the time proportion of each vehicle operating condition;

[0018] The operating condition information, average current and remaining charge change of the prediction segment are input into the mileage prediction model to obtain the remaining driving range of the vehicle.

[0019] The method for predicting the remaining driving range of a pure electric vehicle of the present invention can predict the remaining driving range of the vehicle in combination with a low temperature environment and a battery degradation condition, thereby improving the prediction accuracy of the remaining driving range of the electric vehicle.

[0020] In some embodiments, the step of collecting and processing raw data of the vehicle before a preset time span includes:

[0021] Acquiring a collection time according to a first preset time interval, collecting characteristic parameters corresponding to the collection time, and forming the raw data;

[0022] Determine whether the original data has duplication anomalies;

[0023] According to the duplicate anomaly in the original data, retain one piece of data in the duplicate data and delete the other data in the duplicate data;

[0024] Comparing the time intervals between adjacent collection times with the first preset time interval;

[0025] According to the time interval of the acquisition time being less than the first preset time interval, it is determined that there is redundant acquisition time, and according to the compared acquisition time and the acquisition time adjacent to the compared acquisition time, the redundant acquisition time and the characteristic parameters corresponding to the redundant acquisition time are acquired and deleted;

[0026] Based on the time interval of the acquisition time being greater than the first preset time interval, it is determined that the acquisition time is missing, the missing acquisition time is obtained based on the compared acquisition time and the first preset time interval, and the characteristic parameters corresponding to the missing acquisition time are supplemented based on the characteristic parameters corresponding to the compared acquisition time.

[0027] In some embodiments, after obtaining the collection time according to the first preset time interval, collecting the characteristic parameters corresponding to the collection time, and forming the raw data, the step of collecting and processing the raw data of the vehicle before the preset time span further includes:

[0028] Screen and clean abnormal feature parameters based on the box plot method and basic logic;

[0029] Determine whether there is data missing for the characteristic parameter;

[0030] If there is missing data in the characteristic parameter, the missing data is supplemented by interpolation, or the missing data is deleted.

[0031] In some embodiments, the step of extracting a plurality of pure driving kinematic segments from the raw data includes:

[0032] dividing the raw data into a plurality of unit segments according to a complete charge cycle and a complete discharge cycle;

[0033] dividing each of the unit segments into a plurality of kinematic segments according to a second preset time interval;

[0034] The kinematic segments are screened for pure driving kinematic segments.

[0035] In some embodiments, the step of obtaining historical prediction features affecting the vehicle's driving range based on each of the pure driving kinematic segments, and obtaining the vehicle's current charging segment and current prediction features affecting the vehicle's driving range, includes:

[0036] Read the date, time and vehicle position information of each pure driving kinematic segment;

[0037] Obtaining historical weather conditions, historical temperature levels, and historical congestion levels of a city where the vehicle has historically been located based on the date, the time, and the location information, and determining a historical working day status based on the date, using the historical weather conditions, the historical temperature levels, the historical congestion levels, and the historical working day status as the historical prediction features, wherein the historical working day status is used to indicate whether the vehicle is on a working day;

[0038] obtaining a current weather condition, a current temperature level, a current congestion level, and a current weekday status of a city where the vehicle is currently located, and using the current weather condition, the current temperature level, the current congestion level, and the current weekday status as the current prediction features, wherein the current weekday status is used to indicate whether the vehicle is on a weekday;

[0039] Get the vehicle's current charging section.

[0040] In some embodiments, the step of training the joint prediction model based on the original data includes:

[0041] Obtaining the average battery voltage, average battery current, instantaneous battery voltage, instantaneous battery current, battery temperature, maximum battery cell voltage, minimum battery cell voltage, remaining battery capacity, and battery health status of each kinematic segment;

[0042] using the average battery voltage, the average battery current, the instantaneous battery voltage, the instantaneous battery current, and the battery temperature as first training parameters for training a prediction model for predicting the remaining battery capacity, and performing a Pearson correlation analysis on each of the first training parameters and the remaining battery capacity to obtain a correlation coefficient between the first training parameter and the remaining battery capacity;

[0043] using the average battery voltage, the average battery current, the battery temperature, the maximum battery cell voltage, and the minimum battery cell voltage as second training parameters for training a prediction model for predicting a battery state of health, and performing a Pearson correlation analysis on each of the second training parameters and the battery state of health to obtain a correlation coefficient between the second training parameter and the battery state of health;

[0044] A first training parameter and a second training parameter having a correlation coefficient greater than or equal to a correlation limit are screened out, and the joint prediction model is trained according to the screened first training parameter and the second training parameter.

[0045] In some embodiments, the step of training a mileage prediction model based on the combined prediction model and the original data includes:

[0046] Inputting the first training parameter into the joint prediction model, and obtaining a change in the remaining battery power of each kinematic segment;

[0047] Obtaining a proportion of the speed below a preset value, a deceleration time proportion, and a mileage of each kinematic segment, using the proportion of the speed below the preset value, the deceleration proportion, the change in the remaining battery charge, and the average current as third training parameters for training the mileage prediction model, and performing a Pearson correlation analysis on each of the third training parameters and the mileage to obtain a correlation coefficient between the third training parameter and the mileage;

[0048] A third training parameter having a correlation coefficient greater than or equal to the correlation limit is screened out, and the mileage prediction model is trained according to the screened third training parameter.

[0049] In some embodiments, the step of obtaining the current ambient temperature and obtaining the predicted non-driving demand energy according to the raw data and the current ambient temperature includes:

[0050] Extracting a plurality of stationary segments from the raw data, wherein the vehicle is stationary in each stationary segment and generates non-driving energy consumption;

[0051] Obtaining the ambient temperature and the non-driving energy consumption in each of the stationary segments, and building a non-driving energy consumption prediction model based on the ambient temperature and the non-driving energy consumption;

[0052] The current ambient temperature is obtained and input into the non-driving energy consumption prediction model to obtain the predicted non-driving required energy.

[0053] In some embodiments, the step of obtaining the ambient temperature and the non-driving energy consumption in each stationary segment, and building a non-driving energy consumption prediction model based on the ambient temperature and the non-driving energy consumption, includes:

[0054] Acquire the ambient temperature of each of the static segments;

[0055] Obtaining resting energy consumption in each stationary segment, where the resting energy consumption includes energy consumption for heating the cabin by air conditioning and energy consumption for maintaining the battery at an operating temperature;

[0056] Regression fitting is performed on the resting energy consumption and the ambient temperature of the resting segment to obtain a non-driving energy consumption prediction model.

[0057] In some embodiments, the plurality of still segments are a plurality of consecutive still segments. BRIEF DESCRIPTION OF THE DRAWINGS

[0058] Various other advantages and benefits will become apparent to those skilled in the art upon reading the detailed description of the preferred embodiment below. The accompanying drawings are for illustration purposes only and are not to be considered as limiting the present invention. The same reference numerals are used throughout the accompanying drawings to denote the same components. In the accompanying drawings:

[0059] Figure 1 Schematic diagram of a method for predicting the remaining driving range of a pure electric vehicle according to an embodiment of the present invention;

[0060] Figure 2 A schematic diagram of mileage;

[0061] Figure 3 is a schematic diagram of the total battery voltage;

[0062] Figure 4 Schematic diagram of battery cell voltage;

[0063] Figure 5 Schematic diagram of function regression fitting. DETAILED DESCRIPTION

[0064] Exemplary embodiments of the present disclosure will be described in more detail below with reference to the accompanying drawings. Although exemplary embodiments of the present disclosure are shown in the accompanying drawings, it should be understood that the present disclosure can be implemented in various forms and should not be limited by the embodiments described herein. Rather, these embodiments are provided to enable a more thorough understanding of the present disclosure and to fully convey the scope of the present disclosure to those skilled in the art.

[0065] It should be understood that the terms used herein are for the purpose of describing specific example embodiments only and are not intended to be limiting. Unless the context clearly indicates otherwise, the singular forms "one", "an" and "said" as used herein may also be meant to include plural forms. The terms "comprise", "include", "contain" and "have" are inclusive and therefore specify the presence of stated features, steps, operations, elements and / or parts, but do not exclude the presence or addition of one or more other features, steps, operations, elements, parts, and / or combinations thereof. The method steps, processes, and operations described herein are not to be construed as necessarily requiring them to be performed in the specific order described or illustrated, unless the order of execution is clearly indicated. It should also be understood that additional or alternative steps may be used.

[0066] Although the terms first, second, third, etc. can be used in the text to describe multiple elements, components, regions, layers and / or sections, these elements, components, regions, layers and / or sections should not be limited by these terms. These terms can only be used to distinguish an element, component, region, layer or section from another region, layer or section. Unless the context clearly indicates otherwise, terms such as "first", "second" and other numerical terms do not imply order or sequence when used in the text. Therefore, the first element, component, region, layer or section discussed below can be referred to as the second element, component, region, layer or section without departing from the teaching of the example embodiments.

[0067] For ease of description, spatially relative terms may be used herein to describe the relationship of one element or feature relative to another element or feature as shown in the figures, such as "inside," "outside," "inside," "outside," "below," "beneath," "above," and the like. Such spatially relative terms are intended to include different orientations of the device in use or operation in addition to the orientation depicted in the figures. For example, if the device in the figures is flipped, an element described as "below" or "below" another element or feature would then be oriented as "above" or "above" another element or feature. Thus, the example term "below" can include both above and below orientations. The device may be otherwise oriented (rotated 90 degrees or in other orientations) and the spatially relative descriptors used herein are interpreted accordingly.

[0068] With the continuous advancement of battery technology, the number of electric vehicles in low-temperature environments continues to grow rapidly. However, existing technologies fail to consider the impact of low temperatures and battery degradation, resulting in overestimation of the remaining driving range of electric vehicles in low-temperature environments. Low-temperature environments refer to temperatures of 0°C or below.

[0069] The present invention aims to at least solve the problem of low accuracy in predicting the remaining cruising range of electric vehicles in low-temperature environments. To this end, an embodiment of the present invention proposes a method for predicting the remaining cruising range of a pure electric vehicle, which can predict the remaining cruising range of the vehicle in combination with the low-temperature environment and the battery degradation condition, thereby improving the accuracy of predicting the remaining cruising range of the electric vehicle.

[0070] The following describes a method for predicting the remaining driving range of a pure electric vehicle according to an embodiment of the present invention with reference to the accompanying drawings.

[0071] like Figure 1 As shown, the method for predicting the remaining driving range of a pure electric vehicle according to an embodiment of the present invention includes:

[0072] S100, collecting and processing raw data of the vehicle for a preset time span, wherein the raw data is used to represent the vehicle status and battery status, and the preset time span is greater than or equal to 1 year;

[0073] S200, extracting a plurality of pure driving kinematic segments from the original data, wherein the vehicle only has driving motion in each pure driving kinematic segment;

[0074] S300, obtaining historical prediction features that affect the vehicle's driving range based on each pure driving kinematic segment, and obtaining the vehicle's current charging segment and current prediction features that affect the vehicle's driving range;

[0075] S400, screening out pure driving kinematics segments whose similarity between historical prediction features and current prediction features is greater than or equal to the target similarity, and forming a template library;

[0076] S500: training a joint prediction model based on the original data, wherein the joint prediction model is used to predict the remaining battery power and the battery health status;

[0077] S600: Input characteristic parameters of the current charging segment into a joint prediction model to obtain a predicted battery health state;

[0078] S700: Training a mileage prediction model based on the combined prediction model and the original data, where the mileage prediction model is used to predict the remaining battery driving range;

[0079] S800: Obtain a change in the state of charge during the charging phase, and calculate actual energy consumption of the battery based on the change in the state of charge and the predicted battery health status;

[0080] S900: Obtain the current ambient temperature, and obtain a predicted non-driving energy demand based on the original data and the current ambient temperature;

[0081] S1000, calculating driving energy demand based on actual battery energy consumption and predicted non-driving energy demand;

[0082] S1100: Obtain prediction segments from a template library based on the required driving energy, where the prediction segments are one or more pure driving kinematic segments in the template library, and total energy consumption generated by the prediction segments is less than or equal to the required driving energy;

[0083] S1200: Obtain operating condition information, average current, and remaining charge change of the predicted segment, wherein the operating condition information is used to characterize the type of vehicle operating condition and the time proportion of each vehicle operating condition;

[0084] S1300: Input the operating condition information, average current, and remaining charge change of the prediction segment into a mileage prediction model to obtain the remaining driving range of the vehicle.

[0085] S100. Collect and process raw data of a vehicle within a preset time span, wherein the raw data is used to characterize a vehicle state and a battery state, and the preset time span is greater than or equal to 1 year.

[0086] The raw data spans a year or more, meaning it spans a relatively long time span. Analyzing this data over a long time span can capture the vehicle's driving patterns and energy consumption characteristics under different conditions, thereby improving the accuracy of range predictions. This data encompasses a wide range of driving conditions, such as road conditions, speeds, and driving behavior. These factors all affect a vehicle's energy consumption and remaining range. Integrating these factors can provide a more comprehensive prediction. This data over a long time span helps the model learn the characteristics of different driving styles and road types, enhancing the model's adaptability to driver behavior and improving prediction accuracy.

[0087] As an example, raw data is collected according to Table 1.

[0088]

[0089] Table 1

[0090] It should be noted that Table 1 exemplarily lists some characteristic parameters of the original data. The original data may also include other characteristic parameters. Those skilled in the art may also select the type of characteristic parameters to be collected according to actual conditions.

[0091] In some embodiments, S100, the step of collecting and processing raw data of the vehicle before a preset time span, includes:

[0092] Acquiring a collection time according to a first preset time interval, collecting characteristic parameters corresponding to the collection time, and forming raw data;

[0093] Determine whether there are duplicate anomalies in the original data;

[0094] According to the duplicate anomaly in the original data, one piece of data in the duplicate data is retained and the other data in the duplicate data is deleted;

[0095] comparing the time interval between adjacent collection times with a first preset time interval;

[0096] Determining that there is excess acquisition time based on the time interval of the acquisition times being less than the first preset time interval, and acquiring and deleting the excess acquisition time and characteristic parameters corresponding to the excess acquisition time based on the compared acquisition time and the acquisition time adjacent to the compared acquisition time;

[0097] Based on the fact that the time interval of the acquisition time is greater than the first preset time interval, it is determined that there is a missing acquisition time. The missing acquisition time is obtained based on the compared acquisition time and the first preset time interval, and the characteristic parameters corresponding to the missing acquisition time are supplemented based on the characteristic parameters corresponding to the compared acquisition time.

[0098] The acquisition time is acquired according to the first preset time interval, the characteristic parameters corresponding to the acquisition time are acquired, and raw data are formed.

[0099] When data is collected at fixed time intervals, the model can better capture patterns and trends in the time series, which helps the model learn the intrinsic structure of the time series to improve the accuracy of predictions.

[0100] Training models on data at fixed time intervals reduces the complexity of handling irregular sampling intervals. The model can more directly utilize the continuity and periodicity characteristics of time series data, thereby simplifying the model training and prediction process.

[0101] By collecting data at regular intervals, the model can generalize better to future data, thereby improving the model's predictive power for future data.

[0102] As some examples, the first preset time interval is 10 seconds. By setting the first preset time interval to 10 seconds, it is possible to avoid the time for collecting data being too long or too short.

[0103] Determine whether the original data contains duplicate anomalies.

[0104] Determine whether the duplicate content in the original data needs to be processed based on whether there are duplicate anomalies in the original data.

[0105] According to the duplicate anomaly in the original data, one piece of data in the duplicate data is retained and the other data in the duplicate data is deleted.

[0106] Keep one piece of duplicate data, which can be any piece of duplicate data.

[0107] By retaining one piece of duplicate data and deleting the other pieces of duplicate data, the duplicate data can be removed, thereby preventing the duplicate data from affecting subsequent steps.

[0108] As some examples, the replicate data are shown in Table 2.

[0109]

[0110] Table 2

[0111] The time interval between adjacent collection times is compared with the first preset time interval.

[0112] By comparing the time intervals between adjacent acquisition times with the first preset time interval, it is possible to determine whether there is a problem with the time interval for acquiring data, so as to facilitate subsequent processing of the original data.

[0113] Based on the time interval being less than the first preset time interval, it is determined that there is redundant collection time, and based on the compared collection time and the collection time adjacent to the compared collection time, characteristic parameters corresponding to the redundant collection time and the redundant collection time are acquired and deleted.

[0114] If the time interval is less than the first preset time interval, it indicates that there is excess collection time. The excess collection time can be determined by the collection time adjacent to the comparison time, and the excess collection time and the characteristic parameters corresponding to the excess collection time are deleted to prevent the erroneous collection data from affecting subsequent steps.

[0115] As shown in Table 3, taking the first preset time of 10s as an example, the interval between the acquisition time 20240101101354 and the acquisition time 20240101101401 is 7s, and the interval between the acquisition time 20240101101354 and the acquisition time 20240101101404 is 10s. Therefore, it is determined that the acquisition time 20240101101401 is redundant acquisition time, and the redundant acquisition time and the characteristic parameters corresponding to the redundant acquisition time are deleted to avoid the influence of erroneous acquisition data on subsequent steps.

[0116]

[0117] Table 3

[0118] Based on the time interval being greater than the first preset time interval, it is determined that there is a missing acquisition time, the missing acquisition time is obtained based on the compared acquisition time and the first preset time interval, and the characteristic parameters corresponding to the missing acquisition time are supplemented based on the characteristic parameters corresponding to the compared acquisition time.

[0119] If the time interval is greater than the first preset time interval, it indicates that there is a missing acquisition time. The missing acquisition time is obtained based on the compared acquisition time and the first preset time interval, and the characteristic parameters corresponding to the missing acquisition time are supplemented based on the characteristic parameters corresponding to the compared acquisition time to avoid the lost acquisition data from affecting subsequent steps.

[0120] Specifically, the missing acquisition time is obtained by interpolation of characteristic parameters of the compared acquisition time.

[0121] As shown in Table 4, taking the first preset time as 10s as an example, the interval between the acquisition time 20230101010249 and the acquisition time 20230101010309 is 20s. It is determined that there is data loss between the acquisition time 20230101010249 and the acquisition time 20230101010309, and the lost acquisition time is 20230101010259. According to the feature parameters corresponding to the acquisition time 20230101010249 and the feature parameters corresponding to the acquisition time 20230101010309, the feature parameters corresponding to the acquisition time 20230101010259 are obtained by interpolation.

[0122]

[0123] Table 4

[0124] This embodiment can process abnormal acquisition time and characteristic parameters corresponding to the abnormal acquisition time, thereby preventing the abnormal parameters from affecting subsequent steps.

[0125] In some embodiments, after obtaining the collection time according to the first preset time interval, collecting the characteristic parameters corresponding to the collection time, and forming the raw data, the step of S100, collecting and processing the raw data of the vehicle before the preset time span, further includes:

[0126] Screen and clean abnormal feature parameters based on the box plot method and basic logic;

[0127] Determine whether there is data missing for feature parameters;

[0128] If there is missing data for the characteristic parameters, the missing data is filled by interpolation, or the missing data is deleted.

[0129] Combine Figure 2 、 Figure 3 and Figure 4 As shown, the box plot can be used to find abnormal jumps in characteristic parameters that are obviously inconsistent with basic common sense, so as to process the abnormal characteristic parameters.

[0130] This embodiment can process abnormal characteristic parameters and missing characteristic parameters to prevent the abnormal and missing characteristic parameters from affecting subsequent operations.

[0131] Specifically, when there are fewer abnormal feature parameters, the abnormal feature parameters are deleted and the deleted feature parameters are supplemented through interpolation. When there are more abnormal features, the data is deleted. This can reduce the complexity of data processing and improve data processing efficiency.

[0132] Specifically, when there are a small number of missing feature parameters, the missing feature parameters are deleted and the deleted feature parameters are supplemented by interpolation. When there are a large number of missing feature parameters, the data is deleted. This can reduce the complexity of data processing and thus improve the efficiency of data processing.

[0133] S200 : Extract multiple pure driving kinematics segments from the original data, wherein the vehicle only has driving motion in each pure driving kinematics segment.

[0134] The pure driving kinematics segment can reflect the driving information of the vehicle. The pure driving kinematics segment is extracted so as to facilitate the prediction of the vehicle mileage based on the driving information of the pure driving kinematics segment in the subsequent steps.

[0135] In some embodiments, S200, the step of extracting a plurality of pure driving kinematic segments from the raw data, includes:

[0136] Dividing the raw data into multiple unit segments based on complete charge cycles and complete discharge cycles;

[0137] dividing each unit segment into a plurality of kinematic segments according to a second preset time interval;

[0138] Filter the kinematic segments for pure driving kinematic segments.

[0139] By dividing the data according to the complete charging cycle and the complete discharging cycle, the kinematic segment can be a motion segment of a charging cycle or a motion segment of a discharging cycle, thereby avoiding the kinematic segment including both the charging cycle and the discharging cycle.

[0140] The pure driving kinematic segments are obtained by further dividing the unit segments into kinematic segments and screening out the pure driving kinematic segments.

[0141] S300: Obtain historical prediction features that affect the vehicle's driving range based on each pure driving kinematics segment, and obtain the vehicle's current charging segment and current prediction features that affect the vehicle's driving range.

[0142] It should be noted that the current charging section may be the last charging section of an uncharged vehicle, or may be the current charging section of a charged vehicle.

[0143] The pure driving kinematics segments can be used to obtain historical prediction features that affect the vehicle's range. Based on the current charging segment, the similarity between the current prediction features and the historical prediction features that affect the vehicle's range can be judged, and the relevance of each pure driving kinematics segment to the current one can be determined.

[0144] In some embodiments, S300, the step of obtaining historical prediction features affecting the vehicle's driving range based on each pure driving kinematic segment, and obtaining the vehicle's current charging segment and current prediction features affecting the vehicle's driving range, includes:

[0145] Read the date, time and vehicle position information of each pure driving kinematic segment;

[0146] Based on the date, time, and location information, the historical weather conditions, temperature levels, and congestion levels of the city where the vehicle has been historically located are obtained. The historical working day status is determined based on the date, and the historical weather conditions, temperature levels, congestion levels, and working day status are used as historical prediction features. The historical working day status is used to indicate whether the vehicle is on a working day.

[0147] Obtain the current weather conditions, temperature level, congestion level, and weekday status of the city where the vehicle is currently located, and use these conditions as current prediction features. The weekday status is used to indicate whether the vehicle is on a weekday.

[0148] Get the vehicle's current charging section.

[0149] Both historical and current weather conditions include sunny, rainy, heavy rain, light snow, and heavy snow. Both historical and current temperature levels include relatively low, low, and very low temperatures. Both historical and current congestion levels include peak and off-peak traffic. Both historical and current workday status include both working and non-working days.

[0150] Based on the similarity between historical weather conditions, historical temperature levels, historical congestion levels, historical weekday conditions, and current weather conditions, current temperature levels, historical congestion levels, and current weekday conditions, the similarity between the pure kinematic segment and the current vehicle can be determined, thereby facilitating prediction.

[0151] S400 , screening out pure driving kinematics segments whose similarity between historical prediction features and current prediction features is greater than or equal to the target similarity, and forming a template library.

[0152] Preliminary screening is performed through pure driving fragments, and a template library is formed to reduce the difficulty of screening in subsequent steps.

[0153] As an example, the target similarity is 95%.

[0154] Specifically, the operating condition prediction architecture includes a feature matching layer, and executes steps S300 and S400 to build a template library.

[0155] S500: Training a joint prediction model based on original data, wherein the joint prediction model is used to predict the remaining battery power and the battery health status.

[0156] The joint model is trained to predict the battery status.

[0157] Specifically, the joint prediction model is trained according to the WOA-XGBoost algorithm.

[0158] In some embodiments, S500, the step of training the joint prediction model based on the original data, includes:

[0159] Obtain the average battery voltage, average battery current, instantaneous battery voltage, instantaneous battery current, battery temperature, maximum battery cell voltage, minimum battery cell voltage, remaining battery capacity, and battery health status for each kinematic segment;

[0160] using the average battery voltage, the average battery current, the instantaneous battery voltage, the instantaneous battery current, and the battery temperature as first training parameters for training a prediction model for predicting the remaining battery capacity, and performing a Pearson correlation analysis on each first training parameter and the remaining battery capacity to obtain a correlation coefficient between the first training parameter and the remaining battery capacity;

[0161] using the average battery voltage, average battery current, battery temperature, maximum battery cell voltage, and minimum battery cell voltage as second training parameters for training a prediction model for predicting the battery state of health, and performing a Pearson correlation analysis on each second training parameter and the battery state of health to obtain a correlation coefficient between the second training parameter and the battery state of health;

[0162] A first training parameter and a second training parameter having a correlation coefficient greater than or equal to a correlation limit are screened out, and a joint prediction model is trained based on the screened first training parameter and the second training parameter.

[0163] The joint model can be trained using the first training parameters and the second training parameters in the kinematic segment, and the accuracy of the joint model training can be improved by screening out the first training parameters and the second training parameters whose correlation coefficients are greater than or equal to the correlation limit.

[0164] As an example, the correlation bound is 0.8.

[0165] S600: Input characteristic parameters of the current charging segment into a joint prediction model to obtain a predicted battery health state.

[0166] By predicting the battery health status, the degree of degradation of the vehicle's current battery can be reflected, making the predicted driving range more accurate.

[0167] S700: Train a mileage prediction model based on the combined prediction model and the original data. The mileage prediction model is used to predict the remaining battery driving range.

[0168] The mileage prediction model can be used to predict vehicle mileage.

[0169] Specifically, the mileage prediction model is trained using the lightGBM algorithm.

[0170] In some embodiments, S700, the step of training the mileage prediction model based on the combined prediction model and the original data, includes:

[0171] Inputting the first training parameter into the joint prediction model and obtaining a change in the remaining battery power of each kinematic segment;

[0172] Obtaining the proportion of the speed below a preset value, the deceleration time proportion, and the mileage of each kinematic segment; using the proportion of the speed below the preset value, the deceleration proportion, the change in the remaining battery charge, and the average current as third training parameters for training a mileage prediction model; and performing a Pearson correlation analysis on each third training parameter and the mileage to obtain a correlation coefficient between the third training parameter and the mileage;

[0173] A third training parameter having a correlation coefficient greater than or equal to the correlation limit is selected, and the mileage prediction model is trained according to the selected third training parameter.

[0174] By screening out the first training parameter and the second training parameter whose correlation coefficient is greater than or equal to the correlation limit, the accuracy of the joint model training can be improved.

[0175] As an example, the preset value is 20 kilometers per hour. It is understandable that, regardless of the direction of movement, the value of the speed is a positive number, so the proportion of speeds lower than the preset value is the proportion of speeds between 0 and 20 kilometers per hour.

[0176] Specifically, the ratio P of the speed lower than the preset value is calculated by the following formula: low_speed

[0177]

[0178] Among them, n low_speed is the sampling point in the kinematic segment where the speed is lower than the preset value, and n0 is all the sampling points in the kinematic segment.

[0179] The average current I is calculated using the following formula average

[0180]

[0181] Where i is the instantaneous current and n0 is the total number of sampling points in the kinematic segment.

[0182] The deceleration time ratio P is calculated by the following formula reduce_speed

[0183]

[0184] Among them, n reduce_speed is the number of sampling points in the kinematic segment where the speed is in a deceleration state, and n0 is all the sampling points in the kinematic segment.

[0185] S800 , obtaining a state of charge change ΔSOC during the charging stage, and calculating actual battery energy consumption Econ based on the state of charge change ΔSOC and the predicted battery state of health SOH.

[0186] As an example, the actual energy consumption Econ of the battery is calculated using the following formula:

[0187] Econ=C·SOH·△SOC

[0188] Where C is the nominal capacity of the battery.

[0189] S900: Obtain the current ambient temperature, and obtain predicted non-driving energy demand based on the original data and the current ambient temperature.

[0190] By predicting non-driving energy demand, the impact of low-temperature environments on battery rest consumption can be taken into account, making the predicted driving energy demand in low-temperature environments more accurate.

[0191] In some embodiments, S900, obtaining the current ambient temperature, and obtaining the predicted non-driving energy demand based on the raw data and the current ambient temperature, includes:

[0192] Extracting a plurality of stationary segments from the original data, wherein the vehicle is stationary in each stationary segment and generates non-driving energy consumption;

[0193] Obtain the ambient temperature and non-driving energy consumption in each stationary segment, and build a non-driving energy consumption prediction model based on the ambient temperature and non-driving energy consumption;

[0194] The current ambient temperature is obtained and input into the non-driving energy consumption prediction model to obtain the predicted non-driving required energy.

[0195] A prediction model is established based on the relationship between the ambient temperature and non-driving energy consumption in the stationary segment, and the current non-driving energy is predicted by the current ambient temperature, thereby realizing the prediction of the current non-driving energy.

[0196] In some embodiments, the steps of obtaining the ambient temperature and non-driving energy consumption in each stationary segment and building a non-driving energy consumption prediction model based on the ambient temperature and non-driving energy consumption include:

[0197] Get the ambient temperature of each stationary segment;

[0198] Obtain the resting energy consumption in each stationary segment. Resting energy consumption includes the energy consumed by the air conditioner to heat the cabin and the energy consumed to maintain the battery at operating temperature.

[0199] The resting energy consumption and the ambient temperature of the resting segment are regressed and fitted to obtain the non-driving energy consumption prediction model.

[0200] Resting energy consumption is mainly generated by the air conditioner heating the cabin and heating the battery to maintain the operating temperature of the battery. Therefore, building a model based on the energy consumption of the air conditioner heating the cabin and the energy consumption of maintaining the operating temperature of the battery can reduce the workload while ensuring the accuracy of the prediction model, thereby improving the efficiency of building the energy consumption prediction model.

[0201] like Figure 5 As shown, as an example, the resting energy consumption and the ambient temperature of the rest segment are fitted by functional regression.

[0202] In some embodiments, the plurality of static segments are a plurality of continuous static segments, so as to further increase the prediction accuracy of the energy consumption prediction model.

[0203] S1000: Calculate driving energy demand based on actual battery energy consumption and predicted non-driving energy demand.

[0204] The driving energy requirement is the actual battery energy consumption minus the predicted non-driving energy requirement.

[0205] S1100. Obtain prediction segments from a template library according to driving demand energy, wherein the prediction segments are one or more pure driving kinematic segments in the template library, and total energy consumption generated by the prediction segments is less than or equal to the driving demand energy.

[0206] The pure driving kinematics segments in the template library are further screened by the driving demand energy to obtain pure driving kinematics segments similar to the current condition of the vehicle, so as to make more accurate predictions about the vehicle.

[0207] Furthermore, when there are multiple prediction segments, one of the multiple prediction segments is selected for prediction.

[0208] Specifically, the operating condition architecture includes an operating condition prediction layer, which is used to execute step S1100.

[0209] S1200: Obtain operating condition information, average current, and remaining charge change for the predicted segment. The operating condition information is used to characterize the type of vehicle operating condition and the proportion of time spent in each operating condition. For example, the operating condition information is used to characterize whether the vehicle is operating in a highway, suburban, or urban condition, and to characterize the proportion of time spent in the highway, suburban, and urban conditions.

[0210] By using the operating condition information of the predicted segment as the operating condition information of the vehicle during driving, and combining it with the average current and the change in the remaining power, it is possible to facilitate the prediction of the vehicle's mileage in subsequent steps.

[0211] Specifically, the operating condition information of the predicted segment is obtained through K-means clustering.

[0212] S1300: Input the operating condition information, average current, and remaining charge change of the prediction segment into a mileage prediction model to obtain the remaining driving range of the vehicle.

[0213] Specifically, the mileage prediction model includes an energy consumption layer for executing steps S800 , S900 , and S1000 , and a mileage prediction layer for executing steps S700 and S1300 .

[0214] It should be noted that the location information and data involved in the present invention (including but not limited to data used for analysis, stored data, displayed data, etc.) are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of relevant data must comply with the relevant laws, regulations and standards of relevant countries and regions.

[0215] It should be understood that, although the various steps in the flowcharts involved in the various embodiments described above are displayed in sequence according to the instructions of the arrows, these steps are not necessarily executed in sequence in the order indicated by the arrows. Unless otherwise specified herein, there is no strict order restriction on the execution of these steps, and these steps can be executed in other orders. Moreover, at least a portion of the steps in the flowcharts involved in the various embodiments described above can include multiple steps or multiple stages, and these steps or stages are not necessarily executed and completed at the same time, but can be executed at different times, and the execution order of these steps or stages is not necessarily to be carried out in sequence, but can be executed in turn or alternately with other steps or at least a portion of steps or stages in other steps.

[0216] The logic and / or steps represented in the flowcharts or otherwise described herein, for example, can be considered as an ordered list of executable instructions for implementing the logical functions, and can be embodied in any computer-readable storage medium for use by, or in conjunction with, an instruction execution system, apparatus, or device (e.g., a computer-based system, a system including a processor, or other system that can fetch and execute instructions from an instruction execution system, apparatus, or device). For purposes of this specification, a "computer-readable storage medium" can be any device that can contain, store, communicate, propagate, or transport a program for use by, or in conjunction with, an instruction execution system, apparatus, or device. More specific examples (a non-exhaustive list) of computer-readable storage media include the following: an electrical connection with one or more wires (electronic device), a portable computer disk cartridge (magnetic device), a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM, or flash memory), a fiber optic device, and a portable compact disc read-only memory (CDROM). In addition, the computer-readable storage medium may even be paper or other suitable medium on which the program is printed, since the program may be obtained electronically, for example, by optically scanning the paper or other medium, followed by editing, deciphering, or processing in another suitable manner as necessary, and then stored in a computer memory.

[0217] It should be understood that various parts of the present invention can be implemented using hardware, software, firmware, or a combination thereof. In the above-described embodiments, multiple steps or methods can be implemented using software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented using hardware, as in another embodiment, any one of the following technologies known in the art or a combination thereof can be used to implement the hardware: a discrete logic circuit having a logic gate circuit for implementing a logic function on a data signal, an application-specific integrated circuit having a suitable combination of logic gate circuits, a programmable gate array (PGA), a field programmable gate array (FPGA), etc.

[0218] The above description is merely a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any changes or substitutions that can be easily conceived by a person skilled in the art within the technical scope disclosed in the present invention should be included in the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be based on the scope of protection of the claims.

Claims

1. A method for predicting the remaining driving range of a pure electric vehicle, characterized in that: include: Collecting and processing raw data of the vehicle for a preset time span, wherein the raw data is used to characterize the vehicle status and battery status, and the preset time span is greater than or equal to 1 year; extracting a plurality of pure driving kinematics segments from the raw data, wherein the vehicle only has driving motion in each of the pure driving kinematics segments; Obtaining historical prediction features affecting the vehicle's driving range based on each of the pure driving kinematic segments, and obtaining the vehicle's current charging segment and current prediction features affecting the vehicle's driving range; Screening out the pure driving kinematics segments whose similarity between the historical prediction feature and the current prediction feature is greater than or equal to the target similarity, and forming a template library; Training a joint prediction model based on the raw data, wherein the joint prediction model is used to predict the remaining battery power and the battery health status; Inputting the characteristic parameters of the current charging segment into the joint prediction model to obtain a predicted battery health state; A mileage prediction model is trained based on the combined prediction model and the original data, where the mileage prediction model is used to predict the remaining battery range; Obtaining a change in the state of charge during the charging section, and calculating actual energy consumption of the battery based on the change in the state of charge and the predicted battery health status; Acquiring a current ambient temperature, and obtaining a predicted non-driving demand energy based on the raw data and the current ambient temperature; Calculating the required energy for driving according to the actual energy consumption of the battery and the predicted non-driving required energy; Obtaining a prediction segment from the template library according to the required driving energy, wherein the prediction segment is one or more pure driving kinematic segments in the template library, and a total energy consumption generated by the prediction segment is less than or equal to the required driving energy; Obtaining operating condition information, average current, and remaining charge change of the predicted segment, wherein the operating condition information is used to characterize the type of vehicle operating condition and the time proportion of each vehicle operating condition; The operating condition information, average current and remaining charge change of the prediction segment are input into the mileage prediction model to obtain the remaining driving range of the vehicle.

2. The method for predicting the remaining driving range of a pure electric vehicle according to claim 1, characterized in that: The step of collecting and processing raw data of vehicles before a preset time span includes: Acquiring a collection time according to a first preset time interval, collecting characteristic parameters corresponding to the collection time, and forming the raw data; Determine whether the original data has duplication anomalies; According to the duplicate anomaly in the original data, retain one piece of data in the duplicate data and delete the other data in the duplicate data; Comparing the time intervals between adjacent collection times with the first preset time interval; According to the time interval of the acquisition time being less than the first preset time interval, it is determined that there is redundant acquisition time, and according to the compared acquisition time and the acquisition time adjacent to the compared acquisition time, the redundant acquisition time and the characteristic parameters corresponding to the redundant acquisition time are acquired and deleted; Based on the time interval of the acquisition time being greater than the first preset time interval, it is determined that the acquisition time is missing, the missing acquisition time is obtained based on the compared acquisition time and the first preset time interval, and the characteristic parameters corresponding to the missing acquisition time are supplemented based on the characteristic parameters corresponding to the compared acquisition time.

3. The method for predicting the remaining driving range of a pure electric vehicle according to claim 2, characterized in that: After acquiring the collection time according to the first preset time interval, collecting the characteristic parameters corresponding to the collection time, and forming the raw data, the step of collecting and processing the raw data of the vehicle before the preset time span further includes: Screen and clean abnormal feature parameters based on the box plot method and basic logic; Determine whether there is data missing for the characteristic parameter; If there is missing data in the characteristic parameter, the missing data is supplemented by interpolation, or the missing data is deleted.

4. The method for predicting the remaining driving range of a pure electric vehicle according to claim 3, characterized in that: The step of extracting a plurality of pure driving kinematic segments from the raw data comprises: dividing the raw data into a plurality of unit segments according to a complete charge cycle and a complete discharge cycle; dividing each of the unit segments into a plurality of kinematic segments according to a second preset time interval; The kinematic segments are screened for pure driving kinematic segments.

5. The method for predicting the remaining driving range of a pure electric vehicle according to claim 1, characterized in that: The step of obtaining historical prediction features affecting the vehicle's driving range based on each of the pure driving kinematic segments, and obtaining the vehicle's current charging segment and current prediction features affecting the vehicle's driving range, includes: Read the date, time and vehicle position information of each pure driving kinematic segment; Obtaining historical weather conditions, historical temperature levels, and historical congestion levels of a city where the vehicle has historically been located based on the date, the time, and the location information, and determining a historical working day status based on the date, using the historical weather conditions, the historical temperature levels, the historical congestion levels, and the historical working day status as the historical prediction features, wherein the historical working day status is used to indicate whether the vehicle is on a working day; obtaining a current weather condition, a current temperature level, a current congestion level, and a current weekday status of a city where the vehicle is currently located, and using the current weather condition, the current temperature level, the current congestion level, and the current weekday status as the current prediction features, wherein the current weekday status is used to indicate whether the vehicle is on a weekday; Get the vehicle's current charging section.

6. The method for predicting the remaining driving range of a pure electric vehicle according to claim 4, characterized in that: The step of training the joint prediction model based on the original data includes: Obtaining the average battery voltage, average battery current, instantaneous battery voltage, instantaneous battery current, battery temperature, maximum battery cell voltage, minimum battery cell voltage, remaining battery capacity, and battery health status of each kinematic segment; using the average battery voltage, the average battery current, the instantaneous battery voltage, the instantaneous battery current, and the battery temperature as first training parameters for training a prediction model for predicting the remaining battery capacity, and performing a Pearson correlation analysis on each of the first training parameters and the remaining battery capacity to obtain a correlation coefficient between the first training parameter and the remaining battery capacity; using the average battery voltage, the average battery current, the battery temperature, the maximum battery cell voltage, and the minimum battery cell voltage as second training parameters for training a prediction model for predicting a battery state of health, and performing a Pearson correlation analysis on each of the second training parameters and the battery state of health to obtain a correlation coefficient between the second training parameter and the battery state of health; A first training parameter and a second training parameter having a correlation coefficient greater than or equal to a correlation limit are screened out, and the joint prediction model is trained according to the screened first training parameter and the second training parameter.

7. The method for predicting the remaining driving range of a pure electric vehicle according to claim 6, characterized in that: The step of training the mileage prediction model based on the joint prediction model and the original data includes: Inputting the first training parameter into the joint prediction model, and obtaining a change in the remaining battery power of each kinematic segment; Obtaining a proportion of the speed below a preset value, a deceleration time proportion, and a mileage of each kinematic segment, using the proportion of the speed below the preset value, the deceleration proportion, the change in the remaining battery charge, and the average current as third training parameters for training the mileage prediction model, and performing a Pearson correlation analysis on each of the third training parameters and the mileage to obtain a correlation coefficient between the third training parameter and the mileage; A third training parameter having a correlation coefficient greater than or equal to the correlation limit is screened out, and the mileage prediction model is trained according to the screened third training parameter.

8. The method for predicting the remaining driving range of a pure electric vehicle according to claim 1, characterized in that: The step of obtaining the current ambient temperature and obtaining the predicted non-driving demand energy according to the original data and the current ambient temperature includes: Extracting a plurality of stationary segments from the raw data, wherein the vehicle is stationary in each stationary segment and generates non-driving energy consumption; Obtaining the ambient temperature and the non-driving energy consumption in each of the stationary segments, and building a non-driving energy consumption prediction model based on the ambient temperature and the non-driving energy consumption; The current ambient temperature is obtained and input into the non-driving energy consumption prediction model to obtain the predicted non-driving required energy.

9. The method for predicting the remaining driving range of a pure electric vehicle according to claim 8, characterized in that: The step of obtaining the ambient temperature and the non-driving energy consumption in each stationary segment, and building a non-driving energy consumption prediction model based on the ambient temperature and the non-driving energy consumption, includes: Acquire the ambient temperature of each of the static segments; Obtaining resting energy consumption in each stationary segment, where the resting energy consumption includes energy consumption for heating the cabin by air conditioning and energy consumption for maintaining the battery at an operating temperature; Regression fitting is performed on the resting energy consumption and the ambient temperature of the resting segment to obtain a non-driving energy consumption prediction model.

10. The method for predicting the remaining driving range of a pure electric vehicle according to claim 8, characterized in that: The multiple still segments are multiple continuous still segments.

Images