WO2023130659A1 - 商用电动车辆能耗预测方法、装置和计算机设备 - Google Patents

商用电动车辆能耗预测方法、装置和计算机设备 Download PDF

Info

Publication number
WO2023130659A1
WO2023130659A1 PCT/CN2022/096746 CN2022096746W WO2023130659A1 WO 2023130659 A1 WO2023130659 A1 WO 2023130659A1 CN 2022096746 W CN2022096746 W CN 2022096746W WO 2023130659 A1 WO2023130659 A1 WO 2023130659A1
Authority
WO
WIPO (PCT)
Prior art keywords
data
energy consumption
historical
commercial electric
consumption prediction
Prior art date
Application number
PCT/CN2022/096746
Other languages
English (en)
French (fr)
Inventor
段潘婷
赵微
Original Assignee
宁德时代新能源科技股份有限公司
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 宁德时代新能源科技股份有限公司 filed Critical 宁德时代新能源科技股份有限公司
Priority to JP2022548244A priority Critical patent/JP2024508198A/ja
Priority to KR1020227027806A priority patent/KR20230107471A/ko
Priority to EP22746939.2A priority patent/EP4235338A4/en
Priority to US17/929,889 priority patent/US20230222852A1/en
Publication of WO2023130659A1 publication Critical patent/WO2023130659A1/zh

Links

Images

Classifications

    • G—PHYSICS
    • G05—CONTROLLING; REGULATING
    • G05D—SYSTEMS FOR CONTROLLING OR REGULATING NON-ELECTRIC VARIABLES
    • G05D1/00—Control of position, course, altitude or attitude of land, water, air or space vehicles, e.g. using automatic pilots
    • G05D1/02—Control of position or course in two dimensions
    • Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
    • Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
    • Y02T—CLIMATE CHANGE MITIGATION TECHNOLOGIES RELATED TO TRANSPORTATION
    • Y02T10/00—Road transport of goods or passengers
    • Y02T10/60—Other road transportation technologies with climate change mitigation effect
    • Y02T10/70—Energy storage systems for electromobility, e.g. batteries

Definitions

  • the present application relates to the technical field of motor control, in particular to a commercial electric vehicle energy consumption prediction method, device, computer equipment, computer readable storage medium and computer program product.
  • the energy consumption prediction method of commercial electric vehicles in some cases, under fixed conditions such as NEDC (New European Driving Cycle, New European Driving Cycle) simulated vehicle conditions, based on vehicle dynamics models, combined with vehicle frontal area, mass and rolling resistance coefficient and other vehicle characteristic parameters to predict vehicle energy consumption. Since it is difficult to accurately obtain the above-mentioned vehicle characteristic parameters in the actual application process of the vehicle, the traditional energy consumption prediction method of commercial electric vehicles is limited to the simulated working conditions of the laboratory simulation environment, and has the disadvantage of poor accuracy of prediction results.
  • NEDC New European Driving Cycle, New European Driving Cycle
  • a commercial electric vehicle energy consumption prediction method, device, computer equipment, computer readable storage medium and computer program product are provided, so as to improve the accuracy of commercial electric vehicle energy consumption prediction results.
  • a commercial electric vehicle energy consumption prediction method comprising:
  • the above-mentioned commercial electric vehicle energy consumption prediction method substitutes the discharge duration data and driving position characteristic data into the energy consumption prediction model to obtain energy consumption prediction data. Since the driving route of commercial electric vehicles is relatively fixed, the driving position characteristic data corresponding to the driving route can be To a certain extent, it reflects the actual driving environment of commercial electric vehicles, which is conducive to improving the accuracy of energy consumption prediction data.
  • the discharge duration data and driving location feature data into the energy consumption prediction model before substituting the discharge duration data and driving location feature data into the energy consumption prediction model to obtain the energy consumption prediction data of the commercial electric vehicle, it also includes: according to the historical driving data of the commercial electric vehicle, the rated capacity of the battery, and the predicted The model loss function is set, and the model training is carried out based on the machine learning algorithm to obtain the energy consumption prediction model.
  • the discharge duration data and the characteristic data of the driving position are substituted into the energy consumption prediction model to obtain the energy consumption prediction data of commercial electric vehicles, and the energy consumption prediction model is obtained by performing model training based on the historical driving data of commercial electric vehicles, which is beneficial to Improving the science of energy consumption prediction methods for commercial electric vehicles.
  • the model training is performed based on the machine learning algorithm to obtain the energy consumption prediction model, including: obtaining the historical driving data of the commercial electric vehicle and battery rated capacity; historical driving data includes historical charging and discharging data and historical driving position characteristic data; based on historical charging and discharging data and battery rated capacity, calculate historical energy consumption data per unit time of commercial electric vehicles; according to historical energy consumption data per unit time and historical driving location feature data to obtain historical energy consumption data; and based on historical energy consumption data and a preset model loss function, model training is performed based on machine learning algorithms to obtain an energy consumption prediction model.
  • the historical energy consumption data per unit time of commercial electric vehicles is calculated first, and then model training is performed based on the historical energy consumption data per unit time and historical driving position characteristic data to obtain the energy consumption prediction model. It can accurately characterize the relationship between the energy consumption per unit time and the characteristics of the driving position, which is conducive to improving the matching degree between the energy consumption prediction model and the actual working conditions, thereby improving the prediction accuracy of the model.
  • obtaining the historical driving data of the commercial electric vehicle includes: obtaining the historical raw driving data of the commercial electric vehicle; and preprocessing the historical raw driving data to obtain the historical driving data of the commercial electric vehicle.
  • the historical energy consumption per unit time data of the commercial electric vehicle is calculated based on the historical charging and discharging data and the rated capacity of the battery, including: taking the set time as a cycle, and obtaining the historical charging and discharging data of the commercial electric vehicle Charging data and historical discharge data; based on historical charging data and battery rated capacity, the battery health status of commercial electric vehicles within a set time is obtained; and according to battery health status, battery rated capacity, and historical discharge data corresponding to historical charging data, get Historical unit time energy consumption data of commercial electric vehicles.
  • the real-time battery health status of the vehicle is considered, which is beneficial to improve the accuracy of the historical energy consumption per unit time data, thereby improving the prediction accuracy of the model.
  • one charging section corresponds to multiple sub-discharging sections; according to the battery health status, battery rated capacity, and historical discharge data corresponding to historical charging data, the historical energy consumption per unit time of commercial electric vehicles can be obtained
  • the data includes: According to the battery health status, the battery rated capacity, and the historical discharge data of each sub-discharge section, the historical energy consumption data per unit time of the commercial electric vehicle is obtained.
  • the final historical energy consumption per unit time can be realized by combining the historical discharge data of each sub-discharging segment Data analysis can effectively improve the analysis reliability of historical energy consumption data per unit time.
  • obtaining the historical energy consumption data per unit time of the commercial electric vehicle includes: according to The state of health of the battery, the rated capacity of the battery, the historical charging data of each charging segment and the historical discharging data of the discharging segment corresponding to each charging segment are used to obtain the historical energy consumption data per unit time of the commercial electric vehicle.
  • the energy consumption of each discharging segment can be calculated in combination with the historical discharge data of the discharging segment corresponding to each charging segment. After calculation, the analysis of historical energy consumption data per unit time is finally realized, which can further improve the analysis reliability of historical energy consumption data per unit time.
  • the historical energy consumption data per unit time of the commercial electric vehicle is obtained, including: According to the state of health of the battery, the rated capacity of the battery, the historical charging data of each charging segment, and the historical discharging data of the discharging segment corresponding to each charging segment, the energy consumption of the discharging segment corresponding to each charging segment is obtained; In energy consumption, if the maximum energy consumption value is less than the rated capacity of the battery, calculate the historical energy consumption data per unit time within the set time based on the maximum energy consumption value; Or equal to the rated capacity of the battery, and less than the rated battery capacity of the preset multiple, calculate the historical energy consumption data per unit time within the set time according to the average value of the energy consumption of the discharge segment corresponding to each charging segment; if each charging segment corresponds to In the energy consumption of the discharge section,
  • the energy consumption of the discharge section corresponding to each charging section is processed differently, and different historical energy consumption per unit time data analysis is realized in combination with the different intervals where it is located.
  • the discharge data is discarded, which effectively improves the analysis accuracy of historical energy consumption data per unit time.
  • the energy consumption prediction model after substituting the discharge duration data and driving position feature data into the energy consumption prediction model to obtain the energy consumption prediction data of commercial electric vehicles, it further includes: dividing the historical energy consumption data into a training set and a test set, based on the training According to the test set and the energy consumption prediction data corresponding to the test set, the energy consumption prediction data is corrected to obtain the corrected energy consumption prediction data.
  • the energy consumption prediction results are corrected based on the training set and the test set, which can further improve the prediction accuracy of the commercial electric vehicle energy consumption prediction method.
  • substituting the discharge duration data and the characteristic data of the driving position into the energy consumption prediction model includes: performing interpolation processing on the discharge duration data to obtain interpolated discharge duration data, and interpolating the discharge duration data, And the characteristic data of driving position are substituted into the energy consumption prediction model.
  • interpolation processing is performed first, which can ensure the continuity of the discharge duration data input into the model, and then ensure the continuity of the energy consumption prediction data, which is conducive to improving the efficiency of commercial electric vehicles. Flexibility of energy consumption forecasting methods.
  • substituting the discharge duration data and the characteristic data of the driving position into the energy consumption prediction model to obtain the energy consumption prediction data of commercial electric vehicles includes: substituting the discharge duration data and the characteristic data of the driving position into the energy consumption prediction model to obtain the commercial electric vehicle The initial energy consumption prediction data of the electric vehicle; and grouping the initial energy consumption prediction data according to the preset time, and taking the preset percentile of each group as the energy consumption prediction data corresponding to the preset time; the preset percentile The number is greater than 50%.
  • the larger preset percentile in each group is taken as the energy consumption prediction data, which can eliminate the interference of noise values and improve energy consumption.
  • the accuracy of consumption forecast data is taken as the energy consumption prediction data, which can eliminate the interference of noise values and improve energy consumption.
  • the energy consumption prediction model after substituting the discharge duration data and the characteristic data of the driving location into the energy consumption prediction model to obtain the energy consumption prediction data of the commercial electric vehicle, it further includes: obtaining the estimated value of the battery state of health of the commercial electric vehicle, and according to the battery health State estimates and energy consumption prediction data to determine breakdown risk for commercial electric vehicles.
  • the estimated value of the battery state of health of the commercial electric vehicle is also obtained, and according to the estimated value of the battery state of health and the energy consumption prediction data, the breakdown risk of the commercial electric vehicle can be determined. It is convenient for users to find abnormalities in time and take corresponding measures, which is conducive to reducing the breakdown probability of commercial electric vehicles and improving the safety of vehicles.
  • a commercial electric vehicle energy consumption prediction device comprising:
  • the discharge duration acquisition module is used to obtain the discharge duration data of commercial electric vehicles; the driving location feature acquisition module is used to obtain the driving location characteristic data of commercial electric vehicles; and the energy consumption prediction module is used to combine the discharge duration data and driving location characteristics
  • the data is substituted into the energy consumption prediction model to obtain the energy consumption prediction data of commercial electric vehicles; the energy consumption prediction model is obtained based on machine learning algorithms.
  • the energy consumption prediction device for commercial electric vehicles above substitutes the discharge duration data and the characteristic data of the driving position into the energy consumption prediction model to obtain the energy consumption prediction data. Since the driving route of the commercial electric vehicle is relatively fixed, the characteristic data of the driving position corresponding to the driving route can be To a certain extent, it reflects the actual driving environment of commercial electric vehicles, which is conducive to improving the accuracy of energy consumption prediction data.
  • the device for predicting energy consumption of commercial electric vehicles further includes: an energy consumption prediction model training module, configured to use machine learning algorithms based on historical driving data of commercial electric vehicles, battery rated capacity, and preset model loss functions Perform model training to obtain an energy consumption prediction model.
  • an energy consumption prediction model training module configured to use machine learning algorithms based on historical driving data of commercial electric vehicles, battery rated capacity, and preset model loss functions Perform model training to obtain an energy consumption prediction model.
  • the discharge duration data and the characteristic data of the driving position are substituted into the energy consumption prediction model to obtain the energy consumption prediction data of commercial electric vehicles, and the energy consumption prediction model is obtained by performing model training based on the historical driving data of commercial electric vehicles, which is beneficial to Improving the science of energy consumption prediction methods for commercial electric vehicles.
  • the energy consumption prediction model training module includes: a data acquisition unit, which acquires historical driving data and battery rated capacity of commercial electric vehicles; historical driving data includes historical charging and discharging data and historical driving position characteristic data; energy consumption data calculation The unit calculates the historical energy consumption per unit time data of commercial electric vehicles based on the historical charging and discharging data and the rated capacity of the battery; the historical energy consumption data generating unit obtains the historical energy consumption based on the historical energy consumption per unit time data and the historical driving position characteristic data data; and an energy consumption prediction model training unit, which performs model training based on a machine learning algorithm according to historical energy consumption data and a preset model loss function to obtain an energy consumption prediction model.
  • the historical energy consumption data per unit time of commercial electric vehicles is calculated first, and then model training is performed based on the historical energy consumption data per unit time and historical driving position characteristic data to obtain the energy consumption prediction model. It can accurately characterize the relationship between the energy consumption per unit time and the characteristics of the driving position, which is conducive to improving the matching degree between the energy consumption prediction model and the actual working conditions, thereby improving the prediction accuracy of the model.
  • the commercial electric vehicle energy consumption prediction device further includes: a correction module for dividing the historical energy consumption data into a training set and a test set, and obtaining an energy consumption prediction corresponding to the test set based on the training set and the energy consumption prediction model data; according to the test set and the energy consumption prediction data corresponding to the test set, the energy consumption prediction data is corrected to obtain the corrected energy consumption prediction data.
  • the energy consumption prediction results are corrected based on the training set and the test set, which can further improve the prediction accuracy of the commercial electric vehicle energy consumption prediction method.
  • the device for predicting energy consumption of commercial electric vehicles further includes: a breakdown risk determination module, configured to obtain the estimated value of the battery state of health of the commercial electric vehicle, and determine Risk of breakdown of the vehicle.
  • the estimated value of the battery state of health of the commercial electric vehicle is also obtained, and according to the estimated value of the battery state of health and the energy consumption prediction data, the breakdown risk of the commercial electric vehicle can be determined. It is convenient for users to find abnormalities in time and take corresponding measures, which is conducive to reducing the breakdown probability of commercial electric vehicles and improving the safety of vehicles.
  • a computer device comprising a memory and one or more processors, wherein computer readable instructions are stored in the memory, and when the computer readable instructions are executed by the one or more processors, the one or more processors perform the following steps:
  • One or more non-volatile computer-readable storage media storing computer-readable instructions that, when executed by one or more processors, cause the one or more processors to perform the following steps:
  • a computer program product comprising a computer program which, when executed by one or more processors, causes the one or more processors to perform the following steps:
  • the above-mentioned computer equipment, computer-readable storage medium and computer program product substitute the discharge duration data and driving position characteristic data into the energy consumption prediction model to obtain the energy consumption prediction data. Since the driving route of commercial electric vehicles is relatively fixed, the corresponding driving route The characteristic data of driving position can reflect the actual driving environment of commercial electric vehicles to a certain extent, which is conducive to improving the accuracy of energy consumption prediction data.
  • Fig. 1 is a schematic flow chart of a method for predicting energy consumption of a commercial electric vehicle in some embodiments of the present application
  • Fig. 2 is a schematic flow chart of a method for predicting energy consumption of a commercial electric vehicle in some embodiments of the present application
  • Fig. 3 is a schematic diagram of the training process of the energy consumption prediction model in some embodiments of the present application.
  • Fig. 4 is a schematic diagram of the analysis process of historical unit time energy consumption data in some embodiments of the present application.
  • FIG. 5 is a schematic flow chart of a method for predicting energy consumption of a commercial electric vehicle in some embodiments of the present application
  • FIG. 6 is a schematic flow chart of a method for predicting energy consumption of commercial electric vehicles in some embodiments of the present application.
  • Fig. 7 is a schematic flow chart of a method for predicting energy consumption of a commercial electric vehicle in some embodiments of the present application.
  • Fig. 8 is a flowchart of a method for predicting energy consumption of a commercial electric vehicle in some embodiments of the present application.
  • FIG. 9 is a schematic structural diagram of a commercial electric vehicle energy consumption prediction device in some embodiments of the present application.
  • Fig. 10 is a schematic structural diagram of a commercial electric vehicle energy consumption prediction device in some embodiments of the present application.
  • Fig. 11 is a schematic structural diagram of an energy consumption prediction model in some embodiments of the present application.
  • Fig. 12 is a schematic structural diagram of a commercial electric vehicle energy consumption prediction device in some embodiments of the present application.
  • Fig. 13 is a schematic structural diagram of an energy consumption prediction module in some embodiments of the present application.
  • Fig. 14 is a schematic structural diagram of a commercial electric vehicle energy consumption prediction device in some embodiments of the present application.
  • Fig. 15 is a schematic diagram of the internal structure of a computer device in some embodiments of the present application.
  • the commercial electric vehicle energy consumption prediction method, device, computer equipment, computer readable storage medium and computer program products provided in this application can be applied to various commercial electric vehicles, including but not limited to electric buses, electric buses, and subway electric vehicles , railway main line electric vehicles and light rail electric vehicles, etc.
  • the present application provides a method for predicting energy consumption of a commercial electric vehicle.
  • the method can be applied to a terminal or a server, and can also be implemented through the interaction between the terminal and the server.
  • the description below takes the case where the method is applied to a terminal as an example.
  • the method includes step S102 to step S106.
  • Step S102 Obtain the discharge duration data of the commercial electric vehicle.
  • the discharge duration data refers to a data set composed of the discharge duration in each calculation period of the commercial electric vehicle with the set time as the calculation period.
  • the set time can be half a day, one day or two days.
  • the discharge duration data can be extracted from the charge and discharge data of commercial electric vehicles.
  • the charging and discharging data includes the charging and discharging start time, end time, charging capacity, and SOC (State of Charge, the available percentage of the remaining charge in the battery) corresponding to each time, etc.
  • the discharge duration is the time difference between the discharge start time and the discharge end time.
  • the terminal can obtain the charging and discharging data of commercial electric vehicles, and perform data preprocessing on the charging and discharging data to obtain the discharging duration data.
  • the terminal can eliminate data rows that are repeated and incomplete in the charge and discharge data, and extract the discharge data of commercial electric vehicles in the discharge section according to the field information, and then calculate based on the discharge start time and discharge end time in the discharge data. discharge time.
  • the terminal can also eliminate abnormal values in the calculated discharge duration based on statistical rules, so as to avoid the influence of abnormal data and further improve the accuracy of energy consumption prediction. For example, the discharge duration data of ⁇ -1*sigma and >+3*sigma in multiple calculation cycles within the preset duration can be eliminated.
  • the sub-discharge durations of each sub-discharge segment are added to calculate the discharge duration within the set time.
  • Step S104 Obtain the characteristic data of the driving position of the commercial electric vehicle.
  • step S104 may be performed before, after, or synchronously with step S102.
  • the characteristic data of the driving position includes the driving position of the commercial electric vehicle at different times, and the weather data and terrain data corresponding to the driving position.
  • the weather data includes information such as air temperature, humidity, air pressure, and wind speed;
  • the terrain data includes information such as slope, track resistance, and air resistance.
  • the terminal can obtain the characteristic data of the driving position in the same period of history, or first obtain the driving route of the commercial electric vehicle within the expected time, and then according to the vehicle driving position information on the driving route,
  • the weather data and terrain data corresponding to the driving position are associated to obtain the characteristic data of the driving position corresponding to the driving route.
  • the historical weather data corresponding to the driving route and the terrain data corresponding to the driving route are associated to obtain the characteristic data of the driving position corresponding to the driving route.
  • the specific way for the terminal to acquire the discharge duration data and driving location feature data of commercial electric vehicles may be active acquisition or passive reception.
  • Step S106 Substituting the discharge duration data and the characteristic data of the driving position into the energy consumption prediction model to obtain the energy consumption prediction data of the commercial electric vehicle.
  • the energy consumption prediction model is obtained based on machine learning algorithms.
  • the machine learning algorithm may be a neural network algorithm or a decision tree algorithm.
  • the machine learning algorithm is a GBRT (Gradient Boost Regression Tree, progressive gradient regression tree) algorithm.
  • the energy consumption prediction data of commercial electric vehicles can be obtained by substituting the discharge duration data and driving position characteristic data into the energy consumption prediction model based on the machine learning algorithm.
  • the ambient temperature data of the same period in history and the recent discharge duration data can be used as the model self- Variables are substituted into the energy consumption prediction model to improve the accuracy of energy consumption prediction. For example, when it is necessary to predict the energy consumption in the next three months, the discharge duration of the last three months and the ambient temperature in the same period of history are taken as input quantities and substituted into the energy consumption prediction model.
  • the terminal may also output the energy consumption prediction data.
  • the output object of the energy consumption prediction data may be a storage device, a display device or a communication device.
  • the terminal can also output the energy consumption prediction data to other terminals through the communication device.
  • the above-mentioned commercial electric vehicle energy consumption prediction method substitutes the discharge duration data and driving position characteristic data into the energy consumption prediction model to obtain energy consumption prediction data.
  • the driving position corresponding to the driving route Feature data can reflect the actual driving environment of commercial electric vehicles to a certain extent, which is conducive to improving the accuracy of energy consumption prediction data; Advanced mathematical modeling improves work efficiency on the basis of ensuring accuracy.
  • the data in order to obtain the continuous energy consumption prediction data with the set duration as the cycle, can be interpolated before substituting the discharge duration data into the energy consumption prediction model processing to ensure that the discharge duration data input into the model is continuous data with a set duration as the cycle, thereby ensuring the continuity of the energy consumption prediction data; it is also possible to calculate the initial energy consumption prediction data after obtaining the discontinuous initial energy consumption prediction data Interpolation processing is performed to obtain the energy consumption prediction data of continuous commercial electric vehicles with a set period as the period.
  • the method before step S106, further includes step S105: according to the historical driving data of the commercial electric vehicle, the rated capacity of the battery, and the preset model loss function, the model is performed based on a machine learning algorithm. Train the energy consumption prediction model.
  • step S105 may be performed before, after, or synchronously with step S102 and step S104.
  • Historical driving data refers to the actual data of commercial electric vehicles during historical operation.
  • the terminal obtains the actual energy consumption data for model training based on the historical driving data and battery rated capacity of commercial electric vehicles, and uses the preset model loss function to conduct model training based on machine learning algorithms to obtain the energy consumption prediction model .
  • model loss function may specifically be a mean square error function, an absolute loss function, or a quantile loss function.
  • the specific formula of the model loss function can be:
  • y is the actual value
  • f(x) is the predicted value
  • the discharge duration data and the characteristic data of the driving position are substituted into the energy consumption prediction model to obtain the energy consumption prediction data of commercial electric vehicles, and the energy consumption prediction model is obtained by performing model training based on the historical driving data of commercial electric vehicles, which is beneficial to Improving the science of energy consumption prediction methods for commercial electric vehicles.
  • step S105 includes step S302 to step S308.
  • Step S302 Obtain historical driving data and battery rated capacity of the commercial electric vehicle.
  • the historical driving data includes historical charging and discharging data and historical driving position characteristic data. Further, the historical charging and discharging data includes charging and discharging start time, end time, charging capacity, SOC corresponding to each time, and the like.
  • the rated capacity of the battery refers to the capacity of the battery under rated conditions.
  • the historical driving position characteristic data is the driving position characteristic data corresponding to the time node where the historical charging and discharging data is located.
  • Step S304 Based on the historical charging and discharging data and the rated capacity of the battery, calculate the historical energy consumption data per unit time of the commercial electric vehicle.
  • the historical energy consumption data per unit time of commercial electric vehicles refers to a data set composed of energy consumption values per unit time of commercial electric vehicles with a set time period in the past period of time.
  • the terminal can use the set time as the calculation cycle to obtain the historical energy consumption and discharge duration of commercial electric vehicles in each cycle, and then obtain the historical energy consumption data per unit time of commercial electric vehicles.
  • the terminal obtains the historical energy consumption data per unit time of the commercial electric vehicle based on the historical charging and discharging data and the rated capacity of the battery.
  • the terminal multiplies the SOC difference between the start and end of the discharge by the rated capacity of the battery to obtain the historical discharge capacity of the commercial electric vehicle within the set time, combined with the corresponding discharge duration , to obtain the historical unit time energy consumption of commercial electric vehicles within a preset period of time.
  • the preset duration may be one month, two months or three months, which may be specifically determined according to energy consumption prediction requirements.
  • Step S306 Obtain historical energy consumption data according to the historical energy consumption data per unit time and the characteristic data of the driving position.
  • the historical energy consumption data refers to historical data including historical energy consumption data per unit time and historical driving position characteristic data.
  • the historical unit time energy consumption data of multiple calculation cycles obtained in step S404 can be combined with the historical driving position feature data of the same commercial electric vehicle according to the time node where the calculation cycle is located, to form a Historical energy consumption data of time, historical unit time energy consumption data and historical driving position feature data.
  • Step S308 According to the historical energy consumption data and the preset model loss function, perform model training based on a machine learning algorithm to obtain an energy consumption prediction model.
  • the energy consumption prediction model can be obtained by using the preset model loss function and machine learning algorithm for model training.
  • the historical energy consumption data per unit time of commercial electric vehicles is calculated first, and then model training is performed based on the historical energy consumption data per unit time and historical driving position characteristic data to obtain the energy consumption prediction model. It can accurately characterize the relationship between the energy consumption per unit time and the characteristics of the driving position, which is conducive to improving the matching degree between the energy consumption prediction model and the actual working conditions, thereby improving the prediction accuracy of the model.
  • obtaining the historical driving data of the commercial electric vehicle includes: obtaining the historical raw driving data of the commercial electric vehicle; and preprocessing the historical raw driving data to obtain the historical driving data of the commercial electric vehicle.
  • the terminal can acquire historical raw driving data of commercial electric vehicles, and perform data preprocessing on the basis of historical raw driving data to obtain historical driving data corresponding to commercial electric vehicles.
  • the historical original driving data includes historical charge and discharge data, historical vehicle location information data, and weather data and terrain data corresponding to each vehicle location.
  • the terminal can preprocess the historical charge and discharge data, eliminate repeated and incomplete data lines, and extract the charging data of each charging segment of commercial electric vehicles and the discharge of each charging segment corresponding to each charging segment according to the field information.
  • data row with incomplete data information includes the data row with empty date, and the data row with missing data in the charging segment or discharging segment.
  • step S304 includes steps S402 to S406.
  • Step S402 Taking the set time as the calculation cycle, and based on the historical charging and discharging data, obtain the historical charging data and historical discharging data of the commercial electric vehicle.
  • the charging data includes charging start time, charging end time, charging capacity, SOC corresponding to each time, and the like.
  • the discharge data includes discharge start time, discharge end time, SOC corresponding to each time, and the like.
  • the historical charging data and historical discharging data refer to actual charging data and actual discharging data in a period of time in the past respectively.
  • the terminal extracts and classifies the information in the historical charging and discharging data: classifies the data containing the information in the "charging" field as charging data, and classifies the data containing the information in the "discharging” field as discharging data . Then, the secondary classification is performed with the set time as the cycle, and the historical charging data and historical discharging data of commercial electric vehicles in multiple cycles can be obtained.
  • Step S404 Obtain the state of health of the battery of the commercial electric vehicle within a set time based on the historical charging data and the rated capacity of the battery.
  • the state of health of the battery is also called SOH (State of Health), which is used to characterize the ability of the current battery to store electric energy relative to the new battery, and quantitatively describe the performance state of the current battery in the form of a percentage.
  • SOH State of Health
  • the state of health of the battery is affected by various factors such as temperature, current rate, and cut-off voltage, and changes dynamically from the beginning of the battery life to the end of its life. Based on this, in order to improve the accuracy of historical unit time energy consumption data, the terminal calculates the battery health status of commercial electric vehicles within a set time based on historical charging data and battery rated capacity.
  • the formula for calculating the state of health of the battery is:
  • Q is the charging capacity
  • pack cap is the rated capacity of the battery
  • It is the SOC difference between the charging start time and the charging end time.
  • Step S406 According to the state of health of the battery, the rated capacity of the battery, and the historical discharge data corresponding to the historical charging data, the historical energy consumption data per unit time of the commercial electric vehicle is obtained.
  • the discharge data corresponding to the charging data refers to the discharge data between the charging end time of the current charging segment and the charging start time of the next charging segment.
  • the terminal can obtain the historical energy consumption of commercial electric vehicles within the set time according to the battery health status, battery rated capacity, and the SOC difference between the discharge start time and the discharge end time, and then integrate multiple Calculate the historical energy consumption and discharge duration of the cycle, and obtain the energy consumption value per unit time of each calculation cycle of the commercial electric vehicle within the preset duration.
  • a data set consisting of energy consumption values per unit time of multiple computing cycles over a period of time is the historical energy consumption data per unit time.
  • delt_soc_discharge is the SOC difference between the discharge start time and the discharge end time.
  • hour_discharge is the discharge duration, that is, the time difference between the discharge start time and the discharge end time.
  • the real-time battery health status of the vehicle is considered, which is beneficial to improve the accuracy of the historical energy consumption per unit time data, thereby improving the prediction accuracy of the model.
  • Step S406 includes: obtaining the historical data of commercial electric vehicles according to the battery health status, battery rated capacity, and historical discharge data corresponding to historical charging data.
  • the energy consumption data per unit time includes: according to the battery health status, battery rated capacity, and historical discharge data of each sub-discharge section, the historical energy consumption data per unit time of commercial electric vehicles is obtained.
  • the historical energy consumption data per unit time is obtained by combining the discharge data of each sub-discharging segment.
  • the above energy consumption calculation formula can be used to calculate the corresponding energy consumption of each discharge segment, and the energy consumption of each discharge segment can be added to obtain these discharge segments The total energy consumption and the total discharge time, and divide the total energy consumption by the total discharge time to obtain the energy consumption per unit time; it can also be calculated based on the discharge data of each discharge section to obtain the sub-unit time energy of each sub-discharge section Then take the average or median value of the energy consumption per unit time of each sub-unit as the energy consumption per unit time of the discharge segment corresponding to the charging segment.
  • the final historical energy consumption per unit time can be realized by combining the historical discharge data of each sub-discharging segment Data analysis can effectively improve the analysis reliability of historical energy consumption data per unit time.
  • step S406 includes: according to the battery health status, the battery rated capacity, the historical charging data of each charging segment, and the historical discharging data of the discharging segment corresponding to each charging segment, Obtain the historical unit time energy consumption data of commercial electric vehicles.
  • the energy consumption of the discharge section corresponding to each charging section is calculated by using the formula. Then, according to the energy consumption of the discharge section corresponding to each charging section, differential processing is performed to obtain historical energy consumption data per unit time.
  • the energy consumption of each discharging segment can be calculated in combination with the historical discharge data of the discharging segment corresponding to each charging segment. After calculation, the analysis of historical energy consumption data per unit time is finally realized, which can further improve the analysis reliability of historical energy consumption data per unit time.
  • the historical energy consumption data per unit time of the commercial electric vehicle is obtained, Including: According to the battery health status, battery rated capacity, historical charging data of each charging segment, and historical discharging data of each charging segment corresponding to the discharging segment, the energy consumption of each charging segment corresponding to the discharging segment is obtained; In the energy consumption of the discharge section, if the maximum energy consumption value is less than the rated capacity of the battery, calculate the historical energy consumption data per unit time within the set time according to the maximum energy consumption value; If the consumption value is greater than or equal to the rated capacity of the battery and less than the rated capacity of the preset multiple, the historical energy consumption per unit time data within the set time is calculated according to the average value of the energy consumption of the discharge segment corresponding to each charging segment; if each In the energy consumption of the discharge section corresponding
  • the preset multiple is a real number greater than 1, such as 1.1, 1.2 or 1.3.
  • the calculation method of the energy consumption of the discharge section corresponding to each charging section is implemented in combination with the above energy consumption calculation formula, and will not be repeated here.
  • abnormal values of energy consumption and discharge duration can be processed based on statistical rules to avoid the influence of abnormal data and further improve the accuracy of energy consumption prediction. For example, ⁇ 3*sigma or other rules can be used to remove outliers in historical energy consumption; ⁇ 1*sigma or other rules can be used to remove outliers in unit time energy consumption values corresponding to historical energy consumption; 1*sigma and >+3*sigma discharge duration data.
  • the energy consumption of the discharge section corresponding to each charging section is processed differently, and different historical energy consumption per unit time data analysis is realized in combination with the different intervals where it is located.
  • the discharge data is discarded, which effectively improves the analysis accuracy of historical energy consumption data per unit time.
  • step S107 is also included: dividing the historical energy consumption data into a training set and a test set, and obtaining energy consumption prediction data corresponding to the test set based on the training set and the energy consumption prediction model , and according to the test set and the energy consumption prediction data corresponding to the test set, the energy consumption prediction data is corrected to obtain the corrected energy consumption prediction data.
  • step S107 is executed before step S108.
  • the historical energy consumption data can be divided into two parts, a training set and a test set, according to a preset ratio, and the training set is substituted into the energy consumption prediction model to obtain energy consumption prediction data corresponding to the test set. Then compare the actual energy consumption data in the test set with the corresponding energy consumption prediction data to determine the model prediction error sequence, and then determine the correction value of the energy consumption prediction model according to the error sequence. Further, there is no unique way for the terminal to determine the correction value of the energy consumption prediction model according to the error sequence. For example, the median or average value in the error sequence may be used as the correction value of the energy consumption prediction model.
  • error i is the error value corresponding to the i-th data in the test set
  • Ca i is the actual value of the i-th unit time energy consumption data in the test set
  • Cp1 i is the predicted energy consumption value corresponding to Ca i .
  • the energy consumption prediction data is corrected based on the correction value, and the obtained initial energy consumption prediction value is superimposed on the correction value of the energy consumption prediction model to obtain the revised energy consumption prediction value. That is, the predicted value of the final energy consumption corresponding to formula (5) is:
  • error is the correction value of the energy consumption prediction model
  • Cpredict_value i is the ith initial energy consumption prediction value
  • Cpredict_final i is the corrected energy consumption prediction value corresponding to Cpredict_value i .
  • the data set formed by combining the corrected energy consumption prediction values is the energy consumption prediction data of commercial electric vehicles.
  • the value of the preset ratio is not unique. For example, according to the preset ratio of 4:1, 80% of the actual energy consumption data can be taken as the training set, and the remaining 20% can be used as the test set; According to the preset ratio of 3:2, 60% of the actual energy consumption data can be used as a training set, and the remaining 40% can be used as a test set.
  • the energy consumption prediction results are corrected based on the training set and the test set, which can further improve the prediction accuracy of the commercial electric vehicle energy consumption prediction method.
  • substituting the discharge duration data and the characteristic data of the driving position into the energy consumption prediction model includes: performing interpolation processing on the discharge duration data to obtain interpolated discharge duration data, and interpolating the discharge duration data, And the characteristic data of driving position are substituted into the energy consumption prediction model.
  • interpolation refers to interpolating a continuous function on the basis of discrete data, so that this continuous curve passes through all given discrete data points.
  • the approximate value of the function at other points can be estimated through the value status of the function at a limited number of points.
  • the discharge duration data acquired in step S102 may be It is non-continuous data, and the calculation value of one or more periods within the preset time period is missing. Based on this, the discharge duration data is first interpolated to obtain the continuous discharge duration data with the set duration as the cycle after difference processing, and then the continuous discharge duration data and driving position characteristic data are substituted into the energy consumption prediction model.
  • interpolation processing is performed first, which can ensure the continuity of the discharge duration data input into the model, and then ensure the continuity of the energy consumption prediction data, which is conducive to improving the efficiency of commercial electric vehicles. Flexibility of energy consumption forecasting methods.
  • step S106 includes step S602 and step S604.
  • Step S602 Substituting the discharge duration data and the driving position feature data into the energy consumption prediction model to obtain the initial energy consumption prediction data of the commercial electric vehicle.
  • the initial energy consumption prediction data is the predicted value of the energy consumption prediction model.
  • the initial energy consumption prediction data of commercial electric vehicles can be obtained by substituting the discharge duration data and driving position characteristic data into the energy consumption prediction model.
  • Step S604 Group the initial energy consumption prediction data into groups according to the preset time, and take the preset percentile of each group as the energy consumption prediction data corresponding to the preset time.
  • the discharge duration data is the actual data of multiple calculation cycles in the past preset duration.
  • the initial energy consumption prediction data is the prediction data of multiple calculation cycles corresponding to the discharge duration data. That is, the initial energy consumption prediction data includes prediction data of multiple computing cycles within a preset time period in the future. Further, sort a set of data from small to large, and calculate the corresponding cumulative percentile, then the value of the data corresponding to a certain percentile is called the percentile of this percentile. It is not difficult to understand that the initial energy consumption prediction value corresponding to the preset percentile greater than 50% is the larger value in a group of initial energy consumption prediction data.
  • the specific value of the preset percentile is not unique, for example, it may be 85% percentile, 90% percentile or 95% percentile.
  • the terminal groups the initial energy consumption data according to the preset time to obtain multiple groups of forecast data, and takes the larger preset quantile of each group of data as the energy consumption forecast data of the group, and then obtains the entire preset duration Energy consumption prediction data for domestic commercial electric vehicles.
  • the preset time span is the largest, and the time span of the calculation cycle is the smallest.
  • the value of the preset time is not unique, and can be flexibly set according to the preset duration and calculation cycle. For example, when the preset time is three months and the calculation period is one day, the preset time period can be set to 10 days or 15 days.
  • the larger preset percentile in each group is taken as the energy consumption prediction data, which can eliminate the interference of noise values and improve energy consumption.
  • the accuracy of consumption forecast data is taken as the energy consumption prediction data, which can eliminate the interference of noise values and improve energy consumption.
  • the method further includes step S108: acquiring the estimated value of the state of health of the battery of the commercial electric vehicle, and determining Risk of breakdown of the vehicle.
  • the battery state of health also known as SOH
  • SOH battery state of health
  • the state of health of the battery is affected by various factors such as temperature, current rate, and cut-off voltage, and changes dynamically from the beginning of the battery life to the end of its life.
  • the terminal obtains the estimated value of the battery state of health of the commercial electric vehicle, and multiplies the estimated value of the state of health of the battery by the rated capacity of the battery to obtain the battery power in the calculation period, and then combines the energy consumption prediction data in the same calculation period to judge Whether the commercial electric vehicle has a risk of breaking down: If the battery power corresponding to the estimated value of the battery state of health predicted in the calculation cycle of the set number of consecutive times is less than the predicted value of energy consumption in the corresponding calculation cycle, it is considered that the vehicle has a high risk of breakdown.
  • the specific numerical value of the set times can be 4, 5 or 6.
  • a warning message can be output to remind the user to take corresponding risk prevention measures, for example, remind the user to strengthen the monitoring frequency of the vehicle, or remind the user to prepare spare parts.
  • the estimated value of the battery state of health of the commercial electric vehicle is also obtained, and according to the estimated value of the battery state of health and the energy consumption prediction data, the breakdown risk of the commercial electric vehicle can be determined. It is convenient for users to find abnormalities in time and take corresponding measures, which is conducive to reducing the breakdown probability of commercial electric vehicles and improving the safety of vehicles.
  • the terminal obtains historical driving data and battery rated capacity of the commercial electric vehicle
  • the historical driving data includes historical charging and discharging data and historical driving characteristic data.
  • the driving characteristic data includes the city where the commercial electric vehicle is located at different times, and the temperature corresponding to the city.
  • first perform data preprocessing eliminate duplicate and empty data rows, and extract the data segment information of each vehicle's charge and discharge segment, extract the start SOC and end SOC of each charge segment, and charge into Capacity Q, to extract information such as the start SOC and end SOC of each discharge segment, and the duration of the discharge segment.
  • merge the data of the charge and discharge segments assign each charge segment to at least one discharge segment, and discard the data with only the charge segment or only the discharge segment.
  • the discharge segment corresponding to the charging segment refers to the discharge segment corresponding to the discharge data between the charging end time of the current charging segment and the charging start time of the next charging segment.
  • the historical energy consumption data per unit time of the vehicle is obtained with one day as the calculation cycle.
  • the battery state of health SOH of the commercial electric vehicle within the calculation period is obtained, and then the calculated SOH, the battery rated capacity, and the The discharge data corresponding to the charging data is substituted into formulas (3) and (4) to obtain the energy consumption C of the vehicle in the corresponding calculation period and the energy consumption value per unit time Cper.
  • the energy consumption value per unit time is calculated in combination with the discharge data of each sub-discharging segment.
  • the energy consumption of each discharge section can be added to obtain the total energy consumption and the total discharge duration of these discharge sections, and then the total energy consumption can be divided by the total discharge duration to obtain the energy consumption value per unit time; it can also be based on each
  • the discharge data of the discharge section is calculated separately to obtain the energy consumption per unit time of each sub-discharge section, and then the average or median of the energy consumption per unit time of each sub-discharge section is taken as the energy consumption per unit time of the discharge section corresponding to the charging section value.
  • the energy consumption of the discharging section corresponding to each charging section is calculated respectively according to the charging data of each charging section and the discharging data of the discharging section corresponding to the charging section. Then according to the energy consumption of the discharge section corresponding to each charging section, differentiate the processing: if the maximum energy consumption value of the energy consumption of the discharge section corresponding to each charging section is less than the rated capacity of the battery, then calculate the set time based on the maximum energy consumption value The energy consumption value per unit time within; if the maximum energy consumption value in the energy consumption of the discharge section corresponding to each charging section is greater than or equal to the rated capacity of the battery and less than 1.2 times the rated capacity of the battery, then based on the corresponding discharge section of each charging section Calculate the energy consumption value per unit time within the set time; if the maximum energy consumption value in the energy consumption of the discharge section corresponding to each charging section is greater than or equal to 1.2 times the rated capacity of the battery, the energy consumption is considered Abnormal, discard
  • abnormal values of energy consumption and discharge duration can also be processed based on statistical rules, so as to avoid the influence of abnormal data and further improve the accuracy of energy consumption prediction.
  • ⁇ 3*sigma or other rules can be used to remove outliers in historical energy consumption
  • ⁇ 1*sigma or other rules can be used to remove outliers in historical unit time energy consumption values corresponding to historical energy consumption
  • -1*sigma and >+3*sigma historical discharge duration data can be used.
  • the historical energy consumption per unit time data of commercial electric vehicles within the preset time period can be obtained.
  • the model loss function is formula (1).
  • linear interpolation is performed on the historical discharge duration data to obtain the continuous discharge within the latest preset duration.
  • Duration data, and the continuous discharge duration data, and the ambient temperature data of the same period in history are used to predict the future energy consumption of the vehicle.
  • the continuous discharge duration data and the ambient temperature data of the same period in history are substituted into the training to obtain the GBRT model, and energy consumption prediction data for the same period in the future can be obtained.
  • the trained GBRT model can be used to represent the relationship between the energy consumption value per unit time and the ambient temperature. Based on this, the corresponding energy consumption data per unit time can be obtained according to the ambient temperature data in the same period of history, and combined with the latest preset The energy consumption prediction data for the same period in the future can be obtained from the continuous discharge duration data within the duration.
  • the energy consumption value C(t) for the next three months can be obtained;
  • the energy consumption forecast data is grouped according to the preset time, and the preset 95% quantile of each group is taken as the energy consumption forecast data corresponding to the preset time. For example, for the forecast data of energy consumption in the next three months, it can be grouped by 15 days, and the 95% quantile of each group is taken as the forecast value of energy consumption of the group.
  • the historical energy consumption data is divided into two parts: training set and test set, and the training set is substituted into the energy consumption prediction model to obtain the corresponding energy consumption of the test set forecast data. Then compare the actual energy consumption data in the test set with the corresponding energy consumption prediction data, and determine the model prediction error sequence according to formula (5). Finally, the median in the error sequence is determined as the correction value of the energy consumption prediction model, and the energy consumption prediction value is corrected based on formula (6) to obtain the corrected final energy consumption prediction data.
  • the terminal can also obtain the estimated value of the battery state of health of the commercial electric vehicle, and multiply the estimated value of the state of health of the battery by the rated capacity of the battery to obtain the battery power in the calculation cycle, and then Combined with the energy consumption prediction data in the same calculation cycle, it is judged whether there is a risk of breakdown of the commercial electric vehicle: if the battery power corresponding to the estimated value of the battery health state predicted in five consecutive calculation cycles is less than the energy consumption in the corresponding calculation cycle predicted value, it is considered that the car has a high risk of breaking down. Further, after the breakdown risk of the commercial electric vehicle is determined, a warning message can also be output to remind the user to take corresponding risk prevention measures, such as reminding the user to strengthen the monitoring frequency of the vehicle, or reminding the user to prepare spare parts.
  • the above-mentioned commercial electric vehicle energy consumption prediction method substitutes the discharge duration data and driving position characteristic data into the energy consumption prediction model to obtain energy consumption prediction data.
  • the driving position corresponding to the driving route Feature data can reflect the actual driving environment of commercial electric vehicles to a certain extent, which is conducive to improving the accuracy of energy consumption prediction data;
  • the energy consumption prediction model based on machine learning algorithm is used for energy consumption prediction, and the error term is introduced , can save the cumbersome mathematical modeling, and improve work efficiency on the basis of ensuring accuracy.
  • the battery health state estimation value of commercial electric vehicles is also obtained, and according to the battery health state estimation value and energy consumption prediction data, the breakdown risk of commercial electric vehicles is determined, which can facilitate users to timely Finding abnormalities and taking corresponding measures will help reduce the breakdown probability of commercial electric vehicles and improve the safety of vehicles.
  • steps in the flow charts involved in the above embodiments are shown sequentially according to the arrows, these steps are not necessarily executed sequentially 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 some of the steps in the flow charts involved in the above-mentioned embodiments may include multiple steps or stages, and these steps or stages are not necessarily executed at the same time, but may be performed at different times For execution, the execution order of these steps or stages is not necessarily performed sequentially, but may be executed in turn or alternately with other steps or at least a part of steps or stages in other steps.
  • the embodiment of the present application also provides a commercial electric vehicle energy consumption prediction device for implementing the above-mentioned commercial electric vehicle energy consumption prediction method.
  • the solution to the problem provided by the device is similar to the implementation described in the above method, so the specific limitations in one or more embodiments of the commercial electric vehicle energy consumption prediction device provided below can be referred to above for commercial The limitations of the method for predicting the energy consumption of electric vehicles will not be repeated here.
  • a commercial electric vehicle energy consumption prediction device 900 including a discharge duration acquisition module 902, a driving location characteristic acquisition module 904, and an energy consumption prediction 906, wherein:
  • the discharge duration acquisition module 902 is used to obtain the discharge duration data of the commercial electric vehicle;
  • the driving position feature acquisition module 904 is used to obtain the travel position characteristic data of the commercial electric vehicle;
  • the energy consumption prediction module 906 is used to combine the discharge duration data with the driving position
  • the location feature data is substituted into the energy consumption prediction model to obtain the energy consumption prediction data of commercial electric vehicles;
  • the energy consumption prediction model is obtained based on machine learning algorithms.
  • the commercial electric vehicle energy consumption prediction device 700 also includes: an energy consumption prediction model training module 905, which is used to The loss function is based on the machine learning algorithm for model training to obtain the energy consumption prediction model.
  • the energy consumption prediction model training module 905 includes: a data acquisition unit 112 for acquiring historical driving data and battery rated capacity of commercial electric vehicles; historical driving data includes historical charging and discharging data and historical The characteristic data of the driving position; the energy consumption data calculation unit 114, which is used to calculate the historical energy consumption data per unit time of the commercial electric vehicle based on the historical charging and discharging data and the rated capacity of the battery; the historical energy consumption data generation unit 116, which is used to Time energy consumption data and historical driving location feature data to obtain historical energy consumption data; energy consumption prediction model training unit 118, used to perform model training based on machine learning algorithms to obtain energy consumption based on historical energy consumption data and a preset model loss function predictive model.
  • the energy consumption data calculation unit 114 is specifically used to: take the set time as a period, based on historical charging and discharging data, to obtain historical charging data and historical discharging data of commercial electric vehicles; based on historical charging data and battery rated capacity , to obtain the battery health status of the commercial electric vehicle within the set time; according to the battery health status, battery rated capacity, and historical discharge data corresponding to the historical charging data, the historical energy consumption data per unit time of the commercial electric vehicle is obtained.
  • the commercial electric vehicle energy consumption prediction device 700 also includes a correction module 122, which is used to divide the historical energy consumption data into a training set and a test set, and obtain a test based on the training set and the energy consumption prediction model.
  • the energy consumption forecast data corresponding to the test set; according to the test set and the energy consumption forecast data corresponding to the test set, the energy consumption forecast data is corrected to obtain the corrected energy consumption forecast data.
  • the energy consumption prediction module 906 is specifically configured to: perform interpolation processing on the discharge duration data to obtain the interpolated discharge duration data, and substitute the interpolated discharge duration data and the characteristic data of the driving position into the energy consumption forecasting model.
  • the energy consumption prediction module 906 includes: an initial energy consumption prediction data generation unit 132, which is used to substitute the discharge duration data and the driving position characteristic data into the energy consumption prediction model to obtain the initial energy consumption prediction model of the commercial electric vehicle.
  • Energy consumption prediction data the energy consumption prediction data generation unit 134 is used to group the initial energy consumption prediction data according to the preset time, and take the preset percentile of each group as the energy consumption prediction data within the corresponding preset time; The preset percentile is greater than 50%.
  • the commercial electric vehicle energy consumption prediction device 700 further includes: a breakdown risk determination module 142, configured to obtain the estimated value of the battery state of health of the commercial electric vehicle, and according to the estimated value of the battery state of health and the energy Using consumption forecast data to determine the breakdown risk of commercial electric vehicles.
  • a breakdown risk determination module 142 configured to obtain the estimated value of the battery state of health of the commercial electric vehicle, and according to the estimated value of the battery state of health and the energy Using consumption forecast data to determine the breakdown risk of commercial electric vehicles.
  • Each module in the above-mentioned commercial electric vehicle energy consumption prediction device can be fully or partially realized by software, hardware and a combination thereof.
  • the above-mentioned modules can be embedded in or independent of the processor in the computer device in the form of hardware, and can also be stored in the memory of the computer device in the form of software, so that the processor can invoke and execute the corresponding operations of the above-mentioned modules.
  • a computer device is provided.
  • the computer device may be a terminal, and its internal structure may be as shown in FIG. 15 .
  • the computer device includes a processor, a memory, a communication interface, a display screen and an input device connected through a system bus. Wherein, the processor of the computer device is used to provide calculation and control capabilities.
  • the memory of the computer device includes a non-volatile storage medium and an internal memory.
  • the non-volatile storage medium stores an operating system and computer programs.
  • the internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage medium.
  • the communication interface of the computer device is used to communicate with an external terminal in a wired or wireless manner, and the wireless manner can be realized through WIFI, mobile cellular network, NFC (Near Field Communication) or other technologies.
  • WIFI Wireless Fidelity
  • NFC Near Field Communication
  • the computer program is executed by the processor, a commercial electric vehicle energy consumption prediction method is realized.
  • the display screen of the computer device may be a liquid crystal display screen or an electronic ink display screen
  • the input device of the computer device may be a touch layer covered on the display screen, or a button, a trackball or a touch pad provided on the casing of the computer device , and can also be an external keyboard, touchpad, or mouse.
  • Figure 15 is only a block diagram of a partial structure related to the solution of this application, and does not constitute a limitation on the computer equipment on which the solution of this application is applied.
  • the specific computer equipment can be More or fewer components than shown in the figures may be included, or some components may be combined, or have a different arrangement of components.
  • a computer device including a memory and a processor, where a computer program is stored in the memory, and the processor implements the steps in the foregoing method embodiments when executing the computer program.
  • a computer-readable storage medium on which a computer program is stored, and when the computer program is executed by a processor, the steps in the foregoing method embodiments are implemented.
  • a computer program product including a computer program, and when the computer program is executed by a processor, the steps in the foregoing method embodiments are implemented.
  • any reference to storage, database or other media used in the various embodiments provided in the present application may include at least one of non-volatile and volatile storage.
  • Non-volatile memory can include read-only memory (Read-Only Memory, ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive variable memory (ReRAM), magnetic variable memory (Magnetoresistive Random Access Memory, MRAM), Ferroelectric Random Access Memory (FRAM), Phase Change Memory (Phase Change Memory, PCM), graphene memory, etc.
  • the volatile memory may include random access memory (Random Access Memory, RAM) or external cache memory, etc.
  • RAM Random Access Memory
  • RAM Random Access Memory
  • RAM can take many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM).
  • the databases involved in the various embodiments provided in this application may include at least one of a relational database and a non-relational database.
  • the non-relational database may include a blockchain-based distributed database, etc., but is not limited thereto.
  • the processors involved in the various embodiments provided by this application can be general-purpose processors, central processing units, graphics processors, digital signal processors, programmable logic devices, data processing logic devices based on quantum computing, etc., and are not limited to this.

Landscapes

  • Engineering & Computer Science (AREA)
  • Aviation & Aerospace Engineering (AREA)
  • Radar, Positioning & Navigation (AREA)
  • Remote Sensing (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Automation & Control Theory (AREA)
  • Electric Propulsion And Braking For Vehicles (AREA)

Abstract

本申请涉及一种商用电动车辆能耗预测方法、装置、计算机设备、计算机可读存储介质和计算机程序产品。该方法包括：获取商用电动车辆的放电时长数据；获取商用电动车辆的行驶位置特征数据；将放电时长数据和行驶位置特征数据代入能耗预测模型，得到商用电动车辆的能耗预测数据；该能耗预测模型基于机器学习算法得到。

Description

商用电动车辆能耗预测方法、装置和计算机设备
相关申请的交叉引用
本申请引用于2022年1月7日递交的名称为“商用电动车辆能耗预测方法、装置和计算机设备”的第202210016220.8号中国专利申请，其通过引用被全部并入本申请。
技术领域
本申请涉及电机控制技术领域，特别是涉及一种商用电动车辆能耗预测方法、装置、计算机设备、计算机可读存储介质和计算机程序产品。
背景技术
随着能源和环境问题的加剧和车用电池技术的发展，商用电动车辆得到了越来越广泛的应用。商用电动车辆的能耗水平直接反映了商用电动车辆的整体性能水平，并影响商用电动车辆的续驶里程、质保评估、经济效益等关键指标，因此，有必要对商用电动车辆的能耗进行预测。
一些情形下的商用电动车辆能耗预测方法，在固定工况如NEDC(New European Driving Cycle，新欧洲驾驶周期)模拟车辆工况下，基于车辆动力学模型，结合车辆迎风面积、质量和滚动阻力系数等车辆特征参数，进行车辆能耗预测。由于车辆实际应用过程中，很难精确获取上述车辆特征参数，因此，传统的商用电动车辆能耗预测方法，局限于实验室仿真环境的模拟工况，存在预测结果准确性差的缺点。
申请内容
根据本申请公开的各种实施例，提供一种商用电动车辆能耗预测方法、装置、计算机设备、计算机可读存储介质和计算机程序产品，以提高商用电动车辆能耗预测结果的准确性。
一种商用电动车辆能耗预测方法，包括：
获取商用电动车辆的放电时长数据；获取商用电动车辆的行驶位置特征数据；及将放电时长数据和行驶位置特征数据代入能耗预测模型，得到商用电动车辆的能耗预测数据；能耗预测模型基于机器学习算法得到。
上述商用电动车辆能耗预测方法，将放电时长数据和行驶位置特征数据代入能耗预测模型，得到能耗预测数据，由于商用电动车辆的行驶路线相对固定，与行驶路线对应的行驶位置特征数据可以一定程度上反应商用电动车辆的实际驾驶环境，有利于提高能耗预测数据的准确性。
在一些实施例中，将放电时长数据和行驶位置特征数据代入能耗预测模型，得到商用电动车辆的能耗预测数据之前，还包括：根据商用电动车辆的历史行驶数据、电池额定容量，以及预设的模型损失函数，基于机器学习算法进行模型训练得到能耗预测模型。
上述实施例中，将放电时长数据和行驶位置特征数据代入能耗预测模型，得到商用电动车辆的能耗预测数据之前，基于商用电动车辆的历史行驶数据进行模型训练得到能耗预测模型，有利于提高商用电动车辆能耗预测方法的科学性。
在一些实施例中，根据商用电动车辆的历史行驶数据、电池额定容量，以及预设的模型损失函数，基于机器学习算法进行模型训练得到能耗预测模型，包括：获取商用电动车辆的历史行驶数据和电池额定容量；历史行驶数据包括历史充放电数据和历史行驶位置特征数据；基于历史充放电数据和电池额定容量，计算得到商用电动车辆的历史单位时间能耗数据；根据历史单位时间能耗数据和历史行驶位置特征数据，得到历史能耗数据；及根据历史能耗数据和预设的模型损失函数，基于机器学习算法进行模型训练得到能耗预测模型。
上述实施例中，在进行模型训练的过程中，先计算得到商用电动车辆的历史单位时间能耗数据，再基于历史单位时间能耗数据和历史行驶位置特征数据进行模型训练得到能耗预测模型，可以准确表征单位时间能耗与行驶位置特征的关系，有利于提高能耗预测模型与实际工况的匹配度，进而提升模型的预测精度。
在一些实施例中，获取商用电动车辆的历史行驶数据，包括：获取商用电动车辆的历史原始行驶数据；及对历史原始行驶数据进行预处理，得到商用电动车辆的历史行驶数据。
上述实施例中，在获取商用电动车辆的历史原始行驶数据之后，需要对其进行预处理才会得到历史行驶数据，从而避免出现重复或者信息不全的历史行驶数据，可有效提升模型的预测准确度以及预测效率。
在一些实施例中，基于历史充放电数据和电池额定容量，计算得到商用电动车辆的历史单位时间能耗数据，包括：以设定时间为周期，基于历史充放电数据，得到商用电动车辆的历史充电数据和历史放电数据；基于历史充电数据和电池额定容量，得到设定时 间内商用电动车辆的电池健康状态；及根据电池健康状态、电池额定容量，以及历史充电数据对应的历史放电数据，得到商用电动车辆的历史单位时间能耗数据。
上述实施例中，在计算商用电动车辆的历史单位时间能耗数据的过程中，考虑车辆的实时电池健康状态，有利于提高历史单位时间能耗数据的准确性，进而提升模型的预测精度。
在一些实施例中，预设时间内，一个充电段对应有多个子放电段；根据电池健康状态、电池额定容量，以及历史充电数据对应的历史放电数据，得到商用电动车辆的历史单位时间能耗数据包括：根据电池健康状态、电池额定容量，以及各子放电段的历史放电数据，得到商用电动车辆的历史单位时间能耗数据。
上述实施例中，在进行历史单位时间能耗数据分析时，针对设定时间内一个充电段对应多个子放电段的情形，可结合各子放电段的历史放电数据实现最终的历史单位时间能耗数据分析，有效提高历史单位时间能耗数据的分析可靠性。
在一些实施例中，设定时间内，存在多个充电段；根据电池健康状态、电池额定容量，以及历史充电数据对应的历史放电数据，得到商用电动车辆的历史单位时间能耗数据包括：根据电池健康状态、电池额定容量、各充电段的历史充电数据以及各充电段对应的放电段的历史放电数据，得到商用电动车辆的历史单位时间能耗数据。
上述实施例中，在进行历史单位时间能耗数据分析时，针对设定时间内存在多个充电段的情形，可结合各充电段对应的放电段的历史放电数据分别进行各个放电段的能耗计算之后，最终实现历史单位时间能耗数据分析，可进一步提高历史单位时间能耗数据的分析可靠性。
在一些实施例中，根据电池健康状态、电池额定容量、各充电段的历史充电数据，以及各充电段对应的放电段的历史放电数据，得到商用电动车辆的历史单位时间能耗数据，包括：根据电池健康状态、电池额定容量、各充电段的历史充电数据以及各充电段对应的放电段的历史放电数据，得到各充电段对应的放电段的能耗；若各充电段对应的放电段的能耗中，最大能耗值小于电池额定容量，则根据最大能耗值计算设定时间内的历史单位时间能耗数据；若各充电段对应的放电段的能耗中，最大能耗值大于或等于电池额定容量，且小于预设倍数的电池额定容量，则根据各充电段对应的放电段的能耗的平均值，计算设定时间内的历史单位时间能耗数据；若各充电段对应的放电段的能耗中，最大能耗值大于或等于预设倍数的电池额定容量，则舍弃对应设定时间内的历史充放电数据。
上述实施例中，根据各个充电段对应的放电段的能耗进行区别处理，结合其所处 的不同区间实现不同的历史单位时间能耗数据分析，在异常情况下可将对应周期内的历史充放电数据舍弃，有效提高历史单位时间能耗数据的分析准确性。
在一些实施例中，将放电时长数据和行驶位置特征数据代入能耗预测模型，得到商用电动车辆的能耗预测数据之后，还包括：将历史能耗数据划分成训练集和测试集，基于训练集和能耗预测模型得到测试集对应的能耗预测数据；及根据测试集，以及测试集对应的能耗预测数据，对能耗预测数据进行修正处理，得到修正后的能耗预测数据。
上述实施例中，在得到能耗预测数据后，还基于训练集和测试集对能耗预测结果进行修正处理，可以进一步提高商用电动车辆能耗预测方法的预测精度。
在一些实施例中，将放电时长数据和行驶位置特征数据代入能耗预测模型，包括：对放电时长数据进行插值处理，得到插值处理后的放电时长数据，并将插值处理后的放电时长数据，以及行驶位置特征数据，代入能耗预测模型。
上述实施例中，在将放电时长数据代入能耗预测模型之前，先进行插值处理，可以确保输入模型的放电时长数据的连续性，进而确保能耗预测数据的连续性，有利于提高商用电动车辆能耗预测方法的灵活性。
在一些实施例中，将放电时长数据和行驶位置特征数据代入能耗预测模型，得到商用电动车辆的能耗预测数据，包括：将放电时长数据和行驶位置特征数据代入能耗预测模型，得到商用电动车辆的初始能耗预测数据；及将初始能耗预测数据按照预设时间分组，并取各组的预设百分位数作为对应预设时间内的能耗预测数据；预设百分位数大于50％。
上述实施例中，基于能耗预测模型预测得到的初始能耗预测数据，进行分组后取各组中较大的预设百分位数作为能耗预测数据，可以剔除噪声值的干扰，提高能耗预测数据的准确性。
在一些实施例中，将放电时长数据和行驶位置特征数据代入能耗预测模型，得到商用电动车辆的能耗预测数据之后，还包括：获取商用电动车辆的电池健康状态估计值，并根据电池健康状态估计值和能耗预测数据，确定商用电动车辆的抛锚风险。
上述实施例中，在得到商用电动车辆能耗预测数据后，还获取商用电动车辆的电池健康状态估计值，并根据电池健康状态估计值和能耗预测数据，确定商用电动车辆的抛锚风险，可以便于用户及时发现异常并采取相应措施，有利于降低商用电动车辆的抛锚概率，提高车辆的使用安全性。
一种商用电动车辆能耗预测装置，包括：
放电时长获取模块，用于获取商用电动车辆的放电时长数据；行驶位置特征获取 模块，用于获取商用电动车辆的行驶位置特征数据；及能耗预测模块，用于将放电时长数据和行驶位置特征数据代入能耗预测模型，得到商用电动车辆的能耗预测数据；能耗预测模型基于机器学习算法得到。
上述商用电动车辆能耗预测装置，将放电时长数据和行驶位置特征数据代入能耗预测模型，得到能耗预测数据，由于商用电动车辆的行驶路线相对固定，与行驶路线对应的行驶位置特征数据可以一定程度上反应商用电动车辆的实际驾驶环境，有利于提高能耗预测数据的准确性。
在一些实施例中，商用电动车辆能耗预测装置还包括：能耗预测模型训练模块，用于根据商用电动车辆的历史行驶数据、电池额定容量，以及预设的模型损失函数，基于机器学习算法进行模型训练得到能耗预测模型。
上述实施例中，将放电时长数据和行驶位置特征数据代入能耗预测模型，得到商用电动车辆的能耗预测数据之前，基于商用电动车辆的历史行驶数据进行模型训练得到能耗预测模型，有利于提高商用电动车辆能耗预测方法的科学性。
在一些实施例中，能耗预测模型训练模块包括：数据获取单元，获取商用电动车辆的历史行驶数据和电池额定容量；历史行驶数据包括历史充放电数据和历史行驶位置特征数据；能耗数据计算单元，基于历史充放电数据和电池额定容量，计算得到商用电动车辆的历史单位时间能耗数据；历史能耗数据生成单元，根据历史单位时间能耗数据和历史行驶位置特征数据，得到历史能耗数据；及能耗预测模型训练单元，根据历史能耗数据和预设的模型损失函数，基于机器学习算法进行模型训练得到能耗预测模型。
上述实施例中，在进行模型训练的过程中，先计算得到商用电动车辆的历史单位时间能耗数据，再基于历史单位时间能耗数据和历史行驶位置特征数据进行模型训练得到能耗预测模型，可以准确表征单位时间能耗与行驶位置特征的关系，有利于提高能耗预测模型与实际工况的匹配度，进而提升模型的预测精度。
在一些实施例中，商用电动车辆能耗预测装置还包括：修正模块，用于将历史能耗数据划分成训练集和测试集，基于训练集和能耗预测模型得到测试集对应的能耗预测数据；根据测试集，以及测试集对应的能耗预测数据，对能耗预测数据进行修正处理，得到修正后的能耗预测数据。
上述实施例中，在得到能耗预测数据后，还基于训练集和测试集对能耗预测结果进行修正处理，可以进一步提高商用电动车辆能耗预测方法的预测精度。
在一些实施例中，商用电动车辆能耗预测装置还包括：抛锚风险确定模块，用于 获取商用电动车辆的电池健康状态估计值，并根据电池健康状态估计值和能耗预测数据，确定商用电动车辆的抛锚风险。
上述实施例中，在得到商用电动车辆能耗预测数据后，还获取商用电动车辆的电池健康状态估计值，并根据电池健康状态估计值和能耗预测数据，确定商用电动车辆的抛锚风险，可以便于用户及时发现异常并采取相应措施，有利于降低商用电动车辆的抛锚概率，提高车辆的使用安全性。
一种计算机设备，包括存储器及一个或多个处理器，存储器中储存有计算机可读指令，计算机可读指令被一个或多个处理器执行时，使得一个或多个处理器执行以下步骤：
获取商用电动车辆的放电时长数据；获取商用电动车辆的行驶位置特征数据；及将放电时长数据和行驶位置特征数据代入能耗预测模型，得到商用电动车辆的能耗预测数据；能耗预测模型基于机器学习算法得到。
一个或多个存储有计算机可读指令的非易失性计算机可读存储介质，计算机可读指令被一个或多个处理器执行时，使得一个或多个处理器执行以下步骤：
获取商用电动车辆的放电时长数据；获取商用电动车辆的行驶位置特征数据；及将放电时长数据和行驶位置特征数据代入能耗预测模型，得到商用电动车辆的能耗预测数据；能耗预测模型基于机器学习算法得到。
一种计算机程序产品，包括计算机程序，该计算机程序被一个或多个处理器执行时，使得一个或多个处理器执行以下步骤：
获取商用电动车辆的放电时长数据；获取商用电动车辆的行驶位置特征数据；及将放电时长数据和行驶位置特征数据代入能耗预测模型，得到商用电动车辆的能耗预测数据；能耗预测模型基于机器学习算法得到。
上述计算机设备、计算机可读存储介质和计算机程序产品，将放电时长数据和行驶位置特征数据代入能耗预测模型，得到能耗预测数据，由于商用电动车辆的行驶路线相对固定，与行驶路线对应的行驶位置特征数据可以一定程度上反应商用电动车辆的实际驾驶环境，有利于提高能耗预测数据的准确性。
附图说明
为了更清楚地说明本申请实施例中的技术方案，下面将对实施例中所需要使用的附图作简单地介绍，显而易见地，下面描述中的附图仅仅是本申请的一些实施例，对于本领域普通技术人员来讲，在不付出创造性劳动的前提下，还可以根据这些附图获得其它的 附图。
图1为本申请一些实施例中商用电动车辆能耗预测方法流程示意图；
图2为本申请一些实施例中商用电动车辆能耗预测方法流程示意图；
图3为本申请一些实施例中能耗预测模型训练流程示意图；
图4为本申请一些实施例中历史单位时间能耗数据分析流程示意图；
图5为本申请一些实施例中商用电动车辆能耗预测方法流程示意图；
图6为本申请一些实施例中商用电动车辆能耗预测方法流程示意图；
图7为本申请一些实施例中商用电动车辆能耗预测方法流程示意图；
图8为本申请一些实施例中商用电动车辆能耗预测方法流程图；
图9为本申请一些实施例中商用电动车辆能耗预测装置结构示意图；
图10为本申请一些实施例中商用电动车辆能耗预测装置结构示意图；
图11为本申请一些实施例中能耗预测模型结构示意图；
图12为本申请一些实施例中商用电动车辆能耗预测装置结构示意图；
图13为本申请一些实施例中能耗预测模块结构示意图；
图14为本申请一些实施例中商用电动车辆能耗预测装置结构示意图；
图15为本申请一些实施例中计算机设备内部结构示意图。
具体实施方式
为了使本申请的技术方案及优点更加清楚明白，以下结合附图及实施例，对本申请进行进一步详细说明。应当理解，此处描述的具体实施例仅仅用以解释本申请，并不用于限定本申请。
除非另有定义，本文所使用的所有的技术和科学术语与属于本申请的技术领域的技术人员通常理解的含义相同；本文中所使用的术语只是为了描述具体的实施例的目的，不是旨在于限制本申请；本申请的说明书和权利要求书及上述附图说明中的术语“包括”和“具有”以及它们的任何变形，意图在于覆盖不排他的包含。
在本申请实施例的描述中，技术术语“第一”“第二”等仅用于区别不同对象，而不能理解为指示或暗示相对重要性或者隐含指明所指示的技术特征的数量、特定顺序或主次关系。在本申请实施例的描述中，“多个”的含义是两个以上，除非另有明确具体的限定。同时，在本说明书中使用的术语“和/或”包括相关所列项目的任何及所有组合。
本申请提供的商用电动车辆能耗预测方法、装置、计算机设备、计算机可读存储 介质和计算机程序产品，可以应用于各类商用电动车辆，包括但不限于电动公交车、电动巴士、地铁电动车辆、铁路干线电动车辆和轻轨电动车辆等。
第一方面，本申请提供了一种商用电动车辆能耗预测方法，该方法可以应用于终端，也可以应用于服务器，还可以通过终端与服务器的交互实现。为便于理解，下面均以该方法应用于终端的情况为例，进行说明。在一些实施例中，如图1所示，该方法包括步骤S102至步骤S106。
步骤S102：获取商用电动车辆的放电时长数据。
其中，放电时长数据是指商用电动车辆以设定时间为计算周期，在每个计算周期内的放电时长所组成的数据集。该设定时间，可以是半天、一天或两天。具体的，放电时长数据可以从商用电动车辆的充放电数据中提取。充放电数据包括充放电开始时刻、结束时刻、充入容量以及各时刻对应的SOC(State of Charge,电池中剩余电荷的可用百分比)等。放电时长即为放电开始时刻和放电结束时刻之间的时间差。
进一步地，终端可以获取商用电动车辆的充放电数据，并对充放电数据进行数据预处理，得到放电时长数据。例如，终端可以剔除充放电数据中重复和数据信息不全的数据行，并根据字段信息，提取商用电动车辆在放电段的放电数据，再基于放电数据中的放电开始时刻和放电结束时刻，计算得到放电时长。终端还可以基于统计学规则，剔除计算得到的放电时长中的异常值，以避免异常数据的影响，进一步提高能耗预测精度。例如，可以剔除预设时长内的多个计算周期中，＜-1*sigma和＞+3*sigma的放电时长数据。
需要说明的是，若设定时间内存在多个子放电段，则将各子放电段的子放电时长相加，计算得到设定时间内的放电时长。
步骤S104：获取商用电动车辆的行驶位置特征数据。
其中，步骤S104可以在步骤S102之前、之后，或与步骤S102同步执行。进一步地，行驶位置特征数据包括商用电动车辆不同时刻的行驶位置，以及该行驶位置对应的天气数据和地形数据等。该天气数据包括气温、湿度、气压和风速等信息；该地形数据包括坡度、轨道阻力和空气阻力等信息。
具体的，由于商用电动车辆的行驶路线相对固定，终端可以获取历史同期的行驶位置特征数据，也可以先获取待预测时间内商用电动车辆的行驶路线，再根据行驶路线上的车辆行驶位置信息，关联与行驶位置对应的天气数据和地形数据，得到该行驶路线对应的行驶位置特征数据。例如，行驶路线所对应的历史同期的天气数据，以及行驶路线所对应的地形数据。
进一步地，终端获取商用电动车辆的放电时长数据和行驶位置特征数据的具体方式，可以是主动获取，也可以是被动接收。
步骤S106：将放电时长数据和行驶位置特征数据代入能耗预测模型，得到商用电动车辆的能耗预测数据。
其中，能耗预测模型基于机器学习算法得到。该机器学习算法，可以是神经网络算法或决策树算法。在一个实施例中，该机器学习算法为GBRT(Gradient Boost Regression Tree，渐进梯度回归树)算法。
具体的，将放电时长数据和行驶位置特征数据，代入基于机器学习算法得到的能耗预测模型，即可得到商用电动车辆的能耗预测数据。
进一步地，考虑到商用电动车辆的行驶路线相对固定，而使用频次容易受当前业务量和在役车辆数量的影响，因此，可以使用历史同期的环境温度数据，以及近期的放电时长数据作为模型自变量代入能耗预测模型，以提高能耗预测的精度。例如，在需要对未来三个月的能耗进行预测时，取近期三个月的放电时长和历史同期的环境温度作为输入量，代入能耗预测模型。
此外，终端得到能耗预测数据之后，还可以将该能耗预测数据输出。该能耗预测数据的输出对象，可以是存储装置、显示装置或通信装置。此外，终端还可以通过通信装置将能耗预测数据输出至其他终端。
上述商用电动车辆能耗预测方法，将放电时长数据和行驶位置特征数据代入能耗预测模型，得到能耗预测数据，一方面，由于商用电动车辆的行驶路线相对固定，与行驶路线对应的行驶位置特征数据可以一定程度上反应商用电动车辆的实际驾驶环境，有利于提高能耗预测数据的准确性；另一方面，采用基于机器学习算法得到的能耗预测模型进行能耗预测，可以省去繁琐的数学建模，在确保精度的基础上提高工作效率。
需要说明的是，根据机器学习算法得到的能耗预测模型的特点，为获取以设定时长为周期的连续能耗预测数据，可以在将放电时长数据代入能耗预测模型之前，对数据进行插值处理，确保输入模型的放电时长数据为以设定时长为周期的连续数据，进而确保能耗预测数据的连续性；也可以在得到非连续的初始能耗预测数据之后，对初始能耗预测数据进行插值处理，得到以设定时长为周期的，连续的商用电动车辆的能耗预测数据。
在一些实施例中，如图2所示，步骤S106之前，该方法还包括步骤S105：根据商用电动车辆的历史行驶数据、电池额定容量，以及预设的模型损失函数，基于机器学习 算法进行模型训练得到能耗预测模型。其中，步骤S105可以在步骤S102和步骤S104的之前、之后，或者与上述步骤同步执行。
其中，关于机器学习算法的具体限定参见上文，此处不再赘述。历史行驶数据是指商用电动车辆在历史运行过程中的实际数据。具体的，终端基于商用电动车辆的历史行驶数据和电池额定容量，得到用于模型训练的实际能耗数据，并使用预设的模型损失函数，基于机器学习算法进行模型训练，得到能耗预测模型。
进一步地，模型损失函数具体可以是均方差函数、绝对损失函数或分位数损失函数等。以能耗预测模型为GBRT模型为例，模型损失函数的具体公式可以为：
Figure PCTCN2022096746-appb-000001
其中，y为实际值，f(x)为预测值。基于上述模型损失函数，按照预测值与实际值的差值趋向于最小的损失函数进行残差训练，即可拟合出GBRT回归树，进而得到用于能耗预测的GBRT模型。
上述实施例中，将放电时长数据和行驶位置特征数据代入能耗预测模型，得到商用电动车辆的能耗预测数据之前，基于商用电动车辆的历史行驶数据进行模型训练得到能耗预测模型，有利于提高商用电动车辆能耗预测方法的科学性。
在一些实施例中，如图3所示，步骤S105包括步骤S302至步骤S308。
步骤S302：获取商用电动车辆的历史行驶数据和电池额定容量。
其中，历史行驶数据包括历史充放电数据和历史行驶位置特征数据。进一步地，历史充放电数据包括充放电开始时刻、结束时刻、充入容量以及各时刻对应的SOC等。电池额定容量是指额定条件下电池的容量。历史行驶位置特征数据为与历史充放电数据所在时间节点对应的行驶位置特征数据。
步骤S304：基于历史充放电数据和电池额定容量，计算得到商用电动车辆的历史单位时间能耗数据。
其中，商用电动车辆的历史单位时间能耗数据，是指由过去的一段时间内，以设定时间为周期的商用电动车辆的单位时间能耗值组成的数据集。具体的，终端基于历史充放电数据，可以以设定时间为计算周期，得到各周期内商用电动车辆的历史能耗以及放电时长，进而得到商用电动车辆的历史单位时间能耗数据。
进一步地，终端基于历史充放电数据和电池额定容量，得到商用电动车辆的历史单位时间能耗数据的具体方式并不唯一。在一个实施例中，终端基于历史充放电数据， 将放电起始和结束时刻的SOC差值，乘以电池额定容量，得到设定时间内商用电动车辆的历史放电电量，再结合对应的放电时长，得到商用电动车辆在预设时长内的历史单位时间能耗。其中，预设时长可以是一个月、两个月或三个月，具体可以根据能耗预测需求确定。
步骤S306：根据历史单位时间能耗数据和行驶位置特征数据，得到历史能耗数据。
其中，历史能耗数据是指包含历史单位时间能耗数据和历史行驶位置特征数据的历史数据。具体的，可以将步骤S404中得到的，预设时长内多个计算周期的历史单位时间能耗数据，按照计算周期所在的时间节点，与同一商用电动车辆的历史行驶位置特征数据合并，形成包含时间、历史单位时间能耗数据和历史行驶位置特征数据的历史能耗数据。
步骤S308：根据历史能耗数据和预设的模型损失函数，基于机器学习算法进行模型训练得到能耗预测模型。
具体的，基于历史能耗数据，使用预设的模型损失函数和机器学习算法进行模型训练，即可得到能耗预测模型。
上述实施例中，在进行模型训练的过程中，先计算得到商用电动车辆的历史单位时间能耗数据，再基于历史单位时间能耗数据和历史行驶位置特征数据进行模型训练得到能耗预测模型，可以准确表征单位时间能耗与行驶位置特征的关系，有利于提高能耗预测模型与实际工况的匹配度，进而提升模型的预测精度。
在一些实施例中，获取商用电动车辆的历史行驶数据，包括：获取商用电动车辆的历史原始行驶数据；及对历史原始行驶数据进行预处理，得到商用电动车辆的历史行驶数据。
具体的，终端可以获取商用电动车辆的历史原始行驶数据，并在历史原始行驶数据的基础上，进行数据预处理，得到对应商用电动车辆的历史行驶数据。其中，历史原始行驶数据包括历史充放电数据、历史车辆位置信息数据，以及各车辆位置对应的天气数据和地形数据等。
例如，终端可以对历史充放电数据进行预处理，剔除重复和数据信息不全的数据行，根据字段信息，提取商用电动车辆每个充电段的充电数据，和每个充电段对应的放电段的放电数据。其中，数据信息不全的数据行，包括日期为空的数据行、充电段或放电段数据缺失的数据行。
上述实施例中，在获取商用电动车辆的历史原始行驶数据之后，需要对其进行预处理才会得到历史行驶数据，从而避免出现重复或者信息不全的历史行驶数据，可有效提升模型的预测准确度以及预测效率。
在一些实施例中，请参阅图4，步骤S304包括步骤S402至步骤S406。
步骤S402：以设定时间为计算周期，基于历史充放电数据，得到商用电动车辆的历史充电数据和历史放电数据。
其中，充电数据包括充电开始时刻、充电结束时刻、充入容量以及各时刻对应的SOC等。放电数据包括放电开始时刻、放电结束时刻以及各时刻对应的SOC等。历史充电数据和历史放电数据，分别是指由过去的一段时间内的实际充电数据和实际放电数据。
具体的，终端对历史充放电数据中的信息，进行提取和数据归类：将包含“充电”字段信息的数据归类为充电数据，并将包含“放电”字段信息的数据归类为放电数据。再以设定时间为周期进行二次归类，可以得到多个周期内商用电动车辆的历史充电数据和历史放电数据。
步骤S404：基于历史充电数据和电池额定容量，得到设定时间内商用电动车辆的电池健康状态。
其中，电池健康状态又称SOH(State of Health)，用于表征当前电池相对于新电池存储电能的能力，以百分比的形式定量描述当前电池的性能状态。电池的健康状态受温度、电流倍率、截止电压等多种因素影响，从电池寿命开始到寿命结束期间动态变化。基于此，为提高历史单位时间能耗数据的准确性，终端基于历史充电数据和电池额定容量，计算得到设定时间内商用电动车辆的电池健康状态。
具体的，电池健康状态的计算公式为：
Figure PCTCN2022096746-appb-000002
式中，Q为充入容量，pack cap为电池额定容量，
Figure PCTCN2022096746-appb-000003
为充电开始时刻和充电结束时刻的SOC差。
步骤S406：根据电池健康状态、电池额定容量，以及历史充电数据对应的历史放电数据，得到商用电动车辆的历史单位时间能耗数据。
其中，充电数据对应的放电数据，是指当前充电段的充电结束时刻至下一充电段的充电开始时刻之间的放电数据。
具体的，终端先根据电池健康状态、电池额定容量，以及放电开始时刻和放电结束时刻的SOC差，即可得到商用电动车辆在设定时间内的历史能耗，再综合预设时长内多个计算周期的历史能耗和放电时长，得到商用电动车辆在预设时长内各计算周期的单位时间能耗值。一段时间内，多个计算周期的单位时间能耗值组成的数据集，即为历史单位时间能耗数据。
其中，设定时间内的能耗C的计算公式为：
C＝SOH×pack cap×delt_soc_discharge    (3)
式中，delt_soc_discharge为放电开始时刻和放电结束时刻的SOC差。
单位时间能耗值Cper的计算公式为：
Figure PCTCN2022096746-appb-000004
式中，hour_discharge为放电时长，即放电开始时刻和放电结束时刻的时间差。
上述实施例中，在计算商用电动车辆的历史单位时间能耗数据的过程中，考虑车辆的实时电池健康状态，有利于提高历史单位时间能耗数据的准确性，进而提升模型的预测精度。
在一些实施例中，预设时间内，一个充电段对应有多个子放电段，步骤S406包括：根据电池健康状态、电池额定容量，以及历史充电数据对应的历史放电数据，得到商用电动车辆的历史单位时间能耗数据包括：根据电池健康状态、电池额定容量，以及各子放电段的历史放电数据，得到商用电动车辆的历史单位时间能耗数据。
具体的，若设定时间内，一个充电段对应有多个子放电段，则结合各子放电段的放电数据得到历史单位时间能耗数据。可以结合电池健康状态、电池额定容量以及各子放电段的历史放电数据，采用上述能耗计算公式分别计算得到各个放电段对应的能耗，将各放电段的能耗相加，得到这些放电段的总能耗和总放电时长，并将该总能耗除以总放电时长，得到单位时间能耗值；也可以基于各放电段的放电数据，分别计算得到各子放电段的子单位时间能耗值，再取各子单位时间能耗值的平均值或中位数，作为充电段对应的放电段的单位时间能耗值。
上述实施例中，在进行历史单位时间能耗数据分析时，针对设定时间内一个充电段对应多个子放电段的情形，可结合各子放电段的历史放电数据实现最终的历史单位时间能耗数据分析，有效提高历史单位时间能耗数据的分析可靠性。
在一些实施例中，设定时间内，存在多个充电段；步骤S406包括：根据电池健康状态、电池额定容量、各充电段的历史充电数据以及各充电段对应的放电段的历史放电数据，得到商用电动车辆的历史单位时间能耗数据。
具体的，若设定时间内，存在多个充电段，则根据各充电段的充电数据，以及该充电段对应的放电段的放电数据，结合电池健康状态、电池额定容量以及上述的能耗计算公式，分别计算得到各充电段对应的放电段的能耗。再根据各充电段对应的放电段的能耗，进行区别处理，得到历史单位时间能耗数据。
上述实施例中，在进行历史单位时间能耗数据分析时，针对设定时间内存在多个充电段的情形，可结合各充电段对应的放电段的历史放电数据分别进行各个放电段的能耗计算之后，最终实现历史单位时间能耗数据分析，可进一步提高历史单位时间能耗数据的分析可靠性。
进一步地，在一些实施例中，根据电池健康状态、电池额定容量、各充电段的历史充电数据以及各充电段对应的放电段的历史放电数据，得到商用电动车辆的历史单位时间能耗数据，包括：根据电池健康状态、电池额定容量、各充电段的历史充电数据，以及各充电段对应的放电段的历史放电数据，得到各充电段对应的放电段的能耗；若各充电段对应的放电段的能耗中，最大能耗值小于电池额定容量，则根据最大能耗值计算设定时间内的历史单位时间能耗数据；若各充电段对应的放电段的能耗中，最大能耗值大于或等于电池额定容量，且小于预设倍数的电池额定容量，则根据各充电段对应的放电段的能耗的平均值，计算设定时间内的历史单位时间能耗数据；若各充电段对应的放电段的能耗中，最大能耗值大于或等于预设倍数的电池额定容量，则舍弃对应设定时间内的历史充放电数据。
其中，预设倍数为大于1的实数，如1.1、1.2或1.3。各充电段对应的放电段的能耗的计算方式结合上述能耗计算公式实现，在此不再赘述。此外，在计算单位时间能耗值之前，还可以基于统计学规则，进行能耗和放电时长异常值的处理，以避免异常数据的影响，进一步提高能耗预测精度。例如，可以采用±3*sigma或者其它规则，剔除历史能耗中的离群值；采用±1*sigma或者其它规则，剔除历史能耗对应的单位时间能耗值的离群值；剔除＜-1*sigma和＞+3*sigma的放电时长数据。
上述实施例中，根据各个充电段对应的放电段的能耗进行区别处理，结合其所处的不同区间实现不同的历史单位时间能耗数据分析，在异常情况下可将对应周期内的历史充放电数据舍弃，有效提高历史单位时间能耗数据的分析准确性。
请参阅图5，在一些实施例中，步骤S106之后，还包括步骤S107：将历史能耗数据划分成训练集和测试集，基于训练集和能耗预测模型得到测试集对应的能耗预测数据，并根据测试集，以及测试集对应的能耗预测数据，对能耗预测数据进行修正处理，得到修正后的能耗预测数据。其中，步骤S107在步骤S108之前执行。
具体的，可以按照预设比例，将历史能耗数据划分成训练集和测试集两部分，并将训练集代入能耗预测模型，得到测试集对应的能耗预测数据。再将测试集中的实际能耗数据与对应的能耗预测数据进行比较，确定模型预测误差序列，再根据该误差序列确定能耗预测模型的修正值。进一步地，终端根据该误差序列确定能耗预测模型的修正值的方式并不唯一，例如，可以将误差序列中的中位数或平均值作为能耗预测模型的修正值。
其中，模型预测误差序列中的误差值的计算公式为：
error i＝Cp1 i-Ca i        (5)
式中，error i为测试集中第i个数据对应的误差值；Ca i为测试集中第i个单位时间能耗数据的实际值；Cp1 i为Ca i对应的能耗预测值。
得到能耗预测模型的修正值后，再基于该修正值对能耗预测数据进行修正处理，将得到的初始能耗预测值叠加能耗预测模型的修正值，即可得到修正后的能耗预测值。也即，公式(5)对应的最终能耗预测值为：
Cpredict_final i＝Cpredict_value i-error     (6)
式中，error为能耗预测模型的修正值，Cpredict_value i为第i个初始能耗预测值，Cpredict_final i为Cpredict_value i对应的修正后的能耗预测值。各修正后的能耗预测值组合而成的数据集，即为商用电动车辆的能耗预测数据。
需要说明的是，预设比例的取值并不唯一，例如，可以按照4:1的预设比例，取实际能耗数据中的80％作为训练集，剩下的20％作为测试集；也可以按照3:2的预设比例，取实际能耗数据中的60％作为训练集，剩下的40％作为测试集。
上述实施例中，在得到能耗预测数据后，还基于训练集和测试集对能耗预测结果进行修正处理，可以进一步提高商用电动车辆能耗预测方法的预测精度。
在一些实施例中，将放电时长数据和行驶位置特征数据代入能耗预测模型，包括：对放电时长数据进行插值处理，得到插值处理后的放电时长数据，并将插值处理后的放电时长数据，以及行驶位置特征数据，代入能耗预测模型。
其中，插值是指在离散数据的基础上补插连续函数，使得这条连续曲线通过全部给定的离散数据点。通过插值处理，可以通过函数在有限个点处的取值状况，估算出函数在其他点处的近似值。
具体的，由于历史充放电数据可能存在数据缺失，并且在实际工况下，设定时长内可能存在数据信息不全的情况，例如放电段数据缺失，因此，步骤S102中获取的放电时长数据，可能是非连续数据，缺失预设时长内某一个或多个周期的计算值。基于此，先对放电时长数据进行插值处理，得到差值处理后的以设定时长为周期的连续放电时长数据，再将该连续放电时长数据与行驶位置特征数据，代入能耗预测模型。
上述实施例中，在将放电时长数据代入能耗预测模型之前，先进行插值处理，可以确保输入模型的放电时长数据的连续性，进而确保能耗预测数据的连续性，有利于提高商用电动车辆能耗预测方法的灵活性。
请参阅图6，在一些实施例中，步骤S106包括步骤S602和步骤S604。
步骤S602：将放电时长数据和行驶位置特征数据代入能耗预测模型，得到商用电动车辆的初始能耗预测数据。
其中，初始能耗预测数据为能耗预测模型的预测值。具体的，将放电时长数据和行驶位置特征数据代入能耗预测模型，可以得到商用电动车辆的初始能耗预测数据。
步骤S604：将初始能耗预测数据按照预设时间分组，并取各组的预设百分位数作为对应预设时间内的能耗预测数据。
其中，放电时长数据为过去预设时长内多个计算周期的实际数据。初始能耗预测数据为与放电时长数据对应的，多个计算周期的预测数据。也即，初始能耗预测数据中包含未来的预设时长内的多个计算周期的预测数据。进一步地，将一组数据从小到大排序，并计算相应的累计百分位，则某一百分位所对应数据的值就称为这一百分位的百分位数。不难理解，大于50％的预设百分位数对应的初始能耗预测值，为一组初始能耗预测数据中的较大值。该预设百分位数的具体数值并不唯一，例如可以85％分位数、90％分位数或95％分位数。
具体的，终端将初始能耗数据按照预设时间分组，得到多组预测数据，并取各组数据中较大的预设分位数作为该组的能耗预测数据，进而得到整个预设时长内商用电动车辆的能耗预测数据。
需要说明的是，上文中涉及预设时长、预设时间和计算周期这三个时间概念中，预设时长时间跨度最大，计算周期的时间跨度最小。进一步地，预设时间的取值并不唯 一，可以根据预设时长和计算周期灵活设置。例如，预设时间为三个月且计算周期为一天时，可以将预设时长设置为10天或15天。
上述实施例中，基于能耗预测模型预测得到的初始能耗预测数据，进行分组后取各组中较大的预设百分位数作为能耗预测数据，可以剔除噪声值的干扰，提高能耗预测数据的准确性。
在一些实施例中，如图7所示，步骤S106之后，该方法还包括步骤S108：获取商用电动车辆的电池健康状态估计值，并根据电池健康状态估计值和能耗预测数据，确定商用电动车辆的抛锚风险。
其中，电池健康状态又称SOH，用于表征当前电池相对于新电池存储电能的能力，以百分比的形式定量描述当前电池的性能状态。电池的健康状态受温度、电流倍率、截止电压等多种因素影响，从电池寿命开始到寿命结束期间动态变化。
具体的，终端获取商用电动车辆的电池健康状态估计值，并将电池健康状态估计值乘以电池额定容量，得到该计算周期内的电池电量，再结合同一计算周期内的能耗预测数据，判断该商用电动车辆是否存在抛锚风险：若连续设定次数的计算周期内预测的电池健康状态估计值所对应的电池电量，均小于对应计算周期内的能耗预测值，则认为该车存在较高的抛锚风险。该设定次数的具体数值，可以是4、5或6。
进一步地，在确定商用电动车辆的抛锚风险后，还可以输出警示信息，提示用户采取相应的风险预防措施，例如，提示用户加强该车的监控频次，或者提示用户准备备件等。
上述实施例中，在得到商用电动车辆能耗预测数据后，还获取商用电动车辆的电池健康状态估计值，并根据电池健康状态估计值和能耗预测数据，确定商用电动车辆的抛锚风险，可以便于用户及时发现异常并采取相应措施，有利于降低商用电动车辆的抛锚概率，提高车辆的使用安全性。
为便于理解，下面结合图8，对商用电动车辆能耗预测方法进行详细说明。
终端获取商用电动车辆的历史行驶数据和电池额定容量，该历史行驶数据包括历史充放电数据和历史行驶特征数据。其中，行驶特征数据包括商用电动车辆不同时刻所在的城市，以及该城市对应的气温。获取历史行驶数据后，先进行数据预处理，剔除重复以及日期为空的数据行，并提取每辆车的充放电段数据段信息，提取每个充电段的起始SOC和结束SOC、充入容量Q，提取每个放电段的起始SOC和结束SOC、放电段时长等信息。再将充放电段的数据合并，将每个充电段对应至少一个放电段，并舍弃只有充电 段或只有放电段的数据。其中，充电段对应的放电段，是指当前充电段的充电结束时刻至下一充电段的充电开始时刻之间的放电数据所对应的放电段。
得到对应的历史充放电数据后，再基于历史充放电数据，以一天为计算周期，得到车辆的历史单位时间能耗数据。具体的，先基于历史充电数据和电池额定容量，根据电池健康状态的计算公式(2)，得到计算周期内商用电动车辆的电池健康状态SOH，再将计算得到的SOH、电池额定容量，以及该充电数据对应的放电数据，代入公式(3)和(4)，得到对应的计算周期内车辆的能耗C和单位时间能耗值Cper。
若一个计算周期内，一个充电段对应有多个子放电段，则结合各子放电段的放电数据计算单位时间能耗值。具体的，可以将各放电段的能耗相加，得到这些放电段的总能耗和总放电时长，再将该总能耗除以总放电时长，得到单位时间能耗值；也可以基于各放电段的放电数据，分别计算得到各子放电段的子单位时间能耗值，再取各子单位时间能耗值的平均值或中位数，作为充电段对应的放电段的单位时间能耗值。
若一个计算周期内，存在多个充电段，则根据各充电段的充电数据，以及该充电段对应的放电段的放电数据，分别计算得到各充电段对应的放电段的能耗。再根据各充电段对应的放电段的能耗，进行区别处理：若各充电段对应的放电段的能耗中，最大能耗值小于电池额定容量，则基于该最大能耗值计算设定时间内的单位时间能耗值；若各充电段对应的放电段的能耗中，最大能耗值大于或等于电池额定容量，且小于1.2倍的电池额定容量，则基于各充电段对应的放电段的能耗的平均值，计算设定时间内的单位时间能耗值；若各充电段对应的放电段的能耗中，最大能耗值大于或等于1.2倍的电池额定容量，则认为能耗异常，舍弃该计算周期内的充放电数据。
进一步地，在计算单位时间能耗值之前，还可以基于统计学规则，进行能耗和放电时长异常值的处理，以避免异常数据的影响，进一步提高能耗预测精度。例如，可以采用±3*sigma或者其它规则，剔除历史能耗中的离群值；采用±1*sigma或者其它规则，剔除历史能耗对应的历史单位时间能耗值的离群值；剔除＜-1*sigma和＞+3*sigma的历史放电时长数据。
结合预设时长内多个计算周期的历史单位时间能耗值，即可得到商用电动车辆在预设时长内的历史单位时间能耗数据。得到历史单位时间能耗数据后，一方面，将该历史单位时间能耗数据合并历史行驶数据中的对应城市的环境温度数据，得到合并数据集，并基于该合并数据集，和预设的模型损失函数，进行模型训练得到GBRT模型。其中，模型损失函数为公式(1)。另一方面，考虑到商用电动车辆的行驶路线相对固定， 而使用频次容易受当前业务量和在役车辆数量的影响，对历史放电时长数据进行线性插值，得到最近的预设时长内的连续放电时长数据，并将该连续放电时长数据，与历史同期的环境温度数据，用于进行车辆未来能耗的预测。具体的，将连续的放电时长数据，与历史同期的环境温度数据，代入训练得到GBRT模型，即可得到未来同期的能耗预测数据。可以理解，训练得到的GBRT模型可以用于表征单位时间能耗值与环境温度的关系，基于此，可以根据历史同期的环境温度数据，得到对应的单位时间能耗数据，再结合最近的预设时长内的连续放电时长数据，即可得到未来同期的能耗预测数据。
例如，取近期三个月的放电时长hour_discharge t和历史同期的环境温度tempature t作为模型输入的X，可以得到未来三个月的能耗值C(t)；
C(t)＝f(hour_discharge t,tempature t)      (7)
得到能耗预测数据后，将能耗预测数据按照预设时间分组，并取各组的预设95％分位数作为对应预设时间内的能耗预测数据。例如，对于未来三个月的能耗预测数据，可以按15天进行分组，并取各组的95％分位数，作为该组的能耗预测值。
为进一步提高能耗预测数据的准确性，按照4:1的比例，将历史能耗数据划分成训练集和测试集两部分，并将训练集代入能耗预测模型，得到测试集对应的能耗预测数据。再将测试集中的实际能耗数据与对应的能耗预测数据进行比较，根据公式(5)确定模型预测误差序列。最后将误差序列中的中位数确定为能耗预测模型的修正值，并基于公式(6)对能耗预测值进行修正，即可得到修正后的最终能耗预测数据。
此外，得到修正后的最终能耗预测数据后，终端还可以获取商用电动车辆的电池健康状态估计值，并将电池健康状态估计值乘以电池额定容量，得到该计算周期内的电池电量，再结合同一计算周期内的能耗预测数据，判断该商用电动车辆是否存在抛锚风险：若连续5个计算周期内预测的电池健康状态估计值所对应的电池电量，均小于对应计算周期内的能耗预测值，则认为该车存在较高的抛锚风险。进一步地，在确定商用电动车辆的抛锚风险后，还可以输出警示信息，提示用户采取相应的风险预防措施，例如提示用户加强该车的监控频次，或者提示用户准备备件等。
上述商用电动车辆能耗预测方法，将放电时长数据和行驶位置特征数据代入能耗预测模型，得到能耗预测数据，一方面，由于商用电动车辆的行驶路线相对固定，与行驶路线对应的行驶位置特征数据可以一定程度上反应商用电动车辆的实际驾驶环境，有利于提高能耗预测数据的准确性；另一方面，采用基于机器学习算法得到的能耗预测模型进行能耗预测，并引入误差项，可以省去繁琐的数学建模，在确保精度的基础上提高 工作效率。此外，在得到商用电动车辆能耗预测数据后，还获取商用电动车辆的电池健康状态估计值，并根据电池健康状态估计值和能耗预测数据，确定商用电动车辆的抛锚风险，可以便于用户及时发现异常并采取相应措施，有利于降低商用电动车辆的抛锚概率，提高车辆的使用安全性。
应该理解的是，虽然如上所述的各实施例所涉及的流程图中的各个步骤按照箭头的指示依次显示，但是这些步骤并不是必然按照箭头指示的顺序依次执行。除非本文中有明确的说明，这些步骤的执行并没有严格的顺序限制，这些步骤可以以其它的顺序执行。而且，如上所述的各实施例所涉及的流程图中的至少一部分步骤可以包括多个步骤或者多个阶段，这些步骤或者阶段并不必然是在同一时刻执行完成，而是可以在不同的时刻执行，这些步骤或者阶段的执行顺序也不必然是依次进行，而是可以与其它步骤或者其它步骤中的步骤或者阶段的至少一部分轮流或者交替地执行。
基于同样的发明构思，第二方面，本申请实施例还提供了一种用于实现上述所涉及的商用电动车辆能耗预测方法的商用电动车辆能耗预测装置。该装置所提供的解决问题的实现方案与上述方法中所记载的实现方案相似，故下面所提供的一个或多个商用电动车辆能耗预测装置实施例中的具体限定，可以参见上文中对于商用电动车辆能耗预测方法的限定，在此不再赘述。
在一些实施例中，如图9所示，提供了一种商用电动车辆能耗预测装置900，包括放电时长获取模块902、行驶位置特征获取模块904和能耗预测906，其中：
放电时长获取模块902，用于获取商用电动车辆的放电时长数据；行驶位置特征获取模块904，用于获取商用电动车辆的行驶位置特征数据；能耗预测模块906，用于将放电时长数据和行驶位置特征数据代入能耗预测模型，得到商用电动车辆的能耗预测数据；能耗预测模型基于机器学习算法得到。
请参阅图10，在一些实施例中，商用电动车辆能耗预测装置700还包括：能耗预测模型训练模块905，用于根据商用电动车辆的历史行驶数据、电池额定容量，以及预设的模型损失函数，基于机器学习算法进行模型训练得到能耗预测模型。
请参阅图11，在一些实施例中，能耗预测模型训练模块905包括：数据获取单元112，用于获取商用电动车辆的历史行驶数据和电池额定容量；历史行驶数据包括历史充放电数据和历史行驶位置特征数据；能耗数据计算单元114，用于基于历史充放电数据和电池额定容量，计算得到商用电动车辆的历史单位时间能耗数据；历史能耗数据生成单元116，用于根据历史单位时间能耗数据和历史行驶位置特征数据，得到历史能耗数据； 能耗预测模型训练单元118，用于根据历史能耗数据和预设的模型损失函数，基于机器学习算法进行模型训练得到能耗预测模型。
在一些实施例中，能耗数据计算单元114具体用于：以设定时间为周期，基于历史充放电数据，得到商用电动车辆的历史充电数据和历史放电数据；基于历史充电数据和电池额定容量，得到设定时间内商用电动车辆的电池健康状态；根据电池健康状态、电池额定容量，以及历史充电数据对应的历史放电数据，得到商用电动车辆的历史单位时间能耗数据。
请参阅图12，在一些实施例中，商用电动车辆能耗预测装置700还包括修正模块122，用于将历史能耗数据划分成训练集和测试集，基于训练集和能耗预测模型得到测试集对应的能耗预测数据；根据测试集，以及测试集对应的能耗预测数据，对能耗预测数据进行修正处理，得到修正后的能耗预测数据。
在一些实施例中，能耗预测模块906具体用于：对放电时长数据进行插值处理，得到插值处理后的放电时长数据，并将插值处理后的放电时长数据，以及行驶位置特征数据，代入能耗预测模型。
请参阅图13，在一些实施例中，能耗预测模块906包括：初始能耗预测数据生成单元132，用于将放电时长数据和行驶位置特征数据代入能耗预测模型，得到商用电动车辆的初始能耗预测数据；能耗预测数据生成单元134，用于将初始能耗预测数据按照预设时间分组，并取各组的预设百分位数作为对应预设时间内的能耗预测数据；预设百分位数大于50％。
请参阅图14，在一些实施例中，商用电动车辆能耗预测装置700还包括：抛锚风险确定模块142，用于获取商用电动车辆的电池健康状态估计值，并根据电池健康状态估计值和能耗预测数据，确定商用电动车辆的抛锚风险。
上述商用电动车辆能耗预测装置中的各个模块可全部或部分通过软件、硬件及其组合来实现。上述各模块可以硬件形式内嵌于或独立于计算机设备中的处理器中，也可以以软件形式存储于计算机设备中的存储器中，以便于处理器调用执行以上各个模块对应的操作。
在一些实施例中，提供了一种计算机设备，该计算机设备可以是终端，其内部结构图可以如图15所示。该计算机设备包括通过系统总线连接的处理器、存储器、通信接口、显示屏和输入装置。其中，该计算机设备的处理器用于提供计算和控制能力。该计算机设备的存储器包括非易失性存储介质、内存储器。该非易失性存储介质存储有操作 系统和计算机程序。该内存储器为非易失性存储介质中的操作系统和计算机程序的运行提供环境。该计算机设备的通信接口用于与外部的终端进行有线或无线方式的通信，无线方式可通过WIFI、移动蜂窝网络、NFC(近场通信)或其他技术实现。该计算机程序被处理器执行时以实现一种商用电动车辆能耗预测方法。该计算机设备的显示屏可以是液晶显示屏或者电子墨水显示屏，该计算机设备的输入装置可以是显示屏上覆盖的触摸层，也可以是计算机设备外壳上设置的按键、轨迹球或触控板，还可以是外接的键盘、触控板或鼠标等。
本领域技术人员可以理解，图15中示出的结构，仅仅是与本申请方案相关的部分结构的框图，并不构成对本申请方案所应用于其上的计算机设备的限定，具体的计算机设备可以包括比图中所示更多或更少的部件，或者组合某些部件，或者具有不同的部件布置。
在一些实施例中，提供了一种计算机设备，包括存储器和处理器，存储器中存储有计算机程序，该处理器执行计算机程序时实现上述各方法实施例中的步骤。
在一些实施例中，提供了一种计算机可读存储介质，其上存储有计算机程序，该计算机程序被处理器执行时实现上述各方法实施例中的步骤。
在一些实施例中，提供了一种计算机程序产品，包括计算机程序，该计算机程序被处理器执行时实现上述各方法实施例中的步骤。
本领域普通技术人员可以理解实现上述实施例方法中的全部或部分流程，是可以通过计算机程序来指令相关的硬件来完成，所述的计算机程序可存储于一非易失性计算机可读取存储介质中，该计算机程序在执行时，可包括如上述各方法的实施例的流程。其中，本申请所提供的各实施例中所使用的对存储器、数据库或其它介质的任何引用，均可包括非易失性和易失性存储器中的至少一种。非易失性存储器可包括只读存储器(Read-Only Memory，ROM)、磁带、软盘、闪存、光存储器、高密度嵌入式非易失性存储器、阻变存储器(ReRAM)、磁变存储器(Magnetoresistive Random Access Memory，MRAM)、铁电存储器(Ferroelectric Random Access Memory，FRAM)、相变存储器(Phase Change Memory，PCM)、石墨烯存储器等。易失性存储器可包括随机存取存储器(Random Access Memory，RAM)或外部高速缓冲存储器等。作为说明而非局限，RAM可以是多种形式，比如静态随机存取存储器(Static Random Access Memory，SRAM)或动态随机存取存储器(Dynamic Random Access Memory，DRAM)等。本申请所提供的各实施例中所涉及的数据库可包括关系型数据库和非关系型数据库 中至少一种。非关系型数据库可包括基于区块链的分布式数据库等，不限于此。本申请所提供的各实施例中所涉及的处理器可为通用处理器、中央处理器、图形处理器、数字信号处理器、可编程逻辑器、基于量子计算的数据处理逻辑器等，不限于此。
以上实施例的各技术特征可以进行任意的组合，为使描述简洁，未对上述实施例中的各个技术特征所有可能的组合都进行描述，然而，只要这些技术特征的组合不存在矛盾，都应当认为是本说明书记载的范围。
以上所述实施例仅表达了本申请的几种实施方式，其描述较为具体和详细，但并不能因此而理解为对发明专利范围的限制。应当指出的是，对于本领域的普通技术人员来说，在不脱离本申请构思的前提下，还可以做出若干变形和改进，这些都属于本申请的保护范围。因此，本申请专利的保护范围应以所附权利要求为准。

Claims (20)

  1. 一种商用电动车辆能耗预测方法，包括：
    获取商用电动车辆的放电时长数据；
    获取所述商用电动车辆的行驶位置特征数据；及
    将所述放电时长数据和所述行驶位置特征数据代入能耗预测模型，得到所述商用电动车辆的能耗预测数据；所述能耗预测模型基于机器学习算法得到。
  2. 根据权利要求1所述的方法，其中，所述将所述放电时长数据和所述行驶位置特征数据代入能耗预测模型，得到所述商用电动车辆的能耗预测数据之前，还包括：
    根据商用电动车辆的历史行驶数据、电池额定容量，以及预设的模型损失函数，基于机器学习算法进行模型训练得到能耗预测模型。
  3. 根据权利要求2所述的方法，其中，所述根据商用电动车辆的历史行驶数据、电池额定容量，以及预设的模型损失函数，基于机器学习算法进行模型训练得到能耗预测模型，包括：
    获取商用电动车辆的历史行驶数据和电池额定容量；所述历史行驶数据包括历史充放电数据和历史行驶位置特征数据；
    基于所述历史充放电数据和所述电池额定容量，计算得到所述商用电动车辆的历史单位时间能耗数据；
    根据所述历史单位时间能耗数据和所述历史行驶位置特征数据，得到历史能耗数据；
    根据所述历史能耗数据和预设的模型损失函数，基于机器学习算法进行模型训练得到能耗预测模型。
  4. 根据权利要求3所述的方法，其中，所述获取商用电动车辆的历史行驶数据，包括：
    获取商用电动车辆的历史原始行驶数据；及
    对所述历史原始行驶数据进行预处理，得到所述商用电动车辆的历史行驶数据。
  5. 根据权利要求3所述的方法，其中，所述基于所述历史充放电数据和所述电池额定容量，计算得到所述商用电动车辆的历史单位时间能耗数据，包括：
    以设定时间为周期，基于所述历史充放电数据，得到所述商用电动车辆的历史充电数据和历史放电数据；
    基于所述历史充电数据和所述电池额定容量，得到设定时间内所述商用电动车辆的电池健康状态；及
    根据所述电池健康状态、所述电池额定容量，以及所述历史充电数据对应的历史放电数据，得到所述商用电动车辆的历史单位时间能耗数据。
  6. 根据权利要求5所述的方法，其中，所述预设时间内，一个充电段对应有多个子放电段；所述根据所述电池健康状态、所述电池额定容量，以及所述历史充电数据对应的历史放电数据，得到所述商用电动车辆的历史单位时间能耗数据包括：
    根据所述电池健康状态、所述电池额定容量，以及各所述子放电段的历史放电数据，得到所述商用电动车辆的历史单位时间能耗数据。
  7. 根据权利要求5所述的方法，其中，所述设定时间内，存在多个充电段；所述根据所述电池健康状态、所述电池额定容量，以及所述历史充电数据对应的历史放电数据，得到所述商用电动车辆的历史单位时间能耗数据包括：
    根据所述电池健康状态、所述电池额定容量、各所述充电段的历史充电数据，以及各所述充电段对应的放电段的历史放电数据，得到所述商用电动车辆的历史单位时间能耗数据。
  8. 根据权利要求7所述的方法，其中，所述根据所述电池健康状态、所述电池额定容量、各所述充电段的历史充电数据，以及各所述充电段对应的放电段的历史放电数据，得到所述商用电动车辆的历史单位时间能耗数据，包括：
    根据所述电池健康状态、所述电池额定容量、各所述充电段的历史充电数据，以及各所述充电段对应的放电段的历史放电数据，得到各所述充电段对应的放电段的能耗；
    若各所述充电段对应的放电段的能耗中，最大能耗值小于所述电池额定容量，则根据所述最大能耗值计算所述设定时间内的历史单位时间能耗数据；
    若各所述充电段对应的放电段的能耗中，最大能耗值大于或等于所述电池额定容量，且小于预设倍数的所述电池额定容量，则根据各充电段对应的放电段的能耗的平均值，计算所述设定时间内的历史单位时间能耗数据；
    若各所述充电段对应的放电段的能耗中，最大能耗值大于或等于预设倍数的所述电池额定容量，则舍弃对应设定时间内的历史充放电数据。
  9. 根据权利要求3至8任意一项所述的方法，其中，所述将所述放电时长数据和所述行驶位置特征数据代入能耗预测模型，得到所述商用电动车辆的能耗预测数据之后，还包括：
    将所述历史能耗数据划分成训练集和测试集，基于所述训练集和所述能耗预测模型得到所述测试集对应的能耗预测数据；及
    根据所述测试集，以及所述测试集对应的能耗预测数据，对所述能耗预测数据进行修正处理，得到修正后的能耗预测数据。
  10. 根据权利要求1至8中任意一项所述的方法，其中，所述将所述放电时长数据和所述行驶位置特征数据代入能耗预测模型，包括：
    对所述放电时长数据进行插值处理，得到插值处理后的放电时长数据，并将所述插值处理后的放电时长数据，以及所述行驶位置特征数据，代入能耗预测模型。
  11. 根据权利要求1至8中任意一项所述的方法，其中，所述将所述放电时长数据和所述行驶位置特征数据代入能耗预测模型，得到所述商用电动车辆的能耗预测数据，包括：
    将所述放电时长数据和所述行驶位置特征数据代入能耗预测模型，得到所述商用电动车辆的初始能耗预测数据；及
    将所述初始能耗预测数据按照预设时间分组，并取各组的预设百分位数作为对应预设时间内的能耗预测数据；所述预设百分位数大于50％。
  12. 根据权利要求1至8中任意一项所述的方法，其中，所述将所述放电时长数据和所述行驶位置特征数据代入能耗预测模型，得到所述商用电动车辆的能耗预测数据之后，还包括：
    获取所述商用电动车辆的电池健康状态估计值，并根据所述电池健康状态估计值和所述能耗预测数据，确定所述商用电动车辆的抛锚风险。
  13. 一种商用电动车辆能耗预测装置，包括：
    放电时长获取模块，用于获取商用电动车辆的放电时长数据；
    行驶位置特征获取模块，用于获取所述商用电动车辆的行驶位置特征数据；及
    能耗预测模块，用于将所述放电时长数据和所述行驶位置特征数据代入能耗预测模型，得到所述商用电动车辆的能耗预测数据；所述能耗预测模型基于机器学习算法得到。
  14. 根据权利要求13所述的装置，其中，所述装置还包括：
    能耗预测模型训练模块，用于根据商用电动车辆的历史行驶数据、电池额定容量，以及预设的模型损失函数，基于机器学习算法进行模型训练得到能耗预测模型。
  15. 根据权利要求14所述的装置，其中，所述能耗预测模型训练模块包括：
    数据获取单元，获取商用电动车辆的历史行驶数据和电池额定容量；所述历史行驶数据包括历史充放电数据和历史行驶位置特征数据；
    能耗数据计算单元，基于所述历史充放电数据和所述电池额定容量，计算得到所述商用电动车辆的历史单位时间能耗数据；
    历史能耗数据生成单元，根据所述历史单位时间能耗数据和所述历史行驶位置特征数据，得到历史能耗数据；及
    能耗预测模型训练单元，根据所述历史能耗数据和预设的模型损失函数，基于机器学习算法进行模型训练得到能耗预测模型。
  16. 根据权利要求13至15任意一项所述的装置，其中，所述装置还包括：
    修正模块，用于将所述历史能耗数据划分成训练集和测试集，基于所述训练集和所述能耗预测模型得到所述测试集对应的能耗预测数据；根据所述测试集，以及所述测试集对应的能耗预测数据，对所述能耗预测数据进行修正处理，得到修正后的能耗预测数据。
  17. 根据权利要求13至15任意一项所述的装置，其中，所述装置还包括：
    抛锚风险确定模块，用于获取所述商用电动车辆的电池健康状态估计值，并根据所述电池健康状态估计值和所述能耗预测数据，确定所述商用电动车辆的抛锚风险。
  18. 一种计算机设备，包括存储器及一个或多个处理器，所述存储器中储存有计算机可读指令，所述计算机可读指令被所述一个或多个处理器执行时，使得所述一个或多个处理器执行以下步骤：
    获取商用电动车辆的放电时长数据；
    获取所述商用电动车辆的行驶位置特征数据；及
    将所述放电时长数据和所述行驶位置特征数据代入能耗预测模型，得到所述商用电动车辆的能耗预测数据；所述能耗预测模型基于机器学习算法得到。
  19. 一个或多个存储有计算机可读指令的非易失性计算机可读存储介质，所述计算机可读指令被一个或多个处理器执行时，使得所述一个或多个处理器执行以下步骤：
    获取商用电动车辆的放电时长数据；
    获取所述商用电动车辆的行驶位置特征数据；及
    将所述放电时长数据和所述行驶位置特征数据代入能耗预测模型，得到所述商用电动车辆的能耗预测数据；所述能耗预测模型基于机器学习算法得到。
  20. 一种计算机程序产品，包括计算机程序，该计算机程序被一个或多个处理器执行时，使得所述一个或多个处理器执行以下步骤：
    获取商用电动车辆的放电时长数据；
    获取所述商用电动车辆的行驶位置特征数据；及
    将所述放电时长数据和所述行驶位置特征数据代入能耗预测模型，得到所述商用电动车辆的能耗预测数据；所述能耗预测模型基于机器学习算法得到。
PCT/CN2022/096746 2022-01-07 2022-06-02 商用电动车辆能耗预测方法、装置和计算机设备 WO2023130659A1 (zh)

Priority Applications (4)

Application Number Priority Date Filing Date Title
JP2022548244A JP2024508198A (ja) 2022-01-07 2022-06-02 商用電気車両のエネルギー消費予測方法、装置及びコンピュータ機器
KR1020227027806A KR20230107471A (ko) 2022-01-07 2022-06-02 전기 상용차 전력 소비 예측 방법, 장치 및 컴퓨터 장치
EP22746939.2A EP4235338A4 (en) 2022-01-07 2022-06-02 METHOD AND APPARATUS FOR PREDICTING THE ENERGY CONSUMPTION OF AN ELECTRIC UTILITY VEHICLE, AND COMPUTER DEVICE
US17/929,889 US20230222852A1 (en) 2022-01-07 2022-09-06 Commercial electric vehicle energy consumption prediction method and apparatus, and computer device

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN202210016220.8A CN115808922B (zh) 2022-01-07 2022-01-07 商用电动车辆能耗预测方法、装置和计算机设备
CN202210016220.8 2022-01-07

Related Child Applications (1)

Application Number Title Priority Date Filing Date
US17/929,889 Continuation US20230222852A1 (en) 2022-01-07 2022-09-06 Commercial electric vehicle energy consumption prediction method and apparatus, and computer device

Publications (1)

Publication Number Publication Date
WO2023130659A1 true WO2023130659A1 (zh) 2023-07-13

Family

ID=84688546

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2022/096746 WO2023130659A1 (zh) 2022-01-07 2022-06-02 商用电动车辆能耗预测方法、装置和计算机设备

Country Status (2)

Country Link
CN (1) CN115808922B (zh)
WO (1) WO2023130659A1 (zh)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN117863969A (zh) * 2024-03-13 2024-04-12 国网北京市电力公司 一种考虑电池损耗的电动汽车充放电控制方法及系统

Families Citing this family (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN117150276B (zh) * 2023-11-01 2024-04-09 宁德时代新能源科技股份有限公司 机器学习模型构建方法、车辆行驶风险预测方法以及装置

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN104331743A (zh) * 2014-10-11 2015-02-04 清华大学 一种基于多目标优化的电动车辆出行规划方法
CN111832881A (zh) * 2020-04-30 2020-10-27 北京嘀嘀无限科技发展有限公司 基于路况信息预测电动车能耗的方法、介质和电子设备
CN112078431A (zh) * 2020-08-19 2020-12-15 大众问问(北京)信息科技有限公司 一种车辆能耗预测和能源补充方法及相关设备
WO2021044134A1 (en) * 2019-09-02 2021-03-11 Siemens Plc Method and system for predicting battery degradation
CN113428159A (zh) * 2021-05-28 2021-09-24 合肥国轩高科动力能源有限公司 一种车辆能耗分析方法

Family Cites Families (16)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US10693294B2 (en) * 2012-09-26 2020-06-23 Stem, Inc. System for optimizing the charging of electric vehicles using networked distributed energy storage systems
WO2014049878A1 (ja) * 2012-09-28 2014-04-03 パイオニア株式会社 消費エネルギー推定装置、消費エネルギー推定方法、消費エネルギー推定プログラムおよび記録媒体
BR112015025996A2 (pt) * 2013-04-11 2020-05-05 Nissan Motor dispositivo para predizer o consumo de energia e método para predizer o consumo de energia
JP5980170B2 (ja) * 2013-05-30 2016-08-31 三菱重工メカトロシステムズ株式会社 シミュレーション装置、シミュレーション方法及びプログラム
KR20150123540A (ko) * 2014-04-25 2015-11-04 삼성전자주식회사 전력 소비 최적화를 위한 스마트 시스템의 동작방법 및 장치
CN105904981A (zh) * 2016-04-07 2016-08-31 北京现代汽车有限公司 一种电动汽车续航里程估计控制方法、装置及整车控制器
CN106515478B (zh) * 2016-10-27 2017-10-20 合肥工业大学 电动汽车剩余行驶里程在线预测方法及装置
CN106600080B (zh) * 2017-01-24 2020-07-03 东南大学 一种数据网络与电力网络耦合模型参与系统经济调度方法
CN109367433B (zh) * 2018-12-18 2021-09-24 安徽工程大学 一种智能充电桩
CN109740213B (zh) * 2018-12-24 2023-09-05 斑马网络技术有限公司 车辆能耗评估方法、设备及存储介质
CN111452794B (zh) * 2019-01-22 2021-06-15 上海汽车集团股份有限公司 一种确定能耗的方法及装置和确定驾驶策略的方法及装置
CN110222906A (zh) * 2019-06-17 2019-09-10 北京嘀嘀无限科技发展有限公司 电动车能耗预测方法、计算机可读存储介质和电子设备
CN110435429B (zh) * 2019-08-14 2023-03-28 福州大学 一种融合能耗预测的双电机电动汽车续航里程估计方法
CN113340315A (zh) * 2020-03-02 2021-09-03 北京百度网讯科技有限公司 车辆充电方法、装置、电子设备及存储介质
CN111666715B (zh) * 2020-06-05 2023-04-25 北京理工大学 一种电动汽车能耗预测方法及系统
CN112560186B (zh) * 2020-10-15 2022-08-05 吉林大学 电动汽车行程能耗预测方法、装置、设备及可存储介质

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN104331743A (zh) * 2014-10-11 2015-02-04 清华大学 一种基于多目标优化的电动车辆出行规划方法
WO2021044134A1 (en) * 2019-09-02 2021-03-11 Siemens Plc Method and system for predicting battery degradation
CN111832881A (zh) * 2020-04-30 2020-10-27 北京嘀嘀无限科技发展有限公司 基于路况信息预测电动车能耗的方法、介质和电子设备
CN112078431A (zh) * 2020-08-19 2020-12-15 大众问问(北京)信息科技有限公司 一种车辆能耗预测和能源补充方法及相关设备
CN113428159A (zh) * 2021-05-28 2021-09-24 合肥国轩高科动力能源有限公司 一种车辆能耗分析方法

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN117863969A (zh) * 2024-03-13 2024-04-12 国网北京市电力公司 一种考虑电池损耗的电动汽车充放电控制方法及系统
CN117863969B (zh) * 2024-03-13 2024-05-17 国网北京市电力公司 一种考虑电池损耗的电动汽车充放电控制方法及系统

Also Published As

Publication number Publication date
CN115808922B (zh) 2023-10-27
CN115808922A (zh) 2023-03-17

Similar Documents

Publication Publication Date Title
WO2023130659A1 (zh) 商用电动车辆能耗预测方法、装置和计算机设备
WO2017076154A1 (zh) 网络事件预测以及建立网络事件预测模型的方法和装置
CN107037370A (zh) 基于监控数据的电动汽车剩余电量计算方法
CN110675267A (zh) 一种根据实时路况进行车险预警的方法及系统
WO2023221587A1 (zh) 电动车辆的动力电池的健康状态确定方法及服务器
CN116482540A (zh) 电池电压不一致性的分析预测方法、装置及系统
CN113715629B (zh) 基于改进符号回归和XGBoost算法的剩余续驶里程预测方法
CN116992311B (zh) 基于机器学习的储能电源状态分析方法
CN114282704A (zh) 充电站充电负荷预测方法、装置、计算机设备和存储介质
CN117665620A (zh) 基于新能源汽车数据的电池健康度评估方法
US20230305073A1 (en) Method and apparatus for providing a predicted aging state of a device battery based on a predicted usage pattern
Yamaguchi et al. Real-time prediction for future profile of car travel based on statistical data and greedy algorithm
US20230222852A1 (en) Commercial electric vehicle energy consumption prediction method and apparatus, and computer device
CN116562427A (zh) 充电站的充电负荷预测方法及装置、存储介质、设备
CN110988693A (zh) 一种基于建模仿真的化学电池储能应用分析方法
CN112782584B (zh) 电池电量的剩余使用额度的预测方法、系统、介质及设备
CN114066642A (zh) 电力零售的风险预测方法、终端及存储介质
CN106451438A (zh) 一种考虑智能用电行为的负荷区间预测方法
CN117341477B (zh) 磷酸铁锂动力电池内短路预警方法、装置和计算机设备
CN105021997B (zh) 蓄电池容量曲线的绘制方法及装置
CN116774064B (zh) 电池自放电检测方法、装置、设备和存储介质
US20240051429A1 (en) State prediction system, state prediction method, and program
CN112895967B (zh) 换电里程的剩余使用时间的预测方法、系统、介质及设备
CN117775000A (zh) 车辆剩余续航里程估计方法、装置、存储介质及电子装置
CN116976532A (zh) 非参数化的新能源场景-概率-区间一体化预测方法

Legal Events

Date Code Title Description
WWE Wipo information: entry into national phase

Ref document number: 2022548244

Country of ref document: JP

ENP Entry into the national phase

Ref document number: 2022746939

Country of ref document: EP

Effective date: 20220913