CN112860782A - Pure electric vehicle driving range estimation method based on big data analysis - Google Patents

Pure electric vehicle driving range estimation method based on big data analysis Download PDF

Info

Publication number
CN112860782A
CN112860782A CN202110167243.4A CN202110167243A CN112860782A CN 112860782 A CN112860782 A CN 112860782A CN 202110167243 A CN202110167243 A CN 202110167243A CN 112860782 A CN112860782 A CN 112860782A
Authority
CN
China
Prior art keywords
soc
data
driving
variable
model
Prior art date
Legal status (The legal status 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 status listed.)
Pending
Application number
CN202110167243.4A
Other languages
Chinese (zh)
Inventor
高振海
牛万发
陈思言
付振
梁小明
彭凯
刘相超
温文昊
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Jilin University
Original Assignee
Jilin University
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 Jilin University filed Critical Jilin University
Priority to CN202110167243.4A priority Critical patent/CN112860782A/en
Publication of CN112860782A publication Critical patent/CN112860782A/en
Pending legal-status Critical Current

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
    • G06F16/20—Information retrieval; Database structures therefor; File system structures therefor of structured data, e.g. relational data
    • G06F16/26—Visual data mining; Browsing structured data
    • B—PERFORMING OPERATIONS; TRANSPORTING
    • B60—VEHICLES IN GENERAL
    • B60L—PROPULSION OF ELECTRICALLY-PROPELLED VEHICLES; SUPPLYING ELECTRIC POWER FOR AUXILIARY EQUIPMENT OF ELECTRICALLY-PROPELLED VEHICLES; ELECTRODYNAMIC BRAKE SYSTEMS FOR VEHICLES IN GENERAL; MAGNETIC SUSPENSION OR LEVITATION FOR VEHICLES; MONITORING OPERATING VARIABLES OF ELECTRICALLY-PROPELLED VEHICLES; ELECTRIC SAFETY DEVICES FOR ELECTRICALLY-PROPELLED VEHICLES
    • B60L58/00—Methods or circuit arrangements for monitoring or controlling batteries or fuel cells, specially adapted for electric vehicles
    • B60L58/10—Methods or circuit arrangements for monitoring or controlling batteries or fuel cells, specially adapted for electric vehicles for monitoring or controlling batteries
    • B60L58/12—Methods or circuit arrangements for monitoring or controlling batteries or fuel cells, specially adapted for electric vehicles for monitoring or controlling batteries responding to state of charge [SoC]
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
    • G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N20/00—Machine learning
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/04—Architecture, e.g. interconnection topology
    • G06N3/045—Combinations of networks
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/08—Learning methods
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q10/00—Administration; Management
    • G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
    • B—PERFORMING OPERATIONS; TRANSPORTING
    • B60—VEHICLES IN GENERAL
    • B60L—PROPULSION OF ELECTRICALLY-PROPELLED VEHICLES; SUPPLYING ELECTRIC POWER FOR AUXILIARY EQUIPMENT OF ELECTRICALLY-PROPELLED VEHICLES; ELECTRODYNAMIC BRAKE SYSTEMS FOR VEHICLES IN GENERAL; MAGNETIC SUSPENSION OR LEVITATION FOR VEHICLES; MONITORING OPERATING VARIABLES OF ELECTRICALLY-PROPELLED VEHICLES; ELECTRIC SAFETY DEVICES FOR ELECTRICALLY-PROPELLED VEHICLES
    • B60L2260/00—Operating Modes
    • B60L2260/40—Control modes
    • B60L2260/50—Control modes by future state prediction
    • B60L2260/52—Control modes by future state prediction drive range estimation, e.g. of estimation of available travel distance
    • B—PERFORMING OPERATIONS; TRANSPORTING
    • B60—VEHICLES IN GENERAL
    • B60L—PROPULSION OF ELECTRICALLY-PROPELLED VEHICLES; SUPPLYING ELECTRIC POWER FOR AUXILIARY EQUIPMENT OF ELECTRICALLY-PROPELLED VEHICLES; ELECTRODYNAMIC BRAKE SYSTEMS FOR VEHICLES IN GENERAL; MAGNETIC SUSPENSION OR LEVITATION FOR VEHICLES; MONITORING OPERATING VARIABLES OF ELECTRICALLY-PROPELLED VEHICLES; ELECTRIC SAFETY DEVICES FOR ELECTRICALLY-PROPELLED VEHICLES
    • B60L2260/00—Operating Modes
    • B60L2260/40—Control modes
    • B60L2260/50—Control modes by future state prediction
    • B60L2260/54—Energy consumption estimation
    • 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
    • Y02D—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN INFORMATION AND COMMUNICATION TECHNOLOGIES [ICT], I.E. INFORMATION AND COMMUNICATION TECHNOLOGIES AIMING AT THE REDUCTION OF THEIR OWN ENERGY USE
    • Y02D10/00—Energy efficient computing, e.g. low power processors, power management or thermal management
    • 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

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Data Mining & Analysis (AREA)
  • General Physics & Mathematics (AREA)
  • General Engineering & Computer Science (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Artificial Intelligence (AREA)
  • Software Systems (AREA)
  • Evolutionary Computation (AREA)
  • Business, Economics & Management (AREA)
  • Mathematical Physics (AREA)
  • Computing Systems (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Databases & Information Systems (AREA)
  • General Health & Medical Sciences (AREA)
  • Computational Linguistics (AREA)
  • Biophysics (AREA)
  • Strategic Management (AREA)
  • Human Resources & Organizations (AREA)
  • Biomedical Technology (AREA)
  • Molecular Biology (AREA)
  • Economics (AREA)
  • Health & Medical Sciences (AREA)
  • Evolutionary Biology (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Mechanical Engineering (AREA)
  • Medical Informatics (AREA)
  • Power Engineering (AREA)
  • Sustainable Energy (AREA)
  • Development Economics (AREA)
  • Sustainable Development (AREA)
  • Game Theory and Decision Science (AREA)
  • Transportation (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Entrepreneurship & Innovation (AREA)
  • Marketing (AREA)
  • Operations Research (AREA)
  • Quality & Reliability (AREA)
  • Tourism & Hospitality (AREA)

Abstract

The invention discloses a method for estimating the driving range of a pure electric vehicle based on big data analysis, which comprises the following steps: acquiring original driving data of the electric automobile, and performing data preprocessing on the original driving data; establishing a target variable UM _ SOC of the running consumption SOC of the electric automobile under unit mileage, analyzing the correlation between the variable in the original data and the UM _ SOC, and analyzing the correlation weight between each correlation variable and the UM _ SOC; performing dimensionality reduction on the obtained original variable; according to the traffic condition on the current route, selecting a UM _ SOC big data analysis and prediction model in a proper mode, and predicting the SOC consumption value on the future driving route; calculating the driving range on the target trip route, and calculating the driving range on the predicted trip route by combining the current SOC value of the battery residual capacity and the UM _ SOC predicted value; and carrying out variance analysis on the predicted value and the true value of the SOC consumed by the electric automobile on the driving route, and carrying out secondary correction on the result. The driving range of the electric automobile in the future traveling process can be accurately predicted.

Description

Pure electric vehicle driving range estimation method based on big data analysis
Technical Field
The invention relates to a method for estimating the driving range of a pure electric vehicle based on big data analysis, and belongs to the field of driving safety of electric vehicles.
Background
In the current automotive industry, motorization is becoming the main trend in future automotive development. However, compared with the conventional fuel-powered automobile, the electric automobile still has many defects, which causes the market development of the electric automobile to encounter great obstacles. Among them, the insufficient driving range of the electric vehicle is one of the main defects. In addition, in the current industry, the estimation of the driving range of the electric automobile on future travel mostly depends on historical data, and the accuracy is insufficient. This results in the user being unable to be sure whether the electric vehicle can arrive at the destination as required during the trip, thereby causing "mileage anxiety" during the trip of driving the electric vehicle. The driving range of the electric vehicle is influenced by various factors, and how to accurately estimate the driving range of the pure electric vehicle by synthesizing related factors becomes a problem which is more concerned by the current electric vehicle industry.
The electric automobile driving range forecasting framework under pure data driving based on big data analysis fully utilizes related test data in the industry to quantify the importance degree of related influence factors of the driving range.
Most previous researches are used for estimating the driving range of the electric automobile based on physical modeling, and have the problems of high cost and high hardware dependence
In past research, the estimation of the driving range of the electric automobile is mostly based on physical modeling of the vehicle and a power battery. And calibrating the relevant parameters of the energy consumption model of the electric automobile and the relevant parameters of the state estimation model of the power battery by a researcher according to relevant experiments. In the actual vehicle running state, a researcher uses the collected original data of the relevant vehicle and the battery as the input of a model, and the prediction result of the driving range of the electric vehicle is obtained through model processing. However, in the process of model building, a researcher needs to perform a large number of calibration experiments, and a large amount of manpower and financial resources are needed. And the result obtained by calibration is only suitable for the same vehicle type, and has no mobility between different vehicle types. For this purpose, the driving range of the electric vehicle needs to be predicted from another dimension
Under the current industry development condition, historical data and real-time data of automobile driving are easy to obtain, and the automobile serving as an indispensable travel tool in the current society is developed towards becoming an individualized mobile travel terminal, so that a large number of sensors are mounted on the automobile to acquire driving data. In addition, the current communication industry is developed day by day, the realization of related concepts such as vehicle-vehicle communication, vehicle-road communication and the like is on schedule, and the data acquisition technology for vehicles and traffic conditions under the road condition is mature day by day. Therefore, researchers can conveniently obtain data related to automobile driving through a plurality of channels. In these available relational databases, data on driving conditions of the vehicle, weather, traffic information, and the like, which relate to the driving range, are included.
The big data analysis method is developed more and more mature, and the obtained driving range related data can be processed to obtain a result. Nowadays, with the development of artificial intelligence in the internet industry, a big data analysis method gradually permeates into various fields of automobile development. The large data analysis method is to separate from the physical condition constraint of the automobile related system, analyze the influence of each factor on the target variable from the angle of pure data, wherein typical representatives are various regression analysis algorithms in machine learning. Under the past conditions, due to the development level limitation of the computer industry, the computing capability of the vehicle-mounted computing platform is quite limited, and the variable prediction under pure data driving cannot be supported. However, with the leap-type development of hardware level and the increasing maturity of algorithms such as machine learning, it has not been a problem to carry a hardware platform with high computing power under real vehicle conditions to support the relevant machine learning algorithm. And all that is required for the construction of the relevant machine learning framework is the various vehicle driving data mentioned above. According to the analysis, the establishment of the driving range prediction framework of the pure data driven electric automobile has no obstacles in software, hardware and required data materials. The driving range estimation under pure data drive established according to the machine learning algorithm is not only low in cost, but also has self-correcting capability.
The driving range estimation algorithm based on pure data drive established according to the big data mining method is separated from an electric automobile hardware system, stands on the angle of the whole general electric automobile, and estimates the driving range of the electric automobile by using related variables. The whole algorithm framework is built based on the existing original data, the cost is low, the actual experiment calibration needed is less, and the cost in the algorithm development process is low. The established algorithm framework can be migrated in electric automobile BMSs of various models, and can achieve higher precision only by carrying out less self-adaptive correction without being limited by specific electric automobile models. In the using process of the user, the trip data of the user can be stored at the vehicle end to serve as a learning sample of the algorithm frame, and the prediction accuracy of the algorithm is gradually improved along with the increase of the number of the samples. The accurate driving range estimation is beneficial to reducing the 'travel anxiety' of the user on the electric automobile and accelerating the popularization of the electric automobile market.
Disclosure of Invention
The invention designs and develops a pure electric vehicle driving range estimation method based on big data analysis, which utilizes the collected real-time data of relevant variables in the actual driving process of an electric vehicle and achieves the purpose of reducing the 'range anxiety' of a driver to the electric vehicle according to an effective machine learning algorithm trained offline.
The technical scheme provided by the invention is as follows:
a pure electric vehicle driving range estimation method based on big data analysis comprises the following steps:
acquiring original driving data of the electric automobile, and performing data preprocessing on the original driving data;
establishing a target variable UM _ SOC of the running consumption SOC of the electric automobile under unit mileage, analyzing the correlation between the variable in the original data and the UM _ SOC, and analyzing the correlation weight between each correlation variable and the UM _ SOC;
carrying out dimensionality reduction treatment on the original variable subjected to correlation analysis;
calculating the driving range on the target trip route, and calculating the driving range on the predicted trip route by combining the current SOC value of the battery residual capacity and the UM _ SOC predicted value;
and carrying out variance analysis on the predicted value and the true value of the SOC consumed by the electric automobile on the driving route, and carrying out secondary correction on the result.
It is preferable that the first and second liquid crystal layers are formed of,
the relevant weight expression between each relevant variable and UM _ SOC is as follows:
Figure BDA0002937875970000031
in the equation, dist _ car represents an actual distance traveled by the electric vehicle within the sampled data time interval, and Δ SOC represents a value of an amount of charge consumed by the electric vehicle when traveling the dist _ car distance.
Preferably, the method is characterized in that,
the dimension reduction processing comprises the following steps:
performing low variance filtering and high correlation filtering on the original variables subjected to correlation analysis by adopting a variable dimension reduction method, and excluding unrepresentative original data and original data with high correlation between the original variables; and after the correlation weight analysis is carried out, carrying out dimension reduction processing on the feature data set according to the analysis result, and reserving a variable with a higher correlation weight as a pre-operation data set built by a machine learning data model.
Preferably, the method further includes:
on the basis of an original data set, the traffic condition is divided into three levels: unobstructed, relatively congested and congested;
and dividing the historical driving route into a road section data subset according to the traffic condition, training, and outputting a variable as a UM _ SOC predicted value under the condition of the electric automobile branch section.
Preferably, a BP neural network model is selected as the prediction model, and the model training includes:
selecting an input variable of a neural network, and determining the number m of input layer neural nodes;
determining the number of hidden layers and the number of nodes of each hidden layer;
determining a learning rate, an initial network weight and an initial threshold of the model;
determining the number of output layer neural nodes;
determining a training set and a test set classification of an original data subset under the condition of traffic classes;
training a BP neural network;
and (3) carrying out effectiveness analysis on the established model, and setting an inspection standard formula:
Figure BDA0002937875970000041
when e is less than 0.08, the big data analysis model is good in prediction condition;
when e is more than 0.15, retraining the model;
in the formula, riRepresenting the prediction of a big data analysis model, aiRepresenting the driving range of the electric automobile under the actual condition, and the check standard e is an infinite quantityA class decimal representing the average value of the running errors of the n test data; when the error e is less than 0.08, the big data analysis model is good in prediction condition; when the error e is larger than 0.15, retraining the model;
the method for predicting the SOC consumption of the travel route by using the established BP neural network model comprises the following steps:
dividing a trip route into a plurality of road sections, taking variable data recorded in a current vehicle as input variables, and predicting the unit mileage electric quantity consumption set { S ] of each divided road section on line through a BP neural network model1,S2......SlL is the number of divided road sections, and the mileage length set of each road section is { D }1,D2......Dl}。
It is preferable that the first and second liquid crystal layers are formed of,
the calculating of the driving range on the target travel route comprises:
acquiring the state of charge (SOC) of the battery of the electric automobile at the current momentnowAcquiring the unit mileage electric quantity consumption set and each road section mileage length set under the road section, and comparing the SOCnowAnd SOClowestAnd (4) judging:
when SOC is reachednow≤SOClowestIf so, the electric automobile cannot run safely in the current state;
when SOC is reachednow＞SOClowestAnd indicating that the electric automobile can continue to run under the current state.
The invention has the following beneficial effects:
1. the method integrates the influence of multiple factors into one frame, classifies the road sections according to the traffic conditions, and considers the influence of human factors related to drivers, so that the prediction result is more real.
2. The electric automobile driving range estimation algorithm based on big data analysis breaks away from the constraint of a vehicle physical framework, so that enterprises can reduce a large number of experiments in the aspect of development of related modules, and manpower and financial resources are saved. Moreover, the prediction frame can be properly transferred among different types of vehicle models, and the model has good applicability;
drawings
Fig. 1 is a diagram of dynamic estimation factor analysis of the driving range of the human-vehicle-road-ring-battery according to the present invention.
FIG. 2 is a schematic diagram of a box-type distribution of data according to the present invention.
Detailed Description
The present invention is further described in detail below with reference to the attached drawings so that those skilled in the art can implement the invention by referring to the description text.
As shown in fig. 1-2, the present invention provides a pure electric vehicle driving range based on big data analysis, and relevant influence factors of an electric vehicle during a traveling process include factors such as weather conditions, traveling routes, vehicle structures, traffic conditions, driving and charging habits of a driver, and a battery, and these factors can be roughly classified into relevant categories such as the driver, the vehicle, the road, and the environment, so as to form an electric vehicle driving range analysis framework under the multi-factor information of "human-vehicle-road-ring-battery".
As is known, the driving range of an electric vehicle depends mainly on two aspects, namely the energy consumption of the vehicle and the remaining energy of the battery. Since the automobile is a complex nonlinear time-varying system, the energy consumption during driving depends on a plurality of physical quantities, and the physical quantities are mostly in random time-varying states, such as the driving speed of the automobile, the road traffic flow and the like, it is difficult to establish a clear functional relationship between the driving energy consumption of the automobile and a plurality of variables. The reason for this is that there is a contradiction between the accuracy of the battery physical model and the algorithm efficiency in the actual vehicle state. Therefore, in order to get rid of the inefficient description of the complex physical model with high precision, researchers can utilize a large number of real databases of vehicles in the actual running process to search the relationship between a plurality of state variables and the driving range of the electric automobile from the pure data perspective.
The data fusion technology is a technology for comprehensively processing and optimizing the acquisition, representation and internal connection of various kinds of information. The data information fusion technology is used for processing and integrating from the view angle of multi-information to obtain the internal relation and rule of various information, thereby eliminating useless and wrong information, reserving correct and useful components and finally realizing the optimization of the information. The data packet in a single aspect can only obtain partial information segments of the object to be tested, and the multi-data information can perfectly and accurately reflect the target characteristics after being fused. The information fusion adopts methods such as artificial neural network and random forest. The development direction of data fusion is to synthesize and correlate information of different properties of nonlinear and complex environmental factors and describe targets from different angles. The driving range of the electric automobile has strong nonlinear characteristics with all relevant factors, and the driving range of the electric automobile can be well predicted by integrating data of all the factors, and the method specifically comprises the following steps:
acquiring original driving data of the electric automobile, and performing data preprocessing on the original driving data;
establishing a target variable UM _ SOC of the running consumption SOC of the electric automobile under unit mileage, analyzing the correlation between the variable in the original data and the UM _ SOC, and analyzing the correlation weight between each correlation variable and the UM _ SOC;
performing dimensionality reduction on the obtained original variable;
according to the traffic condition of each road section on the current route, selecting a UM _ SOC big data analysis and prediction model under the corresponding traffic condition, and predicting the SOC consumption value on the future driving route;
calculating the driving range on the target trip route, and calculating the driving range on the predicted trip route by combining the current SOC value of the battery residual capacity and the UM _ SOC predicted value;
and carrying out error analysis on the predicted value and the real value of the SOC consumed by the electric automobile on the driving route, firstly obtaining the difference value between the predicted value and the real value, carrying out correlation analysis on the difference value and each feature, selecting the feature of the top five positions of the correlation ranking, and establishing a multiple linear regression model between the feature and the difference value. And then, taking the future characteristic value as model input, and outputting a prediction result value of the difference value for correcting the SOC consumption value.
Collecting original data: the method comprises the steps of collecting driving condition data and battery operation data of a plurality of electric vehicles by using an open source database, and preprocessing data of an original data set.
Introducing key target variables and carrying out correlation analysis: and establishing a target variable UM _ SOC of the running consumption SOC of the electric automobile under the unit mileage, and analyzing the correlation between the variable in the original data set and the UM _ SOC.
Performing dimensionality reduction on the acquired multiple original variables: and eliminating the original variable based on the obtained correlation coefficient to achieve the purpose of reducing the dimension.
UM _ SOC target variable online prediction: and dividing the historical data set into a plurality of subsets according to the traffic conditions for model training. And selecting a UM _ SOC big data analysis and prediction model in a proper mode based on the real-time working condition information recorded by the vehicle-mounted terminal in the driving process of the electric vehicle according to the traffic condition on the current route, and predicting the SOC consumption value on the future driving route.
Calculating the driving range on the target travel route: and calculating the driving range on the predicted travel route by combining the current battery residual capacity and the UM _ SOC predicted value.
And carrying out variance analysis on the predicted value and the true value of the SOC consumed by the electric automobile on the driving route, and carrying out secondary correction on the result.
Firstly, acquiring a related original database, and processing open source data in a database on a network by using a related program, wherein the open source data comprises a GB32960 related database, a traffic information database, a weather condition database and the like, and the open source data comprises related information such as vehicle state, charging state, vehicle speed, accumulated mileage, total voltage, total current, SOC and the like. After the original data are acquired, the original data need to be cleaned and preprocessed, wherein the processing principle comprises the principles of eliminating data in the vehicle charging process, eliminating data with the interval of data acquired by the vehicle being less than 10 seconds, eliminating abnormal data in the acquisition process, eliminating excessive data lost in the acquisition process, performing interpolation processing on partial missing data, and the like. The preprocessed data is saved in a csv file format. The method of exception data culling is as follows:
and checking the data distribution condition based on the box type graph, and removing abnormal data. Points that deviate significantly from the upper and lower edges of the boxed graph are considered as outliers and should be rejected.
Box graph top edge definition: xU＝QU+1.5I
Box graph lower edge definition: xL＝QL-1.5I
Wherein I represents the upper and lower quartile range, i.e. I ═ QU-QL
And secondly, introducing an electric vehicle consumption SOC concept (UM _ SOC) under unit mileage, and analyzing the correlation weight between each correlation variable and the UM _ SOC by using a correlation analysis method on the preprocessed data.
Figure BDA0002937875970000081
Where dist _ car represents an actual mileage traveled by the electric vehicle over a certain time interval, and Δ SOC represents a value of an amount of charge consumed by the electric vehicle for the distance traveled dist _ car. The variable integrates the related factors such as the energy consumption of the whole vehicle, the battery and the like, and the electric energy consumption condition of the vehicle in the driving process is intuitively reflected.
Variable correlation analysis method: because the correlation analysis is to analyze all data items of the sample, and the result of the correlation analysis does not need to be popularized outwards, a fixed effect regression analysis method is adopted.
The specific process of the fixed effect regression analysis is as follows:
UM_SOCit＝θ+α·Rit+β·Dit+γ·Cit+δ·Vi+η·St；
in the formula UM _ SOCitRepresenting the SOC of the unit mileage consumption of the ith sample vehicle at the sampling time of t, and each parameter of theta, alpha, beta, gamma, delta and eta is a characteristic Rit、Dit、Cit、Vi、StAnd UM _ SOCitThe regression coefficient of the regression equation of (1). Wherein R isitRepresenting the road condition row vector changed along with time and individuals, including the average speed, the maximum speed, the standard deviation of the speed, DitRepresentation over time and individualsThe line vector of the changed driving behavior, including the maximum acceleration, maximum deceleration, acceleration standard deviation, average acceleration, average deceleration, average jerk, average deceleration of the vehicle within the driving segment, CitA row vector representing the charging behaviour of the user, varying with time and individually, comprising the average power and the standard deviation of the power, V, of the last charging before the driving sectioniRepresenting the vehicle' S own attribute row vector, S, which does not change over time, which changes with the individualtLine vectors representing attributes of the external environment that change over time, including temperature and season
Elimination of V using averagingiInfluence of the Properties:
Figure BDA0002937875970000082
in the formula (I), the compound is shown in the specification,
Figure BDA0002937875970000091
represents the unit mileage consumption SOC of the ith sample vehicle within the sampling time,
Figure BDA0002937875970000092
represents the average value of the road condition row vector varying with individuals in the sampling time,
Figure BDA0002937875970000093
the average over the sampling time of a row vector representing the individual-varying driving behavior,
Figure BDA0002937875970000094
the average over the sampling time of a row vector representing the charging behavior of the user as a function of the individual,
Figure BDA0002937875970000095
a row vector representing the attributes of the external environment, including temperature and season, is averaged over the sampling time.
And constructing a multiple linear regression model based on the correlation coefficient, and calculating the correlation coefficient of each factor.
And thirdly, performing low variance filtering and high correlation filtering on the original variables by using a variable dimension reduction method, and removing unrepresentative original data and original data with high correlation between the original variables from the original data. And then, performing dimensionality reduction on a plurality of variables according to the correlation weight analysis result in the second step, and reserving the variables with higher correlation weights for building a specific machine learning data model.
Low variance filtering: firstly, normalization processing is carried out, then the variance of the variable is solved, and the variable with the variance smaller than a certain limit threshold value is removed;
and (3) high correlation filtering: and when the correlation coefficient between the two variables is more than 0.8, rejecting one group of variables.
And arranging the related variables from large to small according to the correlation coefficient, and selecting the eight variables which are the top ones as an input data set constructed by a machine learning data model.
And fourthly, establishing a prediction model of the electric automobile UM _ SOC according to the following steps, and dividing the traffic condition into 3 levels, namely smoothness, congestion and congestion on the basis of the original data set. And dividing the historical driving route into a road section data subset according to the traffic condition, and training on the basis of the subset. And selecting a mature BP neural network model by the prediction model. And (3) the input variable of the model is the variable obtained in the step 3) by dimensionality reduction, and the output variable is the predicted value of UM _ SOC under the condition of the shunt section of the electric automobile.
The model training specifically comprises:
selecting an input variable of a neural network, and determining the number m of input layer neural nodes;
determining the number of hidden layers and the number of nodes of each hidden layer;
determining learning rates, initial network weights, and initial thresholds for a model
Determining the number of output layer neural nodes;
determining training set and test set classifications for raw data subsets in traffic class-by-traffic conditions
Training BP neural network
And carrying out effectiveness analysis on the established model. Setting a checking standard
Figure BDA0002937875970000101
Wherein r isiRepresenting the prediction of a big data analysis model, aiThe driving range of the electric automobile under the actual condition is represented, and the inspection standard e is a dimensionless decimal and represents the average value of the driving errors of the n driving ranges. If the error is 0<e<0.08, the big data analysis model is good in prediction. If the error e>0.15, the model is retrained. Predicting SOC consumption of the travel route by using the established BP neural network prediction model for traffic grades:
predicting SOC consumption of the travel route by using the established BP neural network prediction model for traffic grades:
based on real-time traffic information on a network platform, path points with the same traffic level characteristics on a pre-trip route are divided into a road section, and the whole route is divided into a plurality of road sections.
Using the data values of all variables recorded by the current vehicle-mounted terminal as input variables of the model, and predicting the unit mileage electric quantity consumption set { S ] of each road section on line through a BP neural network model1,S2......SlL is the number of divided road sections, and the mileage length of each road section is recorded as { D }1,D2......Dl}。
And fifthly, calculating the driving range on the current travel route:
acquiring state of charge (SOC) of electric vehicle battery at current moment through BMS system of electric vehiclenow. Obtaining the unit mileage electric quantity consumption [ S ] under the shunt segment obtained in the previous step1,S2......Sl]And mileage length of each road section [ D ]1,D2......Dl]。
First, the SOC is determinednowWhether it is greater than SOClowest，SOClowestElectric quantity is minimum for ensuring safe driving of electric automobileAnd (4) a threshold value. If SOCnow≤SOClowestIf so, the electric automobile cannot run safely in the current state;
if SOCnow＞SOClowestAnd indicating that the electric automobile can continue to run under the current state. Then, the condition SOC is judgednow-S1×D1＞SOClowestIf the determination is true, the first element of the driving range set is recorded as RDD1And ending the driving range estimation:
Figure BDA0002937875970000102
[S1,S2......Sl]is the unit mileage electric quantity consumption vector of each road section under the shunt section [ D1,D2......Dl]And the mileage length vector of each road section.
If yes, recording the first element of the driving range set as RDD1＝S1And judging the condition SOCnow-S1×D1-S2×D2＞SOClowestIf the determination is true, the second element of the driving range set is recorded as RDD2And ending the driving range estimation:
Figure BDA0002937875970000111
if yes, recording the second element of the driving range set as RDD2＝S2,... so on, until the nth road section, judging the condition SOCnow-S1×D1-S2×D2-...-Sl×Dl＞SOClowestIf the determination is positive, the nth element of the driving range set is recorded as RDDnAnd ending the driving range estimation:
Figure BDA0002937875970000112
if yes, recording the second element of the driving range set as RDDn＝SnAnd finishes the driving range estimation.
The predicted driving range of the electric automobile is the sum of all elements in the set RDD.
Figure BDA0002937875970000113
Where k is the number of elements in the set RDD.
And sixthly, analyzing the difference value between the predicted value of the driving range of the electric automobile in the original data test set and the verification set and the driving range value under the real condition. Firstly, a difference value between a predicted value and a true value is obtained, correlation analysis is carried out between the difference value and each feature, the features of the top five places of the correlation ranking are selected, and a multiple linear regression model between the features and the difference value is established. And then, taking the future characteristic value as model input, and outputting a prediction result value of the difference value for correcting the SOC consumption value.
While embodiments of the invention have been described above, it is not limited to the applications set forth in the description and the embodiments, which are fully applicable in various fields of endeavor to which the invention pertains, and further modifications may readily be made by those skilled in the art, it being understood that the invention is not limited to the details shown and described herein without departing from the general concept defined by the appended claims and their equivalents.

Claims (6)

1. A pure electric vehicle driving range estimation method based on big data analysis is characterized by comprising the following steps:
acquiring original driving data of the electric automobile, and performing data preprocessing on the original driving data;
establishing a target variable UM _ SOC of the running consumption SOC of the electric automobile under unit mileage, analyzing the correlation between the variable in the original data and the UM _ SOC, and analyzing the correlation weight between each correlation variable and the UM _ SOC;
carrying out dimensionality reduction treatment on the original variable subjected to correlation analysis;
calculating the driving range on the target trip route, and calculating the driving range on the predicted trip route by combining the current SOC value of the battery residual capacity and the UM _ SOC predicted value;
and carrying out variance analysis on the predicted value and the true value of the SOC consumed by the electric automobile on the driving route, and carrying out secondary correction on the result.
2. The pure electric vehicle range estimation method based on big data analysis according to claim 1,
the relevant weight expression between each relevant variable and UM _ SOC is as follows:
Figure FDA0002937875960000011
in the equation, dist _ car represents an actual distance traveled by the electric vehicle within the sampled data time interval, and Δ SOC represents a value of an amount of charge consumed by the electric vehicle when traveling the dist _ car distance.
3. The pure electric vehicle range estimation method based on big data analysis according to claim 2,
the dimension reduction processing comprises the following steps:
performing low variance filtering and high correlation filtering on the original variables subjected to correlation analysis by adopting a variable dimension reduction method, and excluding unrepresentative original data and original data with high correlation between the original variables; and after the correlation weight analysis is carried out, carrying out dimension reduction processing on the feature data set according to the analysis result, and reserving a variable with a higher correlation weight as a pre-operation data set built by a machine learning data model.
4. The pure electric vehicle range estimation method based on big data analysis according to claim 3, further comprising:
on the basis of an original data set, the traffic condition is divided into three levels: unobstructed, relatively congested and congested;
and dividing the historical driving route into a road section data subset according to the traffic condition, training, and outputting a variable as a UM _ SOC predicted value under the condition of the electric automobile branch section.
5. The pure electric vehicle range estimation method based on big data analysis according to claim 4, wherein a BP neural network model is selected as a prediction model, and model training comprises:
selecting an input variable of a neural network, and determining the number m of input layer neural nodes;
determining the number of hidden layers and the number of nodes of each hidden layer;
determining a learning rate, an initial network weight and an initial threshold of the model;
determining the number of output layer neural nodes;
determining a training set and a test set classification of an original data subset under the condition of traffic classes;
training a BP neural network;
and (3) carrying out effectiveness analysis on the established model, and setting an inspection standard formula:
Figure FDA0002937875960000021
when e is less than 0.08, the big data analysis model is good in prediction condition;
when e is more than 0.15, retraining the model;
in the formula, riRepresenting the prediction of a big data analysis model, aiThe driving range of the electric automobile under the actual condition is represented, and the inspection standard e is a dimensionless decimal and represents the average value of the driving errors of n test data; when the error e is less than 0.08, the big data analysis model is good in prediction condition; when the error e is larger than 0.15, retraining the model;
the method for predicting the SOC consumption of the travel route by using the established BP neural network model comprises the following steps:
dividing a trip route into a plurality of road sections, taking variable data recorded in a current vehicle as input variables, and predicting the unit mileage electric quantity consumption set { S ] of each divided road section on line through a BP neural network model1,S2......SlL is the number of divided road sections, and the mileage length set of each road section is { D }1,D2......Dl}。
6. The pure electric vehicle range estimation method based on big data analysis according to claim 5,
the calculating of the driving range on the target travel route comprises:
acquiring the state of charge (SOC) of the battery of the electric automobile at the current momentnowAcquiring the unit mileage electric quantity consumption set and each road section mileage length set under the road section, and comparing the SOCnowAnd SOClowestAnd (4) judging:
when SOC is reachednow≤SOClowestIf so, the electric automobile cannot run safely in the current state;
when SOC is reachednow＞SOClowestAnd indicating that the electric automobile can continue to run under the current state.
CN202110167243.4A 2021-02-07 2021-02-07 Pure electric vehicle driving range estimation method based on big data analysis Pending CN112860782A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202110167243.4A CN112860782A (en) 2021-02-07 2021-02-07 Pure electric vehicle driving range estimation method based on big data analysis

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202110167243.4A CN112860782A (en) 2021-02-07 2021-02-07 Pure electric vehicle driving range estimation method based on big data analysis

Publications (1)

Publication Number Publication Date
CN112860782A true CN112860782A (en) 2021-05-28

Family

ID=75988876

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202110167243.4A Pending CN112860782A (en) 2021-02-07 2021-02-07 Pure electric vehicle driving range estimation method based on big data analysis

Country Status (1)

Country Link
CN (1) CN112860782A (en)

Cited By (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN114274840A (en) * 2022-01-07 2022-04-05 江苏大学 Intelligent internet electric vehicle electric quantity consumption prediction method and system based on S-CRANN
CN114940132A (en) * 2022-07-27 2022-08-26 中汽研汽车检验中心（天津）有限公司 Electric vehicle endurance mileage prediction method, test method and system
WO2023159562A1 (en) * 2022-02-28 2023-08-31 华为技术有限公司 Driving range determination method and apparatus and vehicle

Citations (8)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
KR20020088708A (en) * 2001-05-21 2002-11-29 황재호 Motor Vehicle Management System Based on Fuel Charge Information by Using the Oil Company Card
JP2007026351A (en) * 2005-07-21 2007-02-01 Sumitomo Wiring Syst Ltd Method and system for managing vehicle information
CN103660984A (en) * 2013-11-26 2014-03-26 北京航空航天大学 BEV driving range estimation device and method based on remote data transmission
CN105904981A (en) * 2016-04-07 2016-08-31 北京现代汽车有限公司 Electric car driving mileage estimation control method and device, and vehicle control unit
CN106427589A (en) * 2016-10-17 2017-02-22 江苏大学 Electric car driving range estimation method based on prediction of working condition and fuzzy energy consumption
CN107933317A (en) * 2017-10-18 2018-04-20 宝沃汽车（中国）有限公司 Method, apparatus, equipment and the pure electric automobile of the remaining continual mileage of estimation
CN110667434A (en) * 2019-09-11 2020-01-10 南京航空航天大学 Working condition-adaptive pure electric vehicle driving mileage estimation method and system
CN111439128A (en) * 2020-03-12 2020-07-24 广汽蔚来新能源汽车科技有限公司 Electric vehicle remaining mileage estimation method and device and computer equipment

Patent Citations (8)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
KR20020088708A (en) * 2001-05-21 2002-11-29 황재호 Motor Vehicle Management System Based on Fuel Charge Information by Using the Oil Company Card
JP2007026351A (en) * 2005-07-21 2007-02-01 Sumitomo Wiring Syst Ltd Method and system for managing vehicle information
CN103660984A (en) * 2013-11-26 2014-03-26 北京航空航天大学 BEV driving range estimation device and method based on remote data transmission
CN105904981A (en) * 2016-04-07 2016-08-31 北京现代汽车有限公司 Electric car driving mileage estimation control method and device, and vehicle control unit
CN106427589A (en) * 2016-10-17 2017-02-22 江苏大学 Electric car driving range estimation method based on prediction of working condition and fuzzy energy consumption
CN107933317A (en) * 2017-10-18 2018-04-20 宝沃汽车（中国）有限公司 Method, apparatus, equipment and the pure electric automobile of the remaining continual mileage of estimation
CN110667434A (en) * 2019-09-11 2020-01-10 南京航空航天大学 Working condition-adaptive pure electric vehicle driving mileage estimation method and system
CN111439128A (en) * 2020-03-12 2020-07-24 广汽蔚来新能源汽车科技有限公司 Electric vehicle remaining mileage estimation method and device and computer equipment

Cited By (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN114274840A (en) * 2022-01-07 2022-04-05 江苏大学 Intelligent internet electric vehicle electric quantity consumption prediction method and system based on S-CRANN
CN114274840B (en) * 2022-01-07 2022-11-18 江苏大学 Intelligent internet electric vehicle electric quantity consumption prediction method and system based on S-CRANN
WO2023159562A1 (en) * 2022-02-28 2023-08-31 华为技术有限公司 Driving range determination method and apparatus and vehicle
CN114940132A (en) * 2022-07-27 2022-08-26 中汽研汽车检验中心（天津）有限公司 Electric vehicle endurance mileage prediction method, test method and system

Similar Documents

Publication Publication Date Title
CN111452619B (en) Online energy consumption prediction method and system for electric vehicle
CN107045788B (en) Traffic road condition prediction method and device
CN108806021B (en) Electric vehicle target road section energy consumption prediction method based on physical model and road characteristic parameters
CN108269401B (en) Data-driven viaduct traffic jam prediction method
CN111301426B (en) Method for predicting energy consumption in future driving process based on GRU network model
CN112860782A (en) Pure electric vehicle driving range estimation method based on big data analysis
CN111832225B (en) Method for constructing driving condition of automobile
JP6961012B2 (en) Customized safe speed determination for the vehicle
CN108556682B (en) Driving range prediction method, device and equipment
CN111738337B (en) Driver distraction state detection and identification method in mixed traffic environment
CN111191824B (en) Power battery capacity attenuation prediction method and system
CN109697854A (en) The urban road traffic state assessment method of various dimensions
CN113335131B (en) Vehicle endurance mileage prediction method, device, equipment and storage medium
CN110738855A (en) Road traffic flow condition prediction method for data sparse time periods
CN113379099B (en) Machine learning and copula model-based highway traffic flow self-adaptive prediction method
CN116523093A (en) Grid demand sensing system and method of energy system based on random source load prediction
CN113642768A (en) Vehicle running energy consumption prediction method based on working condition reconstruction
CN115907076A (en) Mountain city charging load probability prediction system and method
CN113674524A (en) LSTM-GASVR-based multi-scale short-time traffic flow prediction modeling and prediction method and system
CN117238126A (en) Traffic accident risk assessment method under continuous flow road scene
CN112036598A (en) Charging pile use information prediction method based on multi-information coupling
Zhang et al. Neural network based vehicle speed prediction for specific urban driving
CN115830855B (en) Highway section transit time prediction method based on double-state division
CN116151472A (en) Electric automobile short-term energy consumption prediction method based on Internet of vehicles data mining
CN113255958B (en) Urban rail transit passenger flow volume change interval prediction method for describing fluctuation characteristics

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
RJ01 Rejection of invention patent application after publication
RJ01 Rejection of invention patent application after publication

Application publication date: 20210528