CN109131350A - A kind of hybrid vehicle energy management method and system - Google Patents
A kind of hybrid vehicle energy management method and system Download PDFInfo
- Publication number
- CN109131350A CN109131350A CN201810964452.XA CN201810964452A CN109131350A CN 109131350 A CN109131350 A CN 109131350A CN 201810964452 A CN201810964452 A CN 201810964452A CN 109131350 A CN109131350 A CN 109131350A
- Authority
- CN
- China
- Prior art keywords
- model
- prediction
- speed
- autoregression
- gradient
- 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.)
- Granted
Links
Classifications
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B60—VEHICLES IN GENERAL
- B60W—CONJOINT CONTROL OF VEHICLE SUB-UNITS OF DIFFERENT TYPE OR DIFFERENT FUNCTION; CONTROL SYSTEMS SPECIALLY ADAPTED FOR HYBRID VEHICLES; ROAD VEHICLE DRIVE CONTROL SYSTEMS FOR PURPOSES NOT RELATED TO THE CONTROL OF A PARTICULAR SUB-UNIT
- B60W50/00—Details of control systems for road vehicle drive control not related to the control of a particular sub-unit, e.g. process diagnostic or vehicle driver interfaces
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B60—VEHICLES IN GENERAL
- B60W—CONJOINT CONTROL OF VEHICLE SUB-UNITS OF DIFFERENT TYPE OR DIFFERENT FUNCTION; CONTROL SYSTEMS SPECIALLY ADAPTED FOR HYBRID VEHICLES; ROAD VEHICLE DRIVE CONTROL SYSTEMS FOR PURPOSES NOT RELATED TO THE CONTROL OF A PARTICULAR SUB-UNIT
- B60W40/00—Estimation or calculation of non-directly measurable driving parameters for road vehicle drive control systems not related to the control of a particular sub unit, e.g. by using mathematical models
- B60W40/02—Estimation or calculation of non-directly measurable driving parameters for road vehicle drive control systems not related to the control of a particular sub unit, e.g. by using mathematical models related to ambient conditions
- B60W40/06—Road conditions
- B60W40/076—Slope angle of the road
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B60—VEHICLES IN GENERAL
- B60W—CONJOINT CONTROL OF VEHICLE SUB-UNITS OF DIFFERENT TYPE OR DIFFERENT FUNCTION; CONTROL SYSTEMS SPECIALLY ADAPTED FOR HYBRID VEHICLES; ROAD VEHICLE DRIVE CONTROL SYSTEMS FOR PURPOSES NOT RELATED TO THE CONTROL OF A PARTICULAR SUB-UNIT
- B60W40/00—Estimation or calculation of non-directly measurable driving parameters for road vehicle drive control systems not related to the control of a particular sub unit, e.g. by using mathematical models
- B60W40/10—Estimation or calculation of non-directly measurable driving parameters for road vehicle drive control systems not related to the control of a particular sub unit, e.g. by using mathematical models related to vehicle motion
- B60W40/105—Speed
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B60—VEHICLES IN GENERAL
- B60W—CONJOINT CONTROL OF VEHICLE SUB-UNITS OF DIFFERENT TYPE OR DIFFERENT FUNCTION; CONTROL SYSTEMS SPECIALLY ADAPTED FOR HYBRID VEHICLES; ROAD VEHICLE DRIVE CONTROL SYSTEMS FOR PURPOSES NOT RELATED TO THE CONTROL OF A PARTICULAR SUB-UNIT
- B60W50/00—Details of control systems for road vehicle drive control not related to the control of a particular sub-unit, e.g. process diagnostic or vehicle driver interfaces
- B60W2050/0001—Details of the control system
- B60W2050/0019—Control system elements or transfer functions
- B60W2050/002—Integrating means
Landscapes
- Engineering & Computer Science (AREA)
- Automation & Control Theory (AREA)
- Transportation (AREA)
- Mechanical Engineering (AREA)
- Physics & Mathematics (AREA)
- Mathematical Physics (AREA)
- Human Computer Interaction (AREA)
- Traffic Control Systems (AREA)
- Hybrid Electric Vehicles (AREA)
- Electric Propulsion And Braking For Vehicles (AREA)
Abstract
The present invention discloses a kind of hybrid vehicle energy management method and system.This method comprises: predicting using the neural network based on history speed and driving behavior training the following speed, prediction speed is obtained；Road grade is predicted using the gradient prediction model for integrating moving average model(MA model) based on autoregression that the road grade data by acquisition are established, obtains prediction road grade；According to prediction speed and prediction road slope calculation demand power；Power utilization dynamic programming algorithm calculates the torque and revolving speed of each power part according to demand.Hybrid vehicle energy management method of the invention and system, can be improved fuel economy.
Description
Technical field
The present invention relates to technical field of automobile control, more particularly to a kind of hybrid vehicle energy management method and are
System.
Background technique
It restructures the use of energy, develops and uses new energy and power-saving technology to reduce the dependence to non-renewable energy resources, subtract
Few atmosphere pollution has become the common recognition of various countries' development.In recent years, the automobile demand that China increases rapidly is brought to consumption of petroleum
Sharp increase, while the energy security problem for also being faced China is more prominent.However fuel economy is improved now
Method is limited only to improve the hardware technology of automobile to improve the utilization rate of petroleum, however improves petroleum by improving hardware
The effect of the method for utilization rate is not obvious.
Summary of the invention
The object of the present invention is to provide a kind of hybrid vehicle energy management method and systems, improve fuel economy.
To achieve the above object, the present invention provides following schemes:
A kind of hybrid vehicle energy management method, comprising:
The following speed is predicted using the neural network based on history speed and driving behavior training, is predicted
Speed；
Utilize the gradient that moving average model(MA model) is integrated based on autoregression of the road grade data by acquisition established
Prediction model predicts road grade, obtains prediction road grade；
According to the prediction speed and the prediction road slope calculation demand power；
The torque and revolving speed of each power part are calculated using dynamic programming algorithm according to the demand power.
Optionally, the training process of neural network based on history speed and driving behavior training includes:
Obtain history speed of each vehicular data acquisition device automobile collected in each road condition downward driving
And driving behavior；
By the history speed and the driving behavior, segment length is divided at preset timed intervals, obtain input sample and
Export sample；
Neural network is trained using the input sample and output sample, obtains trained neural network.
Optionally, neural network is trained using the input sample and output sample described, is trained
Neural network after, further includes:
The history speed and the driving behavior are updated in real time；
The parameter of the trained neural network is adjusted using updated history speed and driving behavior.
Optionally, the gradient for integrating moving average model(MA model) based on autoregression by the foundation of the road grade data of acquisition is pre-
Survey model detailed process include:
By linear relationship between any one sampled point in road grade data and the sampling point sequence before the sampled point
It is established as the p rank autoregression part of autoregression integral moving average model(MA model)；
The linear weighted function relationship of the white noise of road grade data kind is established as to the q of autoregression integral moving average model(MA model)
Rank rolling average part；
It is that autoregression integrates returning certainly for moving average model(MA model) by the autoregression part and the rolling average thin consolidation
Return rolling average part；
Parameter calculating is carried out to autoregression integral moving average model(MA model), determines autoregression integral moving average model(MA model)
P value, d value and q value, obtain gradient prediction model；Wherein p is the order of autoregression part, and q is the order of rolling average part, d
For non-stationary value of slope it is Sequence Transformed be difference number required for stationary sequence.
Optionally, it is described using by acquisition road grade data established based on autoregression integrate rolling average
The gradient prediction model of model predicts road grade, obtains prediction road grade, specifically includes:
By the prediction speed multiplied by the prediction speed corresponding time, Prediction distance is obtained；
Gradient prediction is carried out according to the Prediction distance using the gradient prediction model, obtains prediction road grade.
Invention additionally discloses a kind of Energy Management System for Hybrid Electric Vehicle, comprising:
Speed prediction module, for utilizing the neural network based on history speed and driving behavior training to the following speed
It is predicted, obtains prediction speed；
Gradient prediction module, for using by acquisition road grade data established based on autoregression integrate move
The gradient prediction model of dynamic averaging model predicts road grade, obtains prediction road grade；
Power computation module, for according to the prediction speed and the prediction road slope calculation demand power；
Power distribution module, for calculating the torsion of each power part using dynamic programming algorithm according to the demand power
Square and revolving speed.
Optionally, which further includes neural metwork training module, for being based on history
Speed and driving behavior training neural network；The neural metwork training module includes:
Data capture unit, for obtaining each vehicular data acquisition device automobile collected under each road condition
History speed and driving behavior when driving；
Sample division unit, for segment length to be drawn at preset timed intervals by the history speed and the driving behavior
Point, obtain input sample and output sample；
Training unit is trained for being trained using the input sample and output sample to neural network
Neural network.
Optionally, the neural metwork training module further include:
Data updating unit, for being updated in real time to the history speed and the driving behavior；
Parameter adjustment unit, for utilizing updated history speed and driving behavior to the trained nerve net
The parameter of network is adjusted.
Optionally, which further includes that gradient prediction model establishes module, for passing through
Gradient prediction model of the foundation of the road grade data of acquisition based on autoregression integral moving average model(MA model)；
The gradient prediction model establishes module and includes:
Unit is established in autoregression part, for will be before any one sampled point in road grade data and the sampled point
Linear relationship is established as the p rank autoregression part of autoregression integral moving average model(MA model) between sampling point sequence；
Unit is established in rolling average part, for the linear weighted function relationship of the white noise of road grade data kind to be established as
The q rank rolling average part of autoregression integral moving average model(MA model)；
Integral unit, for being that autoregression integral is mobile flat by the autoregression part and the rolling average thin consolidation
The auto regressive moving average part of equal model；
Model parameter calculation unit determines certainly for carrying out parameter calculating to autoregression integral moving average model(MA model)
P value, d value and the q value of Regression-Integral moving average model(MA model), obtain gradient prediction model；Wherein p is the order of autoregression part, q
For the order of rolling average part, d be non-stationary value of slope it is Sequence Transformed be difference number required for stationary sequence.
Optionally, the gradient prediction module, specifically includes:
Range prediction unit, for the prediction speed multiplied by the prediction speed corresponding time, to be obtained pre- ranging
From；
Gradient predicting unit is obtained for carrying out gradient prediction according to the Prediction distance using the gradient prediction model
To prediction road grade.
The specific embodiment provided according to the present invention, the invention discloses following technical effects: the present invention is based on nerve nets
Network on-line study predicts automobile speed from the angle of human-vehicle-environment system in real time, comprehensively considers vehicle status parameters, drives
Member's driving style and road ahead environment and traffic state data, improve the accuracy of speed prediction.And based on this utilization
Autoregression integrates moving average model(MA model), predicts the gradient, obtains distance s with the speed that neural network prediction goes out and makes slope
Degree prediction is more convenient accurate, is then based on the gradient and vehicle speed information and is allocated to the torque and revolving speed of each power part,
To obtain more preferably fuel economy.
Detailed description of the invention
It in order to more clearly explain the embodiment of the invention or the technical proposal in the existing technology, below will be to institute in embodiment
Attached drawing to be used is needed to be briefly described, it should be apparent that, the accompanying drawings in the following description is only some implementations of the invention
Example, for those of ordinary skill in the art, without any creative labor, can also be according to these attached drawings
Obtain other attached drawings.
Fig. 1 is the method flow diagram of hybrid vehicle energy management method embodiment of the present invention；
Fig. 2 is the system construction drawing of Energy Management System for Hybrid Electric Vehicle embodiment of the present invention.
Specific embodiment
Following will be combined with the drawings in the embodiments of the present invention, and technical solution in the embodiment of the present invention carries out clear, complete
Site preparation description, it is clear that described embodiments are only a part of the embodiments of the present invention, instead of all the embodiments.It is based on
Embodiment in the present invention, it is obtained by those of ordinary skill in the art without making creative efforts every other
Embodiment shall fall within the protection scope of the present invention.
The object of the present invention is to provide a kind of hybrid vehicle energy management method and systems, are based on radial basis function
RBF Artificial Neural Network predicts short-term speed, and combines the short-term speed predicted, with ARIMA model pair
Road grade is predicted, accuracy and the fuel economy of speed prediction are improved.
In order to make the foregoing objectives, features and advantages of the present invention clearer and more comprehensible, with reference to the accompanying drawing and specific real
Applying mode, the present invention is described in further detail.
Fig. 1 is the method flow diagram of hybrid vehicle energy management method embodiment of the present invention.
Referring to Fig. 1, the hybrid vehicle energy management method, comprising:
Step 101: the following speed is predicted using the neural network based on history speed and driving behavior training,
Obtain prediction speed.Concrete mode are as follows: in last moment 30s history speed and driving behavior to subsequent time 30s
The interior following speed is predicted；In real vehicle driving process, last moment 30s is constantly obtained based on vehicle carried data collecting system
Interior real-time vehicle speed data and driving behavior form neural network input vector, realize speed prediction.
Step 102: using by acquisition road grade data established based on autoregression integrate rolling average mould
The gradient prediction model of type predicts road grade, obtains prediction road grade；
Step 103: according to the prediction speed and the prediction road slope calculation demand power；
Step 104: calculating the torque and revolving speed of each power part using dynamic programming algorithm according to the demand power.
As an alternative embodiment, the neural network based on history speed and driving behavior training was trained
Journey includes:
Obtain history speed of each vehicular data acquisition device automobile collected in each road condition downward driving
And driving behavior.Concrete mode are as follows: different drivers are obtained based on vehicle carried data collecting system or wireless data acquisition system
History vehicle speed information on different road conditions when driving acquires different drivers in difference based on vehicle carried data collecting system
Driving behavior on road condition when driving is stored in formation sample operating condition in vehicle speed data library.Driving behavior includes: to add
Oil receives oil, touches on the brake or get off the brakes, shifts gears, turning to.
By the history speed and the driving behavior, segment length is divided at preset timed intervals, obtain input sample and
Export sample.Concrete mode are as follows: effective actual measurement of automobilism and each moment point t of each sample operating condition is extracted from vehicle speed data library
Data vehicle instantaneous velocity v (t).Clustering is carried out to driving behavior, choose history speed in upper period 30s with
Input of the driving behavior Q as neural network speed prediction model, is defined asN in formulainFor the input of prediction model, VkFor current time speed,
Vk-1For the speed of subsequent time；K is current time；The speed of next 30s period in future is used as output, H in formulahIt is history
Speed vector length, that is, the length of neural network model input vector.Assuming that fnIt is the nonlinear function of neural network prediction,
Then have
Neural network is trained using the input sample and output sample, obtains trained neural network.Choosing
Take radial basis function (RadialBasisFunction, RBF) artificial neural network as sigmoidal function to automobile future
Speed is predicted, the following speed prediction model based on RBF kernel function is constructed, and RBF neuron number takes
It is 100, activation primitive a1=exp (- | | n-c | |2/2b2), n=Wa0+ b, a in formula1And a0It is upper one layer on current layer respectively
Neuron output, n is accumulative output, and W is weighted value, and c is neuron node center, and b is the diffusion of neuron radial basis function
Width.
Training step specifically:
History speed and driving behavior are arranged according to neural network input and output, then will input parameter vector
It is input in RBF kernel function model with output parameter vector and forms training sample progress off-line training, established
Stable RBF kernel function structure；
It determines that RBF kernel function is the connection type of n-h-m, that is, has n input, h hidden layer and m
A output；
To weight ωijAssigning initial value is the random number between 0 to 1, and the number of hidden layer neuron is h, initial network error E
0 is set, worst error ε is set as a positive decimal；
The center c of basic function is determined based on fuzzy K mean cluster algorithmiAnd variances sigmai, i=1,2 ..., h；
Using the weight ω of gradient descent method adjustment network hidden layer to output layerijUntil network error E < ε, terminate；Its
Middle network error indicates that expression formula is as follows using mean square error:
In formula, E indicates network error,For corresponding to input x1Reality output, y (x1) table be corresponding to input
Desired output, 1 is total sample number.
Self-organizing is selected to choose the RBF neural learning method of middle line, core is to solve for hidden layer Basis Function Center, base
To the weight of output unit, thus obtaining j-th of output in RBF neural is indicated for the variance of function and implicit layer unit are as follows:
In formula,For the 1st input sample, l=1,2 ..., l, with ciAs network hidden layer
The center of node, using i as the number of nodes of hidden layer, | | xl-ci||2For European norm, σiFor the width of basic function, ωijIt is hidden
Connection weight containing layer to output layer, j=1,2 ..., m are the number of nodes of output layer, yiFor nerve net corresponding with input sample
The reality output of j-th of output node of network.
As an alternative embodiment, being carried out using the input sample and output sample to neural network described
Training, after obtaining trained neural network, further includes:
The history speed and the driving behavior are updated in real time；Concrete mode are as follows: in an acquisition upper period
(period can be one week, January or 1 year) vehicle speed data and driving behavior data update sample database；
The parameter of the trained neural network is adjusted using updated history speed and driving behavior.
Concrete mode are as follows: in the non-traveling task of vehicle, neural network prediction model is learnt again.
As an alternative embodiment, the detailed process of acquisition road grade data are as follows:
Road grade data are acquired using sensor, or road slope information is obtained based on GIS-Geographic Information System, are stored in
Road grade sample G is formed in database；
The value of slope of extraction is subjected to tranquilization inspection, calculates auto-correlation function (Auto Correlation
Function, ACF):
In formula, μ is the expectation of gradient G, and σ is the standard deviation of gradient G, GtAnd Gt+τThe respectively sample slope at t and t+ τ moment
Angle value；
It calculates partial autocorrelation function (PartialAuto CorrelationFunction, PACF):
αp, p=1,2 ...
αpFor the last one coefficient of p rank autoregression model；
If the ACF and PACF of sampling Gradient meet stationarity boundary condition, which is stable；
If sampling Gradient ACF and PACF be unsatisfactory for stationarity boundary condition, the data be it is jiggly, to sample
Data carry out first-order difference, re-start stationary test, until data meet stationarity requirement, number of repetition is denoted as d；
Road grade data building autoregression based on tranquilization integrates moving average model(MA model) (Autoregressive
Integrated MovingAverage Model, ARIMA), gradient sequence data are as follows:
G={ g1, g2..., gn}
G in formula1, g2..., gnIndicate the value of slope at the 1st, 2 ..., n sampled point.
ARIMA model, which refers to, converts stationary time series for nonstationary time series, then to the lagged value of dependent variable
And present worth, the lagged value of random error are returned, and then establish corresponding prediction model.ARIMA model is according to former sequence
No steady, autoregression item and rolling average item contained by model difference, can be divided into autoregression model (AR), rolling average mould
Type (MA), ARMA model (ARMA) and autoregression integrate moving average model(MA model) (ARIMA) four types.
As an alternative embodiment, being based on autoregression integral movement by the foundation of the road grade data of acquisition
The detailed process of the gradient prediction model of averaging model includes:
By linear relationship between any one sampled point in road grade data and the sampling point sequence before the sampled point
It is established as the p rank autoregression part (AR) of autoregression integral moving average model(MA model)；The gradient g of sampled point zzIt is sampled with first p
Linear relationship between point sequence is as follows:
gz=α1gz-1+α2gz-2+…+αpgz-p+δz
Random sequence { δzBe and { gzIncoherent white noise.
The linear weighted function relationship of the white noise of road grade data kind is established as to the q of autoregression integral moving average model(MA model)
Rank rolling average part；Moving average model(MA model) describes sequence { gzIn, gradient gzIt is expressed as the linear weighted function of several white noises
With.The expression formula that autoregression integrates the q rank rolling average part of moving average model(MA model) is as follows:
gz=ε1δz-1+ε2δz-2+…+εqxz-q+δz
It is that autoregression integrates returning certainly for moving average model(MA model) by the autoregression part and the rolling average thin consolidation
Return rolling average part；The model takes into account the characteristics of two models in front, with parameter describing stable time series as few as possible
The change procedure of data, expression formula are as follows:
gz=α1gz-1+α2gz-2+…+αpgz-p+ε1δz-1+ε2δz-2+…+εqδz-q
In formula, p indicates that autoregression part order, q indicate moving average order, and ε and α indicate corresponding weight coefficient, according to
This model is denoted as ARMA (p, q).
Parameter calculating is carried out to autoregression integral moving average model(MA model), determines autoregression integral moving average model(MA model)
P value, d value and q value, obtain gradient prediction model；Wherein p is the order of autoregression part, and q is the order of rolling average part, d
For non-stationary value of slope it is Sequence Transformed be difference number required for stationary sequence.Concrete mode are as follows: calculate non-stationary value of slope
It is Sequence Transformed to obtain d value for difference number required for stationary sequence；By ARMA model (ARMA) auto-correlation letter
The function that number is decayed when going to zero (hangover) executes step number and determines p value；By ARMA model (ARMA) partial correlation letter
The function that number is decayed when going to zero (hangover) executes step number and determines q value.
As an alternative embodiment, it is described using by acquisition road grade data established based on from
The gradient prediction model of Regression-Integral moving average model(MA model) predicts road grade, obtains prediction road grade, specific to wrap
It includes:
By the prediction speed multiplied by the prediction speed corresponding time, Prediction distance is obtained；
Gradient prediction is carried out according to the Prediction distance using the gradient prediction model, obtains prediction road grade.
As an alternative embodiment, torque arithmetic formula used by step 103 are as follows:
In formula, m is car mass, ToutFor output torque, TbreakBraking torque, RwheelFor radius of wheel, f is road surface resistance
Force coefficient, CdFor air resistance coefficient, A is vehicle front face area.
As an alternative embodiment, the process of step 104 specifically:
Division stage and choice phase variable k；
Select state variable λk；
Trade-off decision variable and determining permission decision set at different levels；
Determine state transition equation, as follows:
λk+1=T (λk, vk)
Determine the form of phase targets function, objective function there must be separability, and meet recurrence relation；
Using r function as cost function, recurrence equation and end-point condition that fundamental equation i.e. optimal value function meets are established:
r*(λk)=min [r (λk, vk)+r*(λk+1)]
Terminal condition: r*(λn+1)=0, vkFor the speed at kth moment.
Backward calculates optimum value function and corresponding optimal solution in state space；
According to optimum value function and optimal solution, sequence calculates the optimal control policy under given original state, i.e., optimal
Motor torque, motor torque etc. control variable.
Fig. 2 is the system construction drawing of Energy Management System for Hybrid Electric Vehicle embodiment of the present invention.
Referring to fig. 2, the Energy Management System for Hybrid Electric Vehicle, comprising:
Speed prediction module 201, for utilizing the neural network based on history speed and driving behavior training to future
Speed is predicted, prediction speed is obtained；
Gradient prediction module 202, for long-pending based on autoregression using being established for the road grade data by acquisition
Divide the gradient prediction model of moving average model(MA model) to predict road grade, obtains prediction road grade；
Power computation module 203, for according to the prediction speed and the prediction road slope calculation demand power；
Power distribution module 204, for calculating each power part using dynamic programming algorithm according to the demand power
Torque and revolving speed.
As an alternative embodiment, the Energy Management System for Hybrid Electric Vehicle further includes neural metwork training mould
Block, for based on history speed and driving behavior training neural network；The neural metwork training module includes:
Data capture unit, for obtaining each vehicular data acquisition device automobile collected under each road condition
History speed and driving behavior when driving；
Sample division unit, for segment length to be drawn at preset timed intervals by the history speed and the driving behavior
Point, obtain input sample and output sample；
Training unit is trained for being trained using the input sample and output sample to neural network
Neural network.
As an alternative embodiment, the neural metwork training module further include:
Data updating unit, for being updated in real time to the history speed and the driving behavior；
Parameter adjustment unit, for utilizing updated history speed and driving behavior to the trained nerve net
The parameter of network is adjusted.
As an alternative embodiment, the Energy Management System for Hybrid Electric Vehicle further includes that gradient prediction model is built
Formwork erection block, the gradient that the foundation for the road grade data by acquisition integrates moving average model(MA model) based on autoregression predict mould
Type；
The gradient prediction model establishes module and includes:
Unit is established in autoregression part, for will be before any one sampled point in road grade data and the sampled point
Linear relationship is established as the p rank autoregression part of autoregression integral moving average model(MA model) between sampling point sequence；
Unit is established in rolling average part, for the linear weighted function relationship of the white noise of road grade data kind to be established as
The q rank rolling average part of autoregression integral moving average model(MA model)；
Integral unit, for being that autoregression integral is mobile flat by the autoregression part and the rolling average thin consolidation
The auto regressive moving average part of equal model；
Model parameter calculation unit determines certainly for carrying out parameter calculating to autoregression integral moving average model(MA model)
P value, d value and the q value of Regression-Integral moving average model(MA model), obtain gradient prediction model；Wherein p is the order of autoregression part, q
For the order of rolling average part, d be non-stationary value of slope it is Sequence Transformed be difference number required for stationary sequence.
As an alternative embodiment, the gradient prediction module 202, specifically includes:
Range prediction unit, for the prediction speed multiplied by the prediction speed corresponding time, to be obtained pre- ranging
From；
Gradient predicting unit is obtained for carrying out gradient prediction according to the Prediction distance using the gradient prediction model
To prediction road grade.
Technical effect of the invention is as follows:
1. short-term to realization automobile from the angle of human-vehicle-environment system the present invention is based on RBF neural on-line study
Speed is predicted to be studied in real time, proposes one kind and comprehensively considers vehicle status parameters, driver's driving style and front road
Road environment and the following short-term speed prediction method of the vehicle of traffic state data, improve the accuracy of speed prediction.
2. using ARIMA model, a kind of gradient prediction technique based on short-term forecast speed is proposed, it is pre- with neural network
The short-term speed measured obtains distance and makes gradient prediction more convenient accurate, applies to fuel consumption control strategy and obtains more preferably
Fuel economy.
For the system disclosed in the embodiment, since it is corresponded to the methods disclosed in the examples, so the ratio of description
Relatively simple, reference may be made to the description of the method.
Used herein a specific example illustrates the principle and implementation of the invention, and above embodiments are said
It is bright to be merely used to help understand method and its core concept of the invention；At the same time, for those skilled in the art, foundation
Thought of the invention, there will be changes in the specific implementation manner and application range.In conclusion the content of the present specification is not
It is interpreted as limitation of the present invention.
Claims (10)
1. a kind of hybrid vehicle energy management method characterized by comprising
The following speed is predicted using the neural network based on history speed and driving behavior training, obtains pre- measuring car
Speed；
Mould is predicted using the gradient for integrating moving average model(MA model) based on autoregression that the road grade data by acquisition are established
Type predicts road grade, obtains prediction road grade；
According to the prediction speed and the prediction road slope calculation demand power；
The torque and revolving speed of each power part are calculated using dynamic programming algorithm according to the demand power.
2. a kind of hybrid vehicle energy management method according to claim 1, which is characterized in that be based on history speed
The training process of trained neural network includes: with driving behavior
It obtains history speed of each vehicular data acquisition device automobile collected in each road condition downward driving and drives
The person's of sailing behavior；
By the history speed and the driving behavior, segment length is divided at preset timed intervals, obtains input sample and output
Sample；
Neural network is trained using the input sample and output sample, obtains trained neural network.
3. a kind of hybrid vehicle energy management method according to claim 2, which is characterized in that utilize institute described
It states input sample and output sample is trained neural network, after obtaining trained neural network, further includes:
The history speed and the driving behavior are updated in real time；
The parameter of the trained neural network is adjusted using updated history speed and driving behavior.
4. a kind of hybrid vehicle energy management method according to claim 1, which is characterized in that pass through the road of acquisition
The detailed process of gradient prediction model of the foundation of road Gradient based on autoregression integral moving average model(MA model) include:
Linear relationship between any one sampled point in road grade data and the sampling point sequence before the sampled point is established
The p rank autoregression part of moving average model(MA model) is integrated for autoregression；
The q rank that the linear weighted function relationship of the white noise of road grade data kind is established as autoregression integral moving average model(MA model) is moved
Dynamic average portion；
It is the autoregression shifting that autoregression integrates moving average model(MA model) by the autoregression part and the rolling average thin consolidation
Dynamic average portion；
To the autoregression integral moving average model(MA model) carry out parameter calculating, determine autoregression integral moving average model(MA model) p value,
D value and q value, obtain gradient prediction model；Wherein p is the order of autoregression part, and q is the order of rolling average part, and d is non-
Steady value of slope is Sequence Transformed for difference number required for stationary sequence.
5. a kind of hybrid vehicle energy management method according to claim 4, which is characterized in that described using by adopting
The gradient prediction model based on autoregression integral moving average model(MA model) of the road grade data of collection established is to road grade
It is predicted, obtains prediction road grade, specifically include:
By the prediction speed multiplied by the prediction speed corresponding time, Prediction distance is obtained；
Gradient prediction is carried out according to the Prediction distance using the gradient prediction model, obtains prediction road grade.
6. a kind of Energy Management System for Hybrid Electric Vehicle characterized by comprising
Speed prediction module, for being carried out using the neural network based on history speed and driving behavior training to the following speed
Prediction obtains prediction speed；
Gradient prediction module, for mobile flat based on autoregression integral using being established for the road grade data by acquisition
The gradient prediction model of equal model predicts road grade, obtains prediction road grade；
Power computation module, for according to the prediction speed and the prediction road slope calculation demand power；
Power distribution module, for according to the demand power using dynamic programming algorithm calculate each power part torque and
Revolving speed.
7. a kind of Energy Management System for Hybrid Electric Vehicle according to claim 6, which is characterized in that further include nerve net
Network training module, for based on history speed and driving behavior training neural network；The neural metwork training module includes:
Data capture unit, for obtaining each vehicular data acquisition device automobile collected in each road condition downward driving
When history speed and driving behavior；
Sample division unit, for by the history speed and the driving behavior, segment length to be divided at preset timed intervals,
Obtain input sample and output sample；
Training unit obtains trained mind for being trained using the input sample and output sample to neural network
Through network.
8. a kind of Energy Management System for Hybrid Electric Vehicle according to claim 7, which is characterized in that the neural network
Training module further include:
Data updating unit, for being updated in real time to the history speed and the driving behavior；
Parameter adjustment unit, for utilizing updated history speed and driving behavior to the trained neural network
Parameter is adjusted.
9. a kind of Energy Management System for Hybrid Electric Vehicle according to claim 6, which is characterized in that further include that the gradient is pre-
Model building module is surveyed, the foundation for the road grade data by acquisition integrates the slope of moving average model(MA model) based on autoregression
Spend prediction model；
The gradient prediction model establishes module and includes:
Unit is established in autoregression part, for by the sampling before any one sampled point in road grade data and the sampled point
Linear relationship is established as the p rank autoregression part of autoregression integral moving average model(MA model) between point sequence；
Unit is established in rolling average part, for the linear weighted function relationship of the white noise of road grade data kind to be established as returning certainly
Return the q rank rolling average part of integral moving average model(MA model)；
Integral unit, for being that autoregression integrates rolling average mould by the autoregression part and the rolling average thin consolidation
The auto regressive moving average part of type；
Model parameter calculation unit determines autoregression for carrying out parameter calculating to autoregression integral moving average model(MA model)
P value, d value and the q value for integrating moving average model(MA model), obtain gradient prediction model；Wherein p is the order of autoregression part, and q is to move
The order of dynamic average portion, d be non-stationary value of slope it is Sequence Transformed be difference number required for stationary sequence.
10. a kind of Energy Management System for Hybrid Electric Vehicle according to claim 9, which is characterized in that the gradient prediction
Module specifically includes:
Range prediction unit, for the prediction speed multiplied by the prediction speed corresponding time, to be obtained Prediction distance；
Gradient predicting unit obtains pre- for carrying out gradient prediction according to the Prediction distance using the gradient prediction model
Survey road grade.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810964452.XA CN109131350B (en) | 2018-08-23 | 2018-08-23 | Energy management method and system for hybrid electric vehicle |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810964452.XA CN109131350B (en) | 2018-08-23 | 2018-08-23 | Energy management method and system for hybrid electric vehicle |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN109131350A true CN109131350A (en) | 2019-01-04 |
| CN109131350B CN109131350B (en) | 2020-04-03 |
Family
ID=64791251
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810964452.XA Active CN109131350B (en) | 2018-08-23 | 2018-08-23 | Energy management method and system for hybrid electric vehicle |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109131350B (en) |
Cited By (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110610611A (en) * | 2019-09-26 | 2019-12-24 | 江苏大学 | Driving safety evaluation method for intelligent network-connected vehicle in mixed-driving traffic flow |
| CN111009134A (en) * | 2019-11-25 | 2020-04-14 | 北京理工大学 | Short-term vehicle speed working condition real-time prediction method based on interaction between front vehicle and self vehicle |
| CN111198501A (en) * | 2020-01-14 | 2020-05-26 | 浙江工业大学 | Method for determining fuel equivalent factor by RBF neural network |
| CN111833600A (en) * | 2020-06-10 | 2020-10-27 | 北京嘀嘀无限科技发展有限公司 | Method and device for predicting transit time and data processing equipment |
| CN112433476A (en) * | 2021-01-27 | 2021-03-02 | 北京理工大学 | Robust prediction control device and robust prediction control method for networked control system of electric vehicle |
| CN112572404A (en) * | 2020-12-24 | 2021-03-30 | 吉林大学 | Heavy commercial vehicle hybrid power energy management method based on front road information |
| US20210107449A1 (en) * | 2020-08-11 | 2021-04-15 | Beijing Institute Of Technology | Cyber-physical energy optimization control system and control method for hybrid electric vehicle |
| US20220080980A1 (en) * | 2020-09-15 | 2022-03-17 | Hyundai Motor Company | Device for predicting speed of vehicle and method thereof |
| CN114771462A (en) * | 2022-05-26 | 2022-07-22 | 中国第一汽车股份有限公司 | Vehicle anti-theft system and method |
| CN115591493A (en) * | 2022-09-22 | 2023-01-13 | 海龙智能技术研究（珠海）有限公司（Cn） | Reaction kettle temperature control method |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102717797A (en) * | 2012-06-14 | 2012-10-10 | 北京理工大学 | Energy management method and system of hybrid vehicle |
| KR20170041018A (en) * | 2015-10-06 | 2017-04-14 | 현대자동차주식회사 | Method and apparatus of controlling hybrid electric vehicle |
| CN107458380A (en) * | 2017-08-03 | 2017-12-12 | 重庆大学 | A kind of road grade real-time estimation method being applied under comprehensive driving cycles |
| CN108382204A (en) * | 2017-02-03 | 2018-08-10 | 福特全球技术公司 | Speed control for vehicle |
-
2018
- 2018-08-23 CN CN201810964452.XA patent/CN109131350B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102717797A (en) * | 2012-06-14 | 2012-10-10 | 北京理工大学 | Energy management method and system of hybrid vehicle |
| KR20170041018A (en) * | 2015-10-06 | 2017-04-14 | 현대자동차주식회사 | Method and apparatus of controlling hybrid electric vehicle |
| CN108382204A (en) * | 2017-02-03 | 2018-08-10 | 福特全球技术公司 | Speed control for vehicle |
| CN107458380A (en) * | 2017-08-03 | 2017-12-12 | 重庆大学 | A kind of road grade real-time estimation method being applied under comprehensive driving cycles |
Non-Patent Citations (2)
| Title |
|---|
| 唐启义 冯明光: "《实用统计分析及其计算机处理平台》", 30 September 1997, 中国农业出版社 * |
| 孙超: ""混合动力汽车预测能量管理研究"", 《中国博士学位论文全文数据库（工程科技Ⅱ辑）》 * |
Cited By (13)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110610611A (en) * | 2019-09-26 | 2019-12-24 | 江苏大学 | Driving safety evaluation method for intelligent network-connected vehicle in mixed-driving traffic flow |
| WO2021103625A1 (en) * | 2019-11-25 | 2021-06-03 | 北京理工大学 | Short-term vehicle speed condition real-time prediction method based on interaction between vehicle ahead and current vehicle |
| CN111009134A (en) * | 2019-11-25 | 2020-04-14 | 北京理工大学 | Short-term vehicle speed working condition real-time prediction method based on interaction between front vehicle and self vehicle |
| CN111198501A (en) * | 2020-01-14 | 2020-05-26 | 浙江工业大学 | Method for determining fuel equivalent factor by RBF neural network |
| CN111198501B (en) * | 2020-01-14 | 2022-05-13 | 浙江工业大学 | Method for determining fuel equivalent factor by RBF neural network |
| CN111833600A (en) * | 2020-06-10 | 2020-10-27 | 北京嘀嘀无限科技发展有限公司 | Method and device for predicting transit time and data processing equipment |
| US20210107449A1 (en) * | 2020-08-11 | 2021-04-15 | Beijing Institute Of Technology | Cyber-physical energy optimization control system and control method for hybrid electric vehicle |
| US20220080980A1 (en) * | 2020-09-15 | 2022-03-17 | Hyundai Motor Company | Device for predicting speed of vehicle and method thereof |
| CN112572404A (en) * | 2020-12-24 | 2021-03-30 | 吉林大学 | Heavy commercial vehicle hybrid power energy management method based on front road information |
| CN112433476A (en) * | 2021-01-27 | 2021-03-02 | 北京理工大学 | Robust prediction control device and robust prediction control method for networked control system of electric vehicle |
| CN114771462A (en) * | 2022-05-26 | 2022-07-22 | 中国第一汽车股份有限公司 | Vehicle anti-theft system and method |
| CN115591493A (en) * | 2022-09-22 | 2023-01-13 | 海龙智能技术研究（珠海）有限公司（Cn） | Reaction kettle temperature control method |
| CN115591493B (en) * | 2022-09-22 | 2024-05-14 | 北京理工大学珠海学院 | Reaction kettle temperature control method |
Also Published As
| Publication number | Publication date |
|---|---|
| CN109131350B (en) | 2020-04-03 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109131350A (en) | A kind of hybrid vehicle energy management method and system | |
| WO2021103625A1 (en) | Short-term vehicle speed condition real-time prediction method based on interaction between vehicle ahead and current vehicle | |
| Guo et al. | ARIMA-based road gradient and vehicle velocity prediction for hybrid electric vehicle energy management | |
| CN103914985B (en) | A kind of hybrid power passenger car following speed of a motor vehicle trajectory predictions method | |
| CN103246943B (en) | Vehicle operating condition multi-scale predicting method based on Markov chain | |
| US8706416B2 (en) | System and method for determining a vehicle route | |
| CN112896186B (en) | Automatic driving longitudinal decision control method under cooperative vehicle and road environment | |
| US8965597B2 (en) | Road grade auto-mapping | |
| CN111267830B (en) | Hybrid power bus energy management method, device and storage medium | |
| Schoen et al. | A machine learning model for average fuel consumption in heavy vehicles | |
| DE102012216115A1 (en) | A vehicle and method for estimating a range for the vehicle | |
| CN108791270A (en) | A kind of power part operating point regulation and control method that hybrid vehicle is predicted based on operating mode | |
| CN106503840A (en) | Parking stall Forecasting Methodology and system can be used in parking lot | |
| CN106095963A (en) | Vehicle drive behavior analysis big data public service platform under the Internet+epoch | |
| US20210237745A1 (en) | Device, method and machine learning system for determining a state of a transmission for a vehicle | |
| CN104183134A (en) | Expressway short-time traffic flow forecast method based on intelligent car type classification | |
| CN109398363A (en) | A kind of pavement grade determines method, apparatus, storage medium and automobile | |
| CN113283460A (en) | Machine learning system and method for operating the system for determining a time series | |
| CN104851280A (en) | Vehicle driving control method, device, system and related equipment | |
| CN110979342A (en) | Working condition information acquisition method for vehicle global energy management control | |
| Grubwinkler et al. | A modular and dynamic approach to predict the energy consumption of electric vehicles | |
| EP2906908A1 (en) | Method and computer program product for estimating a travel time for a vehicle | |
| CN108846571A (en) | A kind of net connectionization electric car macroscopic view energy consumption estimation method | |
| US20210241104A1 (en) | Device, method and machine learning system for determining a velocity for a vehicle | |
| Lee et al. | Dynamic traffic simulation for traffic congestion problem using an enhanced algorithm |
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 | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |