CN108519557A - A kind of power battery parameter identification method suitable for sparse data - Google Patents
A kind of power battery parameter identification method suitable for sparse data Download PDFInfo
- Publication number
- CN108519557A CN108519557A CN201810342607.6A CN201810342607A CN108519557A CN 108519557 A CN108519557 A CN 108519557A CN 201810342607 A CN201810342607 A CN 201810342607A CN 108519557 A CN108519557 A CN 108519557A
- Authority
- CN
- China
- Prior art keywords
- ocv
- identification
- power battery
- parameter
- soc
- 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
Landscapes
- Tests Of Electric Status Of Batteries (AREA)
- Secondary Cells (AREA)
Abstract
The present invention provides a kind of power battery parameter identification methods suitable for sparse data, based on Intensive treatment algorithm, when the statistical distribution characteristic of real system noise is difficult to determine, the Statistical Distribution Characteristics of system noise need not be made to assume, it need to only know the boundary of system noise, the error of introducing is measured by sensor, the rounding error and modeling error of number of machines can regard the form of Bounded Errors as.The algorithm has the ability of identification redundant data simultaneously, when increasing in the battery management system sampling interval, also can guarantee identification precision, thus the case where especially suitable for sparse data, with many advantageous effects for significantly improving power battery management system reliability etc..
Description
Technical field
The present invention relates to electrokinetic cell system field more particularly to a kind of parameter identification techniques of vehicle mounted dynamic battery.
Background technology
For power battery as the important component in new-energy automobile, the estimation of battery remaining power and battery are old
The assessment of change degree can be realized by battery management system, so that the SOC and SOH of battery are maintained in normal working level,
The super-charge super-discharge due to battery, overcurrent permanent damage caused by battery are prevented, to improve the service life of battery, reduces electricity
The use cost of electrical automobile, and accurate power battery parameter identification estimation is to ensure that the base of the accurate online Prediction of internal state
Plinth.
On-line parameter identification usually can directly obtain the parameter under corresponding SOC in real time using information such as electric current, terminal voltages
Value is suitable for real vehicle system complicated and changeable, however the existing parameter identification method stability based on model is poor, in complexity
The phenomenon that will produce large error under operating mode, or even dissipating, and with the increase in onboard sensor sampling interval, identification knot
The precision of fruit decreases.Commonly the recursive least-squares discrimination method based on forgetting factor, random error need to obey zero
The normal distribution of mean value, zero covariance is obviously difficult to be met in practical application.Therefore, it is still necessary to want a kind of in this field
It is not only restricted to noise profile, and the power battery parameter that can reduce the sparse data of required data volume to a certain extent is distinguished
Knowledge method.
Invention content
For technical problem present in above-mentioned this field, the present invention provides a kind of power electrics suitable for sparse data
Pond parameter identification method, specifically includes following steps:
Step 1: real-time online obtains and stores the electric current in power battery operational process, terminal voltage information;
Step 2: establishing state-space model to the power battery;
Step 3: carrying out on-line parameter identification and update to the state-space model based on Intensive treatment algorithm;
Step 4: utilizing the parameter of OCV estimated values and current time by the obtained previous moment of the on-line identification
Vectorial estimated value obtains the OCV estimated results under current time, while obtaining OCV's according to the correspondence interpolation of OCV-SOC
Change curve;
It is inserted Step 5: comparing the OCV on-line identifications result obtained based on the step 3 with being passed through based on the step 4
The OCV change curves being worth to verify the on-line identification result.
Further, state-space model is established to the power battery in the step 2, specifically includes:It is based on
Thevenin equivalent-circuit models simultaneously have following input, output relation:
Wherein, Ut, i indicates terminal voltage, electric current respectively, and k indicates that moment, Φ (k) indicate by known input quantity and output
The measurement vector constituted is measured, θ (k) indicates that the unknown parameter vector for needing to recognize, e (k) indicate model interference or noise sequence, Uoc
(k) the OCV values at k moment, a are indicated1,a2,a3For the component of parameter vector to be identified.
Further, on-line parameter is carried out to the state-space model based on Intensive treatment algorithm in the step 3 to distinguish
Know and update, specifically includes:
3.1, to the measurement vector Φ (k) in the state-space model, unknown parameter vector theta (k), in identification algorithm
The radius vectors σ of the boundary γ and ellipsoid of covariance matrix P (k) and system noise2(k) it is initialized, obtains observation arrow
Measure the radius vectors σ of initial value Φ (0), unknown parameter vector initial value θ (0), covariance matrix initial value P (0) and ellipsoid2
(0).Under normal conditions according to convergence and the parameter characteristic of battery, Φ (0), θ (0) can be set as 0, P (0) is set
Forμ is a smaller positive number, and general value is μ=10-4, I indicate n tie up unit matrix, take ellipsoid radius be more than zero,
σ can be enabled2(0)=1, system noise boundary value γ is determined according to factors such as practical measuring apparatus and modeling errors.
3.2, it carries out recursion update using following formula to the parameter inscribed when current k to calculate, wherein k ∈ (1,2 ...):
Wherein, δ (k)=y (k)-Φ (k) θ (k), G (k)=Φ (k)TP(k-1)Φ(k);
Using ellipsoid Optimality Criteria to parameter alpha (k), β (k) is solved.
Further, the selection of above-mentioned parameter α (k), β (k) value is related to the accuracy of final identification result, therefore can be with
It takes minimum volume ellipsoid criterion min det (P (k)), minimum mark ellipsoid criterion min tr (P (k)) and minimizes parameter σ2(k)
Optimality Criterias are waited to calculate.
Further, the utilization described in the step 4 is estimated by the OCV of the obtained previous moment of the on-line identification
The parameter vector estimated value of evaluation and current time obtains the OCV estimated results under current time, specifically includes:
According to the unknown parameter vector theta (k)=[Uoc(k)-a1Uoc(k-1)a1 a2 a3] on-line identification as a result, passing through
The OCV estimated values at k-1 moment and the component a in the parameter vector to be identified1, obtain the OCV estimated results at k moment:
Uoc(k)=θ1(k)+a1(k)Uoc(k-1)
Wherein, θ1(k) one-component of the unknown parameter vector theta (k) is represented.
Further, the correspondence interpolation according to OCV-SOC described in the step 4 obtains the variation song of OCV
Line specifically includes:
4.1, open voltage test is carried out, multiple OCV values under different state-of-charges are respectively obtained;
4.2, using the electric current i (k) acquired in the step 1, since power battery is opposite in temperature and ageing state
In the case of stabilization, there are one-to-one mapping relations with OCV by SOC, therefore can obtain the corresponding moment based on current integration method
Under SOC value;
4.3, it is based on the multiple OCV values and the SOC value, the relation curve of OCV-SOC is obtained by linear interpolation.
The above-mentioned method that power battery on-line parameter identification is carried out based on Intensive treatment algorithm, in the system of real system noise
When meter distribution character is difficult to determine, the Statistical Distribution Characteristics of system noise need not be made to assume, it is only necessary to know that system noise
Boundary, the error of introducing is measured by sensor, and the rounding error and modeling error of number of machines can regard bounded mistake as
The form of difference.There is the algorithm ability of identification redundant data can also be protected when increasing in the battery management system sampling interval simultaneously
Identification precision is demonstrate,proved, thus the case where especially suitable for sparse data, has and significantly improves power battery management system reliability etc.
Many advantageous effects.
Description of the drawings
Fig. 1 is the flow diagram according to method provided by the present invention
Fig. 2 is Thevenin equivalent-circuit model schematic diagram schematic diagrames
Fig. 3 is the comparison of 1 second sampling interval OCV of UDDS operating modes identification result
Fig. 4 is the comparison of 10 second sampling interval OCV of UDDS operating modes identification result
Fig. 5 is the comparison of 1 second sampling interval OCV of DST operating modes identification result
Fig. 6 is the comparison of 10 second sampling interval OCV of DST operating modes identification result
Specific implementation mode
Below in conjunction with the accompanying drawings to a kind of power battery parameter identification method suitable for sparse data provided by the present invention,
It makes and further illustrating in detail.
A kind of power battery parameter identification method suitable for sparse data provided by the present invention, as shown in Figure 1, specifically
Include the following steps:
Step 1: real-time online obtains and stores the electric current in power battery operational process, terminal voltage information;
Step 2: establishing state-space model to the power battery;
Step 3: carrying out on-line parameter identification and update to the state-space model based on Intensive treatment algorithm;
Step 4: utilizing the parameter of OCV estimated values and current time by the obtained previous moment of the on-line identification
Vectorial estimated value obtains the OCV estimated results under current time, while obtaining OCV's according to the correspondence interpolation of OCV-SOC
Change curve;
It is obtained by interpolation with based on the step 4 Step 5: comparing the on-line identification result obtained based on the step 3
The OCV change curves arrived verify the on-line identification result.
In the preferred embodiment of the application, it is research object, rated capacity to select nickel-cobalt-manganese ternary battery NMC
For 25Ah, charge and discharge blanking voltage is respectively 4.2V and 2.5V, rated current 7.5A.Operating condition of test is dynamic stress operating mode
(DST) and Metro cycle operating mode (UDDS).Use the terminal voltage value that battery testing systematic survey obtains as with reference to value and institute
The terminal voltage estimated value comparison for stating algorithm is used as error, to verify the stability of algorithm in the case of sparse data, by described
The OCV results that OCV-SOC curve interpolations obtain are used as with reference to value comparison estimated value as error, are calculated in the case of sparse data to verify
The reliability of method.Meanwhile by the recursive least-squares based on forgetting factor in the case of algorithm proposed by the present invention and sparse data
Identification algorithm is made comparisons, to illustrate applicability and stability of the algorithm in sparse data.
It is equivalent based on Thevenin as shown in Figure 2 that state-space model is established to the power battery in the step 2
Circuit model.
Fig. 3 and Fig. 4 is respectively under UDDS operating modes, and the least square method of recursion based on forgetting factor proposes to calculate with the present invention
OCV identification result comparison of the method in sampling interval 1s and 10s, Fig. 5 and Fig. 6 show the OCV identification result ratios under DST operating modes
Compared with by attached drawing it is found that when the sampling interval is larger, parameter identification method proposed by the present invention stands good in sparse data, phase
Than in traditional least square method of recursion based on forgetting factor, in the power battery low SOC stages, the identification result of OCV still may be used
To keep convergence, stability is strong, is not in larger concussion spike.
Under UDDS operating modes, the root mean square statistical property of OCV identification results is as shown in table 1 under the different sampling intervals:
The root mean square statistical property of OCV identification results under the different sampling intervals of table 1
By table 1 as it can be seen that the power battery parameter identification method proposed by the present invention for sparse data, precision is high, just
Root evaluated error can control within 5%.
It although an embodiment of the present invention has been shown and described, for the ordinary skill in the art, can be with
Understanding without departing from the principles and spirit of the present invention can carry out these embodiments a variety of variations, modification, replace
And modification, the scope of the present invention is defined by the appended.
Claims (6)
1. a kind of power battery parameter identification method suitable for sparse data, it is characterised in that:Specifically include following steps:
Step 1: real-time online obtains and stores the electric current in power battery operational process, terminal voltage information;
Step 2: establishing state-space model to the power battery;
Step 3: carrying out on-line parameter identification and update to the state-space model based on Intensive treatment algorithm;
Step 4: utilizing the parameter vector of OCV estimated values and current time by the obtained previous moment of the on-line identification
Estimated value obtains the OCV estimated results under current time, while obtaining the variation of OCV according to the correspondence interpolation of OCV-SOC
Curve;
It is obtained by interpolation with based on the step 4 Step 5: comparing the OCV on-line identifications result obtained based on the step 3
The OCV change curves arrived verify the on-line identification result.
2. the method as described in claim 1, it is characterised in that:State space is established to the power battery in the step 2
Model specifically includes:Based on Thevenin equivalent-circuit models and with following input, output relation:
Wherein, Ut, i indicates terminal voltage, electric current respectively, and k indicates that moment, Φ (k) expressions are made of known input quantity and output quantity
Measurement vector, θ (k) indicates that the unknown parameter vector that recognizes, e (k) is needed to indicate model interference or noise sequence, Uoc(k) table
Show the OCV values at k moment, a1,a2,a3For the component of parameter vector to be identified.
3. method as claimed in claim 2, it is characterised in that:Based on Intensive treatment algorithm to the state in the step 3
Spatial model carries out on-line parameter identification and update, specifically includes:
3.1, to the measurement vector Φ (k) in the state-space model, unknown parameter vector theta (k), the association side in identification algorithm
The radius vectors σ of the boundary γ and ellipsoid of poor matrix P (k) and system noise2(k) it is initialized, at the beginning of obtaining measurement vector
The radius vectors σ of initial value Φ (0), unknown parameter vector initial value θ (0), covariance matrix initial value P (0) and ellipsoid2(0)。
3.2, it carries out recursion update using following formula to the parameter inscribed when current k to calculate, wherein k ∈ (1,2 ...):
Wherein, δ (k)=y (k)-Φ (k) θ (k), G (k)=Φ (k)TP(k-1)Φ(k);
Using ellipsoid Optimality Criteria to parameter alpha (k), β (k) is solved.
4. method as claimed in claim 3, it is characterised in that:The ellipsoid Optimality Criteria uses minimum volume ellipsoid criterion
Min det (P (k)), minimum mark ellipsoid criterion min tr (P (k)) and minimum parameter σ2(k) etc..
5. method as claimed in claim 2, it is characterised in that:Utilization described in the step 4 is by the on-line identification institute
The OCV estimated values of obtained previous moment and the parameter vector estimated value at current time obtain the OCV estimation knots under current time
Fruit specifically includes:
According to the unknown parameter vector theta (k)=[Uoc(k)-a1Uoc(k-1) a1 a2 a3] on-line identification as a result, passing through k-1
The OCV estimated values at moment and the component a in the parameter vector to be identified1, obtain the OCV estimated results at k moment:
Uoc(k)=θ1(k)+a1(k)Uoc(k-1)
Wherein, θ1(k) one-component of the unknown parameter vector theta (k) is represented.
6. the method as described in claim 1, it is characterised in that:Being closed according to the corresponding of OCV-SOC described in the step 4
It is the change curve that interpolation obtains OCV, specifically includes:
4.1, open voltage test is carried out, multiple OCV values under different state-of-charges are respectively obtained;
4.2, using the electric current i (k) acquired in the step 1, since power battery is stablized relatively in temperature and ageing state
In the case of, there are one-to-one mapping relations with OCV by SOC, therefore can obtain corresponding based on current integration method when inscribes
SOC value;
4.3, it is based on the multiple OCV values and the SOC value, the relation curve of OCV-SOC is obtained by linear interpolation.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201810342607.6A CN108519557B (en) | 2018-04-17 | 2018-04-17 | A kind of power battery parameter identification method suitable for sparse data |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201810342607.6A CN108519557B (en) | 2018-04-17 | 2018-04-17 | A kind of power battery parameter identification method suitable for sparse data |
Publications (2)
Publication Number | Publication Date |
---|---|
CN108519557A true CN108519557A (en) | 2018-09-11 |
CN108519557B CN108519557B (en) | 2019-11-05 |
Family
ID=63428742
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201810342607.6A Active CN108519557B (en) | 2018-04-17 | 2018-04-17 | A kind of power battery parameter identification method suitable for sparse data |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN108519557B (en) |
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109884550A (en) * | 2019-04-01 | 2019-06-14 | 北京理工大学 | A kind of identification of electrokinetic cell system on-line parameter and retrogressive method |
-
2018
- 2018-04-17 CN CN201810342607.6A patent/CN108519557B/en active Active
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109884550A (en) * | 2019-04-01 | 2019-06-14 | 北京理工大学 | A kind of identification of electrokinetic cell system on-line parameter and retrogressive method |
CN109884550B (en) * | 2019-04-01 | 2020-01-17 | 北京理工大学 | Online parameter identification and backtracking method for power battery system |
Also Published As
Publication number | Publication date |
---|---|
CN108519557B (en) | 2019-11-05 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN106249171B (en) | A kind of electrokinetic cell system identification and method for estimating state for the wide sampling interval | |
CN108196200B (en) | Combined simulation evaluation method for health and state of charge of lithium battery | |
CN104569835B (en) | A kind of method of the state-of-charge for the electrokinetic cell for estimating electric automobile | |
CN106291381B (en) | A kind of method of Combined estimator electrokinetic cell system state-of-charge and health status | |
CN103399277B (en) | A kind of power battery actual capacity estimation method | |
CN104569829B (en) | Battery SOC estimates and automatically corrects | |
CN107991623A (en) | It is a kind of to consider temperature and the battery ampere-hour integration SOC methods of estimation of degree of aging | |
CN105425153B (en) | A kind of method of the state-of-charge for the electrokinetic cell for estimating electric vehicle | |
CN105572596B (en) | Lithium battery SOC estimation method and system | |
CN106291393B (en) | A method of for online recognition battery model parameter | |
CN105223487B (en) | A kind of multimode decoupling method of estimation of lithium ion battery | |
CN108279385A (en) | State of charge evaluation method, device and the electronic equipment of battery | |
JP5474993B2 (en) | Method for determining the state of charge of a battery in the charge or discharge phase | |
CN108896916A (en) | Battery open circuit voltage and health status method for solving based on constant current charge-discharge voltage curve | |
JP2011214843A (en) | Device and method for estimating internal state of battery | |
CN108896924A (en) | The charge states of lithium ion battery estimation method merged based on depth confidence network and Kalman filtering | |
CN105929340B (en) | A method of battery SOC is estimated based on ARIMA | |
CN111812532B (en) | Method for determining aging parameters, state of charge parameters and temperature of a battery | |
CN109884550A (en) | A kind of identification of electrokinetic cell system on-line parameter and retrogressive method | |
CN105738826B (en) | Ice storing time system and method | |
CN106772067A (en) | The method that Multiple Time Scales IAPF filters estimated driving force battery charge state and health status | |
CN110095732A (en) | A kind of lithium battery SOH estimation method for considering ambient humidity and internal resistance being influenced | |
Nejad et al. | Sensitivity of lumped parameter battery models to constituent parallel-RC element parameterisation error | |
CN110412472B (en) | Battery state of charge estimation method based on normal gamma filtering | |
CN115754724A (en) | Power battery state of health estimation method suitable for future uncertainty dynamic working condition discharge |
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 |