GB2386709A - A neural network for estimating battery residual capacity in an electric vehicle - Google Patents
A neural network for estimating battery residual capacity in an electric vehicle Download PDFInfo
- Publication number
- GB2386709A GB2386709A GB0206316A GB0206316A GB2386709A GB 2386709 A GB2386709 A GB 2386709A GB 0206316 A GB0206316 A GB 0206316A GB 0206316 A GB0206316 A GB 0206316A GB 2386709 A GB2386709 A GB 2386709A
- Authority
- GB
- United Kingdom
- Prior art keywords
- current
- soac
- estimation
- model
- brc
- 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
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01R—MEASURING ELECTRIC VARIABLES; MEASURING MAGNETIC VARIABLES
- G01R31/00—Arrangements for testing electric properties; Arrangements for locating electric faults; Arrangements for electrical testing characterised by what is being tested not provided for elsewhere
- G01R31/36—Arrangements for testing, measuring or monitoring the electrical condition of accumulators or electric batteries, e.g. capacity or state of charge [SoC]
- G01R31/367—Software therefor, e.g. for battery testing using modelling or look-up tables
Landscapes
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Secondary Cells (AREA)
Abstract
A Neural Network (NN) based estimation method for Battery Residual Capacity (BRC) is disclosed. The invention uses the discharged and regenerative capacity distribution to represent the discharge current profile for the BRC estimation in terms of the State of Available Capacity (SOAC), and is implemented by the following steps: (1) measuring the current from the battery in EVs; (2) integrating the current based on the selected current ranges and the selected values of the lower and upper current bounds to form the discharged and regenerative capacity; (3) putting the capacity distribution and temperature together to form one vector; (4) normalizing this vector to be suitable for the inputs of the NN model for the BRC estimation; and (5) making the BRC estimation in terms of the SOAC. The major advantage of this method is that the influence of discharge current profile is considered, hence to obtain accurate BRC.
Description
COMPLETE SPECIFICATION
Method for Battery Residual Capacity Estimation in Electric Vehicles I, CHINO CHUEN CHAT, a professor of Electrical Engineering do hereby declare the invention, for which I pray that a patent may be granted to me for this patent by myself and my PhD student WEI XIANG SHEN, and the method by it is to be performed, to be particularly described in and by the following statements:
FIELD OF TO INVENTION
The present invention relates to a method for the battery residual capacity (BRC) estimation in electric vehicles (EVs).
BACKGROUND OF TED INVENTION
In a world where environmental protection and energy conservation are growing concerns, the research and development of various technologies in EVs is being actively conducted. However, the BRC indicator technology can not keep pace with the development of other EV technologies. The key problem arises from the highly nonlinear characteristics of the battery in EVs, which leads to the difficulty in the BRC estimation and thus the BRC indicator for EVs.
Previously, the BRC for the battery in EVs has been estimated by the state of charge (SOC). The purpose of this substitution is to avoid the difficulty in defining the battery available capacity (BAC) at the fully charged state for various EV discharge current profiles.
However, the SOC is different from the BRC in that the SOC indicates the state a battery lies in, rather than giving the capacity remaining of which the EV driving range is dependent.
Although the higher the SOC the more the BRC can be exhibited at the same discharge current, they have no elicit quantitative relationship. For instance, increasing battery temperature or lowering discharge current will cause an increase in the BAC even at the same SOC. FIG. 1 shows the effect of discharge current and temperature on the BAC at the fully charged state, where the SOC is generally defined as unity.
On the other hand, the BRC estimation for the battery in EVs has been explored by using the ampere-hour counter. This method is originated from the following basic equation: Cr = Ca q(t) (1) q(t) = |Id (t)dt (2) to where Cr denotes the BRC, Ca refers to the BAC at the fully charged state for a certain EV discharge current profile, q(t) is the discharged capacity, and Id(t) is the instantaneous discharge current. Since the BAC for the battery in EVs generally alters considerably, it has to be assumed as an appropriate value before the BRC estimation is performed. So far, two approaches have been adopted to approximate the value of the BAC at the fully charged state for the battery in EVs either based on the average discharge current or based on the reference discharge current. For the case of average discharge current, the BRC is estimated by: cr Cave (t) q(t) (3) where Cave (t) is the average BAC corresponding to the average discharge current. The BRC estimation by using (3) will cause an error unless the discharge current does not vary significantly. Table 1 shows a comparison of BACs under different EV discharge current profiles including the US federal urban driving schedule (FUDS), the US federal highway driving schedule (FHDS), the European standard reference cycle (ECE) and the Japanese mode 10.15 (JM10.15) as shown in FIG. 2. It can be found that although their average discharge currents are all approximately equal to 13 A, their BACs are very different.
Table 1 C _ a == Profiles -: FUDS 13.08 15.96
FHDS 13.11 25.05
ECE 13.21 13.05
J M 1 O. I 1 3 1 __1 5 43
For the case of reference discharge current, the BRC is estimated by: Cr = Cref - a(I)q(t) (4) where Cref is the reference BAC corresponding to the reference discharge current, such as the BAC for the 3-hour or 5-hour discharge rate, and x(Id) is the corrective coefficient which is used to calculate the equivalent discharged capacity if the discharge current is either higher or lower than the reference discharge current. To obtain the corrective coefficient, namely the
ratio of the BAC for the reference discharge current to that for the discharge current under test, one discharge current either individually or in combination with the reference discharge current is used to test the battery during the whole discharge period. By doing so, the effect of discharge current profiles on the BAC is ignored, leading to create a significant error. Table 2 contains this error even when the battery is under 2-step discharge current profiles. Another drawback of this method is that the temperature can not be formulated in the calculation of the corrective coefficient because of the nonlinear relationship between the BAC and the temperature under different discharge currents as shown in FIG. 1.
Table 2 Comparison of BACs under 2-step discharge current profiles Profiles BAC(Ah) 8 A for first 3 h and 20 A for other 0.38 h 31.66 20 A for first O.38 h and 8 A for other 3.17 h 33.00 8 A for first 3.17h and 20 A for other 0.1 h 27.33..
Moreover, the neural network (NN) is also applied to the BRC estimation in EVs. Two NN models with three layers (input, hidden and output layers) have been proposed. In the input layer, one NN model has the four neurons to represent the battery terminal voltage, discharge current, temperature and internal impedance; the other NN model has the four neurons to represent the battery terminal voltage, discharge current, temperature and discharged capacity. In the output layer, both two NN models have one neuron to indicate the BRC in terms of the SOC. It can be found that the inputs of these two NN models have no representation of the discharge current profile, so this method suffers from the same problem as the aforementioned methods that the discharge current profile has not yet been taken into account in the BRC estimation.
SUl\Il\dARY OF THE INVENTION The present invention provides a new method for the BRC estimation in EVs by using a NN model. The inputs of the NN model are the temperature and the discharged and regenerative capacity distribution. The output of the NN model is the state of available capacity (SOAC) to represents the BRC.
In an aspect of the invention, an embodiment of the invention uses the discharged and regenerative capacity distribution to represent the discharge current profile for the BRC estimation. So, this invention can take the discharge current profile into account in the BRC estimation that overcomes the drawback identified in the previous methods.
Lo In another aspect, an embodiment of the invention uses the SOAC to represent the BRC, which is defined as a percentage of the BAC at the fully charged state for an EV discharge current profile. Mathematically, the SOAC Pa (t) is written as: P a (t) = 1-q (t) / Ca Thus, this invention can provide the actual BRC in terms of the SOAC, which really governs the EV driving range.
Moreover, this invention offers the more general framework for the BRC estimation in EVs by appropriately forming the capacity distribution through selecting the number of the current ranges and the values of the upper and lower current bounds. Consequently, it can be easily adjusted to adapt the BRC estimation for various types of batteries in the presence of very different discharge current profiles in EVs.
BRIEF DESCRIPTION OF TO DRAWINGS
The present invention is described in more detail with reference to the accompanying drawings, of which: FIG. 1 shows the effect of discharge current and temperature on BAC FIG. 2 shows different EV discharge current profiles FIG. 3 shows two examples of 29 tests corresponding to FADS and ECE FIG. 4 shows the NN model for BRC estimation FIG. 5 shows comparison between actual and estimated SOAC for training data set FIG. 6 shows comparison between actual and estimated SOAC for testing data set FIG. 7 shows the ARPEs for all 29 testing data sets DETAILED DESCRIPTION OF TEIE INVENTION
As far as the battery in EVs concerned, the BAC for different EV discharge current profiles is the key value to govern the EV driving range. So, the EV discharge current profiles that emulate different EV operating conditions are designed for the BAC tests, such as FHDS, FUDS, ECE and JM10.15 as shown in FIG. 2. To carry out the BAC tests, the BAC is defined as the quantity of electricity that can be delivered by the fully charged battery at a certain EV discharge current profile and temperature until the specified cutoff voltage is reached. Mathematically, it can be written as: Ca = f(V(t),Id (t),T(t)) Iv=vo (6)
where V(t) is the battery terminal voltage, T(t) is the temperature, and VOW is the specified cutoff voltage. According to this definition, different combinations of the EV discharge current profiles and temperatures are used to test the battery, where the battery at the fully charged state ( Pa (t) = 1) is discharged until the specified cutoff voltage is reached ( Pa (t) = 0). In this preferred embodiment, total 29 tests are carried out, and the experimental data are recorded. FIG. 3 shows two examples of the 29 tests at the temperature of 25 C, namely the discharge current profiles corresponding to the FIJDS and ECE. From the discharged capacities for all the tests, the BACs are obtained and the corresponding SOACs are calculated by using (5). As a result, the relationship between the SOAC and the EV discharge current profile under different temperatures can be expressed by the data representation. To describe such relationship using the NN model for the BRC estimation in terms of the SOAC, the discharged and regenerative capacity distribution is used to represent the discharge current profile for the SOAC estimation. In this preferred embodiment, the capacity distribution based on the lower and upper current bounds of five current ranges as shown in Table 3, namely I; and I,U (i = 1,...,5), is adopted, where CN is the rated capacity of the battery under test.
Table 3 Lower and upper current bounds for discharged and regenerative capacity distribution 1 2 3 4 5
I, (A) O CN / S CN /3 CN /2 CN /1
I, (A) CN /5 CN IS CN /2 CN /1 100
Then, the three-layered AN model for the SOAC estimation can be shown in FIG. 4. The first layer, namely the input layer, has seven neurons: X (t)discharged capacity for It < Id (t) < If; X2 (t)-discharged capacity for I' < Id (t) < I2; X3 (t)-discharged capacity for I3 < Id (t) < If; X4(t)discharged capacity for I4 < Id (t) < Is; X5 (t)-discharged capacity for I5 < Id (t) < I5; X6 (t)-regenerative capacity for regenerative current; X7 (t)-temperature
Considering the vector X(t) = [Xl(t) X2(t) X3(t) X4(t) X5(f) X6(t) X7(t)], this NN model can then be described as a Unction that maps the input vector X(t) to the output vector Pa (I), namely the SOAC at time I. Mathematically, it can be described as: Pa (t) = WiF(y;) + blO l+exp(-2y) (8) where Pa (t) represents the value of the SOAC estimation, n is the number of neurons in the hidden layer, Wi (i = 1,...,n) are the weights between the hidden layer and the output layer, bl is the bias at the output layer, Hi (i = 1,..., n) is the input to the ith neuron in the hidden layer, and F(Yi) is the tangent-sigmoid function. It is given by: y, = WijXj (I) + b,h (9) j=] where Wij (i = 1,...,n, j = 1,...,7) are the weights between the input layer and the hidden layer, and b,h (i = 1,..., n) are the biases at the hidden layer. To identify the necessary number of hidden layer neurons, eight NN candidates with ranging from 8 to 15 are examined.
Consequently, the NN model with eleven hidden layer neurons is chosen because there is no significant improvement in the estimation accuracy for n greater than 11 under those EV discharge current profiles.
The learning algorithm of the NN is a numerical process that determines the connection strength, such as the weights among the layers and the biases in the neurons. In the learning process, the validation data set is included for the improvement of the generality of the NN. Under this condition, the learning process will be terminated when the error function on the validation data set begins to increase or the error Unction is smaller than the convergence tolerance, whichever is reached first. In this preferred embodiment, the convergence tolerance is set at 10-5. The error function E is defined as: m E =- (Pa (k)-Pa (k)) ( 10) 2 k=1 where m is the number of the training data, pa(k) is the SOAC calculated from the experimental data for the kth training data, and Pa (k) is the corresponding value of the SOAC estimated by the NN model.
The parameters of this NN model are optimized by the Levenberg-Marquardt algorithm, one of the improved back-propagation algorithms. This algorithm is a variation of Newton's method designed for minimizing functions that are sums of squares of other nonlinear functions. So, it is well suited to minimize the error function as defined in (10).
With this algorithm, E can be expressed as a function of the parameters of the NN: H = ii,b,Wij,b,} (i = 1,...,n, j = 1,...,7) (11) The optimum parameters of the NN can be obtained through the following iterative process: Hr+ =Hr-Ar gr (12) where Ar - V2E(H) IN=H and go - VE(H) IH=H are the Hessian matrix and the gradient vector of E with respect to the r th iteration, respectively.
To train the NN by the back-propagation algorithm effectively, the inputs of the NN are generally normalized by the following equation: Xj (t) = j() me (j = 1,..,7) (13) Xjmax Xjmin where Xjn(t) is the normalized value, XjaX and Xj,,,,r, are the maximum and minimum values of Xj (I), respectively. After normalization, the data for each test are put together to form the whole data set, which is then divided into the training, validation and testing data sets. After the data preparation is finished, the training data set is used to train the NN while the testing data set is used to verify the accuracy and effectiveness of the trained NN for the SOAC estimation. For comparisons, the average relative percentage error (ARPE) is adopted.
It is defined as: ARPE = N I Pae I i) (P;1 ( j) 1 100% (14) where N is the number of the training data or the testing data for each test, Pae and Pac refer to the estimated SOAC by the trained NN and the actual SOAC calculated from the experimental data, respectively. The ARPEs for both the training data set and the testing data set for each test are calculated. FIG. 5 shows the estimated SOAC and the actual SOAC for the training data set. It can be found that the SOAC estimation is of high accuracy and the corresponding ARPE is only of 1.27%. To testify the trained NN for the SOAC estimation effectively, the testing data for each test are used to verify the trained NN. The results corresponding to the FADS and the ECE are of 1.22% and 1.28%, respectively (shown in
FIG. 6). It should be noted that the ARPEs of the SOAC estimation for the aforementioned 29 tests in this example are within 2% as summarized in FIG. 7, which illustrate that the proposed approach can provide highly accurate estimation of the SOAC for different EV discharge current profiles.
Therefore, the NN model for the BRC estimation in terms of the SOAC in EVs can be realized by the following procedures. The current that includes either discharge current or regenerative current from the battery in EVs is measured. This current is summed to produce the discharged and regenerative capacity distribution based on the selected current ranges and selected values of the lower and upper current bounds as shown in Table 3. This capacity distribution together with the temperature is formed as one vector. The raw data of this vector is normalized by (13). The normalized vector is then used as the inputs of the NN model for the BRC estimation in terms of the SOAC by (7)-(9).
A method for the BRC estimation in terms of the SOAC in EVs using the NN model is disclosed. This method uses the discharged and regenerative capacity distribution to represent the discharge current profiles. Consequently, the capacity distribution, which is properly formed by selecting the number of the current ranges and the values of the lower and upper current bounds, can be easily adjusted to adapt the complexity and nonlinearity of the characteristics of different types of batteries in EVs. Furthermore, the SOAC is used, instead of the SOC, to estimate the BRC, leading to the actual BRC on which the EV driving range really depends.
It will be appreciated that the various features described herein may be used singly or in any combination thereof. Thus, the present invention is not limited to only the embodiments specifically described herein. While the foregoing description and drawings
represent a preferred embodiment of the present invention, it will be understood that various additions, modifications, and substitutions may be made therein without departing from the spirit and scope of the present invention in its broader aspects. Therefore, the disclosed embodiment is therefore to be considered in all respects as illustrative and not restrictive, the scope of the invention being indicated by the appended claims, and not limited to the foregoing description.
Claims (6)
1. The NN model for the BRC estimation in terms of the SOAC in EVs consists of three layers. The first layer is an input layer including a plurality of processing units that represent the discharged and regenerative capacity distribution and the temperature. The second layer is a hidden layer including a plurality of non-linear processing units. The third layer is an output layer including a linear processing unit to represent the SOAC.
2. The NN model for the BRC estimation in terms of the SOAC in EVs according to claim 1, wherein the SOAC refers to the percentage of the BAC for a certain EV discharge current profile. The values of the SOAC ranges from 0 (corresponding to the battery at the fully discharged state) to 1 (corresponding to the battery at the fully charged state).
3. The NN model for the BRC estimation in terms of the SOAC in BVs according to claim I, wherein the discharged and regenerative capacity distribution is formed by summing the current based on the selected current ranges and selected values of the lower and upper current bounds.
4. The NN model for the BRC estimation in terms of the SOAC in EVs according to claim 3, wherein the current includes either discharge current or regenerative current.
5. The NN model for the BRC estimation in terms of the SOAC in EVs according to claim 1, wherein the temperature refers to that surrounding the battery.
6. The method for the battery residual capacity estimation in electric vehicles by using the neural network model according to claim 1 comprises steps of: measuring the current and the temperature from the battery in electric vehicles, summing the current based on the selected current ranges and the selected values of the lower and upper current bounds to form the discharged and regenerative capacity distribution, putting the discharged and regenerative capacity distribution and temperature together to form the input vector, using the input vector as the input of the neural network model to estimate the battery residual capacity in terms of the state of available capacity in electric vehicles.
6. The NN model for the BRC estimation in terms of the SOAC in BVs according to claim 1 comprises full inter-layer connection and no intralayer connections.
7. The NN model for the BRC estimation in terms of the SOAC in EVs according to claim 1, the input layer passes the inputs directly into the hidden layer.
8. The NN model for the BRC estimation in terms of the SOAC in EVs according to claim 1, the non-linear processing units in the hidden layer comprise the tangent-sigmoid function.
9. The NN model for the BRC estimation in terms of the SOAC in EVs according to claim 1, the linear processing unit in the output layer comprises the linear function.
10. The NN model for the BRC estimation in terms of the SOAC in EVs according to claim 1, its parameters are determined by the steps of: designing the EV discharging current profiles in according with the EV operating conditions; using these discharge current profiles to test the battery under different temperatures;
collecting the experimental data that include the discharge current, regenerative current and temperature for all EV discharge current profiles; calculating the SOAC for each EV discharge current profile; selecting the number of the current ranges in accordance with the current variation ranges of the EV discharge current profiles.
selecting the values of the lower and upper current bounds in accordance with the effect of discharge current on BAC among the EV discharge current profiles; forming the discharged and regenerative capacity distribution based on the selected current ranges and the selected values of the lower and upper current bounds; putting the experimental data of the capacity distribution, the temperature and the SOAC together to form a whole data set; normalizing the raw data in the whole data set to produce the training data set and the testing data set; establishing the relationship between the SOAC and the capacity distribution under different temperatures by using the NN model; training the NN model by the using the training data set through learning algorithm to determine the parameters of the NN model; and verifying the NN model by using the testing data set.
11. A method for the BRC estimation in terms of the SOAC by using the NN model according to claim 1 comprises steps of: measuring the temperature and the current from the battery in EVs; summing the current based on the selected current ranges and the selected values of the lower and upper current bounds to form the discharged and regenerative capacity distribution; putting the capacity distribution and temperature together to form one vector; normalizing this vector to be suitable for the inputs of the NN model; and using the NN model to estimate the BRC in terms of the SOAC.
Amendments to the claims have been filed as follows il I claim: 1. The method for the battery residual capacity estimation in electric vehicles by using the neural network model, wherein the neural network model consists of three layers: an input layer including a plurality of processing units to represent the discharged and regenerative capacity distribution and the temperature, a hidden layer including a plurality of non-linear processing units, and an output layer including a linear processing unit to represent the battery residual capacity in terms of the state of available capacity.
2. The method for the battery residual capacity estimation in electric vehicles by using the neural network model according to claim 1, wherein the value of the battery residual capacity in terms of the state of available capacity ranges from O to 1.
3. The method for the battery residual capacity estimation in electric vehicles by using the neural network model according to claim 1, wherein the discharge current profile is represented by the discharged and regenerative capacity distribution and not by instantaneous current.
4. The method for the battery residual capacity estimation in electric vehicles by using the neural network model according to claim 2, wherein the discharged and regenerative capacity distribution is formed by summing the current based on the selected current ranges and selected values of the lower and upper current bounds.
5. The method for the battery residual capacity estimation in electric vehicles by using the neural network model according to claim 1 comprises steps of: providing a neural network structure with an input layer, a nonlinear hidden layer and a linear output layer, calculating the state of available capacity from the experimental data obtained under the discharge current profile in according with electric vehicle operating conditions, forming the discharged and regenerative capacity distribution based on the selected current ranges and the selected values of lower and upper current bounds,
putting the discharged and regenerative capacity distribution, temperature and corresponding state of available capacity together to obtain a training data set, training the neural network model by using the training data set, establishing the relationship between the discharged and regenerative capacity distribution and the state of available capacity under different temperatures by using the neural networl; model.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
GB0206316A GB2386709B (en) | 2002-03-18 | 2002-03-18 | Battery residual capacity estimation for electric vehicles |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
GB0206316A GB2386709B (en) | 2002-03-18 | 2002-03-18 | Battery residual capacity estimation for electric vehicles |
Publications (3)
Publication Number | Publication Date |
---|---|
GB0206316D0 GB0206316D0 (en) | 2002-05-01 |
GB2386709A true GB2386709A (en) | 2003-09-24 |
GB2386709B GB2386709B (en) | 2004-03-17 |
Family
ID=9933164
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
GB0206316A Expired - Fee Related GB2386709B (en) | 2002-03-18 | 2002-03-18 | Battery residual capacity estimation for electric vehicles |
Country Status (1)
Country | Link |
---|---|
GB (1) | GB2386709B (en) |
Cited By (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
GB2444511A (en) * | 2006-12-06 | 2008-06-11 | Iti Scotland Ltd | Neural network battery management |
CN105974327A (en) * | 2016-06-12 | 2016-09-28 | 广州市香港科大***研究院 | Lithium battery pack SOC prediction method based on neural network and UKF |
DE102006018208B4 (en) * | 2005-04-20 | 2017-11-23 | Denso Corporation | A method and apparatus for detecting a charged state of a secondary battery based on a neural network calculation |
Families Citing this family (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111191824B (en) * | 2019-12-20 | 2024-03-26 | 北京理工新源信息科技有限公司 | Power battery capacity attenuation prediction method and system |
CN113554200B (en) * | 2020-04-23 | 2024-02-27 | 广州汽车集团股份有限公司 | Method, system and equipment for predicting voltage inconsistency of power battery |
CN116577673B (en) * | 2023-07-12 | 2023-09-12 | 深圳先进储能材料国家工程研究中心有限公司 | Distributed neural network-based energy storage power station fault diagnosis method and system |
Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JPH09236641A (en) * | 1995-12-28 | 1997-09-09 | Yazaki Corp | Battery residual quantity measurement device |
US5936385A (en) * | 1995-10-31 | 1999-08-10 | U.S. Philips Corporation | System monitoring the discharging period of the charging/discharging cycles of a rechargeable battery, and host device including a smart battery |
US6064180A (en) * | 1996-10-29 | 2000-05-16 | General Motors Corporation | Method and apparatus for determining battery state-of-charge using neural network architecture |
EP1081499A1 (en) * | 1998-05-28 | 2001-03-07 | Toyota Jidosha Kabushiki Kaisha | Means for estimating charged state of battery and method for estimating degraded state of battery |
-
2002
- 2002-03-18 GB GB0206316A patent/GB2386709B/en not_active Expired - Fee Related
Patent Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US5936385A (en) * | 1995-10-31 | 1999-08-10 | U.S. Philips Corporation | System monitoring the discharging period of the charging/discharging cycles of a rechargeable battery, and host device including a smart battery |
JPH09236641A (en) * | 1995-12-28 | 1997-09-09 | Yazaki Corp | Battery residual quantity measurement device |
US6064180A (en) * | 1996-10-29 | 2000-05-16 | General Motors Corporation | Method and apparatus for determining battery state-of-charge using neural network architecture |
EP1081499A1 (en) * | 1998-05-28 | 2001-03-07 | Toyota Jidosha Kabushiki Kaisha | Means for estimating charged state of battery and method for estimating degraded state of battery |
Cited By (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
DE102006018208B4 (en) * | 2005-04-20 | 2017-11-23 | Denso Corporation | A method and apparatus for detecting a charged state of a secondary battery based on a neural network calculation |
GB2444511A (en) * | 2006-12-06 | 2008-06-11 | Iti Scotland Ltd | Neural network battery management |
GB2444511B (en) * | 2006-12-06 | 2008-10-22 | Iti Scotland Ltd | Battery Management System |
CN105974327A (en) * | 2016-06-12 | 2016-09-28 | 广州市香港科大***研究院 | Lithium battery pack SOC prediction method based on neural network and UKF |
Also Published As
Publication number | Publication date |
---|---|
GB2386709B (en) | 2004-03-17 |
GB0206316D0 (en) | 2002-05-01 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
Hannan et al. | Neural network approach for estimating state of charge of lithium-ion battery using backtracking search algorithm | |
Shen et al. | Adaptive neuro-fuzzy modeling of battery residual capacity for electric vehicles | |
CN112838631B (en) | Dynamic charge management and control device for power battery and charge diagnosis method for power battery | |
CN105974327A (en) | Lithium battery pack SOC prediction method based on neural network and UKF | |
CN103675707B (en) | Lithium ion battery peak power online evaluation method | |
CN109061520A (en) | A kind of power battery health and power rating estimation on line method and system | |
CN110068774A (en) | Estimation method, device and the storage medium of lithium battery health status | |
CN106443453A (en) | Lithium battery SOC estimation method based on BP neural network | |
CN104833922B (en) | The computational methods and device of battery charging and discharging current limit | |
CN111007399A (en) | Lithium battery state of charge prediction method based on improved generation countermeasure network | |
CN111983459B (en) | Health state test evaluation method based on vehicle lithium ion battery | |
CN110350517B (en) | Electric automobile grid-connected power distribution network reconstruction method based on operation risk | |
CN112632850A (en) | Method and system for detecting abnormal battery in lithium battery pack | |
CN106501728A (en) | A kind of battery equivalent model parameter identification method based on multi-objective genetic algorithm | |
CN106777786A (en) | A kind of lithium ion battery SOC estimation method | |
CN112858916A (en) | Battery pack state of charge estimation method based on model and data drive fusion | |
CN106998086A (en) | MW class energy-accumulating power station battery management method and its system | |
CN111426962B (en) | Battery parameter identification method and terminal | |
Zhao et al. | Estimation of the SOC of energy-storage lithium batteries based on the voltage increment | |
CN111325402A (en) | Method for predicting charging behavior of electric vehicle user based on BP neural network | |
GB2386709A (en) | A neural network for estimating battery residual capacity in an electric vehicle | |
CN116298904A (en) | Method for jointly estimating SOC and SOH of lithium battery based on fractional order multi-innovation double unscented Kalman filtering | |
Naguib et al. | Comparative Study between equivalent circuit and recurrent neural network battery voltage models | |
CN114545279A (en) | Lithium battery health state estimation method based on neural network ordinary differential equation | |
CN111999651B (en) | Power performance test method of lithium ion battery for pure electric automobile |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PCNP | Patent ceased through non-payment of renewal fee |
Effective date: 20080318 |