CN109884548B - Method for predicting remaining life of lithium battery based on GASVM-AUKF algorithm - Google Patents
Method for predicting remaining life of lithium battery based on GASVM-AUKF algorithm Download PDFInfo
- Publication number
- CN109884548B CN109884548B CN201910160991.2A CN201910160991A CN109884548B CN 109884548 B CN109884548 B CN 109884548B CN 201910160991 A CN201910160991 A CN 201910160991A CN 109884548 B CN109884548 B CN 109884548B
- Authority
- CN
- China
- Prior art keywords
- algorithm
- lithium battery
- aukf
- residual
- data
- 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.)
- Active
Links
Images
Landscapes
- Secondary Cells (AREA)
- Tests Of Electric Status Of Batteries (AREA)
Abstract
The embodiment of the invention provides a method for predicting the residual life of a lithium battery based on a GASVM-AUKF algorithm, which comprises the following steps: acquiring capacity data of a lithium battery, and establishing a state space equation according to the capacity data; acquiring residual error data corresponding to the lithium battery based on the state space equation and the Adaptive Unscented Kalman Filter (AUKF) algorithm; calculating a predicted value of the residual data based on a Support Vector Machine (SVM) algorithm; and predicting the residual life of the battery of the lithium battery based on the AUKF algorithm and the predicted value of the residual data. According to the lithium battery residual life prediction method and system provided by the embodiment of the invention, the AUKF algorithm is adopted, so that the self-adaptive updating of the process noise covariance and the observation noise covariance can be realized, the influence of noise on the whole filtering effect is reduced, the filtering precision is improved, and the accurate residual life prediction result is realized.
Description
Technical Field
The embodiment of the invention relates to the technical field of batteries, in particular to a method for predicting the residual life of a lithium battery based on a GASVM-AUKF algorithm.
Background
With the development of the economy and the increase of the population of countries in the world, the energy problem becomes the focus of people's attention. Lithium batteries are widely used in people's daily life due to their advantages of high capacity density, low self-discharge rate, high safety, long cycle life, etc.
However, during the use of the lithium battery, the function and performance of the lithium battery are inevitably degraded due to the complexity and various uncertain factors, and finally the lithium battery is out of work. Failure of a lithium battery can cause both economic losses and significant accident disasters. At present, in the prior art, the capacity of a battery is predicted and the remaining service life of the battery is estimated in a mode of realizing the estimation of the RUL of the lithium battery by adopting a data model combined idea.
However, the method provided by the prior art has high noise, so that the filtering effect precision is low, and the prediction result is inaccurate, so that a method for predicting the residual life of the lithium battery based on the GASVM-AUKF algorithm is urgently needed to solve the problems.
Disclosure of Invention
In order to solve the above problems, embodiments of the present invention provide a method for predicting remaining life of a lithium battery based on a gassvm-auckf algorithm, which overcomes or at least partially solves the above problems.
In a first aspect, an embodiment of the present invention provides a method for predicting a remaining life of a lithium battery based on a gassvm-auckf algorithm, including:
acquiring capacity data of a lithium battery, and establishing a state space equation according to the capacity data;
acquiring residual error data corresponding to the lithium battery based on the state space equation and the Adaptive Unscented Kalman Filter (AUKF) algorithm;
calculating a predicted value of the residual data based on a Support Vector Machine (SVM) algorithm;
and predicting the residual life of the battery of the lithium battery based on the AUKF algorithm and the predicted value of the residual data.
According to the lithium battery residual life prediction method based on the GASVM-AUKF algorithm, provided by the embodiment of the invention, the AUKF algorithm is adopted, so that the self-adaptive updating of the process noise covariance and the observation noise covariance can be realized, the influence of noise on the whole filtering effect is reduced, the filtering precision is improved, and the accurate residual life prediction result is realized.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the embodiments or the prior art will be briefly described below, and it is obvious that the drawings in the following description are some embodiments of the present invention, and those skilled in the art can also obtain other drawings according to the drawings without creative efforts.
Fig. 1 is a schematic flow chart of a method for predicting the remaining life of a lithium battery based on a gassvm-AUKF algorithm according to an embodiment of the present invention.
Detailed Description
In order to make the objects, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are some embodiments, but not all embodiments, of the present invention. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
At present, the traditional idea of combining data models is still adopted for a prediction mode of the residual life of a battery in the prior art, but the mode in the prior art does not correct the influence of noise on the whole algorithm, so that the precision of a filtering effect is low.
To solve the problems in the prior art, fig. 1 is a schematic flow chart of a method for predicting the remaining life of a lithium battery based on a gassvm-AUKF algorithm according to an embodiment of the present invention, as shown in fig. 1, including:
101. acquiring capacity data of a lithium battery, and establishing a state space equation according to the capacity data;
102. acquiring residual error data corresponding to the lithium battery based on the state space equation and the Adaptive Unscented Kalman Filter (AUKF) algorithm;
103. calculating a predicted value of the residual data based on a Support Vector Machine (SVM) algorithm;
104. and predicting the residual life of the battery of the lithium battery based on the AUKF algorithm and the predicted value of the residual data.
Specifically, in step 101, battery capacity data in the use process of the lithium battery is first obtained in the embodiment of the present invention, and it should be noted that the battery capacity data is stored data corresponding to time and capacity, and then a state space equation with a time state as a reference may be established according to the battery capacity data. Specifically, the embodiment of the present invention may select different initial prediction points for different battery data, and establish a state space equation by using the battery capacity data recorded before the prediction points.
Further, in step 102, the data is filtered by using an Adaptive Unscented Kalman Filter (AUKF) algorithm, and it can be understood that the AUKF algorithm is a novel nonlinear filtering algorithm, and estimates system parameters by using an estimated state according to a known system parameter estimation state in a loop iteration manner, and has good adaptability. By combining the state space equation obtained in step 101, the AUKF algorithm provided by the embodiment of the invention can calculate and obtain the residual data corresponding to the lithium battery, wherein the residual data refers to the difference between the observed data and the predicted data, so that the residual data at each moment before the initial prediction point is obtained.
In step 103, according to the obtained residual data, the value of the residual data is predicted by using the SVM algorithm, it should be noted that the SVM algorithm is actually a model trained by using the phase space reconstruction method, and the corresponding predicted value can be output after the residual data is input into the model. Preferably, the residual data obtained in step 102 needs to be screened, and then predicted after removing the abnormal value.
Finally, in step 104, the embodiment of the present invention predicts a new model parameter, that is, a state estimation value at the time of predicting T + i, by combining the predicted value of the residual data with the AUKF algorithm
And T is the initial predicted point time, so that the predicted capacity of the battery at the time T + i (i is 1,2, …, n) can be obtained according to the prediction result and by combining the calculation method of the measurement estimation value in the AUKF algorithm.
According to the lithium battery residual life prediction method based on the GASVM-AUKF algorithm, provided by the embodiment of the invention, the AUKF algorithm is adopted, so that the self-adaptive updating of the process noise covariance and the observation noise covariance can be realized, the influence of noise on the whole filtering effect is reduced, the filtering precision is improved, and the accurate residual life prediction result is realized.
On the basis of the above embodiment, the method further includes:
and optimizing algorithm parameters in the SVM algorithm based on a genetic algorithm when calculating the predicted value of the residual data.
From the content of the above embodiment, the method provided by the embodiment of the present invention adopts an SVM algorithm to predict residual data.
Preferably, the embodiment of the present invention adds a genetic algorithm (genetic algorithm) to the kernel function K (x) in the SVM algorithm in the predictioni,xt) And optimizing the punishment coefficient C to finally obtain more accurate residual data
Specifically, the genetic algorithm optimization SVM algorithm comprises the following steps:
1. and selecting a part of residual data obtained by an AUKF algorithm as a training set, and using the latter part as a test set to establish a training model.
2. And then setting the population number, the maximum evolution algebra, the cross probability, the mutation probability and the upper and lower limits of the parameter pair, and randomly generating an initial variable when the time t is 0.
3. Training individuals in the current population as hyper-parameters, and training the SVM by using the verification data set to obtain the target function, the position of the target point and the optimal target value of the current position.
4. Selecting, crossing and mutating the population, using SVM to train and update the position of the solution, the position of the target point and the optimal target value, and setting the evolution generation t as t + 1.
5. If the maximum algebra is reached, the optimal parameters of sigma, C and gamma are used as hyper-parameters, an optimal model is obtained through a training data set, and if not, the step 3 is executed for continuous iteration.
6. New residual data is trained using the model obtained in step 5.
The genetic algorithm and the SVM algorithm obtain predicted residual data, and n-step prediction effects can be obtained by using n one-step predictions:
on the basis of the above embodiment, the establishing a state space equation according to the capacity data includes:
determining a starting prediction point in the capacity data;
and establishing the state space equation according to the capacity data before the initial prediction point and a least square method.
According to the content of the embodiment, the state space equation is established according to the capacity data, in the actual operation process, different initial prediction points are selected according to different battery data, the corresponding state space equation is established by using the experimental data before the prediction points and combining the least square method, and the data after the prediction points are used as test samples, so that the prediction accuracy of the embodiment is verified.
Specifically, in the simulation experiment process, four groups of battery data of B0005, B0006, B0007 and B0018 in NASA are selected to perform the simulation experiment, and the established state space equation is as follows:
wherein x isk＝[ak,bk,ck,dk]K denotes the number of cycles, xkRepresenting state model parameters, QkAn observed value representing a capacity, and wkAnd vkRespectively representing state gaussian white noise and observation gaussian white noise.
On the basis of the above embodiment, obtaining residual data corresponding to the lithium battery based on the state space equation and the Adaptive Unscented Kalman Filter (AUKF) algorithm includes:
carrying out lossless transformation on the state space equation;
and predicting and updating the noise covariance matrix based on the result of the lossless transformation to obtain residual data corresponding to the lithium battery.
On the basis of a state space equation, the embodiment of the invention obtains T residual error data before an initial prediction point by using an AUKF algorithm, and the specific flow is as follows:
1) based on a state space equation, carrying out lossless UT conversion, wherein the specific process is as follows:
the state space equation is as follows:
2n +1 sigma points, i.e. sample points, are calculated, where n refers to the dimension of the state. Preferably, in the embodiment of the present invention, since the state vector dimension is 4, n is 4.
Wherein the content of the first and second substances,
the ith column representing the square root of the matrix.
Then calculating the corresponding weight of the sampling point to obtain the result of UT conversion:
2) and a prediction part:
a set of sampling points (called Sigma point set) and their corresponding weights are obtained by the above two formulas:
calculate a one-step prediction of 2n +1 Sigma point sets, i ═ 1,2, …,2n + 1:
X(i)(k+1|k)＝f[k,X(i)(k|k)]；
and calculating a one-step prediction and covariance matrix of the system state quantity, wherein the one-step prediction and covariance matrix is obtained by weighting and summing the predicted values of the Sigma point set:
3) and an updating part:
from the one-step predicted values, the UT transform is used again to generate a new Sigma point set:
substituting the predicted Sigma point set obtained in the above steps into an observation equation to obtain a predicted observed quantity, i is 1,2, …,2n + 1:
Z(i)(k+1|k)＝h[X(i)(k+1|k)]；
and obtaining the average value and covariance of system prediction by weighting and summing the obtained observation predicted value:
the kalman gain is then calculated:
state update and covariance update of the computing system:
calculating a measurement estimation value:
and finally, updating the process noise covariance and the observation noise covariance:
wherein Z iskRepresenting the capacity value of the lithium battery, corresponding to Q in the state space equationkObtaining residual data e before T time through AUKF algorithm1:TI.e. corresponding to the state update part of the AUKF algorithm
Each value of k corresponds to a residual data.
On the basis of the above embodiment, the calculating a prediction value of the residual data based on a support vector machine SVM algorithm includes:
and calculating the predicted value of the residual data according to the support vector machine of the epsilon-SVR and the radial basis kernel function.
From the content of the above embodiment, it can be seen that the embodiment of the present invention trains residual data by using an SVM algorithm, wherein an epsilon-SVM method which is more sensitive to the residual data is preferably selected, the obtained residual data is screened, after removing an abnormal value, is used as an input of an epsilon-SVM, and is trained to obtain new predicted residual data, wherein the selected SVM kernel function is a radial basis kernel function, and a specific expression thereof is as follows:
on the basis of the above embodiment, predicting the remaining battery life of the lithium battery based on the AUKF algorithm and the predicted value of the residual error data includes:
calculating the predicted capacity of the lithium battery based on the AUKF algorithm and the predicted value of the residual data;
and if the predicted capacity of the lithium battery is larger than a preset capacity threshold, judging that the prediction is accurate so as to obtain the remaining battery life of the lithium battery.
It can be understood that the obtained new residual value is substituted into the AUKF algorithm again, and at this time, a new measured value does not need to be added, so that the defect that the AUKF algorithm can only perform one-step prediction is changed, and the purpose of multi-step prediction is achieved. And after obtaining a new estimated value by using an AUKF algorithm, comparing the new estimated value with a capacity threshold, and when the previously defined capacity threshold is reached, taking the RUL as the time from the prediction starting point to the failure threshold point, and analyzing and comparing all estimated capacity values between the prediction starting point and the failure point with indexes such as root mean square error, absolute error percentage and the like of a test set, namely a real capacity value, so as to verify the accuracy of prediction.
The remaining battery life of the lithium battery can be obtained by accurately estimating the capacity value at the subsequent moment.
Through the above description of the embodiments, those skilled in the art will clearly understand that each embodiment can be implemented by software plus a necessary general hardware platform, and certainly can also be implemented by hardware. With this understanding in mind, the above-described technical solutions may be embodied in the form of a software product, which can be stored in a computer-readable storage medium such as ROM/RAM, magnetic disk, optical disk, etc., and includes instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) to execute the method according to each embodiment or some parts of the embodiments.
Finally, it should be noted that: the above examples are only intended to illustrate the technical solution of the present invention, but not to limit it; although the present invention has been described in detail with reference to the foregoing embodiments, it will be understood by those of ordinary skill in the art that: the technical solutions described in the foregoing embodiments may still be modified, or some technical features may be equivalently replaced; and such modifications or substitutions do not depart from the spirit and scope of the corresponding technical solutions of the embodiments of the present invention.
Claims (3)
1. A method for predicting the residual life of a lithium battery based on a GASVM-AUKF algorithm is characterized by comprising the following steps:
acquiring capacity data of a lithium battery, and establishing a state space equation according to the capacity data;
acquiring residual error data corresponding to the lithium battery based on the state space equation and the Adaptive Unscented Kalman Filter (AUKF) algorithm;
calculating a predicted value of the residual data based on a Support Vector Machine (SVM) algorithm;
predicting the residual life of the battery of the lithium battery based on the AUKF algorithm and the predicted value of the residual data;
the obtaining of the residual data corresponding to the lithium battery based on the state space equation and the Adaptive Unscented Kalman Filter (AUKF) algorithm includes:
carrying out lossless transformation on the state space equation;
predicting and updating a noise covariance matrix based on a result of lossless transformation to obtain residual data corresponding to the lithium battery;
the method further comprises the following steps:
optimizing algorithm parameters in the SVM algorithm based on a genetic algorithm when calculating the predicted value of the residual data;
the calculating the predicted value of the residual data based on the SVM algorithm comprises the following steps:
and calculating the predicted value of the residual data according to the support vector machine of the epsilon-SVR and the radial basis kernel function.
2. The method of claim 1, wherein establishing a state space equation from the capacity data comprises:
determining a starting prediction point in the capacity data;
and establishing the state space equation according to the capacity data before the initial prediction point and a least square method.
3. The method of claim 1, wherein predicting the battery remaining life of the lithium battery based on the AUKF algorithm and the predicted values of residual data comprises:
calculating the predicted capacity of the lithium battery based on the AUKF algorithm and the predicted value of the residual data;
and if the predicted capacity of the lithium battery is larger than a preset capacity threshold, judging that the prediction is accurate so as to obtain the remaining battery life of the lithium battery.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910160991.2A CN109884548B (en) | 2019-03-04 | 2019-03-04 | Method for predicting remaining life of lithium battery based on GASVM-AUKF algorithm |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910160991.2A CN109884548B (en) | 2019-03-04 | 2019-03-04 | Method for predicting remaining life of lithium battery based on GASVM-AUKF algorithm |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN109884548A CN109884548A (en) | 2019-06-14 |
| CN109884548B true CN109884548B (en) | 2021-07-20 |
Family
ID=66930441
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910160991.2A Active CN109884548B (en) | 2019-03-04 | 2019-03-04 | Method for predicting remaining life of lithium battery based on GASVM-AUKF algorithm |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109884548B (en) |
Families Citing this family (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113125969B (en) * | 2020-01-14 | 2022-07-15 | 比亚迪股份有限公司 | Battery data processing method, device and medium based on AUKF |
| CN116466244B (en) * | 2023-06-20 | 2023-08-29 | 中国华能集团清洁能源技术研究院有限公司 | Battery state of charge anomaly detection method based on support vector machine |
Family Cites Families (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US7197487B2 (en) * | 2005-03-16 | 2007-03-27 | Lg Chem, Ltd. | Apparatus and method for estimating battery state of charge |
| CN102680907B (en) * | 2012-05-31 | 2015-02-11 | 浙江大学 | Battery charging stress optical coefficient (SOC) detection method in photovoltaic system |
| DE102013018405A1 (en) * | 2013-11-02 | 2015-05-07 | Daimler Ag | Determination of the condition of a battery characteristic estimation parameters |
| CN103744026A (en) * | 2013-12-19 | 2014-04-23 | 广西科技大学 | Storage battery state of charge estimation method based on self-adaptive unscented Kalman filtering |
| CN104502851A (en) * | 2014-12-12 | 2015-04-08 | 广西科技大学 | SOC (Stress Optical Coefficient) estimation method based on AUKF (Adaptive Unscented Kalman Filter) algorithm |
| CN104714188B (en) * | 2015-03-31 | 2017-05-24 | 桂林电子科技大学 | Method and system for estimating measured noise variance matrix matched battery state of charge (SOC) |
| CN106093783A (en) * | 2016-06-03 | 2016-11-09 | 哈尔滨工业大学 | The battery SOC method of estimation that Kalman filtering merges with data-driven |
| CN106291376A (en) * | 2016-07-29 | 2017-01-04 | 华晨汽车集团控股有限公司 | Lithium battery SOC method of estimation based on supporting vector machine model and Kalman filtering |
| CN107064811A (en) * | 2017-03-01 | 2017-08-18 | 华南理工大学 | A kind of lithium battery SOC On-line Estimation methods |
| CN108872790A (en) * | 2018-07-10 | 2018-11-23 | 武汉科技大学 | A kind of MMC method for diagnosing faults based on Kalman filtering and support vector machines |
-
2019
- 2019-03-04 CN CN201910160991.2A patent/CN109884548B/en active Active
Also Published As
| Publication number | Publication date |
|---|---|
| CN109884548A (en) | 2019-06-14 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110798365B (en) | Neural network-based traffic prediction method and device | |
| KR102239464B1 (en) | Methods and apparatuses for forecasting power demand using deep structure | |
| CN109884548B (en) | Method for predicting remaining life of lithium battery based on GASVM-AUKF algorithm | |
| CN114580493B (en) | Heavy haul railway bridge health monitoring method based on AI | |
| CN110082682A (en) | A kind of lithium battery charge state estimation method | |
| CN115291116A (en) | Energy storage battery health state prediction method and device and intelligent terminal | |
| CN112288137A (en) | LSTM short-term load prediction method and device considering electricity price and Attention mechanism | |
| CN108961460B (en) | Fault prediction method and device based on sparse ESGP (Enterprise service gateway) and multi-objective optimization | |
| CN114297036A (en) | Data processing method and device, electronic equipment and readable storage medium | |
| CN112365033B (en) | Wind power interval prediction method, system and storage medium | |
| CN112819225A (en) | Carbon market price prediction method based on BP neural network and ARIMA model | |
| CN111523727B (en) | Method for predicting remaining life of battery by considering recovery effect based on uncertain process | |
| CN114881318A (en) | Lithium battery health state prediction method and system based on generation countermeasure network | |
| CN114660497A (en) | Lithium ion battery service life prediction method aiming at capacity regeneration phenomenon | |
| CN114611372A (en) | Industrial equipment health prediction method based on Internet of things edge calculation | |
| CN113408138B (en) | Lithium battery SOH estimation method and system based on secondary fusion | |
| CN111400964B (en) | Fault occurrence time prediction method and device | |
| JP7069732B2 (en) | Estimator program, estimation method and estimation device | |
| CN114066405A (en) | Data processing method and device | |
| CN117155806A (en) | Communication base station flow prediction method and device | |
| CN112488399A (en) | Power load prediction method and device | |
| CN110322342A (en) | Borrow or lend money construction method, system and the debt-credit Risk Forecast Method of risk forecast model | |
| CN113151842B (en) | Method and device for determining conversion efficiency of wind-solar complementary water electrolysis hydrogen production | |
| CN112685900B (en) | Power load simulation method for representing impact load power characteristics | |
| CN114648178A (en) | Operation and maintenance strategy optimization method of electric energy metering device based on DDPG 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 |