CN115808633A - POA-SVR-based method for monitoring state of energy storage battery in island environment - Google Patents
POA-SVR-based method for monitoring state of energy storage battery in island environment Download PDFInfo
- Publication number
- CN115808633A CN115808633A CN202211630297.0A CN202211630297A CN115808633A CN 115808633 A CN115808633 A CN 115808633A CN 202211630297 A CN202211630297 A CN 202211630297A CN 115808633 A CN115808633 A CN 115808633A
- Authority
- CN
- China
- Prior art keywords
- energy storage
- storage battery
- state
- value
- monitoring
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Images
Classifications
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02E—REDUCTION OF GREENHOUSE GAS [GHG] EMISSIONS, RELATED TO ENERGY GENERATION, TRANSMISSION OR DISTRIBUTION
- Y02E60/00—Enabling technologies; Technologies with a potential or indirect contribution to GHG emissions mitigation
- Y02E60/10—Energy storage using batteries
Landscapes
- Secondary Cells (AREA)
Abstract
The invention discloses a method for monitoring the state of an energy storage battery in a sea island environment based on POA-SVR (point of arrival-support vector regression), which effectively solves the problem of difficult selection of the kernel parameters of the SVR algorithm through the POA algorithm. Meanwhile, in consideration of the particularity of the island climate environment, the invention extracts 11 groups of multidimensional characteristic vectors from the basic monitoring quantity of the battery by using the energy storage battery data obtained in the island environment, trains a POA-SVR prediction model, and monitors the SOH of the energy storage battery in real time on line by using the trained model; meanwhile, a POA-SVR charge state prediction model is trained to perform online SOC monitoring on the energy storage battery in the island environment, so that the purpose of SOC and SOH combined prediction is achieved, and technical support is provided for stable operation of the energy storage battery in the island environment.
Description
Technical Field
The invention belongs to the technical field of energy storage battery state monitoring, and particularly relates to a POA-SVR-based method for monitoring the state of an energy storage battery in an island environment.
Background
As an indispensable part of an industrial system, an energy storage device has a significant influence on the stability of the system due to the state of the energy storage battery, and for the energy storage battery, the state of health (SOH) and the state of charge (SOC) of the battery are the most important state quantities, and accurate SOC prediction plays an important role in preventing overcharge and overdischarge of the battery and delaying the cycle life of the battery. Meanwhile, with the increase of the use times of the battery, the performance of the battery is reduced, the battery is easy to have safety problems such as thermal runaway, internal short circuit and the like, equipment or a system is paralyzed, and catastrophic accidents can be caused seriously, so that accurate estimation of the state of the energy storage battery has very important significance for estimation of an energy storage system.
In the island environment, the natural environment is quite severe, the humidity is high in rainy seasons, a climate environment different from that of the continent is formed, the prediction of the health state and the charge state of the energy storage battery is easily influenced by the external environment, serious safety accidents can be caused due to the fact that the health state of the energy storage battery cannot be predicted in time, and the reliability of equipment is reduced. At present, the research on the state monitoring of the energy storage battery is carried out in a data-driven mode, most of data sets are acquired by laboratories, few scholars consider the influence of the complexity of the field environment on the energy storage battery, and related research technologies in a specific island environment are few.
Meanwhile, in the method for estimating the state of the battery based on data driving, the support vector regression is a very typical algorithm, a large number of students perform optimization research on the support vector regression in the current research, the optimization research is better than that of the support vector regression prediction which is not optimized, but still has larger errors in prediction precision and monitors the state of the energy storage battery, most of the students concentrate on using model driving to perform SOC estimation on the battery and using data driving to perform SOH estimation on the battery, but no student uses a POA-SVR algorithm to perform SOC and SOH joint estimation on the energy storage battery. In the literature [ Xuanning, niyulong, zhuchubo ] lithium battery residual life prediction based on improved support vector regression [ J ] electrotechnical science, 2021,36 (17): 12], the prediction of the residual effective life (RUL) of the battery is carried out by optimizing the support vector regression by using an improved ant lion optimization algorithm, but it can be seen that the optimized support vector regression still has a large prediction error, and the joint estimation of the battery is not considered. Chinese patent application publication No. CN111443293A proposes a method for estimating state of health SOH of lithium battery based on data driving, in which the method uses a data-driven method to estimate the state of health SOH of lithium battery, but does not consider joint prediction of battery SOC using data driving, and at the same time, does not consider diversity of input quantity when extracting input quantity, so that prediction accuracy is greatly reduced.
Disclosure of Invention
In view of the above, the invention provides a method for monitoring the state of an energy storage battery in an island environment based on POA-SVR, which can realize the combined prediction of SOC and SOH of the energy storage battery in the island environment and effectively improve the prediction precision of a model.
A method for monitoring the state of an energy storage battery in an island environment based on POA-SVR comprises the following steps:
(1) Extracting battery cycle data from the island energy storage equipment data set as a health factor, and carrying out normalization pretreatment on the health factor;
(2) Constructing an algorithm model based on the POA-SVR;
(3) Training the algorithm model by using the health factor and the corresponding SOH label to obtain a prediction model H1 of the battery health state (namely the residual life);
(4) Training the algorithm model by using current and voltage data in the health factor and a corresponding SOC label to obtain a prediction model H2 of the state of charge (namely the residual electric quantity) of the battery;
(5) And monitoring the health factor of the energy storage battery in the island environment in real time, and inputting corresponding data into the prediction models H1 and H2 to obtain the prediction results about the health state and the charge state of the energy storage battery.
Further, the health factors extracted in the step (1) include 11 sets of characteristics reflecting the health state of the energy storage battery from four angles of voltage, current, temperature and charging and discharging time, and the characteristics are respectively as follows: the charging maximum temperature, the discharging maximum temperature, the charging temperature average value, the discharging temperature average value, the equal current drop charging time interval, the constant current discharging time interval, the equal pressure rising time interval, the equal pressure drop time interval, the equal pressure rising stage voltage-to-time integral quantity, the equal pressure drop stage voltage-to-time integral quantity, and the equal current drop stage current-to-time integral quantity.
Further, the specific implementation manner of the step (2) is as follows: firstly, an SVR (Support Vector Regression) model is established, and a nonlinear mapping function thereof is defined as:
wherein: x is an input vector, y is an output vector, w is a weight vector, b is an offset value,
as a non-linear transformation function, x i For the ith set of input samples for model training, a i And a i Is' x i Corresponding Lagrange multiplier, K (x) i And x) is with respect to x i A kernel function with x, i is a natural number, i is more than or equal to 1 and less than or equal to n, and n is the number of samples;
and then, optimizing by using a POA (Pelican Optimization Algorithm) Algorithm to find an optimal solution for the penalty coefficient and the kernel function radius involved in the SVR model.
Further, the kernel function K (x) i And x) is as follows:
wherein: g is the kernel function radius, exp () represents an exponential function with a natural constant e as the base.
Further, the Lagrangian multiplier a i And a i The solving expression of' is as follows:
wherein: y is i For input sample x i And c is a penalty coefficient, epsilon is a constant, j is a natural number, and j is more than or equal to 1 and less than or equal to n.
Further, the specific process of the POA algorithm for optimizing is as follows:
2.1 initializing a population with a certain scale quantity at random according to the upper and lower bounds of the hyper-parameter, wherein each member in the population corresponds to a group of candidate solutions about a penalty coefficient and a kernel function radius;
2.2 updating the population for the first time by exploring in the search space and randomly generating a prey position p;
2.3 second update the population by detecting points in the neighborhood of pelican positions and converging them into a hunting zone with further advantage;
and 2.4, taking the two updating processes as one iteration, and finally outputting the optimal candidate solution from the population through repeated multiple iterations.
Further, the mathematical expression of the candidate solution is as follows:
X k ＝[x k,c ,x k,g ]
x k,c ＝l c +rand·(u c -l c )
x k,g ＝l g +rand·(u g -l g )
wherein: x k Represents the candidate solution, x, corresponding to the kth member of the population k,c Representing a solution candidate X k The variable value of (1) regarding the penalty factor, x k,g Representing a solution candidate X k Middle relation kernel functionThe variable value of the radius, rand, is a random number between 0 and 1, u c And l c Upper and lower bound values, u, of penalty factors, respectively g And l g And the radius is an upper bound value and a lower bound value of the kernel function respectively, k is a natural number, k is more than or equal to 1 and less than or equal to N, and N is the scale number of the population.
Further, the specific implementation manner of step 2.2 is as follows: firstly, for any member in the population, calculating the new position P of the member by the following formula 1 ；
Wherein:
indicating a new position P 1 The variable value of (1) with respect to the penalty coefficient,
indicating a new position P 1 In relation to the value of the variable of the radius of the kernel function, the prey position p = [ p ] c ,p g ]，p c Is 1 c ～u c Random value between p g Is 1 g ～u g Random value between, F p For the model objective function value corresponding to prey position p, F k As a candidate solution X k The corresponding model objective function value, I is a random number of 1 or 2;
then calculates a new position P 1 Corresponding model objective function value if the model objective function value is relative to F k Improved, the new position P is set 1 Replacement candidate solution X k Otherwise, the candidate solution X is retained k 。
Further, the specific implementation manner of step 2.3 is as follows: firstly, for any member in the population, the new of the member is calculated by the following formulaPosition P 2 ；
Wherein:
indicating a new position P 2 The value of the variable related to the penalty factor,
indicating a new position P 2 The variable value of the radius of the kernel function is shown in the specification, T represents the iteration number of the round, T represents the maximum iteration number, and R is a constant;
then calculates a new position P 2 Corresponding model objective function value if the model objective function value is relative to F k Improved, the new position P is set 2 Replacement candidate solution X k Otherwise, the candidate solution X is retained k 。
Further, the model objective function value selects a mean square error between a model output prediction result and a corresponding label.
Based on the technical scheme, the invention has the following beneficial technical effects:
1. compared with the traditional optimized SVR algorithm, the method has stronger development capability in local search and convergence capability towards global optimum, has better prediction precision on the state of the energy storage battery, and simultaneously has good expansibility.
2. The invention simultaneously considers the particularity of the island environment, respectively trains the POA-SVR model by the energy storage battery data obtained under the island environment to obtain the POA-SVR battery residual life prediction model and the POA-SVR battery charge state prediction model, and carries out online monitoring on the state of the energy storage battery under the island environment by the trained model, thereby realizing the combined prediction of the SOC and the SOH of the energy storage battery under the island environment and achieving the purpose of real-time online monitoring on the running state of the energy storage battery under the island environment.
Drawings
Fig. 1 is a schematic flow chart of a method for monitoring the state of an energy storage battery in an island environment according to the present invention.
FIG. 2 is a diagram showing the prediction result of the SOH state of the battery under the algorithm model of the invention.
FIG. 3 is a general schematic diagram of online monitoring using the algorithmic model of the present invention.
Detailed Description
In order to more specifically describe the present invention, the following detailed description is provided for the technical solution of the present invention with reference to the accompanying drawings and the specific embodiments.
As shown in FIG. 1, the method for monitoring the state of the energy storage battery in the island environment based on the POA-SVR comprises the following steps:
(1) And (4) extracting battery cycle data from the island energy storage equipment data set to serve as health factors, and performing normalization pretreatment on the health factors.
The method comprises the steps of extracting battery cycle data from island energy storage equipment data in a centralized mode, wherein the battery cycle data are directly acquired by an energy storage battery monitoring BMS system and comprise voltage, current, temperature and charging and discharging time, extracting 11 groups of characteristic vectors in total by extracting multidimensional characteristic vectors in order to improve the accuracy of model prediction, and reflecting the health state of the battery from four angles of voltage, current, temperature and charging and discharging time by the extracted indexes.
The method comprises the steps that temperature-related health factors are extracted, the temperature in a BMS system is easy to monitor and record, four temperature-related health factors are extracted from energy storage battery temperature monitoring data respectively, and the four temperature-related health factors are respectively the highest charging temperature, the highest discharging temperature, the average charging temperature and the average discharging temperature.
And (4) extracting health factors related to the charging and discharging time, and respectively extracting four groups of health factors of equal-current-drop charging time interval, constant-current discharging time interval, equal-voltage rising time interval and equal-voltage drop time interval according to the monitored charging and discharging time.
And extracting current and voltage related health factors, performing current health factor separation on the obtained current and voltage monitoring data of the energy storage battery, and respectively extracting three groups of health factors of voltage to time integral quantity in an isobaric rise stage, voltage to time integral quantity in an isobaric drop stage and current to time integral quantity in an isocurrentdrop stage.
Carrying out normalization pretreatment on the health factors, wherein the normalization mode is as follows:
wherein: x is the data before normalization, x' is the data after normalization, x max Is the maximum value of the sample data, x min Is the minimum value of the sample data.
(2) And dividing the normalized data into a test set and a data set, optimizing the SVR model by adopting a POA algorithm, training the POA-SVR algorithm by using the test set, and establishing a POA-SVR battery residual life prediction model.
This example uses NASA published battery data set B0005 battery for POA-SVR model validation. Firstly, carrying out NASA public battery B0005 data sample division, dividing 80% of data of a total sample into a training set, dividing 20% of data of the total sample into a test set, establishing an SVR model, and giving a battery cycle data training set:
S＝{(x 1 ,y 1 ),(x 2 ,y 2 ),…,(x i ,y i )},y i ∈R
wherein: x is the number of i As an input vector, y i For the corresponding output vector, the nonlinear mapping function is now defined as:
wherein: w is a weight vector, b is a deviation value; the w and b parameters can be solved by the following objective function:
due to the different degrees of relaxation in the hyperplane, a relaxation variable is introduced, and the objective function is converted into the following constraint form:
in the formula: xi shape i And xi i ' for relaxation factor, converting regression problem into solving objective function minimization problem, introducing Lagrange penalty operator at the moment, converting objective function into dual problem:
a in the objective function i And a i ' for the Lagrange multiplier, by minimizing the Lagrange function, a nonlinear mapping SVR expression can be obtained:
in the formula:
as kernel function, K (x) i X) the low-dimensional linear inseparable problem can be transformed into a high-dimensional linear separable problem by inner product operation, and the kernel function selects a radial basis kernel function defined as:
wherein: g is a kernel function radius, the values of the penalty coefficient c and the kernel function radius g directly influence the final prediction effect of the SVR model, but currently, there is no theoretical guidance for the selection of the kernel function radius and the kernel function radius g, the pelican optimization algorithm has very strong competitiveness in high development capacity in local search and convergence capacity to global optimum, and can achieve a good optimization effect on parameters, and the specific process is as follows:
firstly, optimizing by using a POA algorithm, initializing the POA, and randomly initializing population members according to a lower bound and an upper bound of SVR hyper-parameters, wherein each population member represents a candidate solution:
x i,j ＝l j +rand·(u j -l j ),i＝1,2,...,N,j＝1,2,...,m
in the formula: x is the number of i,j Is the j variable value specified by the i candidate solution, N is the overall population number, m is the number of problem variables (in the present invention, m =2, i.e., two problem variables c and g), and rand is the interval [0,1]Random number of (1) j Is the jth lower bound, u, of the problem variable j Is the jth upper bound of the problem variable, defining the POA population matrix as follows, with each row of the matrix representing a candidate solution and the columns of the matrix representing the values of the problem variable:
wherein: x is a pelican population matrix, X i Is the ith pelican, and each population member is a pelican, representing a solution candidate. The objective function for a given problem is built on a per candidate solution basis, and the objective function vector is determined by the objective function value as follows:
wherein: f is the objective function vector, and F i Is the objective function value of the ith candidate solution.
The first stage is as follows: pelicans localized the prey and then moved to this defined area, where the emphasis in POA algorithm was that the prey location was randomly generated in the search space, which increased the exploratory ability of POA algorithm in the problem solving space, and the strategy for pelican homing was modeled as follows:
in the formula:
for the first stage new position generated by the ith pelican in the j dimension, I is a random number of 1 or 2, and I is randomly selected for each iteration and for each member; when the value of I equals 2, a further displacement will be provided and the member can be directed to a newer region of the search space. p is a radical of formula j The position of the prey in j dimension, F p And selecting an objective function as a mean square error MSE between the actual state and the predicted state of the battery for an objective function value corresponding to the position of the prey:
in the formula:
is in an actual state, y i To predict the state, the position is updated if the new position generated improves the value of the objective function, called a valid update, thereby preventing the algorithm from moving to a non-optimal region:
in the formula: x i ' in order to update the new location,
is the objective function value corresponding to the new position in the first stage.
The second stage of the POA algorithm is a development stage, after pelicans reach the water surface, wings are unfolded to push the fish to move upwards on the water surface, and then preys are collected in a laryngeal sac, so that more fishes in the attacked area are captured by the pelicans; modeling this behavior of pelicans makes the algorithm converge to better points in the hunting area, with the mathematical modeling expression as follows:
in the formula:
for the second stage new position produced by the ith pelican in the j dimension, R is a constant with a value of 0.2, and R (1-T/T) represents the neighborhood radius of the population members, and converges to a better solution by performing a local search around each member. In the initial iteration, the value of the coefficient is large, so that the area around each member is considered, and as the number of times of repeating the algorithm is increased, the R (1-T/T) coefficient is reduced, so that the neighborhood radius of each member is reduced, and therefore the area around each member can be scanned in smaller and more accurate steps, so that the solution which is closer to the global optimum is converged; t is the iteration time of this time, and T is the maximum iteration time, wherein the effective position updating rule is as follows:
wherein: x i ' in order to update the new location,
is the objective function value corresponding to the new position in the second stage.
And (2) repeatedly iterating until requirements are met, outputting two super-parameter optimal solutions of a penalty coefficient c and a kernel function radius g, ending the POA algorithm, returning and inputting the obtained optimal super-parameters into the SVR model to construct a POA-SVR model, wherein the Error evaluation index uses Root Mean Square Error (RMSE):
in the formula:
and y i The predicted capacity and the actual capacity of the ith cycle are respectively, and n represents the number of samples.
As shown in fig. 2, RMSE =0.012125 is obtained from the simulation result of the residual life of the B0005 battery, which is more effective.
(3) As shown in fig. 3, the state of the energy storage battery in the island environment is monitored in real time, and the health factor obtained by online monitoring of the energy storage battery is input into the trained POA-SVR model to predict the SOH of the energy storage battery in the island environment in real time.
(4) And (3) carrying out SOC prediction on the charge state of the energy storage battery in the island environment, inputting current and voltage data obtained by a data set of the energy storage battery equipment into the POA-SVR model as characteristic quantities for training, and establishing the POA-SVR charge state prediction model.
(5) As shown in fig. 3, the state of the energy storage battery in the island environment is monitored in real time, and the current and voltage data obtained by online monitoring of the energy storage battery are input into the trained POA-SVR model to predict the state of charge of the energy storage battery in the island environment.
The embodiments described above are presented to enable a person having ordinary skill in the art to make and use the invention. It will be readily apparent to those skilled in the art that various modifications to the above-described embodiments may be made, and the generic principles defined herein may be applied to other embodiments without the use of inventive faculty. Therefore, the present invention is not limited to the above embodiments, and those skilled in the art should make improvements and modifications to the present invention based on the disclosure of the present invention within the protection scope of the present invention.
Claims (10)
1. A method for monitoring the state of an energy storage battery in an island environment based on POA-SVR comprises the following steps:
(1) Extracting battery cycle data from the island energy storage equipment data set as a health factor, and carrying out normalization pretreatment on the health factor;
(2) Constructing an algorithm model based on the POA-SVR;
(3) Training the algorithm model by using the health factor and the corresponding SOH label to obtain a prediction model H1 of the battery health state;
(4) Training the algorithm model by using current-voltage data in the health factors and corresponding SOC labels to obtain a prediction model H2 of the state of charge of the battery;
(5) And monitoring the health factor of the energy storage battery in the island environment in real time, and inputting corresponding data into the prediction models H1 and H2 to obtain the prediction results about the health state and the charge state of the energy storage battery.
2. The method for monitoring the state of the energy storage battery in the island environment according to claim 1, wherein: the health factors extracted in the step (1) comprise 11 groups of characteristics reflecting the health state of the energy storage battery from four angles of voltage, current, temperature and charging and discharging time, and the characteristics are as follows: the charging maximum temperature, the discharging maximum temperature, the charging temperature average value, the discharging temperature average value, the equal current drop charging time interval, the constant current discharging time interval, the equal pressure rising time interval, the equal pressure drop time interval, the equal pressure rising stage voltage-to-time integral quantity, the equal pressure drop stage voltage-to-time integral quantity, and the equal current drop stage current-to-time integral quantity.
3. The method for monitoring the state of the energy storage battery in the island environment according to claim 1, wherein: the specific implementation manner of the step (2) is as follows: firstly, establishing an SVR model, wherein a nonlinear mapping function is defined as:
wherein: x is an input vector, y is an output vector, w is a weight vector, b is an offset value,
as a non-linear transformation function, x i For the ith set of input samples for model training, a i And a i Is' x i Corresponding Lagrange multiplier, K (x) i And x) is with respect to x i A kernel function with x, i is a natural number, i is more than or equal to 1 and less than or equal to n, and n is the number of samples;
and then, optimizing by using a POA algorithm to find an optimal solution for a penalty coefficient and a kernel function radius related in the SVR model.
4. The method for monitoring the state of the energy storage battery in the island environment according to claim 3, wherein: the kernel function K (x) i And x) is as follows:
wherein: g is the kernel function radius, exp () represents an exponential function with a natural constant e as the base.
5. The method for monitoring the state of the energy storage battery in the island environment according to claim 3, wherein: the Lagrange multiplier a i And a i The solving expression of' is as follows:
wherein: y is i For input sample x i Corresponding label, c is penalty coefficient, epsilon is constant, j is selfAnd j is more than or equal to 1 and less than or equal to n.
6. The method for monitoring the state of the energy storage battery in the island environment according to claim 3, wherein: the specific process of optimizing the POA algorithm is as follows:
2.1 initializing a population with a certain scale quantity at random according to the upper and lower bounds of the hyper-parameter, wherein each member in the population corresponds to a group of candidate solutions about a penalty coefficient and a kernel function radius;
2.2 updating the population for the first time by exploring in the search space and randomly generating a prey position p;
2.3 second update the population by detecting points in the neighborhood of pelican positions and converging them into a hunting zone with further advantage;
and 2.4, taking the two updating processes as one iteration, and finally outputting the optimal candidate solution from the population through repeated iterations.
7. The method for monitoring the state of the energy storage battery in the island environment according to claim 6, wherein: the mathematical expression of the candidate solution is as follows:
X k ＝[x k,c ,x k,g ]
x k,c ＝l c +rand·(u c -l c )
x k,g ＝l g +rand·(u g -l g )
wherein: x k Represents the candidate solution, x, corresponding to the kth member of the population k,c Representing a solution candidate X k With respect to the value of the variable of the penalty factor, x k,g Representing candidate solutions X k Wherein, the variable value of the radius of the kernel function, rand is a random number between 0 and 1, u c And l c Upper and lower bound values, u, of penalty coefficients, respectively g And l g And the radius is an upper bound value and a lower bound value of the kernel function respectively, k is a natural number, k is more than or equal to 1 and less than or equal to N, and N is the scale number of the population.
8. The method of claim 7The method for monitoring the state of the energy storage battery in the island environment is characterized by comprising the following steps: the specific implementation manner of the step 2.2 is as follows: firstly, for any member in the population, calculating the new position P of the member by the following formula 1 ；
Wherein:
indicating a new position P 1 The value of the variable related to the penalty factor,
indicating a new position P 1 In the variable value of the radius of the kernel function, the target position p = [ p ] c ,p g ]，p c Is 1 of c ～u c Random value between, p g Is 1 of g ～u g Random value between, F p For the model objective function value corresponding to prey position p, F k As a candidate solution X k The corresponding model objective function value, I is a random number of 1 or 2;
then calculates a new position P 1 Corresponding model objective function value if the model objective function value is relative to F k Improved, the new position P is set 1 Replacement candidate solution X k Otherwise, the candidate solution X is retained k 。
9. The method for monitoring the state of the energy storage battery in the island environment according to claim 7, wherein: the specific implementation manner of the step 2.3 is: firstly, for any member in the population, calculating the new position P of the member by the following formula 2 ；
Wherein:
indicating a new position P 2 The variable value of (1) with respect to the penalty coefficient,
indicating a new position P 2 The variable value of the radius of the kernel function is shown in the specification, T represents the iteration number of the round, T represents the maximum iteration number, and R is a constant;
then calculates a new position P 2 Corresponding model objective function value if the model objective function value is relative to F k Improved, the new position P is set 2 Replacement candidate solution X k Otherwise, the candidate solution X is retained k 。
10. The method for monitoring the state of the energy storage battery in the island environment according to claim 8 or 9, wherein: and the model objective function value selects a mean square error between a model output prediction result and a corresponding label.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211630297.0A CN115808633A (en) | 2022-12-19 | 2022-12-19 | POA-SVR-based method for monitoring state of energy storage battery in island environment |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211630297.0A CN115808633A (en) | 2022-12-19 | 2022-12-19 | POA-SVR-based method for monitoring state of energy storage battery in island environment |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN115808633A true CN115808633A (en) | 2023-03-17 |
Family
ID=85486186
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202211630297.0A Pending CN115808633A (en) | 2022-12-19 | 2022-12-19 | POA-SVR-based method for monitoring state of energy storage battery in island environment |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115808633A (en) |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116572769A (en) * | 2023-05-26 | 2023-08-11 | 淮阴工学院 | New energy automobile wireless charging duration prediction method and wireless charging equipment |
| CN116633433A (en) * | 2023-05-12 | 2023-08-22 | 国网吉林省电力有限公司 | Model-driven OPGW optical cable fault diagnosis and positioning method |
| CN117330963A (en) * | 2023-11-30 | 2024-01-02 | 国网浙江省电力有限公司宁波供电公司 | Energy storage power station fault detection method, system and equipment |
| CN117825975A (en) * | 2024-03-05 | 2024-04-05 | 烟台海博电气设备有限公司 | Data-driven lithium ion battery SOH evaluation method and system |
-
2022
- 2022-12-19 CN CN202211630297.0A patent/CN115808633A/en active Pending
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116633433A (en) * | 2023-05-12 | 2023-08-22 | 国网吉林省电力有限公司 | Model-driven OPGW optical cable fault diagnosis and positioning method |
| CN116633433B (en) * | 2023-05-12 | 2024-03-08 | 国网吉林省电力有限公司 | Model-driven OPGW optical cable fault diagnosis and positioning method |
| CN116572769A (en) * | 2023-05-26 | 2023-08-11 | 淮阴工学院 | New energy automobile wireless charging duration prediction method and wireless charging equipment |
| CN117330963A (en) * | 2023-11-30 | 2024-01-02 | 国网浙江省电力有限公司宁波供电公司 | Energy storage power station fault detection method, system and equipment |
| CN117330963B (en) * | 2023-11-30 | 2024-04-19 | 国网浙江省电力有限公司宁波供电公司 | Energy storage power station fault detection method, system and equipment |
| CN117825975A (en) * | 2024-03-05 | 2024-04-05 | 烟台海博电气设备有限公司 | Data-driven lithium ion battery SOH evaluation method and system |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN115808633A (en) | POA-SVR-based method for monitoring state of energy storage battery in island environment | |
| Sui et al. | A review of non-probabilistic machine learning-based state of health estimation techniques for Lithium-ion battery | |
| Hannan et al. | SOC estimation of li-ion batteries with learning rate-optimized deep fully convolutional network | |
| CN111638465B (en) | Lithium battery health state estimation method based on convolutional neural network and transfer learning | |
| CN112487702B (en) | Method for predicting residual service life of lithium ion battery | |
| CN110888058B (en) | Algorithm based on power battery SOC and SOH joint estimation | |
| Oji et al. | Data-driven methods for battery soh estimation: Survey and a critical analysis | |
| Feng et al. | A health indicator extraction based on surface temperature for lithium-ion batteries remaining useful life prediction | |
| Chen et al. | An overview of data-driven battery health estimation technology for battery management system | |
| CN110443377B (en) | Support vector machine storage battery life prediction method based on immune algorithm optimization | |
| CN113702843B (en) | Lithium battery parameter identification and SOC estimation method based on suburb optimization algorithm | |
| US20220057451A1 (en) | System and method for estimation of battery state and health | |
| CN111366848A (en) | Battery health state prediction method based on PSO-ELM algorithm | |
| Anton et al. | A new predictive model for the state-of-charge of a high-power lithium-ion cell based on a PSO-optimized multivariate adaptive regression spline approach | |
| Wang et al. | Health diagnosis for lithium-ion battery by combining partial incremental capacity and deep belief network during insufficient discharge profile | |
| CN113917334A (en) | Battery health state estimation method based on evolution LSTM self-encoder | |
| CN116679232A (en) | Lithium battery health state estimation method based on hybrid neural network and transfer learning | |
| CN115201686A (en) | Lithium ion battery health state assessment method under incomplete charging and discharging data | |
| Wu et al. | State of health estimation of lithium-ion batteries using Autoencoders and Ensemble Learning | |
| Ren et al. | A novel multiple training-scale dynamic adaptive cuckoo search optimized long short-term memory neural network and multi-dimensional health indicators acquisition strategy for whole life cycle health evaluation of lithium-ion batteries | |
| CN112327165B (en) | Battery SOH prediction method based on unsupervised transfer learning | |
| Satyan et al. | A Survey on data-driven methods for state of charge estimation of battery | |
| CN116804706A (en) | Temperature prediction method and device for lithium battery of electric automobile | |
| CN113884936A (en) | Lithium ion battery health state prediction method based on ISSA coupling DELM | |
| CN112285565B (en) | Method for predicting SOH (State of health) of battery by transfer learning based on RKHS (remote keyless entry) domain matching |
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 |