CN112130086B - Method and system for predicting remaining life of power battery - Google Patents
Method and system for predicting remaining life of power battery Download PDFInfo
- Publication number
- CN112130086B CN112130086B CN202010475847.0A CN202010475847A CN112130086B CN 112130086 B CN112130086 B CN 112130086B CN 202010475847 A CN202010475847 A CN 202010475847A CN 112130086 B CN112130086 B CN 112130086B
- Authority
- CN
- China
- Prior art keywords
- data
- discharge
- charge
- predicted
- power battery
- 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
- 238000000034 method Methods 0.000 title claims abstract description 47
- 238000010845 search algorithm Methods 0.000 claims description 13
- 238000004422 calculation algorithm Methods 0.000 claims description 12
- 230000032683 aging Effects 0.000 claims description 6
- 238000004364 calculation method Methods 0.000 claims description 4
- 238000007600 charging Methods 0.000 claims description 3
- 238000006243 chemical reaction Methods 0.000 claims description 3
- 238000010281 constant-current constant-voltage charging Methods 0.000 claims description 3
- 125000004122 cyclic group Chemical group 0.000 claims description 3
- 238000013473 artificial intelligence Methods 0.000 description 7
- 238000001514 detection method Methods 0.000 description 2
- 238000010586 diagram Methods 0.000 description 2
- 238000013528 artificial neural network Methods 0.000 description 1
- 230000015556 catabolic process Effects 0.000 description 1
- 238000006731 degradation reaction Methods 0.000 description 1
- 230000001419 dependent effect Effects 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000003062 neural network model Methods 0.000 description 1
- 230000000750 progressive effect Effects 0.000 description 1
- 238000007619 statistical method Methods 0.000 description 1
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]
- G01R31/392—Determining battery ageing or deterioration, e.g. state of health
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B60—VEHICLES IN GENERAL
- B60L—PROPULSION OF ELECTRICALLY-PROPELLED VEHICLES; SUPPLYING ELECTRIC POWER FOR AUXILIARY EQUIPMENT OF ELECTRICALLY-PROPELLED VEHICLES; ELECTRODYNAMIC BRAKE SYSTEMS FOR VEHICLES IN GENERAL; MAGNETIC SUSPENSION OR LEVITATION FOR VEHICLES; MONITORING OPERATING VARIABLES OF ELECTRICALLY-PROPELLED VEHICLES; ELECTRIC SAFETY DEVICES FOR ELECTRICALLY-PROPELLED VEHICLES
- B60L58/00—Methods or circuit arrangements for monitoring or controlling batteries or fuel cells, specially adapted for electric vehicles
- B60L58/10—Methods or circuit arrangements for monitoring or controlling batteries or fuel cells, specially adapted for electric vehicles for monitoring or controlling batteries
- B60L58/16—Methods or circuit arrangements for monitoring or controlling batteries or fuel cells, specially adapted for electric vehicles for monitoring or controlling batteries responding to battery ageing, e.g. to the number of charging cycles or the state of health [SoH]
-
- 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
-
- 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
- Y02T—CLIMATE CHANGE MITIGATION TECHNOLOGIES RELATED TO TRANSPORTATION
- Y02T10/00—Road transport of goods or passengers
- Y02T10/60—Other road transportation technologies with climate change mitigation effect
- Y02T10/70—Energy storage systems for electromobility, e.g. batteries
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Sustainable Development (AREA)
- Sustainable Energy (AREA)
- Power Engineering (AREA)
- Transportation (AREA)
- Mechanical Engineering (AREA)
- Charge And Discharge Circuits For Batteries Or The Like (AREA)
- Secondary Cells (AREA)
Abstract
The invention relates to a method and a system for predicting the residual life of a power battery, wherein the method comprises the following steps: acquiring an original charge and discharge data set; denoising the original charge and discharge data set to obtain a training set; training the XGBoost model to obtain a set to be predicted, inputting the set to the trained XGBoost model to obtain battery capacity percentage data of the next charge-discharge cycle, and setting the charge-discharge cycle frequency corresponding to the obtained predicted value as q; judging whether the battery capacity percentage of the q-th charge-discharge cycle is less than or equal to 80, if yes, executing the next step, otherwise, adding the battery capacity percentage of the q-th charge-discharge cycle to the end of the set to be predicted to construct a new set to be predicted, so that q=q+1; and taking the difference between q and k to obtain the residual life of the power battery to be predicted. The method of the invention avoids the influence of noise in the original data on the prediction result, is simple and easy to realize, and has high prediction speed and high prediction precision.
Description
Technical Field
The invention relates to the field of power battery health state detection, in particular to a method and a system for predicting the residual life of a power battery.
Background
Current prediction methods for the remaining life of a power battery are roughly classified into three types, i.e., based on empirical knowledge, based on a physical model, and based on data driving. The method for predicting the residual life of the power battery based on data driving does not need to establish a capacity degradation white box model and only considers the data rule, so that the method is deeper in research and wider in application. The residual life prediction method based on data driving comprises three types of statistical method based, random process based and artificial intelligence based. The method based on artificial intelligence has better adaptability and various implementation means, so the method is also a main research direction for predicting the residual life of the power battery at present and in the future.
Neural networks are often selected as prediction models based on artificial intelligence methods, but the neural network models have the problems that training time is too long, reasonable network structures are difficult to determine and the like. And the power battery residual life prediction method based on artificial intelligence is very dependent on historical charge and discharge data of the battery, and the measured charge and discharge data are often mixed with noise, so that the prediction precision of a residual life prediction model can be influenced. The traditional charge-discharge data denoising method usually uses a wavelet denoising method, but the wavelet denoising method has the problem that the wavelet basis and the threshold parameters are difficult to select. The existing power battery residual life prediction method based on artificial intelligence has the problems that the prediction time is too long, and the training data has noise so as to limit the prediction precision.
Disclosure of Invention
The invention aims to provide a method and a system for predicting the residual life of a power battery, which improve the prediction precision and reduce the prediction time.
In order to achieve the above object, the present invention provides the following solutions:
a method of predicting remaining life of a power battery, the method comprising:
s1: acquiring an original charge and discharge data set; the original charge-discharge dataset includes: the charge and discharge cycle times and the battery capacity percentage data of the power battery;
s2: denoising the original charge-discharge data set, and taking the denoised original charge-discharge data set as a training set;
s3: training the XGBoost limit gradient lifting model based on the training set to obtain a trained XGBoost model;
s4: acquiring actual charge and discharge cycle times and battery capacity percentage data of a power battery to be predicted, and taking the actual charge and discharge cycle times and the battery capacity percentage data as a set to be predicted of an XGBoost model; the to-be-predicted set contains battery capacity percentage data of k charge-discharge cycles;
s5: inputting the set to be predicted into the trained XGBoost model to obtain battery capacity percentage data of the next charge-discharge cycle of the power battery to be predicted, and setting the number of charge-discharge cycles corresponding to the obtained predicted value as the q-th time;
s6: judging whether the battery capacity percentage of the q-th charge-discharge cycle obtained in the step S5 is less than or equal to 80% of the maximum available discharge capacity of the power battery when leaving the factory, if yes, executing the next step, if not, adding the battery capacity percentage of the q-th charge-discharge cycle to the end of a set to be predicted, constructing a new set to be predicted, enabling q=q+1, and returning to the step S5;
s7: and (3) differentiating the charge-discharge cycle times q corresponding to the predicted value and the actual charge-discharge cycle times k to obtain the residual life of the power battery to be predicted.
Optionally, denoising the original charge-discharge data set specifically includes:
identifying noise points;
deleting the noise point;
and replacing the original noise point position by using the average value of the two adjacent data of the noise point.
Optionally, the following formula is specifically adopted for the identifying noise points:
wherein D is the original charge and discharge data set of the power battery, p is any data point in the data set D, C i And (3) utilizing the i-th cluster obtained after DBSCAN clustering for the data set D, wherein m is the number of clusters obtained after clustering.
Optionally, the training the XGBoost limit gradient lifting model based on the training set specifically includes:
and optimizing the XGBoost model parameters by adopting a grid search algorithm, and setting the XGBoost model parameters as optimal parameters obtained by the grid search algorithm.
Optionally, the objective function of the XGBoost model specifically adopts the following formula:
wherein y is j Is the true value, i.e., the jth data value in the training set;the data value obtained by predicting the jth data in the training set is a predicted value; n is the number of data contained in the training set; />The formula is thatOmega is a regularization term representing the complexity of the tree.
The present invention additionally provides a power battery remaining life prediction system, the prediction system comprising:
the original charge and discharge data set acquisition module is used for acquiring an original charge and discharge data set; the original charge-discharge dataset includes: the charge and discharge cycle times and the battery capacity percentage data of the power battery;
the denoising module is used for denoising the original charge-discharge data set, and taking the denoised original charge-discharge data set as a training set;
the training module is used for training the XGBoost limit gradient lifting model based on the training set to obtain a trained XGBoost model;
the to-be-predicted set acquisition module is used for acquiring the actual charge and discharge cycle times and the battery capacity percentage data of the to-be-predicted power battery and taking the actual charge and discharge cycle times and the battery capacity percentage data as a to-be-predicted set of the XGBoost model; the to-be-predicted set contains battery capacity percentage data of k charge-discharge cycles;
the calculation module is used for inputting the set to be predicted into the trained XGBoost model to obtain battery capacity percentage data of the next charge-discharge cycle of the power battery to be predicted, and setting the number of charge-discharge cycles corresponding to the obtained predicted value to be the q-th time;
the judging module is used for judging whether the battery capacity percentage of the q-th charge-discharge cycle obtained in the calculating module is smaller than or equal to 80% of the maximum available discharge capacity of the power battery when leaving the factory, if yes, executing the next module, if not, adding the battery capacity percentage of the q-th charge-discharge cycle to the end of a set to be predicted to construct a new set to be predicted, and returning q=q+1 to the calculating module;
and the difference making module is used for making difference between the charge and discharge cycle times q corresponding to the predicted value and the actual charge and discharge cycle times k to obtain the residual life of the power battery to be predicted.
Optionally, the denoising module specifically includes:
a noise point identification unit for identifying noise points;
a noise point deleting unit configured to delete the noise point;
and the noise point replacing unit is used for replacing the original noise point position by utilizing the average value of the two adjacent data of the noise point.
Optionally, the noise point identifying unit specifically adopts the following formula:
wherein D is the original charge and discharge data set of the power battery, p is any data point in the data set D, C i And (3) utilizing the i-th cluster obtained after DBSCAN clustering for the data set D, wherein m is the number of clusters obtained after clustering.
Optionally, the training module specifically includes:
and optimizing the XGBoost model parameters by adopting a grid search algorithm, and setting the XGBoost model parameters as optimal parameters obtained by the grid search algorithm.
Optionally, the objective function of the XGBoost model specifically adopts the following formula:
wherein y is j Is the true value, i.e., the jth data value in the training set;the data value obtained by predicting the jth data in the training set is a predicted value; n is the number of data contained in the training set; />The formula is thatOmega is a regularization term representing the complexity of the tree.
According to the specific embodiment provided by the invention, the invention discloses the following technical effects:
according to the invention, the DBSCAN algorithm is utilized to denoise the original charge and discharge data, the XGBoost model is selected as the residual life prediction model, and the grid search algorithm is adopted to adjust and optimize parameters in the model training process, so that the influence of noise in the original data on a prediction result is avoided, the method is simple and easy to realize, and the prediction speed and the prediction precision are high.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions of the prior art, the drawings that are needed in the embodiments will be briefly described below, it being obvious that the drawings in the following description are only some embodiments of the present invention, and that other drawings may be obtained according to these drawings without inventive effort for a person skilled in the art.
FIG. 1 is a flowchart of a method for predicting remaining life of a power battery according to an embodiment of the present invention;
fig. 2 is a schematic diagram of a power battery remaining life prediction system according to an embodiment of the present invention.
Detailed Description
The following description of the embodiments of the present invention will be made clearly and completely with reference to the accompanying drawings, in which it is apparent that the embodiments described are only some embodiments of the present invention, but not all embodiments. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention.
The invention aims to provide a method and a system for predicting the residual life of a power battery, which improve the prediction precision and reduce the prediction time.
In order that the above-recited objects, features and advantages of the present invention will become more readily apparent, a more particular description of the invention will be rendered by reference to the appended drawings and appended detailed description.
The invention aims to solve the problems that the existing power battery residual life prediction method based on artificial intelligence is too long in prediction time and the prediction precision is limited due to noise of training data, and provides a power battery residual life prediction method based on Density clustering (Density-Based Spatial Clustering of Applications with Noise, DBSCAN) noise reduction and limit gradient lifting (Extreme Gradient Boosting, XGBoost). The DBSCAN can realize denoising of training data, the XGBoost model is high in prediction speed and prediction precision, and the problems of long prediction time consumption and noise of the training data can be solved by combining the advantages of the two models.
Fig. 1 is a flowchart of a method for predicting remaining life of a power battery according to an embodiment of the present invention, as shown in fig. 1, the method includes:
s1: acquiring an original charge and discharge data set; the original charge-discharge dataset includes: the number of charge and discharge cycles of the power battery and the battery capacity percentage data.
Specifically, the battery aging test is used for obtaining the charge and discharge cycle times of the power battery and the corresponding battery capacity data thereof, converting the battery capacity data into battery capacity percentage data, and taking a data set containing the charge and discharge cycle times of the power battery and the corresponding battery capacity percentage data thereof as an original charge and discharge data set of the power battery.
The conversion formula for converting the battery capacity data into the battery capacity percentage data is as follows:
wherein BCP represents the converted battery capacity percentage data, C max Represents the real-time maximum available discharge capacity of the power battery, C omax Indicating the maximum usable discharge capacity of the power battery at the time of shipment.
The test implementation flow of the battery aging test is that the temperature environment of the power battery is set to 25 ℃ firstly, and the power battery is kept stand for 2 hours; then charging to 4.2V at a rate of 0.5C based on a constant-current constant-voltage charging mode, setting the cut-off current to be 0.048A, and standing for 1 hour; then constant-current discharge is carried out to discharge cut-off voltage of 2.5V at 1C multiplying power, and standing is carried out for 1 hour; and finally judging whether the discharge capacity reaches 80% of the maximum capacity of the factory, if so, stopping the test, and if not, continuing the cyclic charge and discharge test.
S2: denoising the original charge-discharge data set, and taking the denoised original charge-discharge data set as a training set.
Denoising the original charge and discharge data set of the power battery obtained in the step 1 by using a DBSCAN algorithm, and constructing the denoised data set as a training set of the XGBoost prediction model.
The DBSCAN algorithm can effectively identify noise points in the data set, and Eps and MinPts are set to be two parameters in the DBSCAN algorithm, wherein Eps represents the radius of a cluster of clusters, and MinPts represents the minimum sample number of each cluster of clusters.
An Eps neighbor of a data point in the dataset refers to a set of points of the point within its radius of field Eps, and assuming that the Eps neighbor is Eps (p), the Eps (p) formula is as follows:
Eps(p)＝{q∈D|dis(p,q)≤Eps}
in the above formula, D represents a data set, p, q represents two data points in the data set D, dis (p, q) represents a distance between the data points p and q.
Wherein the definition formula of noise point is as follows:
wherein D is the original charge and discharge data set of the power battery, p is any data point in the data set D, C i The ith class cluster obtained after DBSCAN clustering is used for the data set D, and m is the number of class clusters obtained after clustering;
the method processes noise points by deleting data classified as noise points from the dataset and then replacing them with an average of two adjacent data points of the noise point.
Wherein the DBSCAN algorithm can realize the controllability of data denoising by adjusting the radius of clustering class clusters and the minimum sample number in each class cluster.
S3: and training the XGBoost limit gradient lifting model based on the training set to obtain a trained XGBoost model.
The training process adopts a grid search algorithm to adjust and optimize the XGBoost model parameters, and the XGBoost model parameters are set as optimal parameters obtained by the grid search algorithm.
Wherein the objective function formula of the XGBoost model is as follows:
wherein y is j Is the true value, i.e., the jth data value in the training set;the data value obtained by predicting the jth data in the training set is a predicted value; n is the number of data contained in the training setA number; />Is a loss function, and the formula isOmega is a regularization term representing the complexity of the tree, and training over-fitting can be avoided by adjusting regularization parameter gamma in the model.
The training process adopts a grid search algorithm to tune the XGBoost model parameters, wherein parameters after participating in tuning are set to be learning_rate=0.1, n_identifiers=230, max_depth=7, min_child_weight=5, subsampler=0.8, colsample=0.8 and gamma=0.1.
S4: acquiring actual charge and discharge cycle times and battery capacity percentage data of a power battery to be predicted, and taking the actual charge and discharge cycle times and the battery capacity percentage data as a set to be predicted of an XGBoost model; the set to be predicted contains the battery capacity percentage data of k charge and discharge cycles.
S5: and inputting the set to be predicted into the trained XGBoost model to obtain battery capacity percentage data of the next charge-discharge cycle of the power battery to be predicted, and setting the number of charge-discharge cycles corresponding to the obtained predicted value as the q-th time.
S6: and judging whether the battery capacity percentage of the q-th charge-discharge cycle obtained in the step S5 is less than or equal to 80% of the maximum available discharge capacity of the power battery when leaving the factory, if yes, executing the next step, if not, adding the battery capacity percentage of the q-th charge-discharge cycle to the end of the set to be predicted to construct a new set to be predicted, and returning q=q+1 to the step S5.
S7: and (3) differentiating the charge-discharge cycle times q corresponding to the predicted value and the actual charge-discharge cycle times k to obtain the residual life of the power battery to be predicted.
The formula is as follows: rul=q-k.
Fig. 2 is a schematic structural diagram of a power battery remaining life prediction system according to an embodiment of the present invention, as shown in fig. 2, the prediction system includes: the device comprises an original charge and discharge data set acquisition module 201, a denoising module 202, a training module 203, a set to be predicted acquisition module 204, a calculation module 205, a judgment module 206 and a difference making module 207.
The original charge and discharge data set acquisition module 201 is configured to acquire an original charge and discharge data set; the original charge-discharge dataset includes: the number of charge and discharge cycles of the power battery and the battery capacity percentage data.
The denoising module 202 is configured to denoise the original charge-discharge data set, and take the denoised original charge-discharge data set as a training set.
The training module 203 is configured to train the XGBoost limit gradient lifting model based on the training set, and obtain a trained XGBoost model.
The to-be-predicted set obtaining module 204 is configured to obtain actual charge-discharge cycle times and battery capacity percentage data of the to-be-predicted power battery, and serve as a to-be-predicted set of the XGBoost model; the set to be predicted contains the battery capacity percentage data of k charge and discharge cycles.
The calculation module 205 is configured to input the set to be predicted to the trained XGBoost model, obtain battery capacity percentage data of a next charge-discharge cycle of the power battery to be predicted, and set the number of charge-discharge cycles corresponding to the obtained predicted value to be the qth number.
The judging module 206 is configured to judge whether the battery capacity percentage of the q-th charge-discharge cycle obtained in the calculating module is less than or equal to 80% of the maximum available discharge capacity of the power battery when the power battery leaves the factory, if yes, execute the next module, if not, add the battery capacity percentage of the q-th charge-discharge cycle to the end of the set to be predicted to construct a new set to be predicted, and return q=q+1 to the calculating module.
The difference making module 207 is configured to make a difference between the number q of charge and discharge cycles corresponding to the predicted value and the number k of actual charge and discharge cycles, so as to obtain the remaining life of the power battery to be predicted.
The invention discloses a power battery remaining life prediction method based on density clustering noise reduction and limit gradient lifting, relates to the technical field of power battery health state detection, and aims to solve the problems that the existing power battery remaining life prediction method based on artificial intelligence is excessively long in prediction time consumption and the prediction accuracy is limited due to noise of training data. According to the invention, the DBSCAN algorithm is utilized to denoise the original charge and discharge data, the XGBoost model is selected as the residual life prediction model, and the grid search algorithm is adopted to adjust and optimize parameters in the model training process, so that the influence of noise in the original data on a prediction result is avoided, the method is simple and easy to realize, and the prediction speed and the prediction precision are high.
In the present specification, each embodiment is described in a progressive manner, and each embodiment is mainly described in a different point from other embodiments, and identical and similar parts between the embodiments are all enough to refer to each other. For the system disclosed in the embodiment, since it corresponds to the method disclosed in the embodiment, the description is relatively simple, and the relevant points refer to the description of the method section.
The principles and embodiments of the present invention have been described herein with reference to specific examples, the description of which is intended only to assist in understanding the methods of the present invention and the core ideas thereof; also, it is within the scope of the present invention to be modified by those of ordinary skill in the art in light of the present teachings. In view of the foregoing, this description should not be construed as limiting the invention.
Claims (6)
1. A method for predicting remaining life of a power battery, the method comprising:
s1: acquiring an original charge and discharge data set; the original charge-discharge dataset includes: the number of charge and discharge cycles and the battery capacity percentage data of the power battery are specific:
acquiring the charge and discharge cycle times of the power battery and corresponding battery capacity data through a battery aging test, and converting the battery capacity data into the battery capacity percentage data;
the conversion formula for converting the battery capacity data into the battery capacity percentage data is as follows:
wherein BCP represents the converted battery capacity percentage data, C max Represents the real-time maximum available discharge capacity of the power battery, C omax Representing the maximum available discharge capacity of the power battery when leaving the factory;
the test implementation flow of the battery aging test is that the temperature environment of the power battery is set to 25 ℃ firstly, and the power battery is kept stand for 2 hours; then charging to 4.2V at a rate of 0.5C based on a constant-current constant-voltage charging mode, setting the cut-off current to be 0.048A, and standing for 1 hour; then constant-current discharge is carried out to discharge cut-off voltage of 2.5V at 1C multiplying power, and standing is carried out for 1 hour; finally judging whether the discharge capacity reaches 80% of the maximum capacity of the factory, if so, stopping the test, otherwise, continuing the cyclic charge and discharge test;
s2: denoising the original charge-discharge data set, and taking the denoised original charge-discharge data set as a training set, wherein the method specifically comprises the following steps of:
identifying noise points;
deleting the noise point;
the average value of two adjacent data of the noise point is used for replacing the original noise point position, and the method specifically comprises the following steps:
the DBSCAN algorithm can effectively identify noise points in the original charge and discharge data set, so that Eps and MinPts are two parameters in the DBSCAN algorithm, wherein Eps represents the radius of a cluster type cluster, and MinPts represents the minimum sample number of each type cluster;
and the Eps neighbor of one data point in the original charge and discharge data set refers to a set of points of the point in the range of the Eps of the radius of the field, and the Eps neighbor is Eps (p), and the Eps (p) is expressed as follows:
Eps(p)＝{q∈D|dis(p,q)≤Eps}
in the above formula, D represents an original charge-discharge data set, p, q represents two data points in the original charge-discharge data set D, dis (p, q) represents a distance between the data points p and q;
the DBSCAN algorithm realizes the controllability of data denoising by adjusting the radius of clustering class clusters and the minimum sample number in each class cluster;
s3: training the XGBoost limit gradient lifting model based on the training set to obtain a trained XGBoost model;
s4: acquiring actual charge and discharge cycle times and battery capacity percentage data of a power battery to be predicted, and taking the actual charge and discharge cycle times and the battery capacity percentage data as a set to be predicted of an XGBoost model; the to-be-predicted set contains battery capacity percentage data of k charge-discharge cycles;
s5: inputting the set to be predicted into the trained XGBoost model to obtain battery capacity percentage data of the next charge-discharge cycle of the power battery to be predicted, and setting the number of charge-discharge cycles corresponding to the obtained predicted value as the q-th time;
s6: judging whether the battery capacity percentage of the q-th charge-discharge cycle obtained in the step S5 is less than or equal to 80% of the maximum available discharge capacity of the power battery when leaving the factory, if yes, executing the next step, if not, adding the battery capacity percentage of the q-th charge-discharge cycle to the end of a set to be predicted to construct a new set to be predicted, and returning q=q+1 to the step S5;
s7: the charge-discharge cycle times q corresponding to the predicted value are differed from the actual charge-discharge cycle times k, and the residual life of the power battery to be predicted is obtained;
the objective function of the XGBoost model specifically adopts the following formula:
wherein y is j Is the true value, i.e., the jth data value in the training set;the data value obtained by predicting the jth data in the training set is a predicted value; n is the number of data contained in the training set; />The formula is thatOmega is a regularization term representing the complexity of the tree, and training over-fitting is avoided by adjusting regularization parameter gamma in the XGBoost model.
2. The method for predicting remaining life of a power battery according to claim 1, wherein the identifying noise point specifically employs the following formula:
wherein D is the original charge-discharge data set, p is any data point in the original charge-discharge data set D, C i And (3) clustering the original charge and discharge data set D by using a DBSCAN to obtain an ith class cluster, wherein m is the number of class clusters obtained after clustering.
3. The method for predicting the remaining life of a power battery according to claim 1, wherein the training of the XGBoost limit gradient boost model based on the training set specifically comprises:
and optimizing the XGBoost model parameters by adopting a grid search algorithm, and setting the XGBoost model parameters as optimal parameters obtained by the grid search algorithm.
4. A power battery remaining life prediction system, the prediction system comprising:
the original charge and discharge data set acquisition module is used for acquiring an original charge and discharge data set; the original charge-discharge dataset includes: the number of charge and discharge cycles and the battery capacity percentage data of the power battery are specific:
acquiring the charge and discharge cycle times of the power battery and corresponding battery capacity data through a battery aging test, and converting the battery capacity data into the battery capacity percentage data;
the conversion formula for converting the battery capacity data into the battery capacity percentage data is as follows:
wherein BCP represents the converted battery capacity percentage data, C max Represents the real-time maximum available discharge capacity of the power battery, C omax Representing the maximum available discharge capacity of the power battery when leaving the factory;
the test implementation flow of the battery aging test is that the temperature environment of the power battery is set to 25 ℃ firstly, and the power battery is kept stand for 2 hours; then charging to 4.2V at a rate of 0.5C based on a constant-current constant-voltage charging mode, setting the cut-off current to be 0.048A, and standing for 1 hour; then constant-current discharge is carried out to discharge cut-off voltage of 2.5V at 1C multiplying power, and standing is carried out for 1 hour; finally judging whether the discharge capacity reaches 80% of the maximum capacity of the factory, if so, stopping the test, otherwise, continuing the cyclic charge and discharge test;
the denoising module is used for denoising the original charge-discharge data set, takes the denoised original charge-discharge data set as a training set, and specifically comprises the following steps:
a noise point identification unit for identifying noise points;
a noise point deleting unit configured to delete the noise point;
the noise point replacing unit is used for replacing the original noise point position by using the average value of two adjacent data of the noise point, and specifically comprises the following steps:
the DBSCAN algorithm can effectively identify noise points in the original charge and discharge data set, so that Eps and MinPts are two parameters in the DBSCAN algorithm, wherein Eps represents the radius of a cluster type cluster, and MinPts represents the minimum sample number of each type cluster;
and the Eps neighbor of one data point in the original charge and discharge data set refers to a set of points of the point in the range of the Eps of the radius of the field, and the Eps neighbor is Eps (p), and the Eps (p) is expressed as follows:
Eps(p)＝{q∈Ddis(p,q)≤Eps}
in the above formula, D represents an original charge-discharge data set, p, q represents two data points in the original charge-discharge data set D, dis (p, q) represents a distance between the data points p and q;
the DBSCAN algorithm realizes the controllability of data denoising by adjusting the radius of clustering class clusters and the minimum sample number in each class cluster;
the training module is used for training the XGBoost limit gradient lifting model based on the training set to obtain a trained XGBoost model;
the to-be-predicted set acquisition module is used for acquiring the actual charge and discharge cycle times and the battery capacity percentage data of the to-be-predicted power battery and taking the actual charge and discharge cycle times and the battery capacity percentage data as a to-be-predicted set of the XGBoost model; the to-be-predicted set contains battery capacity percentage data of k charge-discharge cycles;
the calculation module is used for inputting the set to be predicted into the trained XGBoost model to obtain battery capacity percentage data of the next charge-discharge cycle of the power battery to be predicted, and setting the number of charge-discharge cycles corresponding to the obtained predicted value to be the q-th time;
the judging module is used for judging whether the battery capacity percentage of the q-th charge-discharge cycle obtained in the calculating module is smaller than or equal to 80% of the maximum available discharge capacity of the power battery when leaving the factory, if yes, executing the next module, if not, adding the battery capacity percentage of the q-th charge-discharge cycle to the end of a set to be predicted to construct a new set to be predicted, and returning q=q+1 to the calculating module;
the difference making module is used for making a difference between the charge-discharge cycle times q corresponding to the predicted value and the actual charge-discharge cycle times k to obtain the residual life of the power battery to be predicted;
the objective function of the XGBoost model specifically adopts the following formula:
wherein y is j Is the true value, i.e., the jth data value in the training set;the data value obtained by predicting the jth data in the training set is a predicted value; n is the number of data contained in the training set; />The formula is thatOmega is a regularization term representing the complexity of the tree, and training over-fitting is avoided by adjusting regularization parameter gamma in the XGBoost model.
5. The power battery remaining life prediction system according to claim 4, wherein the noise point identification unit specifically employs the following formula:
wherein D is the original charge-discharge data set, p is any data point in the original charge-discharge data set D, C i And (3) clustering the original charge and discharge data set D by using a DBSCAN to obtain an ith class cluster, wherein m is the number of class clusters obtained after clustering.
6. The power cell remaining life prediction system according to claim 4, wherein the training module specifically comprises:
and optimizing the XGBoost model parameters by adopting a grid search algorithm, and setting the XGBoost model parameters as optimal parameters obtained by the grid search algorithm.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010475847.0A CN112130086B (en) | 2020-05-29 | 2020-05-29 | Method and system for predicting remaining life of power battery |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010475847.0A CN112130086B (en) | 2020-05-29 | 2020-05-29 | Method and system for predicting remaining life of power battery |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN112130086A CN112130086A (en) | 2020-12-25 |
| CN112130086B true CN112130086B (en) | 2023-10-31 |
Family
ID=73851399
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010475847.0A Active CN112130086B (en) | 2020-05-29 | 2020-05-29 | Method and system for predicting remaining life of power battery |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112130086B (en) |
Families Citing this family (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113064091B (en) * | 2021-03-10 | 2023-02-10 | 北京理工新源信息科技有限公司 | Method and system for predicting remaining life of battery |
| CN114066069A (en) * | 2021-11-18 | 2022-02-18 | 国网综合能源服务集团有限公司 | Combined weight byproduct gas generation amount prediction method |
| CN114565172B (en) * | 2022-03-08 | 2023-11-03 | 苏州浪潮智能科技有限公司 | Battery remaining life prediction model construction and battery remaining life prediction method |
| CN115184814A (en) * | 2022-09-07 | 2022-10-14 | 江铃汽车股份有限公司 | Power battery pack service life prediction method and device, readable storage medium and equipment |
| CN116413609B (en) * | 2023-06-08 | 2023-08-29 | 江苏正力新能电池技术有限公司 | Battery diving identification method and device, electronic equipment and storage medium |
Family Cites Families (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103399276B (en) * | 2013-07-25 | 2016-01-20 | 哈尔滨工业大学 | A kind of capacity of lithium ion battery is estimated and cycles left life-span prediction method |
| CN106055775B (en) * | 2016-05-27 | 2018-12-11 | 哈尔滨工业大学 | A kind of service life of secondary cell prediction technique that particle filter is combined with mechanism model |
| CN106686179B (en) * | 2016-12-27 | 2019-06-21 | 东南大学 | A kind of mobile telephone power consumption test macro and method based on test automation |
| CN108896914B (en) * | 2018-05-11 | 2020-08-04 | 浙江工业大学 | Gradient lifting tree modeling and prediction method for health condition of lithium battery |
| CN110942099A (en) * | 2019-11-29 | 2020-03-31 | 华侨大学 | Abnormal data identification and detection method of DBSCAN based on core point reservation |
-
2020
- 2020-05-29 CN CN202010475847.0A patent/CN112130086B/en active Active
Also Published As
| Publication number | Publication date |
|---|---|
| CN112130086A (en) | 2020-12-25 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN112130086B (en) | Method and system for predicting remaining life of power battery | |
| CN111783953B (en) | 24-point power load value 7-day prediction method based on optimized LSTM network | |
| CN109214575B (en) | Ultrashort-term wind power prediction method based on small-wavelength short-term memory network | |
| CN112241608A (en) | Lithium battery life prediction method based on LSTM network and transfer learning | |
| Fan et al. | A novel machine learning method based approach for Li-ion battery prognostic and health management | |
| CN111193256A (en) | Power load prediction method based on variational modal decomposition and gated cyclic unit | |
| CN113554466B (en) | Short-term electricity consumption prediction model construction method, prediction method and device | |
| CN112434848B (en) | Nonlinear weighted combination wind power prediction method based on deep belief network | |
| CN110658459B (en) | Lithium ion battery state of charge estimation method based on bidirectional cyclic neural network | |
| CN115099511A (en) | Photovoltaic power probability estimation method and system based on optimized copula | |
| CN111008726B (en) | Class picture conversion method in power load prediction | |
| CN110837915B (en) | Low-voltage load point prediction and probability prediction method for power system based on hybrid integrated deep learning | |
| CN109934422A (en) | Neural network wind speed prediction method based on time series data analysis | |
| Che et al. | Boosting battery state of health estimation based on self-supervised learning | |
| CN102509026A (en) | Comprehensive short-term output power forecasting model for wind farm based on maximum information entropy theory | |
| CN114297907A (en) | Greenhouse environment spatial distribution prediction method and device | |
| CN114444561A (en) | PM2.5 prediction method based on CNNs-GRU fusion deep learning model | |
| CN109444604A (en) | A kind of DC/DC converter method for diagnosing faults based on convolutional neural networks | |
| CN110991689B (en) | Distributed photovoltaic power generation system short-term prediction method based on LSTM-Morlet model | |
| CN111191398B (en) | SVR-based method for predicting degradation trend of storage battery of direct-current system of transformer substation | |
| CN114021830A (en) | Multi-time-range wind speed prediction method based on CNN-LSTM | |
| CN116646927A (en) | Wind power prediction method based on segmented filtering and longitudinal and transverse clustering | |
| CN116865232A (en) | Wind speed error correction-based medium-and-long-term wind power prediction method and system | |
| CN111626465B (en) | New energy power short-term interval prediction method and system | |
| CN115907000A (en) | Small sample learning method for optimal power flow prediction of power system |
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 |