CN116736130A - Lithium battery residual service life prediction method and system - Google Patents
Lithium battery residual service life prediction method and system Download PDFInfo
- Publication number
- CN116736130A CN116736130A CN202310578010.2A CN202310578010A CN116736130A CN 116736130 A CN116736130 A CN 116736130A CN 202310578010 A CN202310578010 A CN 202310578010A CN 116736130 A CN116736130 A CN 116736130A
- Authority
- CN
- China
- Prior art keywords
- lithium battery
- data
- service life
- model
- dae
- 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
- WHXSMMKQMYFTQS-UHFFFAOYSA-N Lithium Chemical compound [Li] WHXSMMKQMYFTQS-UHFFFAOYSA-N 0.000 title claims abstract description 291
- 229910052744 lithium Inorganic materials 0.000 title claims abstract description 291
- 238000000034 method Methods 0.000 title claims abstract description 69
- 238000012549 training Methods 0.000 claims abstract description 67
- 230000004927 fusion Effects 0.000 claims abstract description 37
- 238000012360 testing method Methods 0.000 claims abstract description 36
- 238000012545 processing Methods 0.000 claims abstract description 27
- 230000009467 reduction Effects 0.000 claims abstract description 10
- 238000007599 discharging Methods 0.000 claims description 27
- 230000008569 process Effects 0.000 claims description 19
- 238000000354 decomposition reaction Methods 0.000 claims description 18
- 230000007246 mechanism Effects 0.000 claims description 18
- 238000005070 sampling Methods 0.000 claims description 9
- 230000004913 activation Effects 0.000 claims description 4
- 238000010276 construction Methods 0.000 claims description 4
- 230000006870 function Effects 0.000 claims description 4
- 230000004931 aggregating effect Effects 0.000 claims description 3
- 238000004590 computer program Methods 0.000 claims description 3
- 230000007774 longterm Effects 0.000 abstract description 6
- 238000004891 communication Methods 0.000 description 6
- 238000010586 diagram Methods 0.000 description 5
- 101001095088 Homo sapiens Melanoma antigen preferentially expressed in tumors Proteins 0.000 description 2
- 102100037020 Melanoma antigen preferentially expressed in tumors Human genes 0.000 description 2
- 230000003287 optical effect Effects 0.000 description 2
- 238000005457 optimization Methods 0.000 description 2
- 238000009825 accumulation Methods 0.000 description 1
- 230000002776 aggregation Effects 0.000 description 1
- 238000004220 aggregation Methods 0.000 description 1
- 238000004458 analytical method Methods 0.000 description 1
- 238000013459 approach Methods 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 230000015556 catabolic process Effects 0.000 description 1
- 230000001186 cumulative effect Effects 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000006731 degradation reaction Methods 0.000 description 1
- 238000004146 energy storage Methods 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 238000011156 evaluation Methods 0.000 description 1
- 238000012423 maintenance Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000010606 normalization Methods 0.000 description 1
- 230000000750 progressive effect Effects 0.000 description 1
- 238000006467 substitution reaction 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]
- G01R31/367—Software therefor, e.g. for battery testing using modelling or look-up tables
-
- 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
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/10—Pre-processing; Data cleansing
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/25—Fusion techniques
-
- 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
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Artificial Intelligence (AREA)
- Bioinformatics & Computational Biology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- General Engineering & Computer Science (AREA)
- Life Sciences & Earth Sciences (AREA)
- Charge And Discharge Circuits For Batteries Or The Like (AREA)
Abstract
The invention provides a method and a system for predicting the residual service life of a lithium battery, which belong to the technical field of lithium batteries and comprise the following steps: carrying out standardization processing on the original data of the lithium battery to obtain standardized data of the lithium battery, and constructing a training data set and a testing data set based on the standardized data of the lithium battery; training the improved DAE-Autoformer fusion model by using a training data set to obtain an initial lithium battery residual service life prediction model; performing precision adjustment on the initial lithium battery residual service life prediction model by using the test data set to obtain a lithium battery residual service life prediction model; and inputting the lithium battery data to be detected into a lithium battery residual service life prediction model, and outputting a lithium battery residual service life prediction result. According to the invention, through improving the DAE-Autoformer fusion model, the input battery data is subjected to noise reduction treatment and effective characteristics are learned, and the long-term prediction of the service life of the battery with high precision and high stability is realized in a shorter training time.
Description
Technical Field
The invention relates to the technical field of lithium batteries, in particular to a method and a system for predicting the residual service life of a lithium battery.
Background
In a new energy automobile, most of energy storage devices configured for the electric automobile are mainly lithium batteries, a lithium battery system is used as a core power device of the electric automobile, and the accidental end of service life often leads to the failure of the overall performance of the system, so that the safety of the whole automobile system is affected. And in the use process of the electric automobile, the electric automobile is influenced by a road environment, working temperature, driving habit of a driver and other factors, and difficulty is brought to accurate prediction of the residual service life (Remaining useful life, RUL) of the lithium battery.
The current lithium battery RUL prediction is mainly divided into a model driving method and a data driving method. The model driving method is based on analysis of physical and chemical principles of the battery, and the process of performance degradation of the lithium battery is characterized by establishing mathematical and physical models. However, the model precision is closely related to the model complexity, and the parameters are difficult to determine, so that the generalization capability is weak. The data driving method is increasingly applied to RUL prediction of the lithium battery by virtue of strong data processing capability, no specific physical model and priori knowledge of a mastering system are needed, and the capability of describing the nonlinear relationship between variables well, however, the method has few data noise reduction processing in the long-term prediction process of the service life of the method, has certain limitation in the long-term prediction of the residual service life of the lithium battery, and has the stability of the model to be improved.
Disclosure of Invention
The invention provides a method and a system for predicting the residual service life of a lithium battery, which are used for solving the defect that the residual service life of the lithium battery is lack of stability in long-term prediction in the prior art.
In a first aspect, the present invention provides a method for predicting remaining service life of a lithium battery, including:
collecting original lithium battery data, carrying out standardization processing on the original lithium battery data to obtain standardized lithium battery data, and constructing a training data set and a testing data set based on the standardized lithium battery data;
an improved DAE-Autoformer fusion model is established, the improved DAE-Autoformer fusion model is trained by utilizing the training data set, and an initial lithium battery residual service life prediction model is obtained, wherein the improved DAE-Autoformer fusion model is obtained by cascading a DAE model and an Autoformer model;
performing precision adjustment on the initial lithium battery residual service life prediction model by using the test data set to obtain a lithium battery residual service life prediction model;
and inputting the lithium battery data to be detected into the lithium battery residual service life prediction model, and outputting a lithium battery residual service life prediction result.
According to the method for predicting the residual service life of the lithium battery, provided by the invention, the original data of the lithium battery is collected, and the method comprises the following steps:
collecting lithium battery charging voltage data, lithium battery charging current data and lithium battery capacity data under each cycle in the lithium battery life attenuation process;
the lithium battery charging voltage data comprise sampling point voltages at all moments of corresponding battery charging stages under all charging and discharging cycle times in the battery life attenuation process;
the lithium battery charging current data comprise sampling point currents at all moments of corresponding battery charging stages under all charging and discharging cycle times in the battery life attenuation process;
the lithium battery capacity data comprise battery capacities of sampling points at initial moments of corresponding discharging stages under all charging and discharging cycle times in the battery life attenuation process.
According to the method for predicting the remaining service life of the lithium battery, provided by the invention, the original data of the lithium battery is standardized to obtain the standardized data of the lithium battery, and the method comprises the following steps:
acquiring the total number of charge and discharge cycles of the full life cycle of the lithium battery;
obtaining standardized data of the charging voltage of any lithium battery in any one cycle data sequence according to the original value of the charging voltage data of any lithium battery in any one cycle data sequence in the total number of charging and discharging cycles, the minimum charging voltage of the lithium battery in any one cycle data sequence and the maximum charging voltage of the lithium battery in any one cycle data sequence;
obtaining standardized data of any lithium battery charging current in any one cycle data sequence according to an original value of any lithium battery charging current in any one cycle data sequence in the total number of charging and discharging cycles, a minimum lithium battery charging current in any one cycle data sequence and a maximum lithium battery charging current in any one cycle data sequence;
and obtaining lithium battery capacity standardized data of any cycle according to the battery capacity data value of any cycle in the total number of charge and discharge cycles, the minimum capacity value of the battery capacity data sequence in all cycles and the maximum capacity value of the battery capacity data sequence in all cycles.
According to the lithium battery remaining service life prediction method provided by the invention, a training data set and a testing data set are constructed based on the lithium battery standardized data, and the method comprises the following steps:
selecting corresponding lithium battery charging voltage standardized data and lithium battery charging current standardized data under the charging and discharging cycle times of N groups of batteries in the lithium battery standardized data as the training data set, wherein N is smaller than the total charging and discharging cycle times M;
and determining lithium battery charging voltage standardized data and lithium battery charging current standardized data corresponding to the M-N battery charging and discharging cycle times in the lithium battery standardized data, and taking the lithium battery charging voltage standardized data and the lithium battery charging current standardized data as the test data set.
According to the method for predicting the remaining service life of the lithium battery, an improved DAE-Autoformer fusion model of a denoising automatic encoder is established, the improved DAE-Autoformer fusion model is trained by utilizing the training data set, and an initial lithium battery remaining service life prediction model is obtained, and the method comprises the following steps:
determining to cascade a DAE model and an Autoformer model, wherein the DAE model is formed by sequentially connecting a mask layer, a DAE coding layer and a DAE decoding layer in series, and the Autoformer model is formed by cascading an encoder and a decoder, wherein the encoder is connected with the DAE coding layer, and the encoder and the decoder both comprise a sequence decomposition unit;
and inputting the training data set into the DAE model to perform data noise reduction to obtain a preset low-dimensional hidden variable, and inputting the preset low-dimensional hidden variable into the Autoformer model to perform training to obtain the initial lithium battery residual service life prediction model.
According to the method for predicting the remaining service life of the lithium battery, the training data set is input into the DAE model to perform data noise reduction to obtain the preset low-dimensional hidden variable, and the method comprises the following steps:
injecting Gaussian white noise into the training data set by the mask layer to obtain noise-containing samples, and inputting the noise-containing samples into the DAE coding layer;
the DAE encoding layer encodes the noise-containing samples to obtain the preset low-dimensional hidden variables, and the preset low-dimensional hidden variables are respectively input to a DAE decoding layer and the encoder through a preset activation function;
and the DAE decoding layer restores the preset low-dimensional hidden variable to the initial dimension of the training data set to obtain a reconstructed noise-free sample.
According to the method for predicting the remaining service life of the lithium battery, provided by the invention, the preset low-dimensional hidden variable is input into the Autoformer model for training to obtain the initial lithium battery remaining service life prediction model, and the method comprises the following steps:
acquiring trend items and period items from the preset low-dimensional hidden variables by the sequence decomposition unit in the encoder, eliminating the trend items by an average filter, reserving the period items, aggregating different period similar subsequences by an autocorrelation mechanism to obtain an aggregate information sequence, and processing the aggregate information sequence by a feedforward layer and the sequence decomposition unit to obtain a historical time sequence;
extracting the inherent time dependence characteristic of the prediction state from the period item by an autocorrelation mechanism in the decoder, extracting information from the historical time sequence by the autocorrelation mechanism, processing the extracted information by a feedforward layer, weighting and accumulating the trend item by a sequence decomposition unit respectively embedded between each autocorrelation mechanism and the feedforward layer, and summing the extracted information output corresponding to the period item to obtain the initial lithium battery residual service life prediction model.
According to the method for predicting the remaining service life of the lithium battery, provided by the invention, the initial lithium battery remaining service life prediction model is subjected to precision adjustment by utilizing the test data set to obtain the lithium battery remaining service life prediction model, and the method comprises the following steps:
inputting the test data set into the initial lithium battery residual service life prediction model to obtain a lithium battery residual service life prediction value;
obtaining a lithium battery residual service life true value, calculating the lithium battery residual service life predicted value, the lithium battery residual service life true value and the total number of samples of the test data set by using an average absolute percentage error, and obtaining model prediction precision;
and if the model prediction precision meets the preset model precision threshold, determining that the initial lithium battery residual service life prediction model is the lithium battery residual service life prediction model, otherwise, optimizing the super-parameters of the initial lithium battery residual service life prediction model, and carrying out model training again until the lithium battery residual service life prediction model is obtained.
In a second aspect, the present invention also provides a lithium battery remaining service life prediction system, including:
the construction module is used for collecting original lithium battery data, carrying out standardization processing on the original lithium battery data to obtain standardized lithium battery data, and constructing a training data set and a testing data set based on the standardized lithium battery data;
the training module is used for building an improved DAE-Autoformer fusion model, training the improved DAE-Autoformer fusion model by utilizing the training data set to obtain an initial lithium battery residual service life prediction model, wherein the improved DAE-Autoformer fusion model is obtained by cascading the DAE model and the Autoformer model;
the adjustment module is used for carrying out precision adjustment on the initial lithium battery residual service life prediction model by utilizing the test data set to obtain a lithium battery residual service life prediction model;
and the prediction module is used for inputting the lithium battery data to be detected into the lithium battery residual service life prediction model and outputting a lithium battery residual service life prediction result.
In a third aspect, the present invention also provides an electronic device, including a memory, a processor, and a computer program stored on the memory and executable on the processor, the processor implementing a method for predicting remaining life of a lithium battery as described in any one of the above when executing the program.
According to the lithium battery remaining service life prediction method and system, the improved DAE-Autoformer fusion model is used for carrying out noise reduction processing on input battery data and learning effective characteristics, high-precision high-stability long-term prediction of the battery life is achieved in a short training time, and in an online lithium battery remaining service life prediction scene, the method and system have important guiding significance for effectively formulating maintenance, management and replacement strategies of a user battery and guaranteeing safe and stable operation of the battery in a full life cycle.
Drawings
In order to more clearly illustrate the invention or the technical solutions of the prior art, the drawings used in the description of the embodiments or the prior art will be briefly described, and it is obvious that the drawings in the description below are some embodiments of the invention, and other drawings can be obtained according to these drawings without inventive effort for a person skilled in the art.
FIG. 1 is a schematic flow chart of a method for predicting the residual service life of a lithium battery;
FIG. 2 is a training flow chart of a lithium battery residual service life prediction model provided by the invention;
FIG. 3 is a schematic diagram of a lithium battery residual service life prediction model provided by the invention;
fig. 4 is a schematic structural diagram of a lithium battery remaining service life prediction system provided by the invention;
fig. 5 is a schematic structural diagram of an electronic device provided by the present invention.
Detailed Description
For the purpose of making the objects, technical solutions and advantages of the present invention more apparent, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings, and it is apparent that the described embodiments are some embodiments of the present invention, 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.
Aiming at the limitation of the residual service life prediction technology of the lithium battery in the prior art, the invention provides an improved residual service life prediction model of the lithium battery of the DAE-Autoformer by fusing a Denoising Auto-Encoder (DAE) model and an Autoformer model, adopts noise reduction processing of lithium battery data to realize online accurate prediction of the residual service life of the lithium battery, and has high prediction stability.
Fig. 1 is a flow chart of a method for predicting remaining service life of a lithium battery according to an embodiment of the present invention, as shown in fig. 1, including:
step 100: collecting original lithium battery data, carrying out standardization processing on the original lithium battery data to obtain standardized lithium battery data, and constructing a training data set and a testing data set based on the standardized lithium battery data;
step 200: an improved DAE-Autoformer fusion model is established, the improved DAE-Autoformer fusion model is trained by utilizing the training data set, and an initial lithium battery residual service life prediction model is obtained, wherein the improved DAE-Autoformer fusion model is obtained by cascading a DAE model and an Autoformer model;
step 300: performing precision adjustment on the initial lithium battery residual service life prediction model by using the test data set to obtain a lithium battery residual service life prediction model;
step 400: and inputting the lithium battery data to be detected into the lithium battery residual service life prediction model, and outputting a lithium battery residual service life prediction result.
Specifically, according to the embodiment of the invention, through collecting lithium battery data, sequentially carrying out standardization processing on the lithium battery data to obtain standardized lithium battery data, constructing a training data set and a test data set for predicting the residual service life of the lithium battery, establishing an improved DAE-Autoformer fusion model, carrying out model training by adopting the training data set to obtain an initial lithium battery residual service life prediction model, then adopting the test data set to adjust model parameters for the evaluation condition of the initial lithium battery residual service life prediction model, obtaining a lithium battery residual service life prediction model after the model precision meets the requirement, and inputting the lithium battery data to be detected into the lithium battery residual service life prediction model to obtain a lithium battery residual service life prediction result.
As shown in fig. 2, first, a certain amount of original data of lithium batteries are obtained, standardized data of the lithium batteries are obtained, including charging voltage data, charging current data and battery capacity, a part of sample data is used as a training data set, a lithium battery residual service life prediction model is obtained after training a constructed model, a comparison calculation is performed according to a prediction result and an actual result output by the model, for example, whether the accuracy of the model meets the requirement is verified by adopting an average absolute percentage error (Mean Absolute Percentage Error, MAPE), if not, super-parameter optimization is performed, after the accuracy requirement is met, a final lithium battery residual service life prediction model is confirmed, and standardized lithium battery data to be predicted is input into the model to obtain a high-accuracy lithium battery residual service life prediction result.
According to the invention, through an improved DAE-Autoformer fusion model, the input battery data is subjected to noise reduction treatment and effective characteristics are learned, and the long-term prediction of the service life of the battery with high precision and high stability is realized in a short training time.
Based on the above embodiment, collecting lithium battery raw data includes:
collecting lithium battery charging voltage data, lithium battery charging current data and lithium battery capacity data under each cycle in the lithium battery life attenuation process;
the lithium battery charging voltage data comprise sampling point voltages at all moments of corresponding battery charging stages under all charging and discharging cycle times in the battery life attenuation process;
the lithium battery charging current data comprise sampling point currents at all moments of corresponding battery charging stages under all charging and discharging cycle times in the battery life attenuation process;
the lithium battery capacity data comprise battery capacities of sampling points at initial moments of corresponding discharging stages under all charging and discharging cycle times in the battery life attenuation process.
Based on the above embodiment, the normalizing process is performed on the original data of the lithium battery to obtain normalized data of the lithium battery, including:
acquiring the total number of charge and discharge cycles of the full life cycle of the lithium battery;
obtaining standardized data of the charging voltage of any lithium battery in any one cycle data sequence according to the original value of the charging voltage data of any lithium battery in any one cycle data sequence in the total number of charging and discharging cycles, the minimum charging voltage of the lithium battery in any one cycle data sequence and the maximum charging voltage of the lithium battery in any one cycle data sequence;
obtaining standardized data of any lithium battery charging current in any one cycle data sequence according to an original value of any lithium battery charging current in any one cycle data sequence in the total number of charging and discharging cycles, a minimum lithium battery charging current in any one cycle data sequence and a maximum lithium battery charging current in any one cycle data sequence;
and obtaining lithium battery capacity standardized data of any cycle according to the battery capacity data value of any cycle in the total number of charge and discharge cycles, the minimum capacity value of the battery capacity data sequence in all cycles and the maximum capacity value of the battery capacity data sequence in all cycles.
Specifically, the embodiment of the invention sequentially performs standardization processing on the charging voltage, the charging current and the capacity data of the lithium battery corresponding to each cycle;
the standardized processing process for the lithium battery charging voltage and charging current data is as follows:
wherein x is ij Represents the original value of the j-th data in the data sequence at the i-th cycle,represents the normalized processing result of the jth data in the data sequence under the ith cycle, x ij,min Representing the minimum value, x, in the data sequence at the ith cycle ij,max Representing the maximum value of the data sequence under the ith cycle, and M represents the total number of charge and discharge cycles undergone by the full life cycle of the battery;
the standardized processing procedure for the lithium battery capacity data is as follows:
wherein Q is i Represents the battery capacity data value at the i-th cycle,represents the battery capacity data value normalization processing result at the ith cycle, Q min Representing the minimum value, Q, of the battery capacity data sequence over all cycles max The maximum value of the battery capacity data sequence in all cycles.
Based on the above embodiment, constructing a training data set and a test data set based on the lithium battery standardized data includes:
selecting corresponding lithium battery charging voltage standardized data and lithium battery charging current standardized data under the charging and discharging cycle times of N groups of batteries in the lithium battery standardized data as the training data set, wherein N is smaller than the total charging and discharging cycle times M;
and determining lithium battery charging voltage standardized data and lithium battery charging current standardized data corresponding to the M-N battery charging and discharging cycle times in the lithium battery standardized data, and taking the lithium battery charging voltage standardized data and the lithium battery charging current standardized data as the test data set.
Based on the above embodiment, an improved de-noising automatic encoder DAE-Autoformer fusion model is established, and the improved DAE-Autoformer fusion model is trained by using the training data set to obtain an initial lithium battery residual service life prediction model, which comprises the following steps:
determining to cascade a DAE model and an Autoformer model, wherein the DAE model is formed by sequentially connecting a mask layer, a DAE coding layer and a DAE decoding layer in series, and the Autoformer model is formed by cascading an encoder and a decoder, wherein the encoder is connected with the DAE coding layer, and the encoder and the decoder both comprise a sequence decomposition unit;
and inputting the training data set into the DAE model to perform data noise reduction to obtain a preset low-dimensional hidden variable, and inputting the preset low-dimensional hidden variable into the Autoformer model to perform training to obtain the initial lithium battery residual service life prediction model.
Inputting the training data set into the DAE model for data denoising to obtain a preset low-dimensional hidden variable, wherein the method comprises the following steps of:
injecting Gaussian white noise into the training data set by the mask layer to obtain noise-containing samples, and inputting the noise-containing samples into the DAE coding layer;
the DAE encoding layer encodes the noise-containing samples to obtain the preset low-dimensional hidden variables, and the preset low-dimensional hidden variables are respectively input to a DAE decoding layer and the encoder through a preset activation function;
and the DAE decoding layer restores the preset low-dimensional hidden variable to the initial dimension of the training data set to obtain a reconstructed noise-free sample.
Inputting the preset low-dimensional hidden variable into the Autoformer model for training to obtain the initial lithium battery residual service life prediction model, wherein the method comprises the following steps of:
acquiring trend items and period items from the preset low-dimensional hidden variables by the sequence decomposition unit in the encoder, eliminating the trend items by an average filter, reserving the period items, aggregating different period similar subsequences by an autocorrelation mechanism to obtain an aggregate information sequence, and processing the aggregate information sequence by a feedforward layer and the sequence decomposition unit to obtain a historical time sequence;
extracting the inherent time dependence characteristic of the prediction state from the period item by an autocorrelation mechanism in the decoder, extracting information from the historical time sequence by the autocorrelation mechanism, processing the extracted information by a feedforward layer, weighting and accumulating the trend item by a sequence decomposition unit respectively embedded between each autocorrelation mechanism and the feedforward layer, and summing the extracted information output corresponding to the period item to obtain the initial lithium battery residual service life prediction model.
Specifically, the structure of the improved DAE-Autoformer fusion model constructed by the embodiment of the invention is shown in figure 3, the DAE fusion model is formed by cascading a DAE model and an Autoformer model, and the DAE model is formed by sequentially connecting a mask layer, a DAE coding layer and a DAE decoding layer in series and is used for noise reduction treatment of input data so as to ensure that the essential characteristics of the data are extracted; the Autoformer model consists of an encoder and a decoder.
In fig. 3, the left square frame is a DAE model part, gaussian noise is injected into a training data set by a mask layer, so as to obtain a noisy "spoiled" sample, and the noise "spoiled" sample is output to a DAE coding layer, namely an h1 part;
the DAE coding layer codes the noisy 'spoiled' samples to obtain low-dimensional hidden variables, and outputs the hidden variables to the DAE decoding layer and an Autoformer model, wherein the activation function adopted by the hidden variables is relu, namely an h2 part;
the DAE decoding layer restores the hidden variable with low dimension to the initial dimension to obtain a reconstructed clean input without noise, namely an h3 part;
and finally, outputting a final prediction result by properly training the low-dimensional hidden variable through an Autoformer model.
Further, the trend term and the period term are removed from the low-dimensional hidden variable of the DAE coding layer by a sequence decomposition unit (Series Decomposition Block) in the encoder of the Autoformer model in fig. 3, and the trend term is processed by an average filter;
the encoder eliminates the trend term to obtain a period term, aggregates similar subprocesses of different periods through an autocorrelation mechanism (Auto-Correlation Mechanism), realizes information aggregation, outputs information containing past periods to a decoder, and helps the decoder to optimize a prediction result as cross information;
the decoder adopts a two-way processing mode, the upper branch firstly uses an autocorrelation mechanism to extract the time dependence in the future prediction state from a period item, then uses an encoder-decoder autocorrelation mechanism to extract information from a historical sequence output by the encoder and having high-order time dependence, finally carries out Feed Forward (Feed Forward) processing, and the lower branch uses weighted addition to add the output of each sub-layer of the upper branch and output a prediction result; it can be appreciated that the trend term and the period term are modeled separately in an Autoformer model, wherein for the period term, the autocorrelation mechanism uses the period property of the sequence to aggregate subsequences with similar processes in different periods; for trend terms, we progressively extract trend information from predicted hidden variables using a cumulative approach.
The encoder and the decoder each include a sequence decomposition unit, and the sequence decomposition unit is embedded in the Autoformer encoder and the encoder, respectively. According to the progressive decomposition architecture, hidden variables can be decomposed step by step in the prediction process by the model, and prediction results of period and trend components are obtained respectively in an autocorrelation mechanism and accumulation mode, so that alternate implementation and mutual promotion of decomposition and prediction result optimization are realized.
Based on the above embodiment, performing accuracy adjustment on the initial lithium battery remaining service life prediction model by using the test data set to obtain a lithium battery remaining service life prediction model, including:
inputting the test data set into the initial lithium battery residual service life prediction model to obtain a lithium battery residual service life prediction value;
obtaining a lithium battery residual service life true value, calculating the lithium battery residual service life predicted value, the lithium battery residual service life true value and the total number of samples of the test data set by using an average absolute percentage error, and obtaining model prediction precision;
and if the model prediction precision meets the preset model precision threshold, determining that the initial lithium battery residual service life prediction model is the lithium battery residual service life prediction model, otherwise, optimizing the super-parameters of the initial lithium battery residual service life prediction model, and carrying out model training again until the lithium battery residual service life prediction model is obtained.
Specifically, the embodiment of the invention adopts MAPE to judge whether the model prediction precision meets the requirement, if so, a prediction result is output, and if not, the network super-parameters are readjusted.
Wherein M-N represents the total number of samples of the test dataset, C tri Representing the true value of the residual service life of the lithium battery, C pri And representing the predicted value of the residual service life of the lithium battery.
The remaining service life prediction system of the lithium battery provided by the invention is described below, and the remaining service life prediction system of the lithium battery described below and the remaining service life prediction method of the lithium battery described above can be referred to correspondingly.
Fig. 4 is a schematic structural diagram of a lithium battery remaining service life prediction system according to an embodiment of the present invention, as shown in fig. 4, including: a construction module 41, a training module 42, an adjustment module 43 and a prediction module 44, wherein:
the construction module 41 is used for collecting original lithium battery data, carrying out standardization processing on the original lithium battery data to obtain standardized lithium battery data, and constructing a training data set and a testing data set based on the standardized lithium battery data; the training module 42 is configured to build an improved DAE-auto former fusion model, and train the improved DAE-auto former fusion model by using the training data set to obtain an initial lithium battery residual service life prediction model, where the improved DAE-auto former fusion model is obtained by cascading a DAE model and an auto former model; the adjustment module 43 is configured to perform precision adjustment on the initial lithium battery remaining service life prediction model by using the test data set, so as to obtain a lithium battery remaining service life prediction model; the prediction module 44 is configured to input lithium battery data to be detected into the lithium battery remaining service life prediction model, and output a lithium battery remaining service life prediction result.
Fig. 5 illustrates a physical schematic diagram of an electronic device, as shown in fig. 5, which may include: processor 510, communication interface (Communications Interface) 520, memory 530, and communication bus 540, wherein processor 510, communication interface 520, memory 530 complete communication with each other through communication bus 540. Processor 510 may invoke logic instructions in memory 530 to perform a lithium battery remaining life prediction method comprising: collecting original lithium battery data, carrying out standardization processing on the original lithium battery data to obtain standardized lithium battery data, and constructing a training data set and a testing data set based on the standardized lithium battery data; an improved DAE-Autoformer fusion model is established, the improved DAE-Autoformer fusion model is trained by utilizing the training data set, and an initial lithium battery residual service life prediction model is obtained, wherein the improved DAE-Autoformer fusion model is obtained by cascading a DAE model and an Autoformer model; performing precision adjustment on the initial lithium battery residual service life prediction model by using the test data set to obtain a lithium battery residual service life prediction model; and inputting the lithium battery data to be detected into the lithium battery residual service life prediction model, and outputting a lithium battery residual service life prediction result.
Further, the logic instructions in the memory 530 described above may be implemented in the form of software functional units and may be stored in a computer-readable storage medium when sold or used as a stand-alone product. Based on this understanding, the technical solution of the present invention may be embodied essentially or in a part contributing to the prior art or in a part of the technical solution, in the form of a software product stored in a storage medium, comprising several instructions for causing a computer device (which may be a personal computer, a server, a network device, etc.) to perform all or part of the steps of the method according to the embodiments of the present invention. And the aforementioned storage medium includes: a U-disk, a removable hard disk, a Read-Only Memory (ROM), a random access Memory (RAM, random Access Memory), a magnetic disk, or an optical disk, or other various media capable of storing program codes.
In another aspect, the present invention also provides a non-transitory computer readable storage medium having stored thereon a computer program which, when executed by a processor, is implemented to perform the method for predicting remaining life of a lithium battery provided by the above methods, the method comprising: collecting original lithium battery data, carrying out standardization processing on the original lithium battery data to obtain standardized lithium battery data, and constructing a training data set and a testing data set based on the standardized lithium battery data; an improved DAE-Autoformer fusion model is established, the improved DAE-Autoformer fusion model is trained by utilizing the training data set, and an initial lithium battery residual service life prediction model is obtained, wherein the improved DAE-Autoformer fusion model is obtained by cascading a DAE model and an Autoformer model; performing precision adjustment on the initial lithium battery residual service life prediction model by using the test data set to obtain a lithium battery residual service life prediction model; and inputting the lithium battery data to be detected into the lithium battery residual service life prediction model, and outputting a lithium battery residual service life prediction result.
The apparatus embodiments described above are merely illustrative, wherein the elements illustrated as separate elements may or may not be physically separate, and the elements shown as elements may or may not be physical elements, may be located in one place, or may be distributed over a plurality of network elements. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution of this embodiment. Those of ordinary skill in the art will understand and implement the present invention without undue burden.
From the above description of the embodiments, it will be apparent to those skilled in the art that the embodiments may be implemented by means of software plus necessary general hardware platforms, or of course may be implemented by means of hardware. Based on this understanding, the foregoing technical solution may be embodied essentially or in a part contributing to the prior art in the form of a software product, which may be stored in a computer readable storage medium, such as ROM/RAM, a magnetic disk, an optical disk, etc., including several instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) to execute the method described in the respective embodiments or some parts of the embodiments.
Finally, it should be noted that: the above embodiments are only for illustrating the technical solution of the present invention, and are not limiting; although the 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 scheme described in the foregoing embodiments can be modified or some technical features thereof can be replaced by equivalents; such modifications and substitutions do not depart from the spirit and scope of the technical solutions of the embodiments of the present invention.
Claims (10)
1. A method for predicting remaining service life of a lithium battery, comprising:
collecting original lithium battery data, carrying out standardization processing on the original lithium battery data to obtain standardized lithium battery data, and constructing a training data set and a testing data set based on the standardized lithium battery data;
an improved DAE-Autoformer fusion model of a denoising automatic encoder is established, the improved DAE-Autoformer fusion model is trained by utilizing the training data set, and an initial lithium battery residual service life prediction model is obtained, wherein the improved DAE-Autoformer fusion model is obtained by cascading a DAE model and an Autoformer model;
performing precision adjustment on the initial lithium battery residual service life prediction model by using the test data set to obtain a lithium battery residual service life prediction model;
and inputting the lithium battery data to be detected into the lithium battery residual service life prediction model, and outputting a lithium battery residual service life prediction result.
2. The method for predicting remaining life of a lithium battery as claimed in claim 1, wherein collecting raw data of the lithium battery comprises:
collecting lithium battery charging voltage data, lithium battery charging current data and lithium battery capacity data under each cycle in the lithium battery life attenuation process;
the lithium battery charging voltage data comprise sampling point voltages at all moments of corresponding battery charging stages under all charging and discharging cycle times in the battery life attenuation process;
the lithium battery charging current data comprise sampling point currents at all moments of corresponding battery charging stages under all charging and discharging cycle times in the battery life attenuation process;
the lithium battery capacity data comprise battery capacities of sampling points at initial moments of corresponding discharging stages under all charging and discharging cycle times in the battery life attenuation process.
3. The method for predicting remaining service life of a lithium battery according to claim 2, wherein the normalizing the raw data of the lithium battery to obtain normalized data of the lithium battery comprises:
acquiring the total number of charge and discharge cycles of the full life cycle of the lithium battery;
obtaining standardized data of the charging voltage of any lithium battery in any one cycle data sequence according to the original value of the charging voltage data of any lithium battery in any one cycle data sequence in the total number of charging and discharging cycles, the minimum charging voltage of the lithium battery in any one cycle data sequence and the maximum charging voltage of the lithium battery in any one cycle data sequence;
obtaining standardized data of any lithium battery charging current in any one cycle data sequence according to an original value of any lithium battery charging current in any one cycle data sequence in the total number of charging and discharging cycles, a minimum lithium battery charging current in any one cycle data sequence and a maximum lithium battery charging current in any one cycle data sequence;
and obtaining lithium battery capacity standardized data of any cycle according to the battery capacity data value of any cycle in the total number of charge and discharge cycles, the minimum capacity value of the battery capacity data sequence in all cycles and the maximum capacity value of the battery capacity data sequence in all cycles.
4. The method of claim 3, wherein constructing a training data set and a test data set based on the lithium battery standardized data comprises:
selecting corresponding lithium battery charging voltage standardized data and lithium battery charging current standardized data under the charging and discharging cycle times of N groups of batteries in the lithium battery standardized data as the training data set, wherein N is smaller than the total charging and discharging cycle times M;
and determining lithium battery charging voltage standardized data and lithium battery charging current standardized data corresponding to the M-N battery charging and discharging cycle times in the lithium battery standardized data, and taking the lithium battery charging voltage standardized data and the lithium battery charging current standardized data as the test data set.
5. The method of claim 1, wherein creating an improved de-noising auto-encoder DAE-auto-former fusion model, training the improved DAE-auto-former fusion model with the training data set to obtain an initial lithium battery remaining lifetime prediction model, comprises:
determining to cascade a DAE model and an Autoformer model, wherein the DAE model is formed by sequentially connecting a mask layer, a DAE coding layer and a DAE decoding layer in series, and the Autoformer model is formed by cascading an encoder and a decoder, wherein the encoder is connected with the DAE coding layer, and the encoder and the decoder both comprise a sequence decomposition unit;
and inputting the training data set into the DAE model to perform data noise reduction to obtain a preset low-dimensional hidden variable, and inputting the preset low-dimensional hidden variable into the Autoformer model to perform training to obtain the initial lithium battery residual service life prediction model.
6. The method of claim 5, wherein inputting the training data set into the DAE model for data denoising to obtain a preset low-dimensional hidden variable comprises:
injecting Gaussian white noise into the training data set by the mask layer to obtain noise-containing samples, and inputting the noise-containing samples into the DAE coding layer;
the DAE encoding layer encodes the noise-containing samples to obtain the preset low-dimensional hidden variables, and the preset low-dimensional hidden variables are respectively input to a DAE decoding layer and the encoder through a preset activation function;
and the DAE decoding layer restores the preset low-dimensional hidden variable to the initial dimension of the training data set to obtain a reconstructed noise-free sample.
7. The method for predicting remaining life of a lithium battery according to claim 5, wherein inputting the preset low-dimensional hidden variable into the Autoformer model for training to obtain the initial lithium battery remaining life prediction model comprises:
acquiring trend items and period items from the preset low-dimensional hidden variables by the sequence decomposition unit in the encoder, eliminating the trend items by an average filter, reserving the period items, aggregating different period similar subsequences by an autocorrelation mechanism to obtain an aggregate information sequence, and processing the aggregate information sequence by a feedforward layer and the sequence decomposition unit to obtain a historical time sequence;
extracting the inherent time dependence characteristic of the prediction state from the period item by an autocorrelation mechanism in the decoder, extracting information from the historical time sequence by the autocorrelation mechanism, processing the extracted information by a feedforward layer, weighting and accumulating the trend item by a sequence decomposition unit respectively embedded between each autocorrelation mechanism and the feedforward layer, and summing the extracted information output corresponding to the period item to obtain the initial lithium battery residual service life prediction model.
8. The method for predicting remaining life of a lithium battery according to claim 1, wherein performing precision adjustment on the initial lithium battery remaining life prediction model by using the test data set to obtain a lithium battery remaining life prediction model comprises:
inputting the test data set into the initial lithium battery residual service life prediction model to obtain a lithium battery residual service life prediction value;
obtaining a lithium battery residual service life true value, calculating the lithium battery residual service life predicted value, the lithium battery residual service life true value and the total number of samples of the test data set by using an average absolute percentage error, and obtaining model prediction precision;
and if the model prediction precision meets the preset model precision threshold, determining that the initial lithium battery residual service life prediction model is the lithium battery residual service life prediction model, otherwise, optimizing the super-parameters of the initial lithium battery residual service life prediction model, and carrying out model training again until the lithium battery residual service life prediction model is obtained.
9. A lithium battery remaining life prediction system, comprising:
the construction module is used for collecting original lithium battery data, carrying out standardization processing on the original lithium battery data to obtain standardized lithium battery data, and constructing a training data set and a testing data set based on the standardized lithium battery data;
the training module is used for building an improved DAE-Autoformer fusion model, training the improved DAE-Autoformer fusion model by utilizing the training data set to obtain an initial lithium battery residual service life prediction model, wherein the improved DAE-Autoformer fusion model is obtained by cascading the DAE model and the Autoformer model;
the adjustment module is used for carrying out precision adjustment on the initial lithium battery residual service life prediction model by utilizing the test data set to obtain a lithium battery residual service life prediction model;
and the prediction module is used for inputting the lithium battery data to be detected into the lithium battery residual service life prediction model and outputting a lithium battery residual service life prediction result.
10. An electronic device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, wherein the processor implements the method for predicting remaining life of a lithium battery as claimed in any one of claims 1 to 8 when executing the program.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310578010.2A CN116736130A (en) | 2023-05-22 | 2023-05-22 | Lithium battery residual service life prediction method and system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310578010.2A CN116736130A (en) | 2023-05-22 | 2023-05-22 | Lithium battery residual service life prediction method and system |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN116736130A true CN116736130A (en) | 2023-09-12 |
Family
ID=87907061
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202310578010.2A Pending CN116736130A (en) | 2023-05-22 | 2023-05-22 | Lithium battery residual service life prediction method and system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116736130A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117252112A (en) * | 2023-11-16 | 2023-12-19 | 江苏林洋亿纬储能科技有限公司 | Method for training driving data model and method for estimating remaining life of battery |
-
2023
- 2023-05-22 CN CN202310578010.2A patent/CN116736130A/en active Pending
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117252112A (en) * | 2023-11-16 | 2023-12-19 | 江苏林洋亿纬储能科技有限公司 | Method for training driving data model and method for estimating remaining life of battery |
| CN117252112B (en) * | 2023-11-16 | 2024-01-30 | 江苏林洋亿纬储能科技有限公司 | Method for training driving data model and method for estimating remaining life of battery |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN112986827B (en) | Fuel cell residual life prediction method based on deep learning | |
| CN110798365B (en) | Neural network-based traffic prediction method and device | |
| CN112990556A (en) | User power consumption prediction method based on Prophet-LSTM model | |
| CN110687450B (en) | Lithium battery residual life prediction method based on phase space reconstruction and particle filtering | |
| CN111899759B (en) | Method, device, equipment and medium for pre-training and model training of audio data | |
| CN112415414A (en) | Method for predicting remaining service life of lithium ion battery | |
| CN113065283A (en) | Battery life prediction method, system, electronic device and storage medium | |
| CN116541790B (en) | New energy vehicle health assessment method and device based on multi-feature fusion | |
| CN113376540B (en) | LSTM battery health state estimation method based on evolutionary attention mechanism | |
| CN112230154A (en) | Lithium battery residual life prediction method | |
| CN114740388A (en) | Lithium battery residual life state evaluation method based on improved TCN | |
| CN116736130A (en) | Lithium battery residual service life prediction method and system | |
| CN113449463A (en) | LSTM-DNN-based equipment life prediction method and device | |
| CN112803398A (en) | Load prediction method and system based on empirical mode decomposition and deep neural network | |
| CN112379274A (en) | Method for predicting residual life of power battery | |
| CN116794543A (en) | Lithium battery performance prediction method based on GBLS boost multitask learning model | |
| CN116540101A (en) | Rapid battery life prediction method for key interval feature extraction | |
| CN114924203A (en) | Battery SOH prediction analysis method and electric automobile | |
| CN116894153A (en) | Landslide displacement prediction method based on singular spectrum analysis and ARIMA-LSTM combined model | |
| CN116298936A (en) | Intelligent lithium ion battery health state prediction method in incomplete voltage range | |
| CN115687322A (en) | Water quality time series missing data completion method based on encoder-decoder and autoregressive generated countermeasure network | |
| CN115640510A (en) | Power time series data feature extraction method, system, equipment and medium | |
| CN116665483A (en) | Novel method for predicting residual parking space | |
| CN113221458B (en) | Multi-step prediction method and system for shield cutter head torque | |
| CN113689039A (en) | Lithium ion battery RUL time sequence prediction method |
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 |