CN115204038A - Energy storage lithium battery life prediction method based on data decomposition and integration model - Google Patents
Energy storage lithium battery life prediction method based on data decomposition and integration model Download PDFInfo
- Publication number
- CN115204038A CN115204038A CN202210711671.3A CN202210711671A CN115204038A CN 115204038 A CN115204038 A CN 115204038A CN 202210711671 A CN202210711671 A CN 202210711671A CN 115204038 A CN115204038 A CN 115204038A
- Authority
- CN
- China
- Prior art keywords
- prediction
- data
- lithium battery
- model
- individuals
- 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.)
- Withdrawn
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/20—Design optimisation, verification or simulation
- G06F30/27—Design optimisation, verification or simulation using machine learning, e.g. artificial intelligence, neural networks, support vector machines [SVM] or training a model
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2111/00—Details relating to CAD techniques
- G06F2111/06—Multi-objective optimisation, e.g. Pareto optimisation using simulated annealing [SA], ant colony algorithms or genetic algorithms [GA]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2119/00—Details relating to the type or aim of the analysis or the optimisation
- G06F2119/04—Ageing analysis or optimisation against ageing
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Evolutionary Computation (AREA)
- Business, Economics & Management (AREA)
- General Physics & Mathematics (AREA)
- Economics (AREA)
- General Engineering & Computer Science (AREA)
- Strategic Management (AREA)
- Human Resources & Organizations (AREA)
- Artificial Intelligence (AREA)
- Software Systems (AREA)
- Geometry (AREA)
- Development Economics (AREA)
- Biomedical Technology (AREA)
- Mathematical Physics (AREA)
- Molecular Biology (AREA)
- General Health & Medical Sciences (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Medical Informatics (AREA)
- Computer Hardware Design (AREA)
- Computational Linguistics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Computing Systems (AREA)
- Health & Medical Sciences (AREA)
- Game Theory and Decision Science (AREA)
- Biophysics (AREA)
- Data Mining & Analysis (AREA)
- Entrepreneurship & Innovation (AREA)
- Marketing (AREA)
- Operations Research (AREA)
- Quality & Reliability (AREA)
- Tourism & Hospitality (AREA)
- General Business, Economics & Management (AREA)
- Secondary Cells (AREA)
Abstract
The invention discloses a method for predicting the service life of an energy storage lithium battery based on a data decomposition and integrated learning model, which is used for the field of state detection and management of an energy storage lithium battery device and comprises the following steps: acquiring historical capacity degradation data sets of energy storage lithium batteries with different parameter characteristics in a new energy high-permeability smart grid environment; respectively preprocessing the capacity degradation data sets, and enhancing the potential regularity characteristics of the data to enable the data to be more easily captured by a neural network; respectively constructing residual life prediction model individuals based on LSTM, and performing parameter optimization by using MOEA/D multi-objective optimization algorithm; integrating the obtained prediction individual models according to the prediction performance requirements; the invention can establish a prediction method for the residual life of the lithium battery on the premise of lower cost, and enhance the prediction performance and generalization capability of a prediction model in the life prediction of a multi-battery pack.
Description
Technical Field
The invention relates to the field of state detection of high-capacity energy storage lithium battery devices, in particular to a method for predicting the service life of an energy storage lithium battery based on a data decomposition and integrated learning model.
Background
Lithium batteries are widely used in battery energy storage systems due to their advantages of high energy density, good safety, long cycle charge and discharge life, and low self-discharge rate.
The energy storage lithium battery is internally provided with a dynamic and time-varying electrochemical system, has nonlinear behavior and a complex internal reaction mechanism, and has the service life cycle times influenced by factors such as working temperature, discharge power, charge-discharge state conversion, discharge depth and the like. After a certain number of charge-discharge cycles is reached, the actual energy storage capacity of the energy storage device is greatly reduced due to aging of the lithium battery pack, and certain potential safety hazards are brought. Predicting the remaining life (RUL) of a lithium energy storage battery helps to improve the reliability and safety of the energy storage battery device.
In practical application, due to different factors such as working environment and use condition, the battery pack has different actual parameters even if the energy storage lithium batteries of the same type are in service for a period of time. The problem of consistency loss of a battery pack is aggravated by high uncertainty caused by the fact that a new energy source with intermittence and volatility is connected into a power grid system on a large scale to charge and discharge behaviors of a high-capacity energy storage lithium battery. The service life prediction model established aiming at single battery pack data is often lack of wide representativeness in a new energy high-permeability power grid environment, and meanwhile, when the single prediction model is used for predicting the service life of a battery pack with missing consistency, the single prediction model is often difficult to perform excellently on each battery pack, so that how to establish the prediction model which has excellent prediction accuracy and good generalization performance is one of the problems to be solved urgently in the field.
Disclosure of Invention
The invention aims to provide a lithium battery life prediction method based on a data decomposition and integrated learning model, which is established to be widely used on the premise of lower cost and solves the problems of unstable prediction performance or lack of generalization performance of a multi-battery pack in a new energy high-permeability smart grid environment in the existing energy storage lithium battery residual life prediction technology.
The invention provides a lithium battery life prediction method based on a data decomposition and integrated learning model, which is used for predicting the residual life (RUL) of a lithium battery;
the method is based on data driving, and a battery capacity physical degradation model is established without analyzing complex physical and chemical principles inside the battery pack, so that the establishment cost and complexity of a prediction model are greatly reduced. The relevant theories and contents include the following:
the lithium battery data acquisition part: the method comprises the steps of collecting energy storage lithium battery pack data under different working conditions and environments, carrying out characteristic analysis on the parameter data, classifying and sorting energy storage batteries with similar parameters, and dividing the energy storage lithium battery pack data into different characteristic groups.
And respectively collecting historical capacity degradation data aiming at the energy storage batteries with different feature groups.
Data preprocessing (decomposition): and decomposing the acquired historical capacity data of the energy storage lithium batteries of each characteristic group based on a fully adaptive noise set empirical mode decomposition algorithm (CEEMDAN), and decomposing the original data into a plurality of intrinsic mode components (IMF).
The components of different frequency bands are trained respectively, so that the potential rule information in the data change process can be easily discovered by the neural network in the training process;
inputting the component data into a long-short term memory neural network (LSTM) network respectively, and training a sub-prediction model;
combining the output of the component in the sub-prediction model with the original data set to generate the historical capacity degradation data of the lithium battery after data preprocessing;
training a prediction model: respectively inputting the processed lithium battery data as the long-term and short-term memory neural networks, and training a lithium battery residual life prediction model;
the accuracy and the diversity are used as optimization targets of the prediction model, and model parameters are optimized by utilizing a multi-objective evolutionary algorithm (MOEA/D);
optimizing model parameters by using a multi-objective evolutionary algorithm: the first objective function is shown in equation (1), which maximizes the accuracy of the prediction model by minimizing the root mean square error. Wherein R is i Is the true value of the remaining capacity of the battery of the ith training sample,
representing the estimated value of the mth predictive model on the ith training set.
The objective function targeting diversity is shown in equation (2), which maximizes model diversity by minimizing the correlation function, where
And
respectively represents the predicted values of the mth LSTM and the jth LSTM on the ith training set,
the average of the output over the ith training set is for all LSTM.
Entering algorithm circulation optimization, firstly randomly generating individuals with the number of P;
generating candidate points by using DE and Gaussian variation, and updating neighborhood points;
after the circulation process is finished, high-quality parameter individuals are screened according to a targeted step-by-step strategy, the method is characterized in that high-quality individuals are screened by taking accuracy as a target, then individuals with large differences are screened by taking generalization as a target, and the method comprises the following specific steps:
dividing a population into a plurality of non-dominant layers according to a non-dominant criterion;
step two, respectively calculating the crowdedness of each individual in each non-dominant layer;
thirdly, sorting the individuals of each non-dominant layer according to the congestion degree;
selecting individuals with advanced non-domination levels;
step five, preferentially selecting individuals with high crowdedness when selecting individuals on the same non-dominant layer;
and step six, continuously selecting according to the step four and the step five until the number of the selected individuals meets the requirement.
Selecting parameters for a single predictive model according to a preference selection for accuracy and generalization performance;
model integration: the following formulas (3) and (4) are used for distributing connection weights to the prediction models trained on different battery pack data sets to form an integrated model, and the integrated parameters can be adjusted according to actual requirements, so that the integrated model is more biased to the accuracy or diversity in the lithium battery life prediction problem.
The prediction average performance and the contribution degree of a high-quality individual to the prediction result are controlled by adjusting the sizes of mu and rho.
Drawings
FIG. 1 is a block diagram of the overall process of the method of the present invention
FIG. 2 is a flowchart of a method for establishing a model for predicting the life of an energy storage lithium battery in the invention.
FIG. 3 is a flow chart of a lithium battery data preprocessing method.
Detailed Description
The present invention will be described in further detail by way of examples with reference to the accompanying drawings.
It should be noted that the following examples are illustrative only and are not to be construed as limiting the present invention.
The embodiment of the invention comprises the following steps: the invention provides a lithium battery life prediction method based on a data decomposition and integrated learning model, which is used for predicting the residual life (RUL) of an energy storage lithium battery, and is shown as an overall flow chart of the method in figure 1, which generally shows the flow of the method in the invention;
fig. 2 is a flow chart of a method for establishing a life prediction model of an energy storage lithium battery in the invention, which shows a method and a process for establishing a life prediction model of an energy storage lithium battery in the invention in more detail based on fig. 1; the method is based on data driving, and a battery capacity physical degradation model is established without analyzing complex physical and chemical principles inside the battery pack, so that the establishment cost and complexity of a prediction model are greatly reduced. The relevant theories and contents include the following:
the lithium battery data acquisition part: as shown in the flow chart of the method for establishing the prediction model of the energy storage lithium battery in fig. 2, firstly, energy storage lithium battery pack data under different working conditions and environments, such as terminal voltage, impedance, discharge power and other parameters, are collected from an energy storage battery bank under the environment of a new energy high-permeability smart grid, characteristic analysis is performed on the parameter data, energy storage batteries with similar parameters are classified and sorted, and the energy storage lithium battery pack data are divided into different characteristic groups.
And respectively collecting historical capacity degradation data aiming at the energy storage batteries with different feature groups.
More specifically, 4 sets of lithium battery data (# 5, #6, #7, and # 18) different from each other in actual capacity and parameters were used as raw data in the present embodiment, which was provided by NASA.
In this example, the raw data was normalized by 0 to 1 using the following expression (5):
as shown in fig. 3, the data preprocessing section: decomposing the acquired data based on a fully adaptive noise set empirical mode decomposition algorithm (CEEMDAN), and decomposing the original data into a plurality of intrinsic mode components (IMF);
inputting the component data into a long-short term memory neural network (LSTM) network respectively, and training a sub-prediction model;
combining the output of the component in the sub-prediction model with the original data set to generate the historical capacity degradation data of the lithium battery after data preprocessing;
training a prediction model: respectively taking the processed lithium battery data as the input of a long-term and short-term memory neural network, and training a residual life prediction model of the lithium battery;
the accuracy and the diversity are used as optimization targets of the prediction model, and model parameters are optimized by utilizing a multi-objective evolutionary algorithm (MOEA/D);
optimizing model parameters by a multi-objective evolutionary algorithm: the first objective function is shown in equation (1), which maximizes the accuracy of the prediction model by minimizing the root mean square error.
The objective function targeting diversity is shown in equation (2), which maximizes model diversity by minimizing the correlation function
More specifically, diversity represents the negative correlation path of each individual relative to other individuals, with increasing degrees of negative correlation providing increased diversity. The improvement of the accuracy can lead the overall output of the generation population to be close to the optimal individual, so that the difference between the overall output and the optimal individual is reduced, namely the diversity is reduced; the improvement of diversity can increase the overall difference of the prediction results of the population, enlarge the parameter difference, and lead the overall result to deviate from the true value, thus leading the accuracy of the individual to be reduced;
in the algorithm loop process, if the numerical distributions of the two objective functions are inconsistent, the search direction of the point in the target space may be shifted. In order to prevent the search direction of the midpoint in the target space from being influenced by objective factors, values of two target functions are subjected to standardization processing, and a standardization processing formula is shown as a formula (6):
z q (Q =1.. Q) represents the qth objective function of the individual, where Q is the objective function dimension.
Is the largest value among the qth objective functions of all individuals. In the initially generated solution, the solution is,
is a number randomly selected within a range not exceeding the maximum value of the objective function.
In the process of running the MOEA/D algorithm, the population size and the population updating strategy can influence the distribution of the population. Since the data size of the lithium battery adopted in the embodiment is not huge and does not generate a large amount of calculation, according to practical considerations, the initial population size is set to 400 in order to ensure the diversity of population distribution when the MOEA/D algorithm is applied in the embodiment. Meanwhile, generating child points by using a Tchebycheff aggregation function to finally obtain a Pareto front edge.
Optimizing a multi-objective evolutionary algorithm: firstly, randomly generating P individuals;
and generating candidate points by using DE and Gaussian variation, and updating the neighborhood points.
Screening high-quality parameter individuals according to a targeted step-by-step strategy, which specifically comprises the following steps:
calculating the non-dominance of each individual and sequencing;
calculating the crowdedness of each individual and sequencing;
selecting individuals with high prediction precision performance according to non-dominant sorting;
in the same non-dominant layer, selecting individuals with large differences according to the congestion degree;
more specifically, non-dominant ranking is a hierarchical mechanism that divides a population into several non-dominant layers according to dominant relationships among individuals, and the crowdedness represents the density of individuals around the individuals in a target space.
Repeating the above two steps until the number of the selected individuals meets the requirement;
more specifically, the main purpose of introducing the strategy is to preferentially screen individuals through non-dominated sorting and crowding degree calculation and remove learning models with similar performance from the individuals;
the structure parameters of the LSTM depend on the complexity and scale of the training data sample, and the parameter ranges are set according to the practical problem processed by the embodiment and the sample scale, the embodiment takes the number of neurons of the LSTM hidden layer as MOEA/D decision variables, the ranges are respectively set to be [40,80], and the learning rate of the network structure is set to be [0.01,0.05];
when the MOEA/D algorithm is finished, selecting specific parameters for a single prediction model according to preference selection on accuracy and generalization performance;
and (3) distributing connection weights for the prediction models trained on different battery pack data sets by using the formulas (3) and (4) to form an integrated model, and adjusting the integrated parameters according to actual requirements to make the integrated model more biased to the accuracy or diversity in the problem of predicting the service life of the energy storage lithium battery.
Claims (6)
1. The invention aims to provide an energy storage lithium battery service life prediction method based on a data decomposition and integrated learning model, which is characterized in that a battery capacity physical degradation model is established completely based on data driving without analyzing the internal physical and chemical principles of a complex battery pack, so that the establishment cost and complexity of the prediction model are greatly reduced;
the method comprises the following steps that a lithium battery data acquisition link is used for acquiring representative historical capacity degradation data of the lithium battery pack under different use conditions and environments;
data preprocessing (decomposition): decomposing the acquired data based on a fully adaptive noise set empirical mode decomposition algorithm (CEEMDAN), and decomposing the original data into a plurality of intrinsic mode components (IMF);
respectively inputting the component data into an LSTM network, and training a sub-prediction model;
combining the output of the component in the sub-prediction model with the original data set to generate the historical capacity degradation data of the lithium battery after data preprocessing;
training a prediction model: respectively taking the processed lithium battery data as the input of a long-term and short-term memory neural network, and training a residual life prediction model of the lithium battery;
the accuracy and the diversity are used as optimization targets of the prediction model, and model parameters are optimized by utilizing a multi-objective evolutionary algorithm (MOEA/D);
selecting parameters for a single predictive model according to a preference selection for accuracy and generalization performance;
model integration: connecting weights are distributed to the prediction models trained on different battery pack data sets by using the formulas (4) and (5) to form an integrated model, and the integrated parameters can be adjusted according to actual requirements, so that the integrated model is more biased to the accuracy or diversity in the lithium battery life prediction problem;
by combining the methods, the integrated prediction model for predicting the remaining life of the lithium battery is obtained, the current situation of unstable prediction performance of multiple battery packs in the existing lithium battery remaining life prediction technology can be effectively solved, and the generalization capability of the prediction model is improved.
2. The data collection process according to claim 1, wherein although the factory parameters of the batteries of the same type are close, the parameters are different due to different factors such as working environment and use condition, so that the historical capacity degradation data of the lithium battery pack, which is representative among the above factors, needs to be collected according to the actual factors such as the working environment, the service time and the power supply object of the battery pack.
3. The data decomposition method of claim 1, wherein the collected original data is decomposed into components and residual components of multiple frequency bands by using a CEEMDAN method, then the components of different frequency bands are trained respectively, and finally the predicted output values of each component are combined and filled back into the original data set, so that the expression of the latent rules can be enhanced relative to the original data set, and the information of the latent rules in the data change process can be easily discovered by a neural network in the training process.
4. The optimization of model parameters based on multi-objective evolutionary algorithm (MOEA/D) as claimed in claim 1, characterized in that the chosen optimization targets are two contradictory targets, namely accuracy and generalization;
the accuracy refers to the error accuracy of the lithium battery life prediction result of the prediction model relative to the true value, and the generalization refers to the overall prediction performance of the prediction model in the face of different battery packs with different parameters.
5. The optimization algorithm of claim 1, wherein after the optimization process is completed, high-quality parameter individuals are screened according to a targeted stepwise strategy, wherein high-quality individuals are screened with accuracy as a target, and then individuals with large differences are screened with generalization as a target, and the method comprises the following specific steps:
dividing a population into a plurality of non-dominant layers according to a non-dominant criterion;
step two, respectively calculating the crowdedness of each individual in each non-dominant layer;
thirdly, sorting the individuals of each non-dominant layer according to the congestion degree;
firstly, selecting individuals with advanced non-domination levels;
when selecting individuals in the same non-dominant layer, preferentially selecting individuals with high crowding degree;
and step six, continuously selecting according to the step four and the step five until the number of the selected individuals meets the requirement.
6. The integrated model of claim 1, wherein after the individual prediction models are integrated, the average performance of prediction and the contribution degree μ of the high-quality individual to the prediction result can be controlled by adjusting the values of μ and ρ, and when ρ is a large negative number, a large weight ratio is given to the high-quality individual, so that the contribution degree of the high-quality individual to the prediction result is increased; conversely, when μ takes a large positive number and ρ a small negative number, the integrated model prediction results may be biased toward average performance, with each individual contributing closely to the set. During design, the integration parameters can be adjusted according to actual requirements, so that the integration model is more biased to the accuracy or diversity in the lithium battery service life prediction problem.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210711671.3A CN115204038A (en) | 2022-06-22 | 2022-06-22 | Energy storage lithium battery life prediction method based on data decomposition and integration model |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210711671.3A CN115204038A (en) | 2022-06-22 | 2022-06-22 | Energy storage lithium battery life prediction method based on data decomposition and integration model |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN115204038A true CN115204038A (en) | 2022-10-18 |
Family
ID=83576887
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210711671.3A Withdrawn CN115204038A (en) | 2022-06-22 | 2022-06-22 | Energy storage lithium battery life prediction method based on data decomposition and integration model |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115204038A (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115983501A (en) * | 2023-03-17 | 2023-04-18 | 河北冠益荣信科技有限公司 | Portable energy storage equipment monitoring system and method based on big data |
| CN116502544A (en) * | 2023-06-26 | 2023-07-28 | 武汉新威奇科技有限公司 | Electric screw press life prediction method and system based on data fusion |
| CN117723999A (en) * | 2024-02-07 | 2024-03-19 | 深圳市东田通利电业制品有限公司 | Battery service life prediction method, device, equipment and storage medium |
-
2022
- 2022-06-22 CN CN202210711671.3A patent/CN115204038A/en not_active Withdrawn
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115983501A (en) * | 2023-03-17 | 2023-04-18 | 河北冠益荣信科技有限公司 | Portable energy storage equipment monitoring system and method based on big data |
| CN116502544A (en) * | 2023-06-26 | 2023-07-28 | 武汉新威奇科技有限公司 | Electric screw press life prediction method and system based on data fusion |
| CN116502544B (en) * | 2023-06-26 | 2023-09-12 | 武汉新威奇科技有限公司 | Electric screw press life prediction method and system based on data fusion |
| CN117723999A (en) * | 2024-02-07 | 2024-03-19 | 深圳市东田通利电业制品有限公司 | Battery service life prediction method, device, equipment and storage medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Li et al. | A merged fuzzy neural network and its applications in battery state-of-charge estimation | |
| CN115204038A (en) | Energy storage lithium battery life prediction method based on data decomposition and integration model | |
| CN110417015A (en) | Micro-capacitance sensor Multiobjective Optimal Operation method and system based on Model Predictive Control | |
| CN112791997B (en) | Method for cascade utilization and screening of retired battery | |
| CN110941929A (en) | Battery health state assessment method based on ARMA and Elman neural network combined modeling | |
| EP4055399A1 (en) | Battery performance prediction | |
| CN111525587B (en) | Reactive load situation-based power grid reactive voltage control method and system | |
| Hu et al. | Short-term photovoltaic power prediction based on similar days and improved SOA-DBN model | |
| Wu et al. | Optimized multi-source fusion based state of health estimation for lithium-ion battery in fast charge applications | |
| CN116826710A (en) | Peak clipping strategy recommendation method and device based on load prediction and storage medium | |
| CN117498400B (en) | Distributed photovoltaic and energy storage data processing method and system | |
| CN114880939A (en) | Intelligent prediction method and device for service life of power battery | |
| Zhang et al. | Interpretable battery cycle life range prediction using early cell degradation data | |
| Eleftheriadis et al. | Hyperband optimization of stacked bidirectional long short-term memory neural network for the state of charge estimation | |
| CN114578234A (en) | Lithium ion battery degradation and capacity prediction model considering causality characteristics | |
| Reza et al. | Recent advancement of remaining useful life prediction of lithium-ion battery in electric vehicle applications: A review of modelling mechanisms, network configurations, factors, and outstanding issues | |
| CN112418504A (en) | Wind speed prediction method based on mixed variable selection optimization deep belief network | |
| Wu et al. | Least-squares support vector machine-based learning and decision making in cognitive radios | |
| Udeogu et al. | Remaining useful life prediction for supercapacitors using an optimized end-to-end deep learning approach | |
| CN112100902B (en) | Lithium ion battery life prediction method based on flow data | |
| Cao et al. | Design of river water quality assessment and prediction algorithm | |
| Liu et al. | Discriminative Signal Recognition for Transient Stability Assessment via Discrete Mutual Information Approximation and Eigen Decomposition of Laplacian Matrix | |
| Basia et al. | Comparison of data driven algorithms for SoH estimation of Lithium-ion batteries | |
| CN117454186B (en) | Model training method, battery performance prediction method, device, equipment and storage medium | |
| Kumarappa et al. | Machine learning-based prediction of lithium-ion battery life cycle for capacity degradation modelling |
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 | ||
| WW01 | Invention patent application withdrawn after publication |
Application publication date: 20221018 |
|
| WW01 | Invention patent application withdrawn after publication |