CN116184210A - Battery abnormality detection method, device, system and electronic device - Google Patents
Battery abnormality detection method, device, system and electronic device Download PDFInfo
- Publication number
- CN116184210A CN116184210A CN202211557848.5A CN202211557848A CN116184210A CN 116184210 A CN116184210 A CN 116184210A CN 202211557848 A CN202211557848 A CN 202211557848A CN 116184210 A CN116184210 A CN 116184210A
- Authority
- CN
- China
- Prior art keywords
- data
- detected
- static
- time sequence
- training
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Images
Classifications
-
- 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
- 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]
- G01R31/396—Acquisition or processing of data for testing or for monitoring individual cells or groups of cells within a battery
-
- 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
Abstract
The application relates to a battery abnormality detection method, a device, a system and an electronic device, wherein the method comprises the following steps: acquiring time sequence data to be detected of the battery in a working state; performing feature decomposition on the time sequence data to be detected to obtain dynamic data to be detected and static data to be detected; the dynamic data to be detected is the data with dynamic change in the time dimension in the detected time sequence data, and the static data to be detected is the data with static state in the time dimension in the detected time sequence data; inputting dynamic data to be tested into a trained target time sequence self-encoder model, and outputting first error information; inputting the static data to be tested into the trained target static self-encoder model, and outputting second error information; and generating an abnormality detection result for the battery according to the first error information and the second error information. According to the method and the device, the problem of low accuracy of battery abnormality detection is solved, and an accurate and efficient battery abnormality detection method is achieved.
Description
Technical Field
The present disclosure relates to the field of battery detection technologies, and in particular, to a method, a device, a system, and an electronic device for detecting battery abnormalities.
Background
With the rapid development of the new energy automobile industry, the market conservation quantity of electric automobiles is also rapidly increasing. The performance and the state of the power battery serving as a core component of the electric automobile greatly influence the condition of the whole electric automobile. Therefore, related technologies of monitoring the real-time operation state of the power battery through a battery management system (Battery management system, abbreviated as BMS) and timely alarming the occurrence of faults and early warning the possible occurrence of faults have become increasingly important.
In the related art, the power battery abnormality detection of the new energy automobile mainly comprises abnormality detection based on an equivalent circuit model, abnormality detection based on a statistical method and abnormality detection based on a data-driven intelligent algorithm model. However, the foregoing abnormality detection methods generally have a problem of poor real-time performance due to a poor convergence of the equivalent circuit model, or require a constant update of a threshold value for determining whether the battery is abnormal, or use a single evaluation dimension when detecting by using an intelligent algorithm model, which results in lower accuracy of detecting the battery abnormality.
At present, no effective solution is proposed for the problem of low accuracy of battery detection in the related art.
Disclosure of Invention
The embodiment of the application provides a battery abnormality detection method, device, system and electronic device, which are used for at least solving the problem of low accuracy of battery abnormality detection in the related art.
In a first aspect, an embodiment of the present application provides a method for detecting battery abnormality, including:
acquiring time sequence data to be detected of the battery in a working state;
performing feature decomposition on the time sequence data to be detected to obtain dynamic data to be detected and static data to be detected; the dynamic data to be detected refers to the data which has dynamic change in the time dimension in the detected time sequence data to be detected, and the static data to be detected refers to the data which has static state in the time dimension in the detected time sequence data to be detected;
inputting the dynamic data to be tested into a trained target time sequence self-encoder model, and outputting first error information; inputting the static data to be tested into a trained target static self-encoder model, and outputting second error information;
and generating an abnormality detection result for the battery according to the first error information and the second error information.
In some embodiments, the performing feature decomposition on the time-series data to obtain dynamic data to be tested and static data to be tested includes:
Calculating to obtain the change trend of the time sequence data to be detected in the time dimension, and carrying out feature decomposition on all the time sequence data to be detected according to the change trend to obtain dynamic features to be detected and static features to be detected;
generating the dynamic data to be detected carrying the time dimension according to the dynamic feature to be detected, and calculating a mean value of time sequence data corresponding to the static feature to be detected in the time dimension to obtain the static data to be detected without the time dimension.
In some embodiments, before inputting the dynamic data to be tested to the trained target timing self-encoder model and outputting the first error information, the method further includes:
acquiring training dynamic data;
inputting the training dynamic data to a time sequence encoder in an initial time sequence self-encoder model to perform time sequence encoding processing, outputting first hidden variable data, inputting the first hidden variable data to a time sequence decoder in the initial time sequence self-encoder model to perform time sequence decoding processing, and outputting dynamic training reconstruction data;
and calculating to obtain a first association result between the training dynamic data and the dynamic training reconstruction data, back-propagating the first association result to the initial time sequence self-encoder model for iterative training, and generating the target time sequence self-encoder model.
In some of these embodiments, the acquiring training dynamic data comprises:
acquiring historical battery data;
searching the historical battery data, reserving cleaning data which are failed to match with a preset threshold value in the historical battery data according to a search result, and obtaining training time sequence data according to the cleaning data;
and performing feature decomposition on the training time sequence data to obtain the training dynamic data.
In some embodiments, before inputting the static data to be tested to the trained target static self-encoder model and outputting the second error information, the method further includes:
acquiring training static data;
inputting the training static data to a static encoder in an initial static self-encoder model for encoding processing, outputting second hidden variable data, inputting the second hidden variable data to a static decoder in the initial static self-encoder model for decoding processing, and outputting static training reconstruction data;
and calculating to obtain a second association result between the training static data and the static training reconstruction data, reversely transmitting the second association result to the initial static self-encoder model for iterative training, and generating the target static self-encoder model.
In some embodiments, the generating the abnormality detection result for the battery according to the first error information and the second error information includes:
acquiring training fusion error information; the training fusion error information is obtained by training time sequence data through the target time sequence self-encoder model and the target static self-encoder model;
and calculating to obtain fusion error information according to the first error information and the second error information, and calculating to obtain the abnormality detection result according to the fusion error information and the training fusion error information.
In some embodiments, the acquiring the time sequence data to be measured of the battery in the working state includes:
acquiring real-time characteristic data of at least two characteristic types of the battery in the working state;
and according to the characteristic type, carrying out time sequence division processing on the real-time characteristic data to obtain the time sequence data to be detected.
In a second aspect, an embodiment of the present application provides a battery abnormality detection apparatus, including: the device comprises an acquisition module, a decomposition module, an output module and a generation module;
the acquisition module is used for acquiring time sequence data to be detected of the battery in a working state;
The decomposition module is used for carrying out characteristic decomposition on the time sequence data to be detected to obtain dynamic data to be detected and static data to be detected; the dynamic data to be detected refers to the data which has dynamic change in the time dimension in the detected time sequence data to be detected, and the static data to be detected refers to the data which has static state in the time dimension in the detected time sequence data to be detected;
the output module is used for inputting the dynamic data to be tested into the trained target time sequence self-encoder model and outputting first error information; inputting the static data to be tested into a trained target static self-encoder model, and outputting second error information;
the generation module is used for generating an abnormality detection result for the battery according to the first error information and the second error information.
In a third aspect, an embodiment of the present application provides a battery abnormality detection system, including: the main control device and the battery body; wherein the main control device is connected with the battery body;
the main control device is used for executing the battery abnormality detection method according to the first aspect.
In a fourth aspect, an embodiment of the present application provides an electronic device, including a memory, a processor, and a computer program stored on the memory and capable of running on the processor, where the processor implements the method for detecting a battery abnormality according to the first aspect when executing the computer program.
Compared with the related art, the battery abnormality detection method, device, system and electronic device provided by the embodiment of the application are used for acquiring the time sequence data to be detected of the battery in the working state; performing feature decomposition on the time sequence data to be detected to obtain dynamic data to be detected and static data to be detected; the to-be-detected dynamic data refers to data with dynamic change in the time dimension in the detected to-be-detected time sequence data, and the to-be-detected static data refers to data with static state in the time dimension in the detected to-be-detected time sequence data; inputting dynamic data to be tested into a trained target time sequence self-encoder model, and outputting first error information; inputting the static data to be tested into the trained target static self-encoder model, and outputting second error information; according to the first error information and the second error information, an abnormality detection result aiming at the battery is generated, the problem of low accuracy of battery abnormality detection is solved, and an accurate and efficient battery abnormality detection method is realized.
Drawings
The accompanying drawings, which are included to provide a further understanding of the application and are incorporated in and constitute a part of this application, illustrate embodiments of the application and together with the description serve to explain the application and do not constitute an undue limitation to the application. In the drawings:
Fig. 1 is an application environment diagram of a battery abnormality detection method according to an embodiment of the present application;
fig. 2 is a flowchart of a battery abnormality detection method according to an embodiment of the present application;
FIG. 3 is a flowchart of another battery anomaly detection method according to an embodiment of the present application;
FIG. 4 is a schematic diagram of a target timing self-encoder model training in accordance with embodiments of the present application;
FIG. 5 is a schematic diagram of a target static self-encoder model training in accordance with embodiments of the present application;
fig. 6 is a block diagram of a battery abnormality detection apparatus according to an embodiment of the present application;
fig. 7 is a block diagram of a battery abnormality detection system according to an embodiment of the present application;
fig. 8 is a block diagram of the interior of a computer device according to an embodiment of the present application.
Detailed Description
In order to make the objects, technical solutions and advantages of the present application more apparent, the present application is described and illustrated below with reference to the accompanying drawings and examples. It should be understood that the specific embodiments described herein are for purposes of illustration only and are not intended to limit the present application. All other embodiments, which can be made by one of ordinary skill in the art without undue burden on the person of ordinary skill in the art based on the embodiments provided herein, are intended to be within the scope of the present application. Moreover, it should be appreciated that while such a development effort might be complex and lengthy, it would nevertheless be a routine undertaking of design, fabrication, or manufacture for those of ordinary skill having the benefit of this disclosure, and thus should not be construed as having the benefit of this disclosure.
Reference in the specification to "an embodiment" means that a particular feature, structure, or characteristic described in connection with the embodiment may be included in at least one embodiment of the application. The appearances of such phrases in various places in the specification are not necessarily all referring to the same embodiment, nor are separate or alternative embodiments mutually exclusive of other embodiments. It is to be expressly and implicitly understood by those of ordinary skill in the art that the embodiments described herein can be combined with other embodiments without conflict.
Unless defined otherwise, technical or scientific terms used herein should be given the ordinary meaning as understood by one of ordinary skill in the art to which this application belongs. Reference to "a," "an," "the," and similar terms herein do not denote a limitation of quantity, but rather denote the singular or plural. The terms "comprising," "including," "having," and any variations thereof, are intended to cover a non-exclusive inclusion; for example, a process, method, system, article, or apparatus that comprises a list of steps or modules (elements) is not limited to only those steps or elements but may include other steps or elements not expressly listed or inherent to such process, method, article, or apparatus. The terms "connected," "coupled," and the like in this application are not limited to physical or mechanical connections, but may include electrical connections, whether direct or indirect. The term "plurality" as used herein means greater than or equal to two. "and/or" describes an association relationship of an association object, meaning that there may be three relationships, e.g., "a and/or B" may mean: a exists alone, A and B exist together, and B exists alone. The terms "first," "second," "third," and the like, as used herein, are merely distinguishing between similar objects and not representing a particular ordering of objects.
The battery abnormality detection method provided by the application can be applied to application scenes such as vehicle driving, wheeled robot control and the like. Fig. 1 is an application environment diagram of a vehicle lateral control method according to an embodiment of the present application, and as shown in fig. 1, a vehicle 102 communicates with a server 104 through a network. The data storage system may store data that the server 104 needs to process. The data storage system may be integrated on the server 104 or may be located on a cloud or other network server. The server 104 obtains time sequence data to be detected of the power battery in the vehicle 102 in a working state, performs feature decomposition on the time sequence data to be detected to obtain dynamic data to be detected and static data to be detected, inputs the dynamic data to be detected into a trained target time sequence self-encoder model, outputs first error information, inputs the static data to be detected into the trained target static self-encoder model, and outputs second error information; the server 104 generates an abnormality detection result for the battery based on the first error information and the second error information. The server 104 may be implemented as a stand-alone server or a server cluster including a plurality of servers.
The present embodiment provides a method for detecting battery abnormality, and fig. 2 is a flowchart of a method for detecting battery abnormality according to an embodiment of the present application, as shown in fig. 2, the flowchart including the steps of:
step S210, obtaining time sequence data to be tested of the battery in a working state.
Further, the step of obtaining the time sequence data to be measured of the battery in the working state further comprises the following steps: acquiring real-time characteristic data of at least two characteristic types of the battery in the working state; and according to the characteristic type, carrying out time sequence division processing on the real-time characteristic data to obtain the time sequence data to be detected. Specifically, the time series data to be measured may be various feature data collected at a series Of time points including various feature types Of the battery in a working State such as charging, including data such as a total voltage Of the power battery pack, a total current Of the power battery pack, a voltage Of each single battery, a temperature Of each battery probe, a State Of Charge (SOC) value, a State Of Health (SOH) value, and a total mileage Of the vehicle. And then, taking time as a new dimension, and performing dimension conversion on the characteristic data to obtain the time sequence data to be detected.
Step S220, carrying out feature decomposition on the time sequence data to be detected to obtain dynamic data to be detected and static data to be detected; the dynamic data to be detected refers to the data which has dynamic change in the time dimension in the detected time sequence data, and the static data to be detected refers to the data which keeps static state in the time dimension in the detected time sequence data.
The feature decomposition is carried out on the time-varying speed of each feature dimension of the time-series data to be detected, and the time-varying speed can be decomposed into dynamic features which relatively rapidly vary with time and static features which relatively slowly vary with time or are fixed; it will be appreciated that feature decomposition for each feature dimension may be performed using a threshold, i.e., a feature dimension may be considered a dynamic feature when the rate of change in time is greater than the threshold; alternatively, an algorithm such as slow feature analysis (Slow Feature Analysis, abbreviated as SFA) may be used for decomposition, which is not described herein. Dividing the time sequence data to be detected according to the dynamic characteristics and the static characteristics, and dividing the time sequence data to be detected into dynamic data to be detected and static data to be detected; further, a mean value is calculated on the time dimension for the time sequence data corresponding to the static feature, so that the static data without the time sequence dimension can be obtained.
Step S230, inputting the dynamic data to be tested into a trained target time sequence self-encoder model, and outputting first error information; and inputting the static data to be tested into the trained target static self-encoder model, and outputting second error information.
In the steps S230 to S240, the dynamic data to be detected is input to the target sequential self-encoder model to obtain the dynamic reconstruction data after decoding reconstruction, the target sequential self-encoder model is used to calculate and obtain the first error information such as the difference value or the ratio between the dynamic data to be detected and the dynamic reconstruction data, the static data to be detected is input to the target static self-encoder model to obtain the static reconstruction data after decoding reconstruction, and the target static self-encoder model is used to calculate and obtain the second error information such as the difference value or the ratio between the static data to be detected and the static reconstruction data.
Step S240, generating an abnormality detection result for the battery according to the first error information and the second error information.
And performing fusion calculation processing according to the first error information and the second error information to generate an abnormality detection result of the battery. Specifically, after the fusion error information is obtained according to the first error information and the second error information, the fusion error information can be compared with a preset error threshold value to detect whether the battery has an abnormality; for example, if the fusion error is less than or equal to the error threshold, the battery is indicated to be in a normal working state, and the operation can be continued; if the fusion error is larger than the error threshold, the battery is in an abnormal state or a potential battery abnormal condition exists, and corresponding alarm prompt information needs to be sent to reminding equipment in real time at the moment, so that workers are reminded of timely performing maintenance or maintenance operation.
Through the steps S210 to S240, feature decomposition is performed on the time sequence data to be detected of the battery in the working state to obtain dynamic data to be detected and static data to be detected, the dynamic data to be detected and the static data to be detected are respectively input into two special models to perform decoding reconstruction to obtain corresponding error information, and finally an abnormal detection result is generated according to the fused error information, so that comprehensive abnormal detection of the battery state is realized.
In some embodiments, the performing feature decomposition on the time-series data to obtain the dynamic data to be tested and the static data to be tested further includes the following steps:
step S221, calculating to obtain the change trend of the time-series data to be tested in the time dimension, and carrying out feature decomposition on all the time-series data to be tested according to the change trend to obtain dynamic features to be tested and static features to be tested.
Step S222, generating the dynamic data to be tested carrying the time dimension according to the dynamic feature to be tested, and calculating the average value of the time sequence data corresponding to the static feature to be tested in the time dimension to obtain the static data to be tested without the time dimension.
In the steps S221 to S222, the feature decomposition is performed on the data to be measured, and the feature decomposition is performed on the data to be measured into a dynamic feature that changes rapidly and a static feature that changes slowly. In particular, features in a pair are decomposed using a feature decomposition method, which decomposes each feature dimension of time-series data according to its rate of change over time, wherein a relatively rapid-change over time portion is defined as a dynamic feature, and a relatively slow-change over time portion is defined as a dynamic featureThe stationary part is defined as a static feature; dividing the data into dynamic data and static data according to the dynamic characteristics and the static characteristics, and then calculating the average value of time sequence data corresponding to the static characteristics in the time dimension to obtain the static data without the time sequence dimension. Further, the feature decomposition method may employ an SFA algorithm. For example, for time series data to be measured X (t) = [ X ] having m feature dimensions 1 (t)，x 2 (t)，x 3 (t)，...，x m (t)] T Constructing a fully connected neural network to output as Y (t) = [ Y ] 1 (t)，y 2 (t)，y 3 (t)，...，y j (t)，...，y J (t)] T Wherein y is j (t)＝g j (X(t))，g j (. Cndot.) is a time series mapping function, the optimization objective of the network is for each y j (t) ∈Y (t) such that it is constrained as follows,
the minimum value can be taken as shown in the following equation 1:
in the above formula, E (& gt) t Representing a mathematical expectation in the time dimension,
a deviation representing data for each adjacent point in time; the output Y (t) of the network is the static characteristic data X st (t) then calculating the characteristic data of X (t) which remain after decomposition, namely dynamic characteristic data X d (t)。
Through the steps S221 to S222, the time sequence data to be detected is subjected to feature decomposition according to the variation trend of the time dimension to obtain the dynamic feature to be detected and the static feature to be detected, and the dynamic data to be detected and the static data to be detected in the time sequence data to be detected are determined according to the features, so that the efficiency and the accuracy of dividing the dynamic data and the static data in the time sequence data to be detected are effectively improved, and further the efficiency and the accuracy of detecting the battery abnormality are improved.
In some of these embodiments, a battery abnormality detection method is also provided. Fig. 3 is a flowchart of another battery abnormality detection method according to an embodiment of the present application, as shown in fig. 3, the flowchart including steps S210 to S240 shown in fig. 2, and further including the steps of:
Step S310, training dynamic data is acquired.
The training dynamic data refers to dynamic characteristics which are obtained by taking each historical characteristic data of a large number of pre-collected battery packs of new energy automobiles during charging as training data and decomposing the training data in characteristics and relatively quick in change with time.
Step S320, inputting the training dynamic data to a time sequence encoder in an initial time sequence self-encoder model for time sequence encoding processing, outputting first hidden variable data, inputting the first hidden variable data to a time sequence decoder in the initial time sequence self-encoder model for time sequence decoding processing, and outputting dynamic training reconstruction data.
Wherein, the initial time sequence self-encoder model can be composed of a time sequence neural network and a self-encoder; the time sequence neural network can be a neural network model capable of realizing time sequence prediction, such as a cyclic neural network (Recurrent Neural Networks, RNN for short-term memory (LSTM) model or a gate control cyclic unit (Gated Recurrent Unit, GRU) model. For example, a long-short-term memory cyclic neural network self-encoder (abbreviated as LSTM-AE) model M can be constructed for dynamic time series data d The method comprises the steps of carrying out a first treatment on the surface of the The model is mainly composed of an encoder En d And a decoder De d The structure of the encoder and the decoder is a long-term and short-term memory cyclic neural network, and then dynamic time sequence data is input into the model M d Training is performed so as to obtain the target time sequence self-encoder model through training.
Specifically, FIG. 4 is a schematic diagram of a target timing self-encoder model training according to an embodiment of the present application, as shown in FIG. 4, the model M d Is specifically designed as follows: inputting training dynamic data into a time sequence encoder, wherein the time sequence encoder is formed by stacking three LSTM networks, each LSTM network consists of a forgetting gate, an input gate and an output gate, and the output time sequence of the former LSTM network is used as the input time sequence of the latter LSTM network; wherein the input feature dimension and the output feature dimension of the first LSTM network are [ m ] d ，128]，m d For the number of dynamic features after feature decomposition, the input feature dimension and output feature dimension of the second LSTM network are [128, 64]The third LSTM network has input feature dimensions and output feature dimensions of [64, 32]The method comprises the steps of carrying out a first treatment on the surface of the The third layer network output is characterized by the first hidden variable data described above. Inputting the first hidden variable data into a time sequence encoder which is also formed by stacking three layers of LSTM networks, wherein the input characteristic dimension and the output characteristic dimension of the first LSTM network are [32, 64 ] ]The second LSTM network has input feature dimensions and output feature dimensions of [64, 128]The third LSTM network has an input feature dimension and an output feature dimension of 128, m d ]And the third layer network output is characterized by the dynamic training reconstruction data.
Step S330, a first correlation result between the training dynamic data and the dynamic training reconstruction data is obtained through calculation, the first correlation result is reversely propagated to the initial time sequence self-encoder model for iterative training, and the target time sequence self-encoder model is generated.
Wherein the training dynamic data X is obtained by d (t) input to the initial timing from the encoder model to obtain dynamic training reconstruction data X' d After (t), a first correlation result, such as a difference or ratio, between the training dynamic data and the dynamic training reconstruction data may be calculated. Specifically, referring to fig. 4, the first correlation result may be a mean square error MSE between the training dynamic data and the dynamic training reconstruction data, as shown in the following formula 2:
the above-mentioned maleWherein x is di (t)∈X d (t)，x′ di (t)∈X′ d (t). After the mean square error MSE is calculated by the formula 2, the parameters in the initial time sequence self-encoder model can be iteratively trained by using the mean square error through a back propagation algorithm, and the process is continuously repeated until the iterative training times or training duration are met, or the model converges, so that the target time sequence self-encoder model with complete training is finally obtained.
Through the steps S310 to S330, iterative training is performed on the neural network model according to the comparison result between the training dynamic data and the dynamic training reconstruction data obtained by using the model training, so as to obtain an optimized target time sequence self-encoder model with complete training, update and obtain a model with stronger characterization capability for the dynamic data, improve the accuracy of model reconstruction, train an anomaly detection model through a large amount of unlabeled data, and avoid specific fault labels, thereby being beneficial to improving the accuracy of battery anomaly detection.
In some embodiments, the acquiring training dynamic data further includes the following steps:
step S311, historical battery data is acquired.
Before training the model, the historical characteristic data of each characteristic of the battery at each time point of the battery pack of a large number of new energy vehicles can be collected in advance as training data before the battery pack of the new energy vehicles is charged, including data such as total voltage of the power battery pack, total current of the power battery pack, voltage of each single battery, temperature of each battery probe, SOC value of the battery, SOH value of the battery, total mileage of the vehicle and the like of each new energy vehicle, and these characteristics are marked as f= { F 1 ，f 2 ，f 3 ，...，f m -where m is the number of feature dimensions. The collected battery history feature data may then be recorded as X 1 ＝[x 1 ，x 2 ，x 3 ，...，x i ，...，x N ]Wherein x is i ∈R m ，x i A piece of data representing a single time point of a single vehicle, R m Represents a real set of feature dimensions m, N represents the total dataNumber of bars.
Step S312, searching the historical battery data, reserving cleaning data which is failed to match with a preset threshold value in the historical battery data according to a searching result, and obtaining training time sequence data according to the cleaning data.
In step S312, the data needs to be cleaned and normalized to be dimension-converted into time-series data. The preset threshold may be preset by a worker in combination with an actual situation, for example, the preset threshold may be set to a null value or a zero value, etc. Specifically, search X 1 ∈R N×m The blank value and unreasonable zero value in the data are screened, the row containing the blank value and unreasonable zero value is deleted, and the data are cleaned; then standardizing the number of data in each charging process of each vehicle, and compressing or deleting redundant data to obtain X 2 To facilitate subsequent dimensional transformations.
Then, the original [ N, m]The dimension data are grouped according to each charging of each vehicle, so that the data in the same group are the data of the same vehicle charged at the same time, dimension conversion is carried out on the grouped data, namely, time is taken as a new dimension, and X is taken as a new dimension 2 ＝[x 1 ，x 2 ，x 3 ，...，x N ]Conversion to X (t) = [ X 1 (t)，x 2 (t)，x 3 (t)，...，x i (t)，...，x n (t)]，x i (t)＝[x i (t 1 )，x i (t 2 )，x i (t 3 )，...，x i (t j )，...，x i (t L )]Wherein x is i (t j ) For the ith time series x i (t) at the jth time point t j And the data is that L is the number of time points contained in each piece of charging time sequence data, and n=N/L is the number of pieces of charging time sequence data.
Step S313, performing feature decomposition on the training time sequence data to obtain the training dynamic data.
Specifically, the characteristic decomposition method is used for X (t) ∈R n×L×m The feature F in (a) is decomposed, and the method can be adopted and implemented in the embodimentThe same decomposition method is adopted in the example, and the training dynamic data is obtained.
Through the steps S311 to S313, the training time sequence data is obtained by cleaning the historical battery data, and the training dynamic data is obtained by performing feature decomposition on the training time sequence data, so that the preprocessing of the training data is realized, the training efficiency of the subsequent model can be effectively improved, and the abnormal battery detection efficiency is further improved.
In some embodiments, before the inputting the static data to be tested to the trained target static self-encoder model and outputting the second error information, the battery abnormality detection method further includes the following steps:
step S201, acquiring training static data.
The training static data refers to static features which are obtained by taking each historical feature data of a large number of pre-collected battery packs of the new energy automobiles during charging as training data and decomposing the training data in a feature mode and are relatively slow or constant in change along with time. It should be noted that, in the process of obtaining the training static data, the obtained historical battery data may be cleaned through the steps S311 to S313 to obtain training time sequence data, and the training time sequence data may be subjected to feature decomposition to obtain the training static data and the training dynamic data.
Step S202, inputting the training static data to a static encoder in the initial static self-encoder model for encoding processing, outputting second hidden variable data, inputting the second hidden variable data to a static decoder in the initial static self-encoder model for decoding processing, and outputting static training reconstruction data.
Wherein the initial static self-encoder model may be composed of a static encoder and a static decoder; the training static data is input to the initial static self-encoder model and the target static self-encoder model can be trained. Specifically, FIG. 5 is a schematic diagram of a target static self-encoder model training according to an embodiment of the present application, as shown in FIG. 5 Self-encoder model M s Is specifically designed as follows: the training static data is input to a self-encoder which can be formed by stacking three layers of fully-connected networks, wherein the data characteristic dimension and the output characteristic dimension of the first fully-connected layer are [ m ] s ，128]，m s For the number of static features after feature decomposition, the second LSTM network has an input feature dimension and an output feature dimension of [128, 64]The third LSTM network has input feature dimensions and output feature dimensions of [64, 32]The method comprises the steps of carrying out a first treatment on the surface of the The third layer network output is characterized by the first hidden variable data described above. Inputting the first hidden variable data into a time sequence encoder which is also formed by stacking three layers of LSTM networks, wherein the input characteristic dimension and the output characteristic dimension of the first LSTM network are [32, 64 ]]The second LSTM network has input feature dimensions and output feature dimensions of [64, 128]The third LSTM network has an input feature dimension and an output feature dimension of 128, m d ]And the third layer network output is characterized by the dynamic training reconstruction data.
Step S203, a second association result between the training static data and the static training reconstruction data is obtained through calculation, the second association result is reversely propagated to the initial static self-encoder model for iterative training, and the target static self-encoder model is generated.
Wherein, the training static data X is used for s Inputting to the initial time sequence self-encoder model to obtain static training reconstruction data X' d After (t), a second correlation result, such as a difference or ratio, between the training static data and the static training reconstruction data may be calculated. Specifically, referring to fig. 5, the second correlation result may be a mean square error MSE between the training static data and the static training reconstruction data, as shown in the following formula 3:
in the above formula, x si ∈X s ，x′ si ∈X′ s . Then the mean square is calculated by equation 3After the error MSE, the parameters in the initial static self-encoder model can be iteratively trained through a back propagation algorithm by utilizing the mean square error, and the process is continuously repeated until the iterative training times or training duration are met or the model converges, so that the target static self-encoder model with complete training is finally obtained.
Through the steps S201 to S203, iterative training is performed on the neural network model according to the comparison result between the training static data and the static training reconstruction data obtained by using model training, so as to obtain an optimized target static self-encoder model with complete training, and the model with stronger characterization capability for the static data can be updated, so that the accuracy of model reconstruction is improved, and the accuracy of battery anomaly detection is further improved.
In some embodiments, the generating the abnormality detection result for the battery according to the first error information and the second error information further includes the steps of:
step S231, acquiring training fusion error information; the training fusion error information is obtained by training the training time sequence data through the target time sequence self-encoder model and the target static self-encoder model.
Specifically, the training time sequence data is subjected to feature decomposition to obtain training dynamic data and training static data. Then calculating training dynamic data, and inputting error vector calculation result between dynamic reconstruction data obtained by decoding, encoding and reconstructing the target time sequence self-encoder model, wherein the error vector can be obtained by using Er d A representation; calculating training static data, inputting error vector calculation results between static reconstruction data obtained after decoding and coding reconstruction of a target static self-encoder model, wherein the error vector can be obtained by using Er s And (3) representing. And then, calculating a fusion error vector according to the two error vectors, namely training fusion error information, wherein the calculation formula is shown in a formula 4:
Er＝Er d +λEr s equation 4
In the above formula, the training fusion error vector Er contains n time series errors in total, and λ is a set error fusion parameter.
Step S232, according to the first error information and the second error information, the fusion error information is obtained through calculation, and according to the fusion error information and the training fusion error information, the abnormal detection result is obtained through calculation.
After the training fusion error information is calculated, the first error information and the second error information may be calculated through the steps S210 to S230, and the fusion error information may be calculated by adding the first error information and the second error information. The fused error information may then be compared to a predetermined error threshold to detect if the battery is abnormal. It should be noted that, the error threshold may be obtained by training and calculating the model according to the pre-acquired training data; specifically, after the training dynamic data and the training static data are respectively input into the target time sequence self-encoder model and the target static self-encoder model, training reconstruction errors are comprehensively calculated to obtain training fusion error vectors Er of the training data, and the average mu of n time sequence errors in the training fusion error vectors Er is calculated Er And standard deviation epsilon Er For example, the statistics of the mean value plus 3 times of the standard deviation can be calculated to obtain the final abnormality detection threshold value theta required to be set E As shown in the following equation 5:
θ E ＝μ Er +3ε Er equation 5
Through the steps S231 to S232, training fusion error information is acquired so as to adaptively determine an error threshold value based on deep learning and finally determine an abnormal detection result of the battery, so that a worker does not need to manually set the error threshold value, and the instantaneity and the efficiency of battery abnormal detection are effectively improved.
It should be noted that the steps illustrated in the above-described flow or flow diagrams of the figures may be performed in a computer system, such as a set of computer-executable instructions, and that, although a logical order is illustrated in the flow diagrams, in some cases, the steps illustrated or described may be performed in an order other than that illustrated herein.
The embodiment also provides a device for detecting battery abnormality, which is used for implementing the foregoing embodiments and preferred embodiments, and is not described in detail. As used below, the terms "module," "unit," "sub-unit," and the like may be a combination of software and/or hardware that implements a predetermined function. While the means described in the following embodiments are preferably implemented in software, implementation in hardware, or a combination of software and hardware, is also possible and contemplated.
Fig. 6 is a block diagram of a battery abnormality detection apparatus according to an embodiment of the present application, as shown in fig. 6, including: an acquisition module 62, a decomposition module 64, an output module 66, and a generation module 68; the acquiring module 62 is configured to acquire time sequence data to be measured of the battery in a working state; the decomposition module 64 is configured to perform feature decomposition on the time-series data to obtain dynamic data to be tested and static data to be tested; the dynamic data to be detected refers to the data which has dynamic change in the time dimension in the detected time sequence data to be detected, and the static data to be detected refers to the data which has static state in the time dimension in the detected time sequence data to be detected; the output module 66 is configured to input the dynamic data to be tested to the trained target timing self-encoder model, and output first error information; inputting the static data to be tested into a trained target static self-encoder model, and outputting second error information; the generating module 68 is configured to generate an abnormality detection result for the battery according to the first error information and the second error information.
Through the above embodiment, the to-be-detected time sequence data of the battery in the working state is subjected to characteristic decomposition through the decomposition module 64 to obtain to-be-detected dynamic data and to-be-detected static data, the to-be-detected dynamic data and the to-be-detected static data are respectively input into two special models through the output module 66 to be subjected to decoding reconstruction to obtain corresponding error information, and finally the generation module 68 generates an abnormal detection result according to the fused error information, so that comprehensive abnormal detection of the battery state is realized, and the static data is fused into the models, so that the embodiment of the application can adapt to abnormal detection scenes under different service lives and different working conditions, can detect the abnormal occurrence or potential battery, solve the problem of low accuracy of abnormal detection of the battery, and realize the accurate and efficient abnormal battery detection device.
In some embodiments, the decomposition module 64 is further configured to calculate a variation trend of the time-series data to be tested in the time dimension, and perform feature decomposition on all the time-series data to be tested according to the variation trend to obtain a dynamic feature to be tested and a static feature to be tested; the decomposition module 64 generates the dynamic data to be tested carrying the time dimension according to the dynamic feature to be tested, calculates the average value of the time sequence data corresponding to the static feature to be tested in the time dimension, and obtains the static data to be tested without the time dimension.
In some embodiments, the battery abnormality detection device further includes a training module; the training module is used for acquiring training dynamic data; the training module inputs the training dynamic data to a time sequence encoder in an initial time sequence self-encoder model to perform time sequence encoding processing, outputs first hidden variable data, inputs the first hidden variable data to a time sequence decoder in the initial time sequence self-encoder model to perform time sequence decoding processing, and outputs dynamic training reconstruction data; the training module calculates a first association result between the training dynamic data and the dynamic training reconstruction data, and reversely propagates the first association result to the initial time sequence self-encoder model for iterative training, and generates the target time sequence self-encoder model.
In some embodiments, the training module is further configured to obtain historical battery data; the training module searches the historical battery data, reserves cleaning data which is failed to match with a preset threshold value in the historical battery data according to a search result, and obtains training time sequence data according to the cleaning data; the training module performs feature decomposition on the training time sequence data to obtain the training dynamic data.
In some embodiments, the training module is further configured to obtain training static data; the training module inputs the training static data to a static encoder in an initial static self-encoder model for encoding processing, outputs second hidden variable data, inputs the second hidden variable data to a static decoder in the initial static self-encoder model for decoding processing, and outputs static training reconstruction data; the training module calculates a second association result between the training static data and the static training reconstruction data, and reversely propagates the second association result to the initial static self-encoder model for iterative training, and generates the target static self-encoder model.
In some embodiments, the generating module 68 is further configured to obtain training fusion error information; the training fusion error information is obtained by training the training time sequence data by the target time sequence self-encoder model and the target static self-encoder model; the generating module 68 calculates the fusion error information according to the first error information and the second error information, and calculates the anomaly detection result according to the fusion error information and the training fusion error information.
In some embodiments, the acquiring module 62 is further configured to acquire real-time feature data of at least two feature types of the battery in the operating state; the obtaining module 62 performs time sequence division processing on the real-time feature data according to the feature type to obtain the time sequence data to be tested.
The above-described respective modules may be functional modules or program modules, and may be implemented by software or hardware. For modules implemented in hardware, the various modules described above may be located in the same processor; or the above modules may be located in different processors in any combination.
The present embodiment also provides a battery abnormality detection system, and fig. 7 is a block diagram of a battery abnormality detection system according to an embodiment of the present application, as shown in fig. 7, the system includes: a master control device 72 and a battery body 74; wherein the main control device 72 is connected to the battery body 74; the master control device 72 is configured to perform the steps of any of the method embodiments described above. The master control device 72 includes, but is not limited to, various chips, microprocessors or other hardware devices for controlling the battery abnormality detection process, which are integrated on the vehicle driven by the battery, or the master control device 72 may be a hardware device such as a server that is in communication connection with the vehicle by a remote or local area network or the like, which is not described herein.
Through the above embodiment, the main control device 72 performs feature decomposition on the time sequence data to be detected of the battery body 74 in the working state to obtain the dynamic data to be detected and the static data to be detected, and inputs the dynamic data to be detected and the static data to be detected into two special models respectively for decoding reconstruction to obtain corresponding error information, and finally generates an abnormal detection result according to the fused error information, so that comprehensive abnormal detection of the battery state is realized.
In some of these embodiments, a computer device is provided, which may be a server, and fig. 8 is a block diagram of an interior of the computer device according to an embodiment of the present application, as shown in fig. 8. The computer device includes a processor, a memory, a network interface, and a database connected by a system bus. Wherein the processor of the computer device is configured to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, computer programs, and a database. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage media. The database of the computer device is used for storing the abnormality detection result. The network interface of the computer device is used for communicating with an external terminal through a network connection. The computer program, when executed by a processor, implements the battery abnormality detection method described above.
It will be appreciated by those skilled in the art that the structure shown in fig. 8 is merely a block diagram of some of the structures associated with the present application and is not limiting of the computer device to which the present application may be applied, and that a particular computer device may include more or fewer components than shown, or may combine certain components, or have a different arrangement of components.
The present embodiment also provides an electronic device comprising a memory having stored therein a computer program and a processor arranged to run the computer program to perform the steps of any of the method embodiments described above.
Optionally, the electronic apparatus may further include a transmission device and an input/output device, where the transmission device is connected to the processor, and the input/output device is connected to the processor.
Alternatively, in the present embodiment, the above-described processor may be configured to execute the following steps by a computer program:
s1, acquiring time sequence data to be detected of the battery in a working state.
S2, carrying out feature decomposition on the time sequence data to be detected to obtain dynamic data to be detected and static data to be detected; the dynamic data to be detected refers to the data which has dynamic change in the time dimension in the detected time sequence data, and the static data to be detected refers to the data which keeps static state in the time dimension in the detected time sequence data.
S3, inputting the dynamic data to be tested into a trained target time sequence self-encoder model, and outputting first error information; and inputting the static data to be tested into the trained target static self-encoder model, and outputting second error information.
S4, generating an abnormality detection result for the battery according to the first error information and the second error information.
It should be noted that, specific examples in this embodiment may refer to examples described in the foregoing embodiments and alternative implementations, and this embodiment is not repeated herein.
In addition, in combination with the battery abnormality detection method in the above embodiment, the embodiment of the present application may be implemented by providing a storage medium. The storage medium has a computer program stored thereon; the computer program, when executed by a processor, implements any of the battery abnormality detection methods of the above embodiments.
Those skilled in the art will appreciate that implementing all or part of the above described methods may be accomplished by way of a computer program stored on a non-transitory computer readable storage medium, which when executed, may comprise the steps of the embodiments of the methods described above. Any reference to memory, storage, database, or other medium used in the various embodiments provided herein may include non-volatile and/or volatile memory. The nonvolatile memory can include Read Only Memory (ROM), programmable ROM (PROM), electrically Programmable ROM (EPROM), electrically Erasable Programmable ROM (EEPROM), or flash memory. Volatile memory can include Random Access Memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in a variety of forms such as Static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double Data Rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous Link DRAM (SLDRAM), memory bus direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM), among others.
It should be understood by those skilled in the art that the technical features of the above-described embodiments may be combined in any manner, and for brevity, all of the possible combinations of the technical features of the above-described embodiments are not described, however, they should be considered as being within the scope of the description provided herein, as long as there is no contradiction between the combinations of the technical features.
The above examples merely represent a few embodiments of the present application, which are described in more detail and are not to be construed as limiting the scope of the invention. It should be noted that it would be apparent to those skilled in the art that various modifications and improvements could be made without departing from the spirit of the present application, which would be within the scope of the present application. Accordingly, the scope of protection of the present application is to be determined by the claims appended hereto.
Claims (10)
1. A battery abnormality detection method, characterized by comprising:
acquiring time sequence data to be detected of the battery in a working state;
performing feature decomposition on the time sequence data to be detected to obtain dynamic data to be detected and static data to be detected; the dynamic data to be detected refers to the data which has dynamic change in the time dimension in the detected time sequence data to be detected, and the static data to be detected refers to the data which has static state in the time dimension in the detected time sequence data to be detected;
Inputting the dynamic data to be tested into a trained target time sequence self-encoder model, and outputting first error information; inputting the static data to be tested into a trained target static self-encoder model, and outputting second error information;
and generating an abnormality detection result for the battery according to the first error information and the second error information.
2. The method for detecting battery abnormality according to claim 1, wherein the performing feature decomposition on the time-series data to be detected to obtain dynamic data to be detected and static data to be detected includes:
calculating to obtain the change trend of the time sequence data to be detected in the time dimension, and carrying out feature decomposition on all the time sequence data to be detected according to the change trend to obtain dynamic features to be detected and static features to be detected;
generating the dynamic data to be detected carrying the time dimension according to the dynamic feature to be detected, and calculating a mean value of time sequence data corresponding to the static feature to be detected in the time dimension to obtain the static data to be detected without the time dimension.
3. The battery abnormality detection method according to claim 1, characterized in that before inputting the dynamic data to be measured to the trained target time-series self-encoder model and outputting the first error information, the method further comprises:
Acquiring training dynamic data;
inputting the training dynamic data to a time sequence encoder in an initial time sequence self-encoder model to perform time sequence encoding processing, outputting first hidden variable data, inputting the first hidden variable data to a time sequence decoder in the initial time sequence self-encoder model to perform time sequence decoding processing, and outputting dynamic training reconstruction data;
and calculating to obtain a first association result between the training dynamic data and the dynamic training reconstruction data, back-propagating the first association result to the initial time sequence self-encoder model for iterative training, and generating the target time sequence self-encoder model.
4. The battery abnormality detection method according to claim 3, wherein the acquiring training dynamic data includes:
acquiring historical battery data;
searching the historical battery data, reserving cleaning data which are failed to match with a preset threshold value in the historical battery data according to a search result, and obtaining training time sequence data according to the cleaning data;
and performing feature decomposition on the training time sequence data to obtain the training dynamic data.
5. The battery abnormality detection method according to claim 1, characterized in that before inputting the static data to be detected to the trained target static self-encoder model and outputting the second error information, the method further comprises:
Acquiring training static data;
inputting the training static data to a static encoder in an initial static self-encoder model for encoding processing, outputting second hidden variable data, inputting the second hidden variable data to a static decoder in the initial static self-encoder model for decoding processing, and outputting static training reconstruction data;
and calculating to obtain a second association result between the training static data and the static training reconstruction data, reversely transmitting the second association result to the initial static self-encoder model for iterative training, and generating the target static self-encoder model.
6. The battery abnormality detection method according to claim 1, wherein the generating of the abnormality detection result for the battery based on the first error information and the second error information includes:
acquiring training fusion error information; the training fusion error information is obtained by training time sequence data through the target time sequence self-encoder model and the target static self-encoder model;
and calculating to obtain fusion error information according to the first error information and the second error information, and calculating to obtain the abnormality detection result according to the fusion error information and the training fusion error information.
7. The battery abnormality detection method according to any one of claims 1 to 6, characterized in that the obtaining of time series data to be measured of the battery in a working state includes:
acquiring real-time characteristic data of at least two characteristic types of the battery in the working state;
and according to the characteristic type, carrying out time sequence division processing on the real-time characteristic data to obtain the time sequence data to be detected.
8. A battery abnormality detection device, characterized by comprising: the device comprises an acquisition module, a decomposition module, an output module and a generation module;
the acquisition module is used for acquiring time sequence data to be detected of the battery in a working state;
the decomposition module is used for carrying out characteristic decomposition on the time sequence data to be detected to obtain dynamic data to be detected and static data to be detected; the dynamic data to be detected refers to the data which has dynamic change in the time dimension in the detected time sequence data to be detected, and the static data to be detected refers to the data which has static state in the time dimension in the detected time sequence data to be detected;
the output module is used for inputting the dynamic data to be tested into the trained target time sequence self-encoder model and outputting first error information; inputting the static data to be tested into a trained target static self-encoder model, and outputting second error information;
The generation module is used for generating an abnormality detection result for the battery according to the first error information and the second error information.
9. A battery abnormality detection system, characterized in that the system comprises: the main control device and the battery body; wherein the main control device is connected with the battery body;
the main control device is configured to execute the battery abnormality detection method according to any one of claims 1 to 7.
10. An electronic device comprising a memory and a processor, wherein the memory has stored therein a computer program, the processor being arranged to run the computer program to perform the battery anomaly detection method of any one of claims 1 to 7.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211557848.5A CN116184210A (en) | 2022-12-06 | 2022-12-06 | Battery abnormality detection method, device, system and electronic device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211557848.5A CN116184210A (en) | 2022-12-06 | 2022-12-06 | Battery abnormality detection method, device, system and electronic device |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN116184210A true CN116184210A (en) | 2023-05-30 |
Family
ID=86431526
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202211557848.5A Pending CN116184210A (en) | 2022-12-06 | 2022-12-06 | Battery abnormality detection method, device, system and electronic device |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116184210A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116628428A (en) * | 2023-07-24 | 2023-08-22 | 华能信息技术有限公司 | Data processing method and system |
| CN116819378A (en) * | 2023-08-29 | 2023-09-29 | 中国华能集团清洁能源技术研究院有限公司 | Energy storage battery abnormality detection method and device |
-
2022
- 2022-12-06 CN CN202211557848.5A patent/CN116184210A/en active Pending
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116628428A (en) * | 2023-07-24 | 2023-08-22 | 华能信息技术有限公司 | Data processing method and system |
| CN116628428B (en) * | 2023-07-24 | 2023-10-31 | 华能信息技术有限公司 | Data processing method and system |
| CN116819378A (en) * | 2023-08-29 | 2023-09-29 | 中国华能集团清洁能源技术研究院有限公司 | Energy storage battery abnormality detection method and device |
| CN116819378B (en) * | 2023-08-29 | 2023-12-26 | 中国华能集团清洁能源技术研究院有限公司 | Energy storage battery abnormality detection method and device |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Li et al. | Big data driven lithium-ion battery modeling method based on SDAE-ELM algorithm and data pre-processing technology | |
| Wang et al. | A data-driven method for battery charging capacity abnormality diagnosis in electric vehicle applications | |
| Meng et al. | A review on prognostics and health management (PHM) methods of lithium-ion batteries | |
| CN116184210A (en) | Battery abnormality detection method, device, system and electronic device | |
| Kim et al. | Estimation of Li-ion battery state of health based on multilayer perceptron: As an EV application | |
| Li et al. | Lithium-ion battery modeling based on Big Data | |
| US8855954B1 (en) | System and method for prognosticating capacity life and cycle life of a battery asset | |
| CN109001640B (en) | Data processing method and device for power battery | |
| Zhang et al. | Cost-effective Lebesgue sampling long short-term memory networks for lithium-ion batteries diagnosis and prognosis | |
| CN111191824B (en) | Power battery capacity attenuation prediction method and system | |
| Jianfang et al. | Multi‐Scale prediction of RUL and SOH for Lithium‐Ion batteries based on WNN‐UPF combined model | |
| CN114791993B (en) | Power battery pack SOH prediction method and system | |
| Chehade et al. | A multioutput convolved Gaussian process for capacity forecasting of li-ion battery cells | |
| Su et al. | Prediction of remaining useful life for lithium-ion battery with multiple health indicators | |
| Harippriya et al. | Battery management system to estimate battery aging using deep learning and machine learning algorithms | |
| CN115902642A (en) | Battery state of charge estimation method and device, electronic equipment and storage medium | |
| DIng et al. | Li-ion battery health estimation based on multi-layer characteristic fusion and deep learning | |
| Xie et al. | Residual life prediction of lithium-ion batteries based on data preprocessing and a priori knowledge-assisted CNN-LSTM | |
| Savargaonkar et al. | Uncorrelated sparse autoencoder with long short-term memory for state-of-charge estimations in lithium-ion battery cells | |
| Pau et al. | Microcontroller architectures for battery state of charge prediction with tiny neural networks | |
| Last et al. | Predictive maintenance with multi-target classification models | |
| Sun et al. | Autoencoder-enhanced regularized prototypical network for new energy vehicle battery fault detection | |
| CN116819328A (en) | Electric automobile power battery fault diagnosis method, system, equipment and medium | |
| CN115808627A (en) | Lithium battery SOH prediction method and device | |
| WO2023052910A1 (en) | System and method for estimating state of health and remaining useful life of a battery |
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 |