CN113344024A - Lithium ion battery thermal runaway grading early warning method and early warning system - Google Patents
Lithium ion battery thermal runaway grading early warning method and early warning system Download PDFInfo
- Publication number
- CN113344024A CN113344024A CN202110435874.XA CN202110435874A CN113344024A CN 113344024 A CN113344024 A CN 113344024A CN 202110435874 A CN202110435874 A CN 202110435874A CN 113344024 A CN113344024 A CN 113344024A
- Authority
- CN
- China
- Prior art keywords
- battery
- thermal runaway
- early warning
- lithium ion
- ion battery
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- 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]
- G01R31/378—Arrangements for testing, measuring or monitoring the electrical condition of accumulators or electric batteries, e.g. capacity or state of charge [SoC] specially adapted for the type of battery or accumulator
-
- 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]
- G01R31/382—Arrangements for monitoring battery or accumulator variables, e.g. SoC
-
- 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
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/044—Recurrent networks, e.g. Hopfield networks
-
- 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
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- General Engineering & Computer Science (AREA)
- Evolutionary Computation (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- Software Systems (AREA)
- Mathematical Physics (AREA)
- Health & Medical Sciences (AREA)
- Biomedical Technology (AREA)
- Computing Systems (AREA)
- Molecular Biology (AREA)
- General Health & Medical Sciences (AREA)
- Evolutionary Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Secondary Cells (AREA)
Abstract
The invention provides a grading early warning method and an early warning system for thermal runaway of a lithium ion battery. The method adopts a fault-free battery characteristic element to train the thermal runaway prediction capability of a prediction model consisting of a long-term and short-term memory network, a time convolution network and a GRU neural network; and then inputting the battery characteristic elements of the battery to be monitored on the time sequence into a prediction model to obtain a predicted value at a certain moment in a normal state, and comparing the predicted value with an acquired true value corresponding to the moment, wherein the larger the difference between the predicted value and the true value is, the larger the thermal runaway risk of the battery is. By the operation, respective advantages are integrated, and the problem of overlarge coefficient deviation trained by respective models is avoided, so that high-precision battery thermal runaway graded detection is realized, and the speed and precision of lithium ion battery thermal runaway early warning are improved.
Description
Technical Field
The invention relates to the technical field of battery safety early warning, in particular to a grading early warning method and an early warning system for thermal runaway of a lithium ion battery.
Background
In recent years, lithium ion battery energy storage systems have been widely used in various areas such as new energy grid-connection, micro-grid, smart grid, and the like. The lithium ion battery has the characteristics of high energy density, high conversion efficiency, quick response and the like, and has wide prospect in a large energy storage system. However, with the large-scale application of the energy storage system of the lithium ion battery, the safety problem of the energy storage system is highly valued. Statistical investigation on accidents shows that thermal runaway of lithium ion batteries caused by self chemical reactions or external influences is a main cause of safety problems. After the lithium battery is out of control thermally, the lithium battery has the characteristics of high diffusion speed, high flame intensity, large amount of toxic gas generation and the like, and even after fire extinguishment, the afterburning phenomenon can be generated, so that the whole energy storage system is seriously threatened.
The thermal runaway of the battery is caused by the heat generated by the battery and the abnormal rise of temperature, which causes a chain-type exothermic reaction, resulting in the combustion and explosion of the battery. According to trigger conditions, causes for causing thermal runaway of the battery are mainly divided into three categories: mechanical abuse, such as mechanical deformation due to needle stick, crushing, weight impact, etc.; electric abuse, which is mainly faults of electric components, such as overcharge and overdischarge, lithium dendrite is generated inside the battery and the diaphragm is punctured; thermal abuse, such as excessive temperature, can cause decomposition of the positive and negative electrode materials. Therefore, the accurate prediction of the lithium ion thermal runaway fault can accelerate the development of the energy storage technology, and has great significance for promoting the transformation of new energy structures in China, guaranteeing the energy safety and realizing the aims of energy conservation and emission reduction. Patent CN20160282373.1 discloses a battery fault detection method and a battery fault detection device, in which a data mining algorithm is used to identify a faulty battery cell in a battery pack according to battery characteristic features of the battery pack. The accuracy of battery failure detection can be improved without using a threshold value preset based on manual experience. However, the method can only detect the fault of the current battery state, cannot realize early warning, and is easy to cause problems of early warning and untimely response.
In view of the above, there is a need to design an improved grading early warning method and an improved grading early warning system for thermal runaway of a lithium ion battery to solve the above problems.
Disclosure of Invention
In order to overcome the defects of the prior art, the invention aims to provide a grading early warning method and an early warning system for thermal runaway of a lithium ion battery. The long-short term memory network, the time convolution network and the GRU neural network are adopted for prediction respectively, and equal weight coupling is carried out, so that the advantages of the long-short term memory network, the time convolution network and the GRU neural network are integrated, and the problem of overlarge coefficient deviation trained by respective models is solved, thereby realizing high-precision battery thermal runaway graded detection and improving the speed and precision of lithium ion battery thermal runaway early warning.
In order to achieve the aim, the invention provides a grading early warning method for thermal runaway of a lithium ion battery, which comprises the following steps:
s1, collecting battery characteristic elements without faults, and respectively training the prediction capabilities of a long-term and short-term memory network, a time convolution network and a GRU neural network on the battery characteristic elements on a time sequence;
s2, collecting battery characteristic elements of a battery to be monitored on a time sequence, respectively inputting the battery characteristic elements into the trained long-short term memory network, the trained time convolution network and the trained GRU neural network to obtain three groups of predicted values, and performing equal weight coupling on the three groups of predicted values to serve as fault-free predicted values;
and S3, continuously acquiring the battery characteristic elements of the battery to be monitored on the time sequence as real values, and comparing the real values with the failure-free predicted values obtained in the step S2, wherein the larger the difference between the real values and the failure-free predicted values is, the larger the thermal runaway risk of the battery is.
As a further improvement of the present invention, the characteristic feature of the battery includes, but is not limited to, one or more of temperature, current, voltage, and state of charge of the battery.
As a further improvement of the present invention, in step S1, the battery characteristic feature without failure is obtained by filtering out the battery characteristic feature with thermal runaway through a gaussian filter.
As a further improvement of the invention, the Gaussian filter is a low-pass filter with the frequency domain width of +/-200 w and the threshold value of pi/100.
As a further improvement of the present invention, in step S1, the training of the prediction capability includes: respectively training the long-short term memory network, the time convolution network and the GRU neural network by adopting a fault-free battery characteristic element on a time sequence, and coupling to obtain a prediction result, namely a change trend value of the battery characteristic element on time; then returning the respective loss values of the three networks, and continuing training.
As a further improvement of the invention, the loss values of the long-short term memory network and the GRU neural network are obtained by an MSE loss function; and the loss value of the time convolution network is obtained by mean absolute deviation MAD.
As a further improvement of the present invention, step S3 further includes: and taking the ratio of the absolute value of the difference between the real value and the no-fault predicted value to the no-fault predicted value as a thermal runaway early warning coefficient, wherein the larger the thermal runaway early warning coefficient is, the larger the thermal runaway risk of the battery is.
As a further improvement of the present invention, step S3 further includes: dividing the thermal runaway early warning coefficient into a plurality of grade intervals, and carrying out grading early warning; wherein, the higher the upper limit value of the grade interval, the higher the thermal runaway risk of the battery is.
In order to achieve the above object, the present invention further provides a lithium ion battery thermal runaway grading early warning system, which includes:
the data acquisition module is used for acquiring characteristic element data of the battery;
the thermal runaway model building module is used for building the long-short term memory network, the time convolution network and the GRU neural network in parallel into a prediction model associated with the characteristic elements of the battery and training the prediction capability of the battery by adopting the characteristic elements of the battery without faults;
and the thermal runaway early warning module is used for inputting the battery characteristic elements of the battery to be monitored on the time sequence into the prediction model to obtain a predicted value at a certain moment, and then comparing the predicted value with the acquired real value corresponding to the moment to obtain a thermal runaway early warning result.
As a further improvement of the invention, the lithium ion battery thermal runaway grading early warning system further comprises an equal weight coupling module, which is used for performing equal weight coupling on the predicted values of the long-short term memory network, the time convolution network and the GRU neural network.
The invention has the beneficial effects that:
1. the lithium ion battery thermal runaway grading early warning method provided by the invention adopts the long and short term memory network, the time convolution network and the GRU neural network to respectively predict and carry out equal weight coupling, thereby not only integrating respective advantages, but also avoiding the problem of overlarge coefficient deviation trained by respective models, and also reducing training period, thereby realizing high-precision battery thermal runaway grading detection, improving the speed and precision of lithium ion battery thermal runaway early warning, being beneficial to promoting the intelligent development of new energy storage safety, and accelerating the application of the fourth industrial revolution in the energy and energy storage industry.
2. According to the grading early warning method for the thermal runaway of the lithium ion battery, provided by the invention, the difference between the predicted characteristic element and the actual value of the characteristic element of the historical battery is gradually corrected and reduced according to the characteristic element of the historical battery of a normal battery as a training set, and the predicted value obtained according to real-time monitoring data is the predicted value of a normal state because a model is trained and checked by adopting normal data during prediction. And comparing with the real value at the moment, wherein the larger the difference between the two values is, the larger the thermal runaway risk of the battery is. By the operation, the working state of the battery can be predicted in advance on a time sequence, and the thermal runaway early warning is realized in a real sense.
3. According to the grading early warning method for the thermal runaway of the lithium ion battery, loss generated by the predicted value and the real value of the training set is returned when the model is trained, the training effect is enhanced, and the training workload of the whole model can be reduced.
Drawings
FIG. 1 is a flow chart of a model training method of the lithium ion battery thermal runaway grading early warning method of the invention.
Fig. 2 is a block diagram of a composition structure of the lithium ion battery thermal runaway stage warning.
FIG. 3 is a block diagram of a long term memory network.
Fig. 4 is a block diagram of a time convolutional network.
Fig. 5 is a block diagram of a GRU neural network.
Fig. 6 is an image of a probability density function of a normal distribution in a frequency domain.
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention more apparent, the present invention is described in detail below with reference to specific embodiments.
It should be noted that, in order to avoid obscuring the present invention with unnecessary details, only the structures and/or processing steps closely related to the scheme of the present invention are shown in the specific embodiments, and other details not closely related to the present invention are omitted.
In addition, it is also to be noted that the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or apparatus.
In the long-term power supply working process of the lithium ion battery, temperature malfunction can be caused due to the service life problem of the lithium ion battery or the external environment problem, and fire accidents can be caused seriously. The occurrence of temperature disorders is not usually sudden and usually shows a trend of change, and if the early warning and the treatment are not timely carried out, the temperature disorders can gradually develop into uncontrollable thermal runaway accidents. Based on the above, the invention provides a high-precision grading early warning method for thermal runaway of a lithium ion battery, which accurately predicts the parameters in the normal state at the subsequent moment according to the historical monitoring parameters of the lithium ion battery, compares the predicted parameters with the parameters monitored in real time, and if the difference between the parameters monitored in real time and the predicted parameters at the corresponding time is too far, the battery is possibly subjected to the thermal runaway risk, so that early warning is realized.
Referring to fig. 1 and 2, the method for grading and early warning thermal runaway of a lithium ion battery provided by the invention comprises the following steps:
s1, collecting characteristic elements (historical battery characteristic elements) of a battery without faults on a time sequence, and respectively training the prediction capabilities of a long-term and short-term memory network, a time convolution network and a GRU neural network on the battery characteristic elements on the time sequence (namely, the difference between the predicted characteristic elements and the actual values of the historical battery characteristic elements is gradually corrected and reduced according to the historical battery characteristic elements, and the predicted values and the actual values which are randomly compared in a training set are values of the same time point, so that the accurate prediction of the variation trend is realized); in this way, the network model is trained and checked by using the historical data of the normal battery, so that the historical data can predict the battery characteristic elements of the normal state at the subsequent time in advance; when the prediction is carried out, the model is trained and checked by adopting normal data, so that the predicted values obtained according to the real-time monitoring data are the predicted values of the normal state.
S2, collecting battery characteristic elements of a battery to be monitored on a time sequence, respectively inputting the battery characteristic elements into the trained long-short term memory network, the trained time convolution network and the trained GRU neural network to obtain three groups of predicted values, and performing equal weight coupling on the three groups of predicted values to serve as fault-free predicted values; the result can be corrected by performing equal weight coupling comprehensively, the coefficient deviation trained by respective models is reduced, the training period is shortened, and the prediction result is more accurate.
And S3, continuously acquiring the battery characteristic elements of the battery to be monitored on the time sequence as real values, and comparing the real values with the failure-free predicted values obtained in the step S2, wherein the larger the difference between the real values and the failure-free predicted values is, the larger the thermal runaway risk of the battery is.
Step S3 further includes: and taking the ratio of the absolute value of the difference between the real value and the no-fault predicted value to the no-fault predicted value as a thermal runaway early warning coefficient, wherein the larger the thermal runaway early warning coefficient is, the larger the thermal runaway risk of the battery is.
Step S3 further includes: dividing the thermal runaway early warning coefficient into a plurality of grade intervals, and carrying out grading early warning; wherein, the higher the upper limit value of the grade interval, the higher the thermal runaway risk of the battery is. For example, the whole interval is 0-100%, and each interval range is 25%, the interval is divided into four grade intervals: [ 0-25% ], [ 26-50% ], [ 51-75% ], and [ 76-100% ]. And when the interval value of the obtained thermal runaway early warning coefficient is larger, the thermal runaway risk is larger, and early warning in different degrees is carried out.
The battery characteristic element includes but is not limited to one or more of temperature, current, voltage and state of charge of the battery. Preferably, the temperature, the current, the voltage and the state of charge can be selected, and the prediction accuracy of the normal state of the battery can be improved through the centralized training of various battery characteristic elements. By adopting the early warning method, the early warning model is trained based on the characteristic elements of the non-thermal runaway fault battery, so that the predicted value of the battery characteristic elements at a certain subsequent time can be sensitively given according to the historical battery characteristic elements of the normal battery. For example: suppose t1-t3At the moment when the battery is working normally, t4At that time, the battery tends to be thermally runaway, or is a starting point for thermal runaway to develop. The prediction model composed of the three networks can be based on t1-t3A characteristic factor of the battery at the time, given as t4Characteristic factor of battery in normal state at time, and t4The real monitoring values at the moment are compared, so that t can be accurately judged4And immediately giving early warning to take corresponding countermeasures when the time is abnormal. But using the prior art at t4The thermal runaway development starting point at a moment may not be able to sensitively judge that it is in a thermal runaway state, and thus advance warning cannot be performed.
In step S1, the non-faulty battery feature is obtained by filtering out the thermal runaway battery feature through a gaussian filter, that is, filtering out the outlier data (i.e., thermal runaway battery data) of the collected battery feature by using the gaussian filter. The Gaussian filter is a low-pass filter with the frequency domain width of +/-200 w and the threshold value of pi/100. For example, using an averaging filter to approximate, the reaction function is:
please refer to fig. 6, which shows a normal distributionThe image of the probability density function in the frequency domain. Sigma2The normal distribution variance is represented, and the graph is wider when the sigma is increased, and the peak value is lower; x is an independent variable representing the main frequency (i.e. frequency peak) of the input time sequence after Fourier transform to the frequency domain, and because the high-frequency thermal runaway battery signal is to be filtered, a Gaussian filter is adopted, and as the main frequency is increased, the weight g (x) is lower and approaches to zero, and then the filtering can be completed.
In step S1, the training of the prediction capability includes: respectively training the long-short term memory network, the time convolution network and the GRU neural network by adopting a fault-free battery characteristic element on a time sequence, and coupling to obtain a prediction result, namely a change trend value of the battery characteristic element on time; then returning the respective loss values of the three networks, and continuing training.
Specifically, please refer to fig. 3 to 5. FIG. 3 is a structure diagram of a Long Short Term Memory (LSTM) network, wherein the structure in the frame A in FIG. 3 represents a cell state (ct), the structure in the frame B in FIG. 3 represents a forgetting gate (ft), wherein σ is a sigmoid function, the output ft is 0-1, and represents a forgetting probability. The structure in box C represents the entry gate, where σ is still the sigmoid function and tanh is the activation function, which collectively update the cell state. The structure in box D represents cell renewal, where the input cell state is first multiplied by the forgotten gate output and then added to the product of it and at of the input gate. The structure in the E box represents a hidden state ht, and the hidden state of the previous node and the input signal of the node are multiplied by the cell state through a sigmoid function.
The loss function of LSTM is the MSE loss function:
loss(xi，yi)＝(xi-yi)2
wherein x isiTo predict the elements, yiThe predicted value is output for the actual element and is returned as the square of the difference.
The long-short term memory network has the following advantages: (1) no gradient vanishing or gradient explosion occurs and the connection weights may be changed every time step. (2) The functions of forgetting gate, sigmoid, tanh and the like are involved in the internal self-circulation of the nerve cell, and for a system for processing time series signals, the input of the time information of the internal cell state of the LSTM is the output of the previous neuron, so that the external manual determination of what information the LSTM should forget or remember is not needed, and the internal part of the neuron is self-loop.
FIG. 4 is a block diagram of a time convolutional network, using a causal convolutional TCN model. The TCN model comprises an input layer, two hidden layers and an output layer, namely, for the value of the previous layer at the moment t, the TCN model only depends on the value of the next layer at the moment t and before the moment t. The first bit of each layer is filled with 0, the data in the future cannot be seen, the data are in a one-way structure, are not two-way and are a strict time sequence model. The loss function uses the mean absolute error, MAD, as shown below:
wherein xi is the true value, and m (x) is the predicted value
The time convolution network has the following characteristics:
(1) the inherent characteristics of the time convolution network enable the TCN to extract different eigenvalues and eigenvectors in the time series signal.
(2) Has the parallelism of data processing. Given a time series signal, the TCN can process it in parallel, without the need for sequential processing as the RNN.
(3) The time convolution network also rarely has the problem of gradient disappearance or gradient explosion (in the process of processing signals, because each layer of the traditional neural network has the influence of an activation function, some characteristic weights are continuously increased along with the increase of the number of layers, and some characteristic weights are continuously reduced), because convolution kernels are mainly adopted for convolution, and the whole connection layer can uniformly map the extracted different characteristic vectors with weights.
(4) The occupied memory is lower. When the RNN is used, information of each step needs to be stored, a large amount of memory is occupied, a convolution kernel in one layer of the TCN is shared, and the memory use is lower.
FIG. 5 is a diagram of a GRU neural network architecture, which combines forgetting gate and input gate into a single update gate, as compared to LSTM. The cellular state and the hidden state are also mixed, plus some other modifications. The final model is simpler than the standard LSTM model. The loss function also employs a MSE function.
The GRU is a kind of LSTM, and combines the forgetting gate and the input gate into an update gate, which reduces many coefficients and calculation requirements, although the accuracy is not as good as the LSTM, but the calculation is fast, and the error of the LSTM caused by too much calculation can be avoided.
Therefore, the three networks are integrated to carry out equal weight coupling, so that the result can be corrected, the coefficient deviation trained by respective models can be reduced, the training period can be shortened, and the prediction result is more accurate.
The invention also provides a lithium ion battery thermal runaway grading early warning system, which comprises:
the data acquisition module is used for acquiring characteristic element data of the battery;
the thermal runaway model building module is used for building the long-short term memory network, the time convolution network and the GRU neural network in parallel into a prediction model associated with the characteristic elements of the battery and training the prediction capability of the battery by adopting the characteristic elements of the battery without faults;
and the thermal runaway early warning module is used for inputting the battery characteristic elements of the battery to be monitored on the time sequence into the prediction model to obtain a predicted value at a certain moment, and then comparing the predicted value with the acquired real value corresponding to the moment to obtain a thermal runaway early warning result.
The lithium ion battery thermal runaway grading early warning system further comprises an equal weight coupling module which is used for performing equal weight coupling on the predicted values of the long-term and short-term memory network, the time convolution network and the GRU neural network. The method can correct the result, reduce the coefficient deviation trained by respective models, reduce the training period and enable the prediction result to be more accurate.
In conclusion, the thermal runaway graded early warning method and the early warning system provided by the invention adopt the long and short term memory network, the time convolution network and the GRU neural network to respectively predict and carry out equal weight coupling, thereby not only integrating respective advantages, but also avoiding the problem of overlarge coefficient deviation trained by respective models, and reducing training period, thereby realizing high-precision graded detection of the thermal runaway of the battery and improving the speed and precision of the thermal runaway early warning of the lithium ion battery; and the working state of the battery can be predicted in advance on the time sequence, so that the thermal runaway early warning is realized in a real sense.
Although the present invention has been described in detail with reference to the preferred embodiments, it will be understood by those skilled in the art that various changes may be made and equivalents may be substituted for elements thereof without departing from the spirit and scope of the present invention.
Claims (10)
1. A grading early warning method for thermal runaway of a lithium ion battery is characterized by comprising the following steps:
s1, collecting battery characteristic elements without faults, and respectively training the prediction capabilities of a long-term and short-term memory network, a time convolution network and a GRU neural network on the battery characteristic elements on a time sequence;
s2, collecting battery characteristic elements of a battery to be monitored on a time sequence, respectively inputting the battery characteristic elements into the trained long-short term memory network, the trained time convolution network and the trained GRU neural network to obtain three groups of predicted values, and performing equal weight coupling on the three groups of predicted values to serve as fault-free predicted values;
and S3, continuously acquiring the battery characteristic elements of the battery to be monitored on the time sequence as real values, and comparing the real values with the failure-free predicted values obtained in the step S2, wherein the larger the difference between the real values and the failure-free predicted values is, the larger the thermal runaway risk of the battery is.
2. The lithium ion battery thermal runaway grading pre-warning method according to claim 1, wherein the battery characteristic elements include but are not limited to one or more of temperature, current, voltage and state of charge of the battery.
3. The grading pre-warning method for thermal runaway of the lithium ion battery according to claim 1, wherein in step S1, the fault-free battery characteristic elements are obtained by filtering out thermal runaway battery characteristic elements through a gaussian filter.
4. The grading early warning method for the thermal runaway of the lithium ion battery as claimed in claim 3, wherein the Gaussian filter is a low-pass filter with a frequency domain width of ± 200w and a threshold value of Π/100.
5. The lithium ion battery thermal runaway grading early warning method according to claim 1, wherein in step S1, the training of the prediction capability comprises: respectively training the long-short term memory network, the time convolution network and the GRU neural network by adopting a fault-free battery characteristic element on a time sequence, and coupling to obtain a prediction result, namely a change trend value of the battery characteristic element on time; then returning the respective loss values of the three networks, and continuing training.
6. The grading pre-warning method for thermal runaway of the lithium ion battery according to claim 5, wherein the loss values of the long-short term memory network and the GRU neural network are obtained through an MSE loss function; and the loss value of the time convolution network is obtained by mean absolute deviation MAD.
7. The grading early warning method for thermal runaway of the lithium ion battery according to claim 1, wherein the step S3 further comprises: and taking the ratio of the absolute value of the difference between the real value and the no-fault predicted value to the no-fault predicted value as a thermal runaway early warning coefficient, wherein the larger the thermal runaway early warning coefficient is, the larger the thermal runaway risk of the battery is.
8. The grading early warning method for thermal runaway of the lithium ion battery according to claim 7, wherein the step S3 further comprises: dividing the thermal runaway early warning coefficient into a plurality of grade intervals, and carrying out grading early warning; wherein, the higher the upper limit value of the grade interval, the higher the thermal runaway risk of the battery is.
9. The utility model provides a hierarchical early warning system of lithium ion battery thermal runaway which characterized in that includes:
the data acquisition module is used for acquiring characteristic element data of the battery;
the thermal runaway model building module is used for building the long-short term memory network, the time convolution network and the GRU neural network in parallel into a prediction model associated with the characteristic elements of the battery and training the prediction capability of the battery by adopting the characteristic elements of the battery without faults;
and the thermal runaway early warning module is used for inputting the battery characteristic elements of the battery to be monitored on the time sequence into the prediction model to obtain a predicted value at a certain moment, and then comparing the predicted value with the acquired real value corresponding to the moment to obtain a thermal runaway early warning result.
10. The grading pre-warning system for the thermal runaway of the lithium ion battery as claimed in claim 9, further comprising an equal weight coupling module for performing equal weight coupling on the predicted values of the long-short term memory network, the time convolution network and the GRU neural network.
Priority Applications (2)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110435874.XA CN113344024B (en) | 2021-04-22 | 2021-04-22 | Lithium ion battery thermal runaway grading early warning method and early warning system |
| AU2021102629A AU2021102629A4 (en) | 2021-04-22 | 2021-05-17 | Lithium battery thermal runaway multi-level detection and early warning system and judgment method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110435874.XA CN113344024B (en) | 2021-04-22 | 2021-04-22 | Lithium ion battery thermal runaway grading early warning method and early warning system |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN113344024A true CN113344024A (en) | 2021-09-03 |
| CN113344024B CN113344024B (en) | 2022-04-12 |
Family
ID=76662634
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110435874.XA Active CN113344024B (en) | 2021-04-22 | 2021-04-22 | Lithium ion battery thermal runaway grading early warning method and early warning system |
Country Status (2)
| Country | Link |
|---|---|
| CN (1) | CN113344024B (en) |
| AU (1) | AU2021102629A4 (en) |
Cited By (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115034312A (en) * | 2022-06-14 | 2022-09-09 | 燕山大学 | Fault diagnosis method for dual neural network model satellite power supply system |
| CN115184808A (en) * | 2022-07-05 | 2022-10-14 | 东莞新能安科技有限公司 | Battery thermal runaway risk detection method, device, equipment and computer storage medium |
| CN115294745A (en) * | 2022-05-23 | 2022-11-04 | 电子科技大学 | Lithium battery thermal runaway layering early warning method based on neural network and data difference |
| CN115420401A (en) * | 2022-08-25 | 2022-12-02 | 上海玫克生储能科技有限公司 | Early warning method, early warning system, storage medium and electronic equipment |
| CN116430245A (en) * | 2023-06-14 | 2023-07-14 | 威海谱跃光电科技有限公司 | Battery thermal runaway prediction method based on gradient optimization multi-physical information neural network |
| WO2023159638A1 (en) * | 2022-02-28 | 2023-08-31 | 宁德时代新能源科技股份有限公司 | Battery fault detection system, method and apparatus |
| CN116720073A (en) * | 2023-08-10 | 2023-09-08 | 江苏金恒信息科技股份有限公司 | Abnormality detection extraction method and system based on classifier |
| CN117388725A (en) * | 2023-03-13 | 2024-01-12 | 中国石油大学(华东) | Early abnormality early warning method based on expansion force of lithium ion battery |
| CN117691227A (en) * | 2024-02-04 | 2024-03-12 | 江苏林洋亿纬储能科技有限公司 | Method and system for safety pre-warning of battery energy storage system and computing device |
Families Citing this family (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113702836B (en) * | 2021-07-23 | 2023-08-18 | 国家电网有限公司西北分部 | Lithium ion battery state of charge estimation method based on EMD-GRU |
| CN113591404B (en) * | 2021-09-29 | 2021-12-07 | 杭州宇谷科技有限公司 | Battery abnormity detection system and method based on deep learning |
| CN115032548B (en) * | 2022-05-25 | 2023-03-21 | 广州汽车集团股份有限公司 | Early warning method and system for automobile storage battery |
| CN117346946B (en) * | 2023-11-29 | 2024-05-03 | 宁德时代新能源科技股份有限公司 | Air pressure sampling circuit, method, battery management system and power utilization device |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108519556A (en) * | 2018-04-13 | 2018-09-11 | 重庆邮电大学 | A kind of lithium ion battery SOC prediction techniques based on Recognition with Recurrent Neural Network |
| WO2019017991A1 (en) * | 2017-07-21 | 2019-01-24 | Quantumscape Corporation | Predictive model for estimating battery states |
| US20190242936A1 (en) * | 2018-02-05 | 2019-08-08 | Wuhan University | Fault diagnosis method for series hybrid electric vehicle ac/dc converter |
| CN111999648A (en) * | 2020-08-20 | 2020-11-27 | 浙江工业大学 | Lithium battery residual life prediction method based on long-term and short-term memory network |
-
2021
- 2021-04-22 CN CN202110435874.XA patent/CN113344024B/en active Active
- 2021-05-17 AU AU2021102629A patent/AU2021102629A4/en not_active Ceased
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2019017991A1 (en) * | 2017-07-21 | 2019-01-24 | Quantumscape Corporation | Predictive model for estimating battery states |
| US20190242936A1 (en) * | 2018-02-05 | 2019-08-08 | Wuhan University | Fault diagnosis method for series hybrid electric vehicle ac/dc converter |
| CN108519556A (en) * | 2018-04-13 | 2018-09-11 | 重庆邮电大学 | A kind of lithium ion battery SOC prediction techniques based on Recognition with Recurrent Neural Network |
| CN111999648A (en) * | 2020-08-20 | 2020-11-27 | 浙江工业大学 | Lithium battery residual life prediction method based on long-term and short-term memory network |
Non-Patent Citations (2)
| Title |
|---|
| XIA B,ET AL.: "A correlation based fault detection method for short circuits in battery packs", 《JOURNAL OF POWER SOURCES》 * |
| 刘月峰 等: "锂离子电池RUL预测方法综述", 《计算机工程》 * |
Cited By (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2023159638A1 (en) * | 2022-02-28 | 2023-08-31 | 宁德时代新能源科技股份有限公司 | Battery fault detection system, method and apparatus |
| CN115294745A (en) * | 2022-05-23 | 2022-11-04 | 电子科技大学 | Lithium battery thermal runaway layering early warning method based on neural network and data difference |
| CN115294745B (en) * | 2022-05-23 | 2023-09-29 | 电子科技大学 | Lithium battery thermal runaway layered early warning method based on neural network and data difference |
| CN115034312A (en) * | 2022-06-14 | 2022-09-09 | 燕山大学 | Fault diagnosis method for dual neural network model satellite power supply system |
| CN115034312B (en) * | 2022-06-14 | 2023-01-06 | 燕山大学 | Fault diagnosis method for dual neural network model satellite power system |
| CN115184808A (en) * | 2022-07-05 | 2022-10-14 | 东莞新能安科技有限公司 | Battery thermal runaway risk detection method, device, equipment and computer storage medium |
| CN115420401A (en) * | 2022-08-25 | 2022-12-02 | 上海玫克生储能科技有限公司 | Early warning method, early warning system, storage medium and electronic equipment |
| CN117388725A (en) * | 2023-03-13 | 2024-01-12 | 中国石油大学(华东) | Early abnormality early warning method based on expansion force of lithium ion battery |
| CN116430245A (en) * | 2023-06-14 | 2023-07-14 | 威海谱跃光电科技有限公司 | Battery thermal runaway prediction method based on gradient optimization multi-physical information neural network |
| CN116720073A (en) * | 2023-08-10 | 2023-09-08 | 江苏金恒信息科技股份有限公司 | Abnormality detection extraction method and system based on classifier |
| CN117691227A (en) * | 2024-02-04 | 2024-03-12 | 江苏林洋亿纬储能科技有限公司 | Method and system for safety pre-warning of battery energy storage system and computing device |
| CN117691227B (en) * | 2024-02-04 | 2024-04-26 | 江苏林洋亿纬储能科技有限公司 | Method and system for safety pre-warning of battery energy storage system and computing device |
Also Published As
| Publication number | Publication date |
|---|---|
| AU2021102629A4 (en) | 2021-07-08 |
| CN113344024B (en) | 2022-04-12 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN113344024B (en) | Lithium ion battery thermal runaway grading early warning method and early warning system | |
| CN111007401A (en) | Electric vehicle battery fault diagnosis method and device based on artificial intelligence | |
| Fan et al. | Research and application of smart grid early warning decision platform based on big data analysis | |
| Zhang et al. | Intelligent state of charge estimation of battery pack based on particle swarm optimization algorithm improved radical basis function neural network | |
| Wang et al. | Health diagnosis for lithium-ion battery by combining partial incremental capacity and deep belief network during insufficient discharge profile | |
| Jiang et al. | A fault diagnosis method for electric vehicle power lithium battery based on wavelet packet decomposition | |
| Shi et al. | Study of wind turbine fault diagnosis and early warning based on SCADA data | |
| Sun et al. | A novel fault prediction method based on convolutional neural network and long short-term memory with correlation coefficient for lithium-ion battery | |
| Li et al. | Battery safety risk assessment in real-world electric vehicles based on abnormal internal resistance using proposed robust estimation method and hybrid neural networks | |
| Yao et al. | An intelligent fault diagnosis method for lithium-ion battery pack based on empirical mode decomposition and convolutional neural network | |
| CN112085043B (en) | Intelligent monitoring method and system for network security of transformer substation | |
| You et al. | Capacity Estimation of lithium battery based on charging data and Long Short-term Memory Recurrent Neural Network | |
| Diao et al. | Research on Electric Vehicle Charging Safety Warning based on A-LSTM Algorithm | |
| Zhang et al. | Remaining useful life prediction of lithium-ion batteries based on TCN-DCN fusion model combined with IRRS filtering | |
| Xin et al. | Prediction of remaining useful life of proton exchange membrane fuel cell based on wavelet-LSTM | |
| Liu et al. | Battery fault diagnosis scheme based on improved rbf neural network | |
| Mehta et al. | A Comprehensive study of Machine Learning Techniques used for estimating State of Charge for Li-ion Battery | |
| LU500227B1 (en) | A system and a method for smart grading and early warning the thermal runaway of a lithium-ion battery | |
| CN113935423A (en) | Power battery fault early warning method and system for coupling fuzzy control rule with Elman neural network | |
| Wang et al. | The Application Research of Deep Learning Based Method for Short-term Wind Speed Forecasting | |
| Wang et al. | A double-layer fault diagnosis strategy for electric vehicle batteries based on Gaussian mixture model | |
| Ren et al. | A study of wind speed prediction based on particle swarm algorithm to optimize the parameters of sparse least squares support vector | |
| Tang et al. | Rul prediction of lithium batteries based on dlukf algorithm | |
| Luciani et al. | A Machine Learning Method for State of Charge Estimation in Lead-Acid Batteries for Heavy-Duty Vehicles | |
| Ning | Comparison of several common intelligent fault diagnosis knowledge-based method under nonlinear small sample conditions |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |