CN115294745B - Lithium battery thermal runaway layered early warning method based on neural network and data difference - Google Patents
Lithium battery thermal runaway layered early warning method based on neural network and data difference Download PDFInfo
- Publication number
- CN115294745B CN115294745B CN202210562350.1A CN202210562350A CN115294745B CN 115294745 B CN115294745 B CN 115294745B CN 202210562350 A CN202210562350 A CN 202210562350A CN 115294745 B CN115294745 B CN 115294745B
- Authority
- CN
- China
- Prior art keywords
- sampling
- neural network
- signal
- risk level
- precision
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Classifications
-
- G—PHYSICS
- G08—SIGNALLING
- G08B—SIGNALLING OR CALLING SYSTEMS; ORDER TELEGRAPHS; ALARM SYSTEMS
- G08B31/00—Predictive alarm systems characterised by extrapolation or other computation using updated historic data
-
- G—PHYSICS
- G08—SIGNALLING
- G08B—SIGNALLING OR CALLING SYSTEMS; ORDER TELEGRAPHS; ALARM SYSTEMS
- G08B29/00—Checking or monitoring of signalling or alarm systems; Prevention or correction of operating errors, e.g. preventing unauthorised operation
- G08B29/18—Prevention or correction of operating errors
- G08B29/185—Signal analysis techniques for reducing or preventing false alarms or for enhancing the reliability of the system
- G08B29/186—Fuzzy logic; neural networks
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02E—REDUCTION OF GREENHOUSE GAS [GHG] EMISSIONS, RELATED TO ENERGY GENERATION, TRANSMISSION OR DISTRIBUTION
- Y02E60/00—Enabling technologies; Technologies with a potential or indirect contribution to GHG emissions mitigation
- Y02E60/10—Energy storage using batteries
Landscapes
- Physics & Mathematics (AREA)
- Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Automation & Control Theory (AREA)
- Evolutionary Computation (AREA)
- Fuzzy Systems (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Computer Security & Cryptography (AREA)
- Artificial Intelligence (AREA)
- Business, Economics & Management (AREA)
- Computing Systems (AREA)
- Emergency Management (AREA)
- Alarm Systems (AREA)
Abstract
The application provides a lithium battery thermal runaway layered early warning method based on neural network and data difference, belongs to the technical field of battery fire control, and aims to improve the prediction and prevention effects while reducing prediction lag by correcting low-fidelity data and difference prediction through the neural network, and has good popularization value.
Description
Technical Field
The application relates to the technical field of battery fire control, in particular to a lithium battery thermal runaway layering early warning method based on neural network and data difference.
Background
The new energy battery thermal runaway early warning is an important line of defense for the safety of new energy battery driving equipment, and is an effective way for reducing the life and property loss caused by fire. In the process of firing a new energy battery, chemical reaction still occurs in the battery, so that a domino effect can be caused, and the traditional early warning difficulty is high; the smoke sensor can be used for effectively early warning the thermal runaway of the lithium battery and taking measures to prevent fire. The prior art generally has great hysteresis for the prediction of thermal runaway, and has poor prevention effect for the prediction of thermal runaway.
Disclosure of Invention
Aiming at the problems in the prior art, the application provides a lithium battery thermal runaway layering early warning method based on a neural network and data difference, and aims to reduce prediction lag and improve prediction prevention effect.
In order to achieve the above purpose, the application adopts the following technical scheme:
a lithium battery thermal runaway layered early warning method based on neural network and data difference at least comprises the following steps: the signal sampling device can acquire an induction signal representing thermal runaway of the battery and comprises two sampling modes of low sampling frequency, high sampling precision and high sampling frequency and low sampling precision;
the central processing unit device can conduct neural network processing according to the induction signals representing the thermal runaway of the battery, predicts to obtain future sampling values, conducts differential processing on the predicted future sampling values, determines a risk level according to the magnitude of the differential values, and outputs execution signals related to the risk level; and
the alarm device can receive the execution signal and respond to the execution signal to give an alarm prompt;
the early warning method at least comprises the following steps:
s1, signal acquisition is carried out by using a signal sampling device;
step S2, the central processing unit device determines a risk level according to the induction signals acquired by the signal acquisition device, and adjusts a sampling mode according to the risk level;
and step S3, when the risk level reaches the preset level requirement, the central processing unit outputs an execution signal related to the risk level, and an alarm prompt is given by using the alarm device.
Preferably, in step S1, a low sampling frequency high sampling precision mode is used for signal acquisition;
step S2 specifically comprises the steps of predicting a plurality of sampling values with a neural network at equal intervals in the future after obtaining the sampling values from the signal sampling device, performing differential calculation on the predicted plurality of sampling values, determining a risk level as a second level when the differential value is higher than a set first threshold value, and adjusting the sampling mode of the sampling device to a high sampling frequency and low sampling precision mode, otherwise, still as a first level, and performing signal acquisition by the signal sampling device by adopting the low sampling frequency and high sampling precision mode; after entering a high sampling frequency low sampling precision mode, performing neural network fitting on the acquired low precision sequence, performing precision lifting processing, taking current data after the precision lifting processing and previous high precision data as input for predicting a subsequent sequence of the neural network, predicting a sampling value sequence at a certain time interval in the future, performing differential calculation, and determining that the risk level is three-level when the differential value is higher than a second threshold value;
step S3 specifically includes that when the central processing unit determines that the risk level is three-level, the central processing unit outputs an execution signal, and an alarm prompt is made by using the alarm device.
Preferably, the differential calculation used includes forward differential, backward differential, deriving the end of the fitting function, or a combination thereof.
Compared with the prior art, the application has at least the following beneficial effects:
1. the layered control system enters different stages, and dynamically adjusts the power consumption and the precision of the measurement system;
2. the neural network is used for carrying out primary correction on the data under the high sampling frequency, then predicting the data with the low sampling frequency, fully utilizing the data, improving the sampling signal-to-noise ratio and laying a theoretical foundation for a layering system taking the result of differential calculation as a layering basis;
3. the differential calculation is used as a layering control basis, so that a good prediction effect is achieved on the abrupt change of the thermal runaway of the lithium battery;
4. based on the basic contradiction that the indexes of the sampling frequency and the sampling precision cannot be improved at the same time in the traditional data, the simple neural network is used for improving the data signal-to-noise ratio, reducing the high-frequency noise of signals, and reducing the contradiction between the sampling frequency and the sampling precision, the differential calculation result for carrying out layering early warning is not excessively interfered by the high-frequency noise, so that the accuracy of using the differential as the layering index is greatly improved. Furthermore, two sensors may be used to sense data simultaneously, sampling with two different frequencies, so that the predicted curve has the highest accuracy.
Drawings
The application will be described in more detail hereinafter on the basis of embodiments and with reference to the accompanying drawings. Wherein:
FIG. 1 is a schematic representation of the post-fitting differential computation of the neural network of the present application;
FIG. 2 is a schematic diagram of the initial sampling results of the present application;
FIG. 3 is a schematic illustration of the neural network fitting result of the present application retaining the original sampling points;
FIG. 4 is a schematic diagram of the neural network fitting result of the application with the original sampling points removed;
FIG. 5 is a schematic representation of neural network fitting results without secondary use of neural network optimization in accordance with the present application;
FIG. 6 is a graph showing a comparison of the results of a neural network fit optimized for a second use of the neural network of the present application;
FIG. 7 is a schematic diagram of a neural network employed in the present application;
in the drawings, like parts are designated with like reference numerals. The figures are not to scale.
Detailed Description
The application will be further described with reference to the accompanying drawings.
The application provides a lithium battery thermal runaway layered early warning method based on neural network and data difference, which adopts an early warning system at least comprising:
the signal sampling device can acquire an induction signal representing thermal runaway of the battery and comprises two sampling modes of low sampling frequency, high sampling precision and high sampling frequency and low sampling precision;
the central processing unit device can conduct neural network processing according to the induction signals representing the thermal runaway of the battery, predicts to obtain future sampling values, conducts differential processing on the predicted future sampling values, determines a risk level according to the magnitude of the differential values, and outputs execution signals related to the risk level; and
the alarm device can receive the execution signal and respond to the execution signal to give an alarm prompt;
the early warning method at least comprises the following steps:
s1, signal acquisition is carried out by using a signal sampling device;
step S2, the central processing unit device determines a risk level according to the induction signals acquired by the signal acquisition device, and adjusts a sampling mode according to the risk level;
and step S3, when the risk level reaches the preset level requirement, the central processing unit outputs an execution signal related to the risk level, and an alarm prompt is given by using the alarm device.
Preferably, in step S1, a low sampling frequency high sampling precision mode is used for signal acquisition;
step S2 specifically comprises the steps of predicting a plurality of sampling values with a neural network at equal intervals in the future after obtaining the sampling values from the signal sampling device, performing differential calculation on the predicted plurality of sampling values, determining a risk level as a second level when the differential value is higher than a set first threshold value, and adjusting the sampling mode of the sampling device to a high sampling frequency and low sampling precision mode, otherwise, still as a first level, and performing signal acquisition by the signal sampling device by adopting the low sampling frequency and high sampling precision mode; after entering a high sampling frequency low sampling precision mode, performing neural network fitting on the acquired low precision sequence, performing precision lifting processing, taking current data after the precision lifting processing and previous high precision data as input for predicting a subsequent sequence of the neural network, predicting a sampling value sequence at a certain time interval in the future, performing differential calculation, and determining that the risk level is three-level when the differential value is higher than a second threshold value;
step S3 specifically includes that when the central processing unit determines that the risk level is three-level, the central processing unit outputs an execution signal, and an alarm prompt is made by using the alarm device.
Preferably, the differential calculation used includes forward differential, backward differential, deriving the end of the fitting function, or a combination thereof. Preferably, when signal sampling is performed, the signal sampling device can perform signal acquisition in a low sampling frequency high sampling precision mode and a high sampling frequency low sampling precision mode.
When data acquisition is carried out on the new energy battery, data with different sampling precision and different sampling frequencies are sequentially acquired, and then difference calculation is carried out through neural network fitting, as shown in fig. 1. Firstly, the acquired sampling data are shown in the diamond part of fig. 2, then, the neural network fitting is carried out on the sampling data (shown in fig. 3), the first section of fitting curve of fig. 4 can be obtained after the difference is carried out, the difference operation (differential derivative here) is carried out on the tail end of the first section of fitting curve, the decision is judged to be in a risk level two at the moment, and the high sampling frequency low sampling precision mode is started; the obtained sampling points are like the punctiform part in fig. 2, the second section fitting curve in fig. 4 can be obtained after fitting by using a neural network, differential operation (differential derivation here) is carried out on the tail end of the second section fitting curve, the decision of the calculation result is judged to be the risk level three, and an alarm is sounded.
Specifically, in one embodiment, the effect of globally fitting the curve as shown in fig. 5 can be obtained by directly using all the sampled data as the input of the neural network fitting, which is similar to the trend of the two-section curve of the graph.
In one embodiment, a quadratic fitting method is used, after neural network fitting is performed on data with high sampling frequency, the obtained data with correction accuracy is sampled again to replace the original low-accuracy data, and then the curve is globally fitted, as shown in fig. 6, compared with the result of the global fitting curve in fig. 5, the curve obtained at the moment is smoother, and the differential operation result is more stable and smoother.
Specifically, it should be noted that the neural network used is a DNN-deep neural network, and the specific framework is an input layer (1 x 50) +a full connection layer (50 x 50) +an output layer (50 x 1) (see fig. 7); the risk level is first-order if the time sequence stops before 0.5 s; the time series is a risk level of two if it stops before 1.5s and after 0.5 s.
Preferably, when using DNN-deep neural networks for fitting, a feed-forward network of multiple layers is provided, each layer comprising 30-50 neurons and each layer having a different architecture (e.g., one layer having three input neurons and another layer having only two input neurons, but only one output neuron), and the use of a ReLU activation function for the output layers of the respective networks uses an identity activation function; further preferably, during training, the collected data points are split into a training set and a verification set, for example, the training set accounts for 90% of the total data points, the verification set accounts for 10%, and then fitting adjustment is performed by using a secondary cost function and a dynamic learning rate; the learning rate is dynamically adjusted according to the fitting condition of the neural network, and is preferably adjusted according to the tolerance value. With this arrangement, a more efficient network can be provided, improving CPU time with different tolerances.
Preferably, when the DNN-deep neural network is adopted for fitting, a sampling algorithm based on a multi-fidelity neural network agent is adopted, wherein the sampling algorithm is provided with two low-fidelity and high-fidelity calculation models, when the calculation is carried out, a group of large low-fidelity data and a group of small high-fidelity data are firstly generated by utilizing the two calculation models, and then two networks, namely a first neural network NN, are constructed on different layers 1 And a second neural network NN 2 Wherein the first neural network NN 1 The correlation between the low fidelity data and the high fidelity data is learned, and additional high fidelity data can be generated according to the correlation; in the first neural network NN 1 After generating the additional high fidelity data, the second neural network NN 2 Training with the additional high fidelity data, both raw and new, as a high fidelity replacement, is then embedded into a classical Markov Chain (MC) sample training model framework to calculate some desired amount of statistics of interest. By adopting the method, the accuracy of neural network simulation can be improved. Wherein the calculation model used comprises:
in the method, in the process of the application,the correlation between low-fidelity data and high-fidelity data is measured, and the method comprises the steps of ++>Andrespectively representing the approximation of the quantity Q (Y) obtained by the low-fidelity and high-fidelity calculation models, Q (Y) being the desired acquired high-fidelity data, with arrow indicating the output of the test data through the response neural network, the corner mark i representing the sampled data point label, Y I 、Y II Is representative of two non-intersecting sets selected from the set of sample points.
In the description of the present application, it should be understood that the terms "upper," "lower," "bottom," "top," "front," "rear," "inner," "outer," "left," "right," and the like indicate or are based on the orientation or positional relationship shown in the drawings, merely to facilitate description of the present application and to simplify the description, and do not indicate or imply that the devices or elements referred to must have a particular orientation, be configured and operated in a particular orientation, and thus should not be construed as limiting the present application.
Although the application herein has been described with reference to particular embodiments, it is to be understood that these embodiments are merely illustrative of the principles and applications of the present application. It is therefore to be understood that numerous modifications may be made to the illustrative embodiments and that other arrangements may be devised without departing from the spirit and scope of the present application as defined by the appended claims. It should be understood that the different dependent claims and the features described herein may be combined in ways other than as described in the original claims. It is also to be understood that features described in connection with separate embodiments may be used in other described embodiments.
Claims (4)
1. A lithium battery thermal runaway layered early warning method based on neural network and data difference is characterized in that an early warning system adopted by the method at least comprises the following steps:
the signal sampling device can acquire an induction signal representing thermal runaway of the battery and comprises two sampling modes of low sampling frequency, high sampling precision and high sampling frequency and low sampling precision;
the central processing unit device can conduct neural network processing according to the induction signals representing the thermal runaway of the battery, predicts to obtain future sampling values, conducts differential processing on the predicted future sampling values, determines a risk level according to the magnitude of the differential values, and outputs execution signals related to the risk level; and an alarm device capable of receiving the execution signal and making an alarm prompt in response to the execution signal;
the early warning method at least comprises the following steps:
s1, signal acquisition is carried out by using a signal sampling device;
step S2, the central processing unit device determines the risk level according to the induction signals acquired by the signal sampling device, and adjusts the sampling mode according to the risk level;
step S3, when the risk level reaches the preset level requirement, the central processing unit device outputs an execution signal related to the risk level, and an alarm prompt is given by using the alarm device;
in the step S1, a low sampling frequency high sampling precision mode is used for signal acquisition;
step S2 specifically comprises the steps of predicting a plurality of sampling values with an equal time interval in the future by using a neural network after obtaining the sampling values from the signal sampling device, performing differential calculation on the predicted plurality of sampling values, determining a risk level as a second level when the differential value is higher than a set first threshold value, and adjusting the sampling mode of the signal sampling device to a high sampling frequency and low sampling precision mode, otherwise, determining the risk level as a first level, and still adopting the low sampling frequency and high sampling precision mode by the signal sampling device; after entering a high sampling frequency low sampling precision mode, performing neural network fitting on the acquired low precision sequence, performing precision lifting processing, taking current data after the precision lifting processing and previous high precision data as input for predicting a subsequent sequence of the neural network, predicting a sampling value sequence at a certain time interval in the future, performing differential calculation, and determining that the risk level is three-level when the differential value is higher than a second threshold value;
step S3 specifically includes that when the central processing unit determines that the risk level is three-level, the central processing unit outputs an execution signal, and an alarm prompt is made by using the alarm device.
2. The method for lithium battery thermal runaway layered early warning based on neural network and data difference as claimed in claim 1, wherein the operation purpose of the difference calculation is to obtain a signal representing the future trend of the sampled value, which includes deriving the end of the prediction curve or differentiating the obtained discrete data, and a forward difference method, a backward difference method or other equivalent difference method can be adopted.
3. The lithium battery thermal runaway layered early warning method based on the neural network and the data difference as claimed in claim 1 is characterized in that the neural network is formed into a 1x50 input layer+50 x50 first full connection layer+50 x50 second full connection layer+50 x1 output layer.
4. The lithium battery thermal runaway layered early warning method based on the neural network and the data difference as claimed in claim 1, wherein the neural network is provided with a feed-forward network of a plurality of layers, each layer comprises 30-50 neurons, the architecture of each layer is different, and the acquired data points are split into a training set and a verification set when training is carried out.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202210562350.1A CN115294745B (en) | 2022-05-23 | 2022-05-23 | Lithium battery thermal runaway layered early warning method based on neural network and data difference |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202210562350.1A CN115294745B (en) | 2022-05-23 | 2022-05-23 | Lithium battery thermal runaway layered early warning method based on neural network and data difference |
Publications (2)
Publication Number | Publication Date |
---|---|
CN115294745A CN115294745A (en) | 2022-11-04 |
CN115294745B true CN115294745B (en) | 2023-09-29 |
Family
ID=83821143
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202210562350.1A Active CN115294745B (en) | 2022-05-23 | 2022-05-23 | Lithium battery thermal runaway layered early warning method based on neural network and data difference |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN115294745B (en) |
Citations (12)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
WO2005078673A1 (en) * | 2004-02-03 | 2005-08-25 | Sys Technologies, Inc. | Remote battery monitoring system having embedded telesensors |
JP2007165040A (en) * | 2005-12-12 | 2007-06-28 | Matsushita Electric Ind Co Ltd | Maintenance control method of secondary battery |
JP2009192311A (en) * | 2008-02-13 | 2009-08-27 | Nippon Telegr & Teleph Corp <Ntt> | Device for predicting generation of thunder, and method therefor, and program |
CN105740578A (en) * | 2016-03-08 | 2016-07-06 | 国网新疆电力公司检修公司 | Maximum entropy-spectrum method applied to disconnecting switch vibration monitoring |
CN107133142A (en) * | 2017-04-18 | 2017-09-05 | 浙江大学 | A kind of monitoring data intellegent sampling method based on association analysis |
CN209071574U (en) * | 2018-04-26 | 2019-07-05 | 桂林电子科技大学 | Battery thermal runaway spreads simulator |
CN111308376A (en) * | 2020-02-24 | 2020-06-19 | 上海蔚来汽车有限公司 | Power battery insulation monitoring method, system and device |
WO2021063136A1 (en) * | 2019-09-30 | 2021-04-08 | 江苏大学 | Data-driven high-precision integrated navigation data fusion method |
CN112946522A (en) * | 2021-02-05 | 2021-06-11 | 四川大学 | On-line monitoring method for short-circuit fault in battery energy storage system caused by low-temperature working condition |
WO2021155851A1 (en) * | 2020-02-07 | 2021-08-12 | 华为技术有限公司 | Neural network circuit and neural network system |
CN113344024A (en) * | 2021-04-22 | 2021-09-03 | 华中科技大学 | Lithium ion battery thermal runaway grading early warning method and early warning system |
CN114312322A (en) * | 2021-12-31 | 2022-04-12 | 中国第一汽车股份有限公司 | Vehicle detection method and device |
Family Cites Families (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
KR101710666B1 (en) * | 2012-12-12 | 2017-02-27 | 한국전자통신연구원 | Apparatus and Method for Monitoring Complex Slope based on Wireless Network |
-
2022
- 2022-05-23 CN CN202210562350.1A patent/CN115294745B/en active Active
Patent Citations (13)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
WO2005078673A1 (en) * | 2004-02-03 | 2005-08-25 | Sys Technologies, Inc. | Remote battery monitoring system having embedded telesensors |
JP2007165040A (en) * | 2005-12-12 | 2007-06-28 | Matsushita Electric Ind Co Ltd | Maintenance control method of secondary battery |
JP2009192311A (en) * | 2008-02-13 | 2009-08-27 | Nippon Telegr & Teleph Corp <Ntt> | Device for predicting generation of thunder, and method therefor, and program |
CN105740578A (en) * | 2016-03-08 | 2016-07-06 | 国网新疆电力公司检修公司 | Maximum entropy-spectrum method applied to disconnecting switch vibration monitoring |
CN107133142A (en) * | 2017-04-18 | 2017-09-05 | 浙江大学 | A kind of monitoring data intellegent sampling method based on association analysis |
CN209071574U (en) * | 2018-04-26 | 2019-07-05 | 桂林电子科技大学 | Battery thermal runaway spreads simulator |
WO2021063136A1 (en) * | 2019-09-30 | 2021-04-08 | 江苏大学 | Data-driven high-precision integrated navigation data fusion method |
WO2021155851A1 (en) * | 2020-02-07 | 2021-08-12 | 华为技术有限公司 | Neural network circuit and neural network system |
CN113255875A (en) * | 2020-02-07 | 2021-08-13 | 华为技术有限公司 | Neural network circuit and neural network system |
CN111308376A (en) * | 2020-02-24 | 2020-06-19 | 上海蔚来汽车有限公司 | Power battery insulation monitoring method, system and device |
CN112946522A (en) * | 2021-02-05 | 2021-06-11 | 四川大学 | On-line monitoring method for short-circuit fault in battery energy storage system caused by low-temperature working condition |
CN113344024A (en) * | 2021-04-22 | 2021-09-03 | 华中科技大学 | Lithium ion battery thermal runaway grading early warning method and early warning system |
CN114312322A (en) * | 2021-12-31 | 2022-04-12 | 中国第一汽车股份有限公司 | Vehicle detection method and device |
Also Published As
Publication number | Publication date |
---|---|
CN115294745A (en) | 2022-11-04 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN105510546B (en) | A kind of biochemical oxygen demand (BOD) BOD intelligent detecting methods based on self-organizing Recurrent RBF Neural Networks | |
CN108898215B (en) | Intelligent sludge bulking identification method based on two-type fuzzy neural network | |
CN107422272B (en) | Intelligent detection device for SOC of power battery of electric automobile | |
CN111625994A (en) | Multi-source information fusion fire prediction method based on dynamic integrated neural network | |
CN110481536B (en) | Control method and device applied to hybrid electric vehicle | |
CN111007399A (en) | Lithium battery state of charge prediction method based on improved generation countermeasure network | |
CN113554200B (en) | Method, system and equipment for predicting voltage inconsistency of power battery | |
CN107192953B (en) | A kind of energy-storage battery soft fault diagnosis method based on GA-Elman | |
CN114580706A (en) | Power financial service wind control method and system based on GRU-LSTM neural network | |
CN112087442B (en) | Time sequence related network intrusion detection method based on attention mechanism | |
CN115291116A (en) | Energy storage battery health state prediction method and device and intelligent terminal | |
CN102567782A (en) | Neural-network-based automobile engine torque estimation method | |
CN110516813A (en) | A method of batteries of electric automobile RDR prediction is carried out based on big data machine learning | |
WO2023226355A1 (en) | Dual-ion battery fault detection method and system based on multi-source perception | |
CN115294745B (en) | Lithium battery thermal runaway layered early warning method based on neural network and data difference | |
CN116699434A (en) | Fuel cell life prediction method based on impedance spectrum relaxation time distribution | |
CN108538301B (en) | Intelligent digital musical instrument based on neural network audio technology | |
CN113791351B (en) | Lithium battery life prediction method based on transfer learning and difference probability distribution | |
CN116503029B (en) | Module data cooperative processing method and system for automatic driving | |
CN114036506A (en) | Method for detecting and defending false data injection attack based on LM-BP neural network | |
CN111626324A (en) | Seabed observation network data heterogeneous analysis integration method based on edge calculation | |
CN115422977A (en) | Radar radiation source signal identification method based on CNN-BLS network | |
CN114757548A (en) | Wind power energy storage equipment adjusting performance evaluation method adopting scene construction | |
CN117031291B (en) | Method and terminal for detecting available battery charging capacity | |
CN117476125B (en) | Dried beancurd stick raffinate recovery data processing system based on data analysis |
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 |