Lithium battery temperature early warning method based on artificial intelligence assisted decision-making
Through artificial intelligence assisting decision-making, combined with machine learning and fuzzy control technology, the problem that traditional lithium battery temperature risk assessment cannot adapt to complex environments is solved, and comprehensive assessment and prediction of lithium battery temperature risks is achieved, which improves safety and reliability.
Patent Information
- Application Number
- CN202411244806.5
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2024-09-06
- Publication Date
- 2024-11-12
- Estimated Expiration
- 2044-09-06
AI Technical Summary
The temperature risk assessment of traditional lithium batteries relies on fixed temperature thresholds, cannot adapt to complex operating environments and different types of batteries, and cannot predict potential temperature anomalies, resulting in passive responses.
Using an artificial intelligence-assisted decision-making method, combined with machine learning and fuzzy control technology, the calculation is carried out by obtaining the operating data of lithium batteries, extracting characteristic parameters, inputting temperature abnormality probability prediction model, defining the input variables and output variables of the fuzzy controller, and performing Fuzzification and fuzzy reasoning, output temperature risk coefficients, and achieve more refined temperature warning and management.
It has achieved a comprehensive assessment and prediction of the temperature risks of lithium batteries, adapted to complex environments and different types of batteries, improved the safety and reliability of lithium batteries, and achieved more refined temperature warning and management.
Smart Images

Figure CN118761017B_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of lithium battery temperature early warning technology, and in particular to a lithium battery temperature early warning method based on artificial intelligence assisted decision-making. Background Art
[0002] As the core power source of modern electronic devices and electric vehicles, the safety of lithium batteries has always been the focus of industry attention. Temperature is a key factor affecting the performance and safety of lithium batteries. Excessive temperature may lead to a decline in battery performance, shortened life, and even cause safety accidents. Therefore, real-time monitoring and evaluation of lithium battery temperature risks are crucial. At present, traditional lithium battery temperature risk assessment mainly relies on setting a fixed temperature threshold. When the battery temperature exceeds the preset threshold, an alarm is triggered. However, the alarm based on the fixed temperature threshold cannot take into account the dynamic environment and complex working conditions of the battery operation, and the fixed temperature threshold is difficult to adapt to batteries of different types and states; on the other hand, it cannot predict potential temperature anomalies and can only respond passively when problems occur. Based on this, in response to the above problems, we designed a lithium battery temperature early warning method based on artificial intelligence-assisted decision-making. Summary of the invention
[0003] The purpose of the present invention is to provide a lithium battery temperature warning method based on artificial intelligence assisted decision-making, which not only realizes the comprehensive assessment and prediction of lithium battery temperature risks by combining machine learning and fuzzy control technology, but also can adapt to complex operating environments and different types of batteries. At the same time, the present invention can achieve more refined temperature warning and management by setting multiple risk levels, greatly improving the safety and reliability of lithium battery use.
[0004] The embodiments of the present invention are implemented by the following technical solutions:
[0005] A lithium battery temperature early warning method based on artificial intelligence assisted decision-making, the method comprising the following steps:
[0006] Acquire the operating data of the lithium battery, preprocess the operating data of the lithium battery, and perform feature extraction on the preprocessed operating data of the lithium battery to obtain characteristic parameters of the lithium battery;
[0007] Input the characteristic parameters of the lithium battery into the trained lithium battery temperature anomaly probability prediction model for calculation, and output the temperature anomaly probability of the lithium battery;
[0008] The input variables of the preset fuzzy controller are defined as the characteristic parameters and temperature anomaly probability of the lithium battery, and the output variable of the preset fuzzy controller is defined as the temperature risk coefficient of the lithium battery. Based on the set fuzzy rules, the input variables are fuzzified, fuzzy inferenced and defuzzified in sequence, and the temperature risk coefficient of the lithium battery is output;
[0009] A temperature warning plan for lithium batteries is implemented based on the temperature risk coefficient of the lithium battery.
[0010] Optionally, the characteristic parameters of the lithium battery specifically include: temperature parameters, current parameters, voltage parameters and thermal imaging image data of the lithium battery.
[0011] Optionally, the step of inputting the characteristic parameters of the lithium battery into a trained lithium battery temperature anomaly probability prediction model for calculation is specifically as follows: inputting the thermal imaging image data of the lithium battery into a trained lithium battery temperature anomaly probability prediction model for calculation, wherein the training process of the lithium battery temperature anomaly probability prediction model is as follows:
[0012] Obtain historical thermal imaging image data of lithium batteries;
[0013] Label the status of historical thermal imaging image data of lithium batteries, perform data enhancement on the historical thermal imaging image data of lithium batteries, and divide them into training sets and test sets;
[0014] A lithium battery temperature anomaly probability prediction model is constructed based on the VGG neural network. The training set is input into the lithium battery temperature anomaly probability prediction model for calculation, and the loss value between the prediction result of the training set and the actual result is calculated. The parameters of the lithium battery temperature anomaly probability prediction model are iteratively updated according to the loss value until the maximum number of iterations is reached, and the training of the lithium battery temperature anomaly probability prediction model is completed.
[0015] The iterative updating of the parameters of the lithium battery temperature anomaly probability prediction model according to the loss value is specifically based on the iterative updating of the parameters of the lithium battery temperature anomaly probability prediction model based on the stochastic gradient descent optimizer:
[0016] Initialize the parameters of the lithium battery temperature abnormality probability prediction model;
[0017] Randomly divide the training set into multiple training subsets of equal size;
[0018] Input the training subset into the lithium battery temperature abnormality probability prediction model to obtain the prediction result;
[0019] Calculate the total loss between the predicted result and the actual result based on the defined loss function, and calculate the gradient of the total loss with respect to the parameters of the lithium battery temperature abnormality probability prediction model;
[0020] The parameters of the lithium battery temperature anomaly probability prediction model are iteratively updated through the stochastic gradient descent algorithm until the preset maximum number of iterations is reached, and the trained lithium battery temperature anomaly probability prediction model is output.
[0021] Optionally, the loss function defined above is calculated as follows:
[0022]
[0023] in, is the loss function, is the total number of samples, For the sample , , For sample The real label, Samples predicted by the lithium battery temperature abnormality probability prediction model The probability of temperature anomaly, is the modulation factor of the focus loss, For sample The deep features of For Category The characteristic center of For sample and Indicator of whether they belong to the same class, For sample and The characteristic distance, is the boundary parameter of contrast loss, is the weight parameter of the lithium battery temperature abnormality probability prediction model, It is the weight parameter set of the lithium battery temperature abnormality probability prediction model.
[0024] Optionally, the input variables of the preset fuzzy controller are specifically defined as: temperature parameters, current parameters, voltage parameters and temperature abnormality probability of the lithium battery; the output variables are specifically defined as: temperature risk coefficient of the lithium battery; at the same time, the first fuzzy set, the second fuzzy set and the third fuzzy set corresponding to the input variables are determined respectively according to the input variables of the fuzzy controller, and the corresponding membership function is defined for each fuzzy set.
[0025] Optionally, the first fuzzy set, the second fuzzy set and the third fuzzy set corresponding to the input variables are determined respectively according to the input variables of the fuzzy controller, and a corresponding membership function is defined for each fuzzy set, which is specifically:
[0026] The membership functions corresponding to the first fuzzy set, the second fuzzy set and the third fuzzy set of the temperature parameter of the lithium battery are respectively:
[0027]
[0028] The membership functions corresponding to the first fuzzy set, the second fuzzy set and the third fuzzy set of the current parameter of the lithium battery are:
[0029]
[0030] The membership functions corresponding to the first fuzzy set, the second fuzzy set and the third fuzzy set of the voltage parameter of the lithium battery are:
[0031]
[0032] The temperature anomaly probability of the lithium battery, the membership functions corresponding to the first fuzzy set, the second fuzzy set and the third fuzzy set are respectively:
[0033]
[0034] in, are the membership functions corresponding to the first fuzzy set of the temperature parameter, current parameter, voltage parameter and temperature anomaly probability of the lithium battery, respectively. The membership functions corresponding to the second fuzzy set of the temperature parameters, current parameters, voltage parameters and temperature anomaly probability of lithium batteries are respectively: are the membership functions corresponding to the third fuzzy set of the temperature parameter, current parameter, voltage parameter and temperature anomaly probability of the lithium battery, respectively. They are the temperature parameters, current parameters, voltage parameters and temperature anomaly probability of lithium batteries. To adjust the parameters, are respectively the first temperature, the first current, the first voltage and the first temperature abnormality probability of the lithium battery, are respectively the second temperature, the second current, the second voltage and the second temperature abnormality probability of the lithium battery, are the third temperature, third current, third voltage and third temperature abnormality probability of the lithium battery, are the minimum thresholds of the temperature, current, voltage and temperature abnormality probability of the lithium battery, are the temperature, current, voltage and the first intermediate threshold of the temperature abnormality probability of the lithium battery, are the temperature, current, voltage and the second intermediate threshold of the temperature abnormality probability of the lithium battery, They are respectively the maximum thresholds of the temperature, current, voltage and temperature abnormality probability of the lithium battery.
[0035] Optionally, the fuzzy rules of the preset fuzzy controller are specifically defined as:
[0036] Real-time acquisition of the temperature parameters, current parameters, voltage parameters of the lithium battery and the temperature anomaly probability output by the lithium battery temperature anomaly probability prediction model;
[0037] Determine respectively whether the temperature parameter, current parameter, voltage parameter and temperature anomaly probability of the lithium battery are in the first fuzzy set, the second fuzzy set or the third fuzzy set;
[0038] If the temperature parameter of the lithium battery is in the first fuzzy set, the current parameter is in the first fuzzy set, the voltage parameter is in the first fuzzy set, and the temperature anomaly probability is in the first fuzzy set, it is determined to be the first risk level;
[0039] If the temperature parameter of the lithium battery is in the second fuzzy set, the current parameter is in the second fuzzy set, the voltage parameter is in the second fuzzy set, and the temperature anomaly probability is in the first fuzzy set or the second fuzzy set, it is determined to be the second risk level;
[0040] If the temperature parameter of the lithium battery is in the second fuzzy set, the current parameter is in the second fuzzy set, the voltage parameter is in the second fuzzy set, and the temperature anomaly probability is in the third fuzzy set, it is determined to be the third risk level;
[0041] If the temperature parameter of the lithium battery is in the third fuzzy set, and the current parameter is in the second fuzzy set or the third fuzzy set, and the voltage parameter is in the second fuzzy set or the third fuzzy set, and the temperature anomaly probability is in the second fuzzy set, it is determined to be the fourth risk level;
[0042] If the temperature parameter of the lithium battery is in the third fuzzy set, the current parameter is in the third fuzzy set, the voltage parameter is in the third fuzzy set, and the temperature anomaly probability is in the third fuzzy set, it is determined to be the fifth risk level;
[0043] If the temperature parameter of the lithium battery is in the third fuzzy set, the current parameter is in the third fuzzy set, the voltage parameter is in the third fuzzy set, and the temperature anomaly probability is in the third fuzzy set, and the membership value of each parameter in its respective fuzzy set is greater than the preset membership threshold, then it is determined to be the sixth risk level;
[0044] If the temperature parameter of the lithium battery is in the first fuzzy set, the current parameter is in the third fuzzy set, the voltage parameter is in the first fuzzy set or the second fuzzy set, and the temperature anomaly probability is in the second fuzzy set or the third fuzzy set, it is determined to be the third risk level;
[0045] If the temperature parameter of the lithium battery is in the second fuzzy set or the third fuzzy set, the current parameter is in the first fuzzy set, the voltage parameter is in the third fuzzy set, and the temperature anomaly probability is in the second fuzzy set or the third fuzzy set, it is determined to be the fourth risk level;
[0046] If any parameter is in the third fuzzy set, and the other parameters are in the first or second fuzzy set, it is determined to be the third risk level;
[0047] If three or more parameters are in the third fuzzy set, it is determined to be the fifth risk level.
[0048] Optionally, the temperature risk coefficient of the output lithium battery is calculated by the following formula:
[0049]
[0050]
[0051] in, is the lithium battery temperature risk coefficient, is the membership value of the temperature parameter, current parameter, voltage parameter and temperature anomaly probability of the lithium battery in their respective corresponding third fuzzy sets, is the dynamic weight of the temperature parameter, current parameter, voltage parameter and temperature anomaly probability of the lithium battery, is the number of parameters in the third fuzzy set, is the lithium battery operating time, is the base time, is the lithium battery temperature risk coefficient at the previous moment.
[0052] The technical solution of the present invention has at least the following advantages and beneficial effects:
[0053] By combining machine learning and fuzzy control technology, the present invention not only realizes the comprehensive assessment and prediction of lithium battery temperature risks, but also can adapt to complex operating environments and different types of batteries. At the same time, by setting multiple risk levels, the present invention can achieve more refined temperature warning and management, greatly improving the safety and reliability of lithium battery use. BRIEF DESCRIPTION OF THE DRAWINGS
[0054] Figure 1 A schematic flow chart of a lithium battery temperature early warning method based on artificial intelligence assisted decision-making provided by the present invention. DETAILED DESCRIPTION
[0055] In order to make the purpose, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Generally, the components of the embodiments of the present invention described and shown in the drawings here can be arranged and designed in various different configurations.
[0056] Reference Figure 1 As shown, Figure 1 A schematic flow chart of a lithium battery temperature early warning method based on artificial intelligence assisted decision-making provided by the present invention.
[0057] In one embodiment, the present invention provides: a lithium battery temperature early warning method based on artificial intelligence assisted decision-making, the method comprising the steps of:
[0058] Acquire the operating data of the lithium battery, preprocess the operating data of the lithium battery, and perform feature extraction on the preprocessed operating data of the lithium battery to obtain characteristic parameters of the lithium battery;
[0059] Input the characteristic parameters of the lithium battery into the trained lithium battery temperature anomaly probability prediction model for calculation, and output the temperature anomaly probability of the lithium battery;
[0060] The input variables of the preset fuzzy controller are defined as the characteristic parameters and temperature anomaly probability of the lithium battery, and the output variable of the preset fuzzy controller is defined as the temperature risk coefficient of the lithium battery. Based on the set fuzzy rules, the input variables are fuzzified, fuzzy inferenced and defuzzified in sequence, and the temperature risk coefficient of the lithium battery is output;
[0061] A temperature warning plan for lithium batteries is implemented based on the temperature risk coefficient of the lithium battery.
[0062] In the application of this embodiment, firstly, through feature extraction and temperature anomaly probability prediction, the influence of various factors on temperature risk can be more comprehensively considered; secondly, the introduction of the fuzzy controller makes risk assessment more flexible and intelligent, and can handle uncertainty and ambiguity; furthermore, this embodiment can realize predictive assessment of risks, rather than just passive response; finally, by setting different risk levels, this embodiment can realize more refined temperature warning and management, thereby improving the safety and efficiency of battery use.
[0063] In this embodiment, the characteristic parameters of the lithium battery specifically include: temperature parameters, current parameters, voltage parameters and thermal imaging image data of the lithium battery.
[0064] In implementation, this embodiment uses thermal imaging image data as input in machine learning applications; and in the subsequent process of building a fuzzy controller, the temperature parameters, current parameters, voltage parameters of the lithium battery and the probability of temperature anomaly of the lithium battery output by machine learning are used as input variables. On the one hand, by using thermal imaging image data, the machine learning model in this embodiment can capture subtle changes in the temperature distribution on the battery surface, thereby more accurately predicting the probability of temperature anomaly; on the other hand, the fuzzy controller constructed in this embodiment comprehensively considers multiple parameters such as temperature, current, voltage, and the probability of temperature anomaly output by the machine learning model, making the temperature risk assessment of lithium batteries more comprehensive and reliable.
[0065] Specifically, the characteristic parameters of the lithium battery are input into the trained lithium battery temperature anomaly probability prediction model for calculation, which is specifically: the thermal imaging image data of the lithium battery is input into the trained lithium battery temperature anomaly probability prediction model for calculation, wherein the training process of the lithium battery temperature anomaly probability prediction model is as follows:
[0066] Obtain historical thermal imaging image data of lithium batteries;
[0067] Label the status of historical thermal imaging image data of lithium batteries, perform data enhancement on the historical thermal imaging image data of lithium batteries, and divide them into training sets and test sets;
[0068] A lithium battery temperature anomaly probability prediction model is constructed based on the VGG neural network. The training set is input into the lithium battery temperature anomaly probability prediction model for calculation, and the loss value between the prediction result of the training set and the actual result is calculated. The parameters of the lithium battery temperature anomaly probability prediction model are iteratively updated according to the loss value until the maximum number of iterations is reached, and the training of the lithium battery temperature anomaly probability prediction model is completed.
[0069] The iterative updating of the parameters of the lithium battery temperature anomaly probability prediction model according to the loss value is specifically based on the iterative updating of the parameters of the lithium battery temperature anomaly probability prediction model based on the stochastic gradient descent optimizer:
[0070] Initialize the parameters of the lithium battery temperature abnormality probability prediction model;
[0071] Randomly divide the training set into multiple training subsets of equal size;
[0072] Input the training subset into the lithium battery temperature abnormality probability prediction model to obtain the prediction result;
[0073] Calculate the total loss between the predicted result and the actual result based on the defined loss function, and calculate the gradient of the total loss with respect to the parameters of the lithium battery temperature abnormality probability prediction model;
[0074] The parameters of the lithium battery temperature anomaly probability prediction model are iteratively updated through the stochastic gradient descent algorithm until the preset maximum number of iterations is reached, and the trained lithium battery temperature anomaly probability prediction model is output.
[0075] The loss function defined in this embodiment has a specific calculation formula as follows:
[0076]
[0077] in, is the loss function, is the total number of samples, For the sample , , For sample The real label, Samples predicted by the lithium battery temperature abnormality probability prediction model The probability of temperature anomaly, is the modulation factor of the focus loss, For sample The deep features of For Category The characteristic center of For sample and Indicator of whether they belong to the same class, For sample and The characteristic distance, is the boundary parameter of contrast loss, is the weight parameter of the lithium battery temperature abnormality probability prediction model, It is the weight parameter set of the lithium battery temperature abnormality probability prediction model.
[0078] This embodiment iteratively updates the parameters of the lithium battery temperature abnormality probability prediction model through a stochastic gradient descent algorithm, and its calculation formula is:
[0079]
[0080]
[0081] in, For the current moment The parameter vector of is the updated parameter vector, is the learning rate, is the loss function About parameters At the current point The gradient of is the current mini-batch, i.e. the training subset, is the size of mini-batch, For the The loss function for each sample.
[0082] In the specific implementation process of this embodiment, the input variables of the preset fuzzy controller are specifically defined as: the temperature parameters, current parameters, voltage parameters and temperature abnormality probability of the lithium battery; the output variables are specifically defined as: the temperature risk coefficient of the lithium battery; at the same time, the first fuzzy set, the second fuzzy set and the third fuzzy set corresponding to the input variables are determined respectively according to the input variables of the fuzzy controller, and the corresponding membership function is defined for each fuzzy set.
[0083] The first fuzzy set, the second fuzzy set and the third fuzzy set corresponding to the input variables are determined according to the input variables of the fuzzy controller, and the corresponding membership function is defined for each fuzzy set, which is specifically:
[0084] The membership functions corresponding to the first fuzzy set, the second fuzzy set and the third fuzzy set of the temperature parameter of the lithium battery are respectively:
[0085]
[0086] The membership functions corresponding to the first fuzzy set, the second fuzzy set and the third fuzzy set of the current parameter of the lithium battery are:
[0087]
[0088] The membership functions corresponding to the first fuzzy set, the second fuzzy set and the third fuzzy set of the voltage parameter of the lithium battery are:
[0089]
[0090] The temperature anomaly probability of the lithium battery, the membership functions corresponding to the first fuzzy set, the second fuzzy set and the third fuzzy set are respectively:
[0091]
[0092] in, are the membership functions corresponding to the first fuzzy set of the temperature parameter, current parameter, voltage parameter and temperature anomaly probability of the lithium battery, respectively. The membership functions corresponding to the second fuzzy set of the temperature parameters, current parameters, voltage parameters and temperature anomaly probability of lithium batteries are respectively: are the membership functions corresponding to the third fuzzy set of the temperature parameter, current parameter, voltage parameter and temperature anomaly probability of the lithium battery, respectively. They are the temperature parameters, current parameters, voltage parameters and temperature anomaly probability of lithium batteries. To adjust the parameters, are respectively the first temperature, the first current, the first voltage and the first temperature abnormality probability of the lithium battery, are respectively the second temperature, the second current, the second voltage and the second temperature abnormality probability of the lithium battery, are the third temperature, third current, third voltage and third temperature abnormality probability of the lithium battery, are the minimum thresholds of the temperature, current, voltage and temperature abnormality probability of the lithium battery, are the temperature, current, voltage and the first intermediate threshold of the temperature abnormality probability of the lithium battery, are the temperature, current, voltage and the second intermediate threshold of the temperature abnormality probability of the lithium battery, They are respectively the maximum thresholds of the temperature, current, voltage and temperature abnormality probability of the lithium battery.
[0093] In implementation, this embodiment defines the first fuzzy set, the second fuzzy set and the third fuzzy set of the input variables, which specifically correspond to the low, medium and high intervals respectively. Therefore, the first temperature actually designed in this embodiment is less than the second temperature less than the third temperature, the first current is less than the second current less than the third current, the first voltage is less than the second voltage less than the third voltage, and the first temperature abnormal probability is less than the second temperature abnormal probability less than the third temperature abnormal probability. Specifically, the second intermediate threshold is greater than the first intermediate threshold.
[0094] The fuzzy rules of the preset fuzzy controller are specifically defined as:
[0095] Real-time acquisition of the temperature parameters, current parameters, voltage parameters of the lithium battery and the temperature anomaly probability output by the lithium battery temperature anomaly probability prediction model;
[0096] Determine respectively whether the temperature parameter, current parameter, voltage parameter and temperature anomaly probability of the lithium battery are in the first fuzzy set, the second fuzzy set or the third fuzzy set;
[0097] If the temperature parameter of the lithium battery is in the first fuzzy set, the current parameter is in the first fuzzy set, the voltage parameter is in the first fuzzy set, and the temperature anomaly probability is in the first fuzzy set, it is determined to be the first risk level;
[0098] If the temperature parameter of the lithium battery is in the second fuzzy set, the current parameter is in the second fuzzy set, the voltage parameter is in the second fuzzy set, and the temperature anomaly probability is in the first fuzzy set or the second fuzzy set, it is determined to be the second risk level;
[0099] If the temperature parameter of the lithium battery is in the second fuzzy set, the current parameter is in the second fuzzy set, the voltage parameter is in the second fuzzy set, and the temperature anomaly probability is in the third fuzzy set, it is determined to be the third risk level;
[0100] If the temperature parameter of the lithium battery is in the third fuzzy set, and the current parameter is in the second fuzzy set or the third fuzzy set, and the voltage parameter is in the second fuzzy set or the third fuzzy set, and the temperature anomaly probability is in the second fuzzy set, it is determined to be the fourth risk level;
[0101] If the temperature parameter of the lithium battery is in the third fuzzy set, the current parameter is in the third fuzzy set, the voltage parameter is in the third fuzzy set, and the temperature anomaly probability is in the third fuzzy set, it is determined to be the fifth risk level;
[0102] If the temperature parameter of the lithium battery is in the third fuzzy set, the current parameter is in the third fuzzy set, the voltage parameter is in the third fuzzy set, and the temperature anomaly probability is in the third fuzzy set, and the membership value of each parameter in its respective fuzzy set is greater than the preset membership threshold, then it is determined to be the sixth risk level;
[0103] If the temperature parameter of the lithium battery is in the first fuzzy set, the current parameter is in the third fuzzy set, the voltage parameter is in the first fuzzy set or the second fuzzy set, and the temperature anomaly probability is in the second fuzzy set or the third fuzzy set, it is determined to be the third risk level;
[0104] If the temperature parameter of the lithium battery is in the second fuzzy set or the third fuzzy set, the current parameter is in the first fuzzy set, the voltage parameter is in the third fuzzy set, and the temperature anomaly probability is in the second fuzzy set or the third fuzzy set, it is determined to be the fourth risk level;
[0105] If any parameter is in the third fuzzy set, and the other parameters are in the first or second fuzzy set, it is determined to be the third risk level;
[0106] If three or more parameters are in the third fuzzy set, it is determined to be the fifth risk level.
[0107] Specifically, the first risk level is determined as a low risk level, the second risk level is determined as a medium-low risk level, the third risk level is determined as a medium risk level, the fourth risk level is determined as a medium-high risk level, the fifth risk level is determined as a high risk level, and the sixth risk level is determined as an extremely high risk level. In the determination of the sixth risk level, the preset membership threshold is specifically set to 0.9. It can be understood that the specific threshold here can be set according to actual conditions. If the membership of each parameter in the corresponding third fuzzy set is greater than the preset membership threshold, it is determined to be an extremely high risk. In addition, if any parameter is in the third fuzzy set, and the other parameters are in the first fuzzy set or the second fuzzy set, it is determined to be at least the third risk level.
[0108] In this embodiment, based on the fuzzy set defined in this embodiment and its corresponding membership function, according to the above fuzzy rules, the temperature risk coefficient of the lithium battery is output through defuzzification processing, and its calculation formula is:
[0109]
[0110]
[0111] in, is the lithium battery temperature risk coefficient, is the membership value of the temperature parameter, current parameter, voltage parameter and temperature anomaly probability of the lithium battery in their respective corresponding third fuzzy sets, is the dynamic weight of the temperature parameter, current parameter, voltage parameter and temperature anomaly probability of the lithium battery, is the number of parameters in the third fuzzy set, is the lithium battery operating time, is the base time, is the lithium battery temperature risk coefficient at the previous moment.
[0112] Specifically, this embodiment can delete , to more directly reflect the current state, this embodiment considers the calculation formula here The purpose is to help capture the cumulative effect of lithium battery temperature risks. Some risks may take some time to manifest, or persistent minor risks may accumulate into major risks. And by comparing current risks with historical risks, it is easier to identify rising or falling trends in risks.
[0113] More specifically, this embodiment defines the early warning logic of R: If R < 0.2, it is identified as the first level (low risk): maintain the normal operation of the lithium battery system, and perform a routine temperature check at a set interval, with the initial setting time being once every 4 hours. If 0.2≤R<0.4, it is identified as the second level (medium-low risk): increase the monitoring frequency of the lithium battery temperature to once every 2 hours, and record the operating data of the lithium battery in detail, including temperature fluctuations, load changes, etc., and notify the technicians to pay attention to the system status of the lithium battery. If 0.4≤R<0.6, it is identified as the third level (medium risk): further increase the monitoring frequency of the lithium battery temperature to once every 1 hour, and start the auxiliary cooling equipment at the same time to enhance the heat dissipation capacity of the lithium battery, reduce the charge and discharge current of the lithium battery to 90% of the rated value, and notify the technicians to pay close attention to the system status of the lithium battery and be ready to intervene at any time. If 0.6≤R<0.8, it is identified as Level 4 (medium-high risk): monitor the temperature of the lithium battery once every 10 minutes, reduce the charge and discharge current of the lithium battery to 70% of the rated value, start all available cooling systems, maximize the heat dissipation efficiency of the lithium battery, and technicians are on standby to prepare isolation measures to prevent the situation from further deteriorating. If 0.8≤R<0.9, it is identified as Level 5 (high risk): monitor the temperature of the lithium battery once every 1 minute, start the emergency cooling equipment, and safety personnel including the fire department are on site to prepare to disconnect the affected lithium battery module and ensure the safety of personnel. If R≥0.9, it is identified as Level 6 (extremely high risk): immediately stop all charging and discharging activities, disconnect all connections between the battery and the external system, immediately evacuate all non-essential personnel in the surrounding area, the fire department will arrive immediately, start the flame retardant materials and fire equipment, and perform physical isolation or suppression measures. It can be understood that the temperature risk warning measures for lithium batteries provided in this embodiment can be adjusted according to the actual application scenario.
[0114] The above are only preferred embodiments of the present invention and are not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.
Claims
1. A lithium battery temperature early warning method based on artificial intelligence assisted decision-making, characterized in that: The steps of the method include: Acquire the operating data of the lithium battery, preprocess the operating data of the lithium battery, and perform feature extraction on the preprocessed operating data of the lithium battery to obtain characteristic parameters of the lithium battery; Input the characteristic parameters of the lithium battery into the trained lithium battery temperature anomaly probability prediction model for calculation, and output the temperature anomaly probability of the lithium battery; The input variables of the preset fuzzy controller are defined as the characteristic parameters and temperature anomaly probability of the lithium battery, and the output variable of the preset fuzzy controller is defined as the temperature risk coefficient of the lithium battery. Based on the set fuzzy rules, the input variables are fuzzified, fuzzy inferenced and defuzzified in sequence, and the temperature risk coefficient of the lithium battery is output; Execute the temperature warning plan for lithium batteries based on the temperature risk coefficient of lithium batteries; The characteristic parameters of the lithium battery specifically include: temperature parameters, current parameters, voltage parameters and thermal imaging image data of the lithium battery; The step of inputting the characteristic parameters of the lithium battery into the trained lithium battery temperature anomaly probability prediction model for calculation is specifically as follows: inputting the thermal imaging image data of the lithium battery into the trained lithium battery temperature anomaly probability prediction model for calculation, wherein the training process of the lithium battery temperature anomaly probability prediction model is as follows: Obtain historical thermal imaging image data of lithium batteries; Label the status of historical thermal imaging image data of lithium batteries, perform data enhancement on the historical thermal imaging image data of lithium batteries, and divide them into training sets and test sets; A lithium battery temperature anomaly probability prediction model is constructed based on the VGG neural network. The training set is input into the lithium battery temperature anomaly probability prediction model for calculation, and the loss value between the prediction result of the training set and the actual result is calculated. The parameters of the lithium battery temperature anomaly probability prediction model are iteratively updated according to the loss value until the maximum number of iterations is reached, and the training of the lithium battery temperature anomaly probability prediction model is completed. The iterative updating of the parameters of the lithium battery temperature anomaly probability prediction model according to the loss value is specifically based on the iterative updating of the parameters of the lithium battery temperature anomaly probability prediction model based on the stochastic gradient descent optimizer: Initialize the parameters of the lithium battery temperature abnormality probability prediction model; Randomly divide the training set into multiple training subsets of equal size; Input the training subset into the lithium battery temperature abnormality probability prediction model to obtain the prediction result; The total loss between the predicted result and the actual result is calculated based on the defined loss function, and the gradient of the total loss with respect to the parameters of the lithium battery temperature abnormality probability prediction model is calculated; The parameters of the lithium battery temperature anomaly probability prediction model are iteratively updated through the stochastic gradient descent algorithm until the preset maximum number of iterations is reached, and the trained lithium battery temperature anomaly probability prediction model is output.
2. The lithium battery temperature early warning method based on artificial intelligence assisted decision-making according to claim 1 is characterized in that: The loss function defined above has a specific calculation formula: in, is the loss function, is the total number of samples, For the sample , , For sample The real label, Samples predicted by the lithium battery temperature abnormality probability prediction model The probability of temperature anomaly, is the modulation factor of the focus loss, For sample The deep features of For Category The characteristic center of For sample and Indicator of whether they belong to the same class, For sample and The characteristic distance of is the boundary parameter of contrast loss, is the weight parameter of the lithium battery temperature abnormality probability prediction model, It is the weight parameter set of the lithium battery temperature abnormality probability prediction model.
3. The lithium battery temperature early warning method based on artificial intelligence assisted decision-making according to claim 2 is characterized in that: The input variables of the preset fuzzy controller are specifically defined as: temperature parameters, current parameters, voltage parameters of the lithium battery and the probability of abnormal temperature of the lithium battery; The output variable is specifically defined as: the temperature risk coefficient of the lithium battery; at the same time, the first fuzzy set, the second fuzzy set and the third fuzzy set corresponding to the input variables are determined according to the input variables of the fuzzy controller, and the corresponding membership function is defined for each fuzzy set.
4. The lithium battery temperature early warning method based on artificial intelligence assisted decision-making according to claim 3 is characterized in that: The first fuzzy set, the second fuzzy set and the third fuzzy set corresponding to the input variables are determined according to the input variables of the fuzzy controller, and the corresponding membership function is defined for each fuzzy set, which is specifically: The membership functions corresponding to the first fuzzy set, the second fuzzy set and the third fuzzy set of the temperature parameter of the lithium battery are respectively: The membership functions corresponding to the first fuzzy set, the second fuzzy set and the third fuzzy set of the current parameter of the lithium battery are: The membership functions corresponding to the first fuzzy set, the second fuzzy set and the third fuzzy set of the voltage parameter of the lithium battery are: The temperature anomaly probability of the lithium battery, the membership functions corresponding to the first fuzzy set, the second fuzzy set and the third fuzzy set are respectively: in, are the membership functions corresponding to the first fuzzy set of the temperature parameter, current parameter, voltage parameter and temperature anomaly probability of the lithium battery, respectively. The membership functions corresponding to the second fuzzy set of the temperature parameters, current parameters, voltage parameters and temperature anomaly probability of lithium batteries are respectively: are the membership functions corresponding to the third fuzzy set of the temperature parameter, current parameter, voltage parameter and temperature anomaly probability of the lithium battery, respectively. They are the temperature parameters, current parameters, voltage parameters and temperature anomaly probability of lithium batteries. To adjust the parameters, are respectively the first temperature, the first current, the first voltage and the first temperature abnormality probability of the lithium battery, are respectively the second temperature, the second current, the second voltage and the second temperature abnormality probability of the lithium battery, are the third temperature, third current, third voltage and third temperature abnormality probability of the lithium battery, are the minimum thresholds of the temperature, current, voltage and temperature abnormality probability of the lithium battery, are the temperature, current, voltage and the first intermediate threshold of the temperature abnormality probability of the lithium battery, are the temperature, current, voltage and the second intermediate threshold of the temperature abnormality probability of the lithium battery, They are respectively the maximum thresholds of the temperature, current, voltage and temperature abnormality probability of the lithium battery.
5. The lithium battery temperature early warning method based on artificial intelligence assisted decision-making according to claim 4 is characterized in that: The fuzzy rules of the preset fuzzy controller are specifically defined as: Real-time acquisition of the temperature parameters, current parameters, voltage parameters of the lithium battery and the temperature anomaly probability output by the lithium battery temperature anomaly probability prediction model; Determine respectively whether the temperature parameter, current parameter, voltage parameter and temperature anomaly probability of the lithium battery are in the first fuzzy set, the second fuzzy set or the third fuzzy set; If the temperature parameter of the lithium battery is in the first fuzzy set, the current parameter is in the first fuzzy set, the voltage parameter is in the first fuzzy set, and the temperature anomaly probability is in the first fuzzy set, it is determined to be the first risk level; If the temperature parameter of the lithium battery is in the second fuzzy set, the current parameter is in the second fuzzy set, the voltage parameter is in the second fuzzy set, and the temperature anomaly probability is in the first fuzzy set or the second fuzzy set, it is determined to be the second risk level; If the temperature parameter of the lithium battery is in the second fuzzy set, the current parameter is in the second fuzzy set, the voltage parameter is in the second fuzzy set, and the temperature anomaly probability is in the third fuzzy set, it is determined to be the third risk level; If the temperature parameter of the lithium battery is in the third fuzzy set, and the current parameter is in the second fuzzy set or the third fuzzy set, and the voltage parameter is in the second fuzzy set or the third fuzzy set, and the temperature anomaly probability is in the second fuzzy set, it is determined to be the fourth risk level; If the temperature parameter of the lithium battery is in the third fuzzy set, the current parameter is in the third fuzzy set, the voltage parameter is in the third fuzzy set, and the temperature anomaly probability is in the third fuzzy set, it is determined to be the fifth risk level; If the temperature parameter of the lithium battery is in the third fuzzy set, the current parameter is in the third fuzzy set, the voltage parameter is in the third fuzzy set, and the temperature anomaly probability is in the third fuzzy set, and the membership value of each parameter in its respective fuzzy set is greater than the preset membership threshold, then it is determined to be the sixth risk level; If the temperature parameter of the lithium battery is in the first fuzzy set, the current parameter is in the third fuzzy set, the voltage parameter is in the first fuzzy set or the second fuzzy set, and the temperature anomaly probability is in the second fuzzy set or the third fuzzy set, it is determined to be the third risk level; If the temperature parameter of the lithium battery is in the second fuzzy set or the third fuzzy set, the current parameter is in the first fuzzy set, the voltage parameter is in the third fuzzy set, and the temperature anomaly probability is in the second fuzzy set or the third fuzzy set, it is determined to be the fourth risk level; If any parameter is in the third fuzzy set, and the other parameters are in the first or second fuzzy set, it is determined to be the third risk level; If three or more parameters are in the third fuzzy set, it is determined to be the fifth risk level.
6. The lithium battery temperature early warning method based on artificial intelligence assisted decision-making according to claim 5 is characterized in that: The temperature risk coefficient of the output lithium battery is calculated as follows: in, is the lithium battery temperature risk coefficient, is the membership value of the temperature parameter, current parameter, voltage parameter and temperature anomaly probability of the lithium battery in their respective corresponding third fuzzy sets, is the dynamic weight of the temperature parameter, current parameter, voltage parameter and temperature anomaly probability of the lithium battery, is the number of parameters in the third fuzzy set, is the lithium battery operating time, is the base time, is the lithium battery temperature risk coefficient at the previous moment.
