CN-121983686-A - Lithium battery thermal runaway alarm method based on multi-mode sensing

CN121983686ACN 121983686 ACN121983686 ACN 121983686ACN-121983686-A

Abstract

The invention provides a lithium battery thermal runaway alarm method based on multi-mode sensing, which comprises the steps of collecting temperature data outside a battery unit and temperature data inside the battery unit in real time, dynamically correcting the temperature data inside the battery unit based on the temperature data outside the battery unit to eliminate external environment interference, taking the corrected temperature data inside the battery unit as key input and cooperatively integrating with other multi-mode sensing data, and driving a hierarchical alarm response mechanism based on the integrated data to match risk severity. The method still maintains high robustness under complex working conditions, effectively reduces false alarm rate, delays thermal runaway chain reaction, and provides full-period guarantee for battery safety management.

Inventors

GUI WANGSHENG
LIU CHEN
WANG DINGGUO
HUANG FUKAI
LI XIANGLIN
Yan Denghe

Assignees

安徽合湃新能源科技有限公司

Dates

Publication Date: 20260505
Application Date: 20251223

Claims (10)

1. A lithium battery thermal runaway alarming method based on multi-mode sensing is characterized by comprising the following steps: S1, acquiring temperature data outside a battery unit and temperature data inside the battery unit in real time; S2, dynamically correcting the temperature data in the battery unit based on the temperature data outside the battery unit so as to eliminate external environment interference; s3, taking the corrected temperature data in the battery unit as key input and carrying out cooperative integration with other multi-mode sensing data; S4, driving a hierarchical alarm response mechanism based on the integrated data to match the risk severity.
2. The multi-mode sensing-based lithium battery thermal runaway alarm method of claim 1, wherein the dynamic correction is implemented by a temperature compensation algorithm that calculates an environmental thermal interference factor using the external temperature data of the battery cell and applies the environmental thermal interference factor to the internal temperature data of the battery cell in real time to generate a correction value, the temperature compensation algorithm creating a mathematical compensation model based on the thermal conductivity characteristics of the battery cell.
3. The multi-mode sensing-based lithium battery thermal runaway alarm method of claim 2, wherein the temperature compensation algorithm adopts a linear regression model as a mathematical compensation model, the linear regression model is generated through training of a historical temperature data set of the battery unit in a stable running state, the input is temperature data outside the battery unit, and the output is temperature data in the battery unit directly superimposed by a correction factor.
4. The multi-mode sensing-based lithium battery thermal runaway alarm method of claim 2, wherein the dynamic correction further comprises a parameter self-adaptive adjustment mechanism, wherein the parameter self-adaptive adjustment mechanism automatically adjusts parameters of the mathematical compensation model according to the real-time change rate of the temperature data outside the battery unit, and the parameter self-adaptive adjustment mechanism analyzes the temperature change trend through a sliding window to update model coefficients.
5. The method for alarming thermal runaway of a lithium battery based on multi-mode sensing according to claim 3, wherein the training process of the linear regression model comprises characteristic engineering pretreatment, the historical temperature dataset is input into the linear regression model after normalization treatment, and a correction factor mapping relation is generated for real-time correction.
6. The multi-mode sensing-based lithium battery thermal runaway alarm method of claim 4, wherein the parameter adaptive adjustment mechanism employs an optimization algorithm to achieve parameter updating, the optimization algorithm calculates a loss function gradient according to a real-time temperature change rate, and iteratively optimizes coefficients of a mathematical compensation model to improve dynamic convergence efficiency.
7. The multi-mode sensing-based lithium battery thermal runaway alarm method of claim 1, wherein the other multi-mode sensing data comprises battery cell voltage data, battery cell current data and battery cell gas sensor data, the collaborative integration is to fuse the corrected temperature data in the battery cell with the multi-mode sensing data through a data fusion engine, and the data fusion engine calculates contribution degrees of all data sources by adopting a weighted average algorithm to generate a comprehensive hot air risk index as an alarm decision reference.
8. The multi-modal sensing-based lithium battery thermal runaway alarm method of claim 7, wherein the data fusion engine is implemented as a kalman filter architecture, the kalman filter architecture takes the corrected temperature data in the battery cells as an observation input, constructs a state estimation equation in combination with the multi-modal sensing data, outputs a fused comprehensive hot air risk index, and updates the state covariance in real time.
9. The multi-modal sensing-based lithium battery thermal runaway alarm method of claim 8, wherein the state estimation equation of the kalman filter architecture integrates a battery thermodynamic physical model, which simulates a heat transfer process based on battery cell material properties and uses multi-modal sensing data as feedback correction state estimation results.
10. The multi-modal awareness based lithium battery thermal runaway warning method of claim 1 wherein the hierarchical warning response mechanism is divided into a three-level response strategy, the primary response is based on triggering an audible and visual early warning when the integrated thermal risk index exceeds a primary temperature threshold, the intermediate response is based on activating the cooling system when the integrated thermal risk index continuously rises above an intermediate temperature threshold, the advanced response is based on performing battery cell power shut-off when the integrated thermal risk index reaches an advanced temperature threshold, and the response level is performed in progression.

Description

Lithium battery thermal runaway alarm method based on multi-mode sensing Technical Field The invention relates to the field of energy storage, in particular to a lithium battery thermal runaway alarming method based on multi-mode sensing. Background The current lithium battery thermal runaway monitoring technology generally relies on a single temperature sensor or a simplified model to perform early warning, is easily influenced by external environment thermal interference (such as equipment cabin temperature fluctuation and local heat source radiation), and causes false judgment of the real thermal state of an internal battery core, thereby causing false alarm or missing alarm risks. The existing method has limitations on the data integration level, lacks a collaborative analysis mechanism of multidimensional sensing data, cannot distinguish progressive stages of thermal runaway, and meanwhile, the alarm response is mostly triggered by a single threshold value, so that the dynamic change of the risk severity is difficult to match, the accident upgrading is possibly caused by response lag, and the resource waste is possibly caused by excessive response. There is a need for a thermal runaway prevention and control scheme that actively eliminates ambient noise, fuses multisource information, and achieves a precise hierarchical response. Disclosure of Invention The invention provides a lithium battery thermal runaway alarm method based on multi-mode sensing, which comprises the following steps: S1, acquiring temperature data outside a battery unit and temperature data inside the battery unit in real time; S2, dynamically correcting the temperature data in the battery unit based on the temperature data outside the battery unit so as to eliminate external environment interference; s3, taking the corrected temperature data in the battery unit as key input and carrying out cooperative integration with other multi-mode sensing data; S4, driving a hierarchical alarm response mechanism based on the integrated data to match the risk severity. The dynamic correction is realized through a temperature compensation algorithm, the temperature compensation algorithm calculates an environmental thermal interference factor by using temperature data outside the battery unit, the environmental thermal interference factor is applied to temperature data inside the battery unit in real time to generate a correction value, and the temperature compensation algorithm establishes a mathematical compensation model based on the thermal conduction characteristic of the battery unit. The temperature compensation algorithm adopts a linear regression model as a mathematical compensation model, the linear regression model is generated through training of a historical temperature data set of the battery unit in a stable running state, the temperature data is input as temperature data outside the battery unit, and the temperature data is output as a correction factor and is directly superposed into temperature data inside the battery unit. The dynamic correction further comprises a parameter self-adaptive adjustment mechanism, the parameter self-adaptive adjustment mechanism automatically adjusts parameters of the mathematical compensation model according to the real-time change rate of the temperature data outside the battery unit, and the parameter self-adaptive adjustment mechanism analyzes the temperature change trend through a sliding window to update model coefficients. The training process of the linear regression model comprises feature engineering preprocessing, the historical temperature dataset is input into the linear regression model after normalization processing, and a correction factor mapping relation is generated for real-time correction. The parameter self-adaptive adjustment mechanism realizes parameter updating by adopting an optimization algorithm, the optimization algorithm calculates a loss function gradient according to the real-time temperature change rate, and the coefficient of the mathematical compensation model is iteratively optimized to improve the dynamic convergence efficiency. The other multi-mode sensing data comprise battery cell voltage data, battery cell current data and battery cell gas sensor data, and the collaborative integration is to fuse the corrected temperature data in the battery cell with the multi-mode sensing data through a data fusion engine, and the data fusion engine calculates contribution degrees of all data sources by adopting a weighted average algorithm to generate a comprehensive hot air risk index as an alarm decision reference. The data fusion engine is implemented as a Kalman filter architecture, the Kalman filter architecture takes the corrected temperature data in the battery unit as an observation input, a state estimation equation is constructed by combining the multi-mode sensing data, a fused comprehensive hot air risk index is output, and a state covariance is updated in re