CN-122025925-A - Immersed battery thermal management system and method based on fiber bragg grating sensing

CN122025925ACN 122025925 ACN122025925 ACN 122025925ACN-122025925-A

Abstract

The application relates to an immersed battery thermal management system and method based on fiber grating sensing, wherein the method comprises the steps of collecting the temperature of a positive electrode lug, the temperature of a negative electrode lug, the temperature of an environment and the temperature of cooling liquid of a battery in a battery pack; and finishing data preprocessing, feature extraction and construction of a prediction model, constructing a correction coefficient model, correcting the battery health state, the thermal runaway risk level and the dynamic early warning threshold value output by the tab temperature driving prediction module by using the correction coefficient model, and constructing a hierarchical control strategy based on the output result of the model self-optimization module so as to realize accurate cooling of the temperature acquisition positions of the positive electrode tab and the negative electrode tab. According to the application, the double fiber bragg grating sensors are adopted to perform one-to-one targeted lamination temperature measurement on the positive electrode lugs and the negative electrode lugs of the battery, so that the direct and accurate transmission of temperature signals is realized, the problems of lagging temperature measurement and easy corrosion of the traditional sensors in an immersed environment are solved, and the safety redundancy, the temperature control precision and the energy utilization efficiency of the immersed battery thermal management system are improved.

Inventors

LU GUI
CUI XINTAO
CAI QINGFENG

Assignees

华北电力大学

Dates

Publication Date: 20260512
Application Date: 20260127

Claims (10)

1. The immersed battery thermal management method based on fiber grating sensing is characterized by comprising the following steps: constructing a prediction model according to the improved double-branch attention model, inputting the prediction model into the positive electrode lug temperature, the negative electrode lug temperature, the ambient temperature and the cooling liquid temperature of each battery in the battery pack, and outputting the prediction model into the battery health state, the thermal runaway risk level and the dynamic early warning threshold; Constructing a correction coefficient model based on the characteristics of the sample library and the marked samples, inputting real-time flow, viscosity, conductivity data and vibration data of the battery pack, and correcting the battery health state, the thermal runaway risk level and the dynamic early warning threshold value by using the correction coefficient model to obtain the corrected battery health state, the corrected thermal runaway risk level and the corrected dynamic early warning threshold value; and constructing a hierarchical control strategy based on the corrected battery health state, the corrected thermal runaway risk level and the corrected dynamic early warning threshold value, outputting a hierarchical control instruction, and controlling the opening of the spray opening corresponding to each battery positive and negative electrode lug according to the hierarchical control instruction.
2. The method for thermal management of an immersed battery based on fiber grating sensing as recited in claim 1, wherein constructing a predictive model comprises the steps of: Receiving positive electrode lug temperature, negative electrode lug temperature, ambient temperature and cooling liquid temperature data transmitted by a double fiber grating device, and preprocessing the data; Based on the preprocessed temperature data, calculating the real-time temperature difference, the temperature difference change rate, the temperature fluctuation range and the compensation fusion characteristics of the positive and negative electrode lugs of the battery, and transmitting the characteristic data to an improved double-branch attention model unit; Based on the improved double-branch attention neural network, predicting the battery health state and the thermal runaway risk level; And constructing a dynamic early warning threshold calculation model by combining the ambient temperature and the cooling liquid temperature.
3. The method for thermal management of an immersed battery based on fiber bragg grating sensing according to claim 2, wherein the two branch weights are dynamically allocated through attention fusion, abnormal characteristic contribution is strengthened, and a prediction model formula is as follows: ; ; In the formula, Is the output feature vector of the first branch of the t moment model; is the output feature vector of the second branch of the t moment model; And Respectively a weight matrix of the two branch feature vectors; And Bias items of the two branch feature vectors respectively; And Respectively the intermediate variables of the two branch feature vectors after linear transformation; And The attention weights of the two branches are respectively dynamically distributed with the contribution degrees of the characteristics of the two branches; Is the final fusion characteristic at the time t; The final fusion characteristic of the current charge-discharge period; And The output layer weight matrix; And Is an output layer bias term, and softmax is a maximum function.
4. The method for thermal management of an immersed battery based on fiber grating sensing according to claim 2, wherein the dynamic early warning threshold calculation model is: ; ; In the formula, Is a basic early warning threshold based on battery state of health; Is the rated maximum heat resistance threshold of the new battery; Is the battery aging threshold decay coefficient; Is the final dynamic early warning threshold; Is an ambient temperature compensation coefficient; The ambient temperature at time t; Is a standard ambient temperature reference value; The temperature of the cooling liquid at the moment t; is a standard coolant temperature reference value.
5. The method for thermal management of an immersed battery based on fiber grating sensing according to claim 2, wherein the temperature fluctuation range calculation model is as follows: , Wherein, the method comprises the steps of, The average temperature of the positive electrode lug and the negative electrode lug of the ith sampling point; The number of sampling points for the current charge-discharge cycle, Is the average temperature over the cycle; for the ith sampling moment, the temperature of the battery positive electrode lug collected by the double fiber bragg grating targeting monitoring module is sigma T ; The ambient temperature and the fluctuation interference of the temperature of the cooling liquid are eliminated through the compensation fusion characteristic, and the compensation fusion characteristic calculation model is as follows: In which, in the process, Is the compensation fusion characteristic at the moment t; is the positive and negative poles of the battery at the moment t real-time temperature difference of the electrode lugs; Is an ambient temperature compensation coefficient; The ambient temperature at time t; Is the temperature compensation coefficient of the cooling liquid; The temperature of the cooling liquid at time t.
6. The method for thermal management of an immersed battery based on fiber grating sensing as recited in claim 1, wherein the modeling of the correction factor comprises the steps of: collecting real-time flow, viscosity and conductivity data of the immersed cooling liquid and real-time vibration data of the battery pack, and performing time stamp alignment treatment on the collected data; Based on the collected multi-source data, constructing an environment and working condition multi-dimensional data sample library and a correction coefficient model, and correcting the battery health state, the thermal runaway risk level and the dynamic early warning threshold value according to the correction coefficient model; Starting a double-drive optimization mechanism of basic period optimization and multi-condition forced optimization, and realizing continuous improvement of model precision; And calling a correction model to finish the correction of the prediction result, and outputting the corrected battery health state, the corrected thermal runaway risk level and the corrected dynamic early warning threshold value.
7. The method for thermal management of an immersed battery based on fiber grating sensing as recited in claim 6, wherein constructing a correction factor model comprises the steps of: The characteristic extraction of the cooling liquid parameter, namely calculating the fluctuation rate of the cooling liquid flow, the viscosity-temperature correlation coefficient and the conductivity baseline drift amount, and converting the original cooling liquid parameter into a heat dissipation efficiency correlation characteristic value; Vibration data feature extraction, namely performing frequency domain analysis on the vibration data, extracting the amplitude and the energy duty ratio of the feature frequency, converting the amplitude and the energy duty ratio into structural thermal resistance equivalent feature values, and generating vibration related quantification basis required by a correction model; The data fusion normalization is carried out on the cooling liquid heat dissipation related characteristics and the vibration data related characteristics to form a multi-dimensional sample library with a unified format, a correction coefficient model is constructed, and the battery health state, the thermal runaway risk level and the dynamic early warning threshold value are corrected according to the correction coefficient model.
8. The method for thermal management of an immersed battery based on fiber grating sensing as set forth in claim 7, wherein the correction factor model is: ; ; ; In the formula, T is a time stamp and is a dynamic correction coefficient; is the ith standardized feature value; a dynamic weight coefficient for the ith feature; is the base weight; Is a characteristic sensitivity coefficient; is the learning rate; Is an abnormal reset value; Is the real-time deviation of the ith feature at time t; is the mean value of the i-th characteristic history standardized value; The correction model of the battery health state, the thermal runaway risk level and the dynamic early warning threshold value is as follows; ; ; In the formula (I), in the formula (II), The corrected battery health state; a corrected thermal runaway risk level; the dynamic early warning threshold value after correction; Is the SOH adaptation coefficient of the battery health state; Is the risk level R adaptation coefficient; Is a dynamic early warning threshold adaptation coefficient.
9. The method for thermal management of an immersed battery based on fiber grating sensing as recited in claim 1, wherein step S5 comprises the steps of: synchronously receiving the data corrected by the quantized correction coefficient and the real-time monitoring data, and performing time sequence alignment and validity check on the data to generate a qualified input data set; constructing a hierarchical control strategy based on the verified multidimensional parameters, and generating opening adjustment instructions of each miniature independent spray opening; the independent spray port array executes opening adjustment instructions and opening real-time monitoring; Based on the corrected battery health state, the corrected thermal runaway risk level and the corrected dynamic early warning threshold value, the accurate cooling of the battery positive electrode tab temperature and the battery negative electrode tab temperature acquisition point is realized; and optimizing a subsequent hierarchical control strategy by combining the deviation analysis result of the feedback correction unit.
10. The submerged battery thermal management system adopting the submerged battery thermal management method based on fiber bragg grating sensing according to any one of claims 1 to 9 comprises a data collection module, a dual fiber bragg grating targeting monitoring module, a lug temperature driving prediction module, a model self-optimization module, a precise spraying control module, an alarm linkage module and a PLC controller, and is characterized in that: The data collection module is used for collecting batteries, gratings, cooling liquid data, historical data and fault data, marking the data and forming a standardized reference sample library; The dual-fiber grating targeted monitoring module adopts dual-fiber grating sensors to perform targeted temperature acquisition on positive electrode lugs and negative electrode lugs of a plurality of batteries in the battery pack, and synchronously acquires the ambient temperature and the temperature of cooling liquid; The tab temperature driving prediction module is used for completing data preprocessing, feature extraction and model prediction based on the positive electrode tab temperature, the negative electrode tab temperature, the ambient temperature and the cooling liquid temperature data acquired by the dual-fiber grating targeting monitoring module and outputting the battery health state, the thermal runaway risk level and the dynamic early warning threshold; The model self-optimizing module is used for iterating and optimizing the model through real-time data, constructing a correction model based on the cooling liquid flow, viscosity, conductivity data and battery pack vibration data, and correcting the output result of the tab temperature driving prediction module; The accurate spray control module is used for constructing a decomposition control strategy based on the corrected battery health state, the corrected thermal runaway risk level and the corrected dynamic early warning threshold value output by the model self-optimization module, and realizing accurate cooling of the temperature of the positive electrode tab and the temperature acquisition position of the negative electrode tab; the alarm linkage module outputs a grading alarm signal aiming at different corrected thermal runaway risk levels; the PLC is connected with the data collection module, the dual fiber bragg grating targeting monitoring module, the lug temperature driving prediction module, the accurate spraying control module, the model self-optimization module and the alarm linkage module in a network manner.

Description

Immersed battery thermal management system and method based on fiber bragg grating sensing Technical Field The application relates to the technical field of storage batteries, in particular to an immersed battery thermal management system and method based on fiber bragg grating sensing. Background With the continuous improvement of the energy density and safety requirements of power batteries in the fields of new energy automobiles, energy storage power stations and the like, the submerged liquid cooling technology has become the core development direction of high-power battery pack thermal management because of the high-efficiency heat dissipation capability. Compared with the traditional air cooling and conventional liquid cooling technologies, the immersed liquid cooling realizes heat exchange by directly contacting the surface of the battery through the cooling liquid, so that the heat dissipation efficiency is improved obviously, and the risk of thermal runaway of the battery can be effectively restrained, but the conventional immersed battery thermal management system still has a technical bottleneck in a core link, and the safety, the reliability and the large-scale application of the system are restricted. In the aspect of temperature monitoring, the traditional temperature detection relies on a multi-beam electric signal sensor, a cooling liquid flow field can be seriously disturbed in a spray type immersion environment, eddy current and local heat accumulation are caused, and the thermal runaway risk is aggravated, but the traditional fiber grating monitoring technology has the advantages of insulating disturbance resistance and simple wiring, is developed for non-immersion scenes, does not realize accurate coverage of a full cell under the immersion environment, particularly fails to target and monitor positive and negative electrode lugs of a heating core of a battery, and cannot provide high-fidelity early warning data. In the aspect of battery state prediction, the existing algorithm is based on multi-parameter fusion such as voltage, current and temperature, the system complexity is high, the voltage and current parameters are easily affected by cooling liquid and electromagnetic interference in an immersed environment, the acquisition accuracy is difficult to ensure, meanwhile, the thermal runaway core characterization parameter of the tab temperature is generally ignored, and the prediction result, the actual health state of the battery and the thermal runaway risk deviation are larger. In the aspect of cooling control, the conventional system adopts a module-level unified cooling strategy, the state difference of a single battery is not considered, accurate targeted cooling cannot be realized, the control logic only depends on real-time temperature feedback, the response is delayed, the battery state prediction result is not combined, local overheating or excessive cooling is easily caused, the high-efficiency advantage of immersed liquid cooling cannot be fully exerted, and energy waste is possibly caused. Therefore, the invention provides an immersed battery thermal management system and method based on fiber grating sensing. Disclosure of Invention Aiming at the core technical bottlenecks of insufficient temperature monitoring adaptability, low state prediction reliability and poor cooling control accuracy, the invention provides an immersed battery thermal management system and method based on fiber bragg grating sensing, which are based on dual-fiber optical grating ear targeted monitoring and combined with state prediction of electrode ear temperature driving to form a closed-loop technical system of single-cell electrode ear level accurate spray control, and key defects of the prior art in terms of immersed scene adaptation, prediction reliability and control accuracy are overcome in a targeted manner. The system not only can remarkably improve the safety redundancy, the temperature control precision and the energy utilization efficiency of the submerged battery thermal management system, but also can provide core technical support for the large-scale landing of scenes such as high-power batteries, large-scale energy storage power stations and the like, enrich the technical system of submerged thermal management and multi-physical field coupling monitoring, provide a new integrated model of monitoring, prediction and control for the industry, and has important academic innovation value, engineering application prospect and industrialization enabling significance. The technical scheme is specifically as follows. An immersed battery thermal management system based on fiber grating sensing comprises the following components. And the data collection module is used for collecting data such as a battery, a grating, cooling liquid and the like, collecting historical data and fault data, and classifying and labeling the data to form a standardized reference sample library. And the dual-fibe