A new energy automobile battery cell short circuit detection system and method

Abstract

The application relates to the technical field of new energy automobile battery safety management, in particular to a new energy automobile battery monomer short circuit detection system and method, the system collects target battery signals and battery physical and chemical data through a battery data acquisition module, filters, normalizes and extracts features of the target battery signals through a battery signal processing module, dynamically adjusts a short circuit feature self-adaptive threshold according to real-time working condition quantitative values through a monomer short circuit monitoring module, predicts a short circuit development trend and provides an intervention window period through a short circuit trend prediction module, and monitors an electric-power coupling value and dynamically adjusts and regulates a regulation strategy through a risk window regulation module, so that early and accurate detection and graded early warning of battery monomer short circuits are realized, and the accuracy of short circuit detection and the timeliness of graded regulation are improved.

Description

Technical Field

[0001] This invention relates to the field of new energy vehicle battery safety management technology, and in particular to a short-circuit detection system and method for a single battery cell in a new energy vehicle. Background Technology

[0002] As the core energy storage and supply unit of new energy vehicles, the safety of the power battery system is directly related to the safety of vehicle operation, the life safety of passengers, and the safety of public transportation. With the rapid increase in the penetration rate of new energy vehicles, the installed capacity of power batteries has shown explosive growth. At the same time, safety accidents caused by battery thermal runaway also occur frequently, becoming a key bottleneck restricting the healthy development of the new energy vehicle industry. Internal short circuits in battery cells are one of the main causes of thermal runaway in power batteries. They are highly concealed, develop rapidly, and are highly destructive. Early weak short circuit signals are difficult to be effectively identified by traditional monitoring methods. Once the best intervention window is missed, the short circuit point will rapidly expand within seconds to minutes. Current technology is unable to meet the high-precision and high-reliability detection requirements for battery cell short circuits under complex real-world vehicle conditions.

[0003] Chinese patent application CN120428125A discloses a short-circuit test cabinet for new energy power batteries, including an explosion-proof cabinet with a transmission compartment and a test compartment. The transmission compartment is equipped with an impact transmission system, a termination system, and a power supply system. To prevent the circuit from exploding during testing and reduce damage to the testing equipment, the impact transmission system and termination system are installed. Utilizing mechanical transmission, the short-circuit connection of the battery is immediately interrupted upon battery explosion, directly converting the impact force generated by the explosion into mechanical action. This eliminates the need for signal conversion, resulting in faster response speed, higher resistance to extreme environments, and lower cost. It also effectively reduces damage to the test compartment from the explosion shock wave during testing. However, this solution still suffers from problems such as a lack of short-circuit trend prediction capability, insufficient accuracy in short-circuit detection due to the failure to consider electro-mechanical coupling effects, and inadequate timeliness of graded control. Summary of the Invention

[0004] To address these issues, the present invention provides a short-circuit detection system and method for single battery cells in new energy vehicles, which overcomes the problems of insufficient short-circuit trend prediction capability, inadequate short-circuit detection accuracy due to failure to consider electro-mechanical coupling effect, and lack of timeliness in graded control in the prior art.

[0005] To achieve the above objectives, on the one hand, the present invention provides a short-circuit detection system for individual battery cells in new energy vehicles, comprising: a battery data acquisition module, a battery signal processing module, a single-cell short-circuit monitoring module, a short-circuit trend prediction module, and a risk window control module, wherein:

[0006] The short-circuit trend prediction module is used to acquire short-circuit conditions based on battery physical data, compare the short-circuit conditions with the comprehensive short-circuit characteristic index to obtain the comparison results, and also to perform window correction on the feature extraction process based on the comparison results, and to predict the short-circuit trend of the comprehensive short-circuit characteristic index to obtain the trend prediction results.

[0007] The risk window control module is used to obtain the intervention window period based on the trend prediction results, and output the graded control strategy according to the intervention window period. It is also used to obtain the electro-mechanical coupling value and adjust the output process of the graded control strategy according to the electro-mechanical coupling value.

[0008] Furthermore, the battery data acquisition module is used to acquire target battery signals and battery physical and chemical data;

[0009] The battery signal processing module is used to process the target battery signal to obtain the processed battery signal, and also to extract features from the processed battery signal using feature extraction methods to obtain the target battery features.

[0010] The single-cell short-circuit monitoring module is used to obtain the comprehensive short-circuit characteristic index based on the target battery characteristics, obtain the real-time operating condition quantification value based on the battery physical data, obtain the short-circuit characteristic adaptive threshold based on the target battery characteristics and the real-time operating condition quantification value, and output the short-circuit graded early warning based on the comprehensive short-circuit characteristic index and the short-circuit characteristic adaptive threshold.

[0011] Furthermore, the battery signal processing module performs signal processing on the target battery signal, specifically: processing the voltage signal through a three-level filtering strategy to obtain a processed voltage signal, processing the temperature signal through Kalman filtering to obtain a processed temperature signal, and processing the gas signal through wavelet denoising to obtain a processed gas signal, and using the processed voltage signal, processed temperature signal, and processed gas signal as the processed battery signal;

[0012] The battery signal processing module extracts features from the processed battery signal using a feature extraction method, which includes:

[0013] Step A01: Calculate the voltage drop rate A based on the current voltage value u(t), the previous voltage value u(t-Δt), and the sampling interval Δt, and set A=[u(t)-u(t-Δt)] / Δt;

[0014] Step A02: Calculate the temperature rise gradient B based on the battery pack's highest temperature Tmax, lowest temperature Tmin, and sampling interval Δt, and set B = (Tmax - Tmin) / Δt.

[0015] Step A03: Calculate the gas concentration accumulation rate C based on the current gas concentration Cgas(t), the previous gas concentration Cgas(t-Δt), and the sampling interval Δt, and set C=[Cgas(t)-Cgas(t-Δt)] / Δt;

[0016] Step A04: The voltage drop rate, temperature rise gradient, and gas concentration accumulation rate are taken as target battery characteristics.

[0017] Furthermore, the single-cell short-circuit monitoring module acquires the comprehensive short-circuit characteristic index based on the target battery characteristics, and normalizes the voltage drop rate, temperature rise gradient, and gas concentration accumulation rate to obtain the voltage drop rate value A1, temperature rise gradient value B1, and gas concentration accumulation rate value C1.

[0018] The comprehensive short-circuit characteristic index SSY is calculated based on the voltage drop rate value A1, temperature rise gradient value B1, gas concentration accumulation rate value C1, voltage drop rate weight w1, temperature rise gradient value weight w2, and gas concentration accumulation rate weight w3. SSY is set as A1×w1+B1×w2+C1×w3, and w1+w2+w3=1.

[0019] The real-time operating condition quantization value Gk is obtained based on the battery physical data, and the short-circuit feature adaptive threshold is obtained based on the target battery characteristics and the real-time operating condition quantization value. The short-circuit feature adaptive threshold includes a first adaptive threshold SSY1 and a second adaptive threshold SSY2.

[0020] The voltage drop rate value A1 and the temperature rise gradient value B1 are fused to obtain the feature fusion index Tz. Based on the real-time operating condition quantization value Gk, the feature fusion index Tz, the basic threshold weight set Wb={wb1,wb2}, and the operating condition sensitivity coefficient set Ks={ks1,ks2}, the first adaptive threshold SSY1 and the second adaptive threshold SSY2 are calculated. The following settings are made:

[0021] SSY1=wb1×[1+ks1×Gk]×[1+Tz], SSY2=wb2×[1+ks2×Gk]×[1+Tz], and wb1<wb2, ks1<ks2, SSY1<SSY2;

[0022] The comprehensive short-circuit characteristic index SSY is compared with the first adaptive threshold SSY1 and the second adaptive threshold SSY2. Based on the comparison result, the state of the comprehensive short-circuit characteristic index is determined, and a short-circuit graded early warning is output based on the determination result. Wherein:

[0023] When SSY≤SSY1, the single-unit short-circuit monitoring module determines the state of the comprehensive short-circuit characteristic index to be low and outputs the primary warning as a short-circuit graded warning.

[0024] When SSY1<SSY≤SSY2, the single-unit short-circuit monitoring module determines the state of the comprehensive short-circuit characteristic index to be moderate and outputs the intermediate warning as a short-circuit graded warning.

[0025] When SSY > SSY2, the single-unit short-circuit monitoring module determines the state of the comprehensive short-circuit characteristic index to be high and outputs the advanced warning as a short-circuit graded warning.

[0026] Furthermore, the short-circuit trend prediction module acquires the short-circuit situation based on the battery physicochemical data, compares the gas proportion factor Qt with the gas composition fingerprint spectrum F={f1,f2,...,fn} to obtain the similarity set S={s1,s2,...,sn}, where n represents the order, n=1,2,...,n;

[0027] The maximum similarity max(S) is obtained based on the similarity set S. The maximum similarity max(S) is then compared with a preset maximum similarity max(S0). Based on the comparison result, the state of the maximum similarity is determined, and the short-circuit condition is output according to the determination result. Where:

[0028] When max(S)≤max(S0), the short-circuit trend prediction module determines the state of maximum similarity as dissimilar and outputs the absence of short-circuit cause as a short-circuit situation;

[0029] When max(S)>max(S0), the short-circuit trend prediction module determines the state of maximum similarity as similar and outputs the existence of a short-circuit cause as a short-circuit situation.

[0030] Furthermore, the short-circuit trend prediction module compares the short-circuit situation with the comprehensive short-circuit characteristic index, specifically as follows:

[0031] When a short circuit occurs because there is a cause for the short circuit:

[0032] If SSY≤SSY1, the comparison result is determined to be conflicting, and the feature extraction process is windowed.

[0033] Otherwise, if the comparison result is determined to be without conflict, no window correction will be performed on the feature extraction process;

[0034] When the short circuit condition is that there is no cause for the short circuit:

[0035] If SSY≤SSY1, the comparison result is determined to be that there is no conflict, and no window correction is performed in the feature extraction process;

[0036] Otherwise, if the comparison result indicates a conflict, a window correction is performed on the feature extraction process.

[0037] The process of window correction for feature extraction is as follows: extend the time window of feature extraction, extend and correct the sampling interval Δt according to the window correction factor kc1 to obtain the extended sampling interval Δt1, set Δt1=Δt×(1+kc1), and 0.2≤kc1≤0.5, and use the extended sampling interval Δt1 as the sampling interval Δt to re-extract the target battery features.

[0038] Furthermore, the short-circuit trend prediction module performs short-circuit trend prediction on the comprehensive short-circuit characteristic index: the target battery characteristics and real-time operating condition quantification values ​​are input into the short-circuit trend prediction model to obtain the trend prediction result Ypv output by the short-circuit trend prediction model, where v represents the order of the values ​​in the trend prediction result, and v=1, 2, 3, ..., v.

[0039] The comprehensive short-circuit characteristic index SSY is compared with the trend prediction result Ypv. Based on the comparison result, the state of the comprehensive short-circuit characteristic index is judged, and the intervention window period is obtained based on the judgment result.

[0040] When SSY≤Ypv, the risk window control module determines that the state of the comprehensive short-circuit characteristic index is safe and does not acquire the intervention window period;

[0041] When SSY > Ypv, the risk window control module determines that the state of the comprehensive short-circuit characteristic index is unsafe and obtains the intervention window period: the intervention window period CM is calculated from the trend prediction time starting point t1 to the time point tv corresponding to Ypv, and CM is set as tv-t1.

[0042] Furthermore, the risk window control module compares the intervention window period CM with the first preset intervention window period CM1 and the second preset intervention window period CM2, where CM1 < CM2. Based on the comparison result, it judges the situation of the intervention window period and outputs a tiered control strategy based on the judgment result, wherein:

[0043] When CM≤CM1, the risk window control module determines that the situation during the intervention window period is highly urgent and outputs the first response strategy as a graded control strategy.

[0044] When CM1<CM≤CM2, the risk window control module determines that the situation during the intervention window period is moderately urgent and outputs the second response strategy as a graded control strategy.

[0045] When CM > CM2, the risk window control module determines that the situation during the intervention window period is of low urgency and outputs the third response strategy as a graded control strategy.

[0046] Furthermore, the risk window control module acquires the electro-mechanical coupling value by: normalizing the shell deformation rate D to obtain the shell deformation rate value D1; calculating the electro-mechanical coupling value MH based on the voltage drop rate value A1, the shell deformation rate value D1, the second weight w4 of the voltage drop rate value, and the weight w5 of the shell deformation rate value, setting MH = A1 × w4 + D1 × w5, and w4 + w5 = 1;

[0047] The electro-force coupling value MH is compared with the preset electro-force coupling value MH0. Based on the comparison result, the state of the electro-force coupling value is judged, and the output process of the graded control strategy is adjusted in stages according to the judgment result.

[0048] When MH≤MH0, the risk window control module determines that the state of the electro-mechanical coupling value is not serious and does not perform graded adjustment on the output process of the graded control strategy.

[0049] When MH > MH0, the risk window control module determines that the state of the electro-mechanical coupling value is severe and adjusts the output process of the graded control strategy in a graded manner.

[0050] On the other hand, the present invention also provides a method for a short-circuit detection system for a single battery cell in a new energy vehicle, comprising:

[0051] Step S1: Collect the target battery signal and battery physical and chemical data;

[0052] Step S2: Perform signal processing on the target battery signal to obtain the processed battery signal, and extract features from the processed battery signal to obtain the target battery features.

[0053] Step S3: Obtain the comprehensive short-circuit characteristic index based on the target battery characteristics, obtain the real-time operating condition quantization value based on the battery physical data, obtain the short-circuit characteristic adaptive threshold based on the target battery characteristics and the real-time operating condition quantization value, and output the short-circuit graded early warning based on the comprehensive short-circuit characteristic index and the short-circuit characteristic adaptive threshold.

[0054] Step S4: Obtain the short circuit condition based on the battery physical data, compare the short circuit condition with the comprehensive short circuit characteristic index to obtain the comparison result, perform window correction on the feature extraction process based on the comparison result, and predict the short circuit trend of the comprehensive short circuit characteristic index to obtain the trend prediction result.

[0055] Step S5: Obtain the intervention window period based on the trend prediction results, output the graded control strategy based on the intervention window period, obtain the electro-mechanical coupling value, and adjust the output process of the graded control strategy in a graded manner based on the electro-mechanical coupling value.

[0056] Compared with existing technologies, the beneficial effects of this invention are as follows: The system acquires target battery signals and physical and chemical data through a battery data acquisition module to obtain comprehensive data on battery electrical, thermal, gaseous, and mechanical characteristics, thus providing a multi-dimensional data foundation for short-circuit detection. The system also filters, normalizes, and extracts features from the target battery signals through a battery signal processing module to eliminate noise interference and extract key short-circuit characteristic indicators, thereby improving the accuracy of subsequent detection. Furthermore, the system dynamically adjusts the adaptive threshold for short-circuit characteristics based on real-time operating condition quantification values ​​through a single-cell short-circuit monitoring module to adapt to changes in battery state under different driving conditions, thereby reducing false alarm and false negative rates. The system also predicts short-circuit development trends and provides an intervention window through a short-circuit trend prediction module to reserve time for handling before thermal runaway occurs, thereby improving the timeliness and proactivity of graded control. Finally, the system monitors the electro-mechanical coupling value and dynamically adjusts the control strategy through a risk window control module to improve detection reliability by utilizing the correlation between mechanical deformation and electrical anomalies, thus constructing a multi-layered safety protection system integrating electrical and mechanical components. Attached Figure Description

[0057] Figure 1 This is a schematic diagram of the structure of a short-circuit detection system for a new energy vehicle battery cell provided in an embodiment of the present invention;

[0058] Figure 2 This is a flowchart illustrating the feature extraction method in the short-circuit detection system for a new energy vehicle battery cell provided in an embodiment of the present invention.

[0059] Figure 3 This is a schematic diagram of the situation comparison process in the short circuit detection system for a new energy vehicle battery cell provided in an embodiment of the present invention;

[0060] Figure 4 This is a flowchart illustrating the method for a short-circuit detection system for a new energy vehicle battery cell provided in an embodiment of the present invention. Detailed Implementation

[0061] To make the objectives and advantages of the present invention clearer, the present invention will be further described below with reference to embodiments; it should be understood that the specific embodiments described herein are merely for explaining the present invention and are not intended to limit the present invention.

[0062] Preferred embodiments of the present invention will now be described with reference to the accompanying drawings. Those skilled in the art should understand that these embodiments are merely illustrative of the technical principles of the present invention and are not intended to limit the scope of protection of the present invention.

[0063] It should be noted that in the description of this invention, the terms "upper", "lower", "left", "right", "inner", "outer", etc., which indicate directions or positional relationships, are based on the directions or positional relationships shown in the accompanying drawings. This is only for the convenience of description and is not intended to indicate or imply that the device or element must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, it should not be construed as a limitation of this invention.

[0064] Furthermore, it should be noted that, in the description of this invention, unless otherwise explicitly specified and limited, the terms "installation," "connection," and "linking" should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral connection; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; and they can refer to the internal connection of two components. Those skilled in the art can understand the specific meaning of the above terms in this invention according to the specific circumstances.

[0065] See Figure 1 This invention provides a short-circuit detection system for individual battery cells in new energy vehicles, comprising:

[0066] The battery data acquisition module 101 is used to acquire target battery signals and battery physical and chemical data;

[0067] The battery signal processing module 102 is used to process the target battery signal to obtain the processed battery signal, and also to extract features from the processed battery signal using a feature extraction method to obtain the target battery features.

[0068] The single-cell short-circuit monitoring module 103 is used to obtain the comprehensive short-circuit characteristic index based on the target battery characteristics, to obtain the real-time operating condition quantification value based on the battery physical data, to obtain the short-circuit characteristic adaptive threshold based on the target battery characteristics and the real-time operating condition quantification value, and to output the short-circuit graded early warning based on the comprehensive short-circuit characteristic index and the short-circuit characteristic adaptive threshold.

[0069] The short-circuit trend prediction module 104 is used to acquire the short-circuit situation based on the battery physical data, compare the situation with the comprehensive short-circuit characteristic index to obtain the situation comparison result, and also to perform window correction on the feature extraction process based on the situation comparison result, and to predict the short-circuit trend of the comprehensive short-circuit characteristic index to obtain the trend prediction result.

[0070] The risk window control module 105 is used to obtain the intervention window period based on the trend prediction results, and output the graded control strategy based on the intervention window period. It is also used to obtain the electro-mechanical coupling value and adjust the output process of the graded control strategy in a graded manner based on the electro-mechanical coupling value.

[0071] Specifically, the new energy vehicle battery cell short-circuit detection system can be applied to the safety management of new energy vehicle power batteries. The system utilizes multi-dimensional signal acquisition, adaptive threshold judgment, trend prediction, and electro-mechanical coupling analysis to achieve early and accurate detection and tiered control of battery cell short circuits, thereby improving the accuracy of short-circuit detection and the timeliness of tiered control. Specifically, the system acquires target battery signals and physical and chemical data through a battery data acquisition module to obtain comprehensive data on the battery's electrical, thermal, gaseous, and mechanical characteristics, providing a multi-dimensional data foundation for short-circuit detection. Furthermore, the system uses a battery signal processing module to filter, normalize, and extract features from the target battery signals to eliminate noise interference. The system extracts key short-circuit characteristic indicators to improve the accuracy of subsequent detection. It also dynamically adjusts the adaptive threshold of short-circuit characteristics based on real-time operating condition quantification values ​​through a single-unit short-circuit monitoring module, adapting to changes in battery state under different driving conditions and reducing false alarm and false alarm rates. Furthermore, the system predicts short-circuit development trends and provides intervention windows through a short-circuit trend prediction module, allowing time for handling before thermal runaway occurs, thus improving the timeliness and proactiveness of graded control. Finally, the system monitors electro-mechanical coupling values ​​and dynamically adjusts control strategies through a risk window control module, leveraging the correlation between mechanical deformation and electrical anomalies to improve detection reliability, thereby constructing a multi-layered safety protection system integrating electrical and mechanical components.

[0072] Specifically, the battery data acquisition module 101 acquires target battery signals and battery physical and chemical data.

[0073] Specifically, the target battery signals include voltage signals, temperature signals, and gas signals. The battery physicochemical data includes vehicle speed, current, gear position, ambient temperature, gas scaling factor, and casing deformation rate. The voltage signal refers to the time-series voltage data obtained by the battery management system through sampling at a sampling frequency fs to the terminal voltage of the battery cells. In this embodiment, fs = 10Hz and the sampling accuracy is set to 0.1mV. The temperature signal refers to the surface temperature data obtained by temperature sensors arranged on the surface of the battery cells through sampling at a sampling frequency fs. The gas signal refers to the concentration data of characteristic gases released by the battery cells, such as carbon monoxide concentration and hydrogen concentration, detected by an electrochemical sensor array. The gas scaling factor refers to the... During the thermal runaway of a single battery cell, the dimensionless ratio of the concentrations of various characteristic gases detected by the electrochemical sensor array is used. In this embodiment, the gas proportionality factor, such as the concentrations of carbon monoxide and hydrogen, is obtained by principal component analysis. The gas proportionality factor Qt is calculated based on the carbon monoxide concentration cco and the hydrogen concentration ch2, and Qt is set to cco / ch2. When ch2 < cco, Qt is set to 0. The shell deformation rate refers to the amount of geometric change in the battery cell shell per unit time. In this embodiment, the displacement of multiple measuring points on the surface of the battery shell is monitored by a laser displacement sensor. The shell deformation rate Ds is calculated based on the displacement change ΔL and the time interval Δt, and Ds is set to ΔL / Δt.

[0074] Specifically, the battery data acquisition module 101 collects target battery signals and battery physical and chemical data to comprehensively perceive the state of individual battery cells, thereby improving the accuracy of short circuit detection and the sensitivity of early warning.

[0075] Specifically, the battery signal processing module 102 performs signal processing on the target battery signal, specifically: processing the voltage signal through a three-level filtering strategy to obtain a processed voltage signal, processing the temperature signal through Kalman filtering to obtain a processed temperature signal, and processing the gas signal through wavelet denoising to obtain a processed gas signal, and using the processed voltage signal, processed temperature signal, and processed gas signal as the processed battery signal;

[0076] Figure 2 This is a flowchart illustrating the feature extraction method in a new energy vehicle battery cell short-circuit detection system provided in an embodiment of the present invention. Figure 2 As shown, the feature extraction method includes:

[0077] Step A01: Calculate the voltage drop rate A based on the current voltage value u(t), the previous voltage value u(t-Δt), and the sampling interval Δt, and set A=[u(t)-u(t-Δt)] / Δt;

[0078] Step A02: Calculate the temperature rise gradient B based on the battery pack's highest temperature Tmax, lowest temperature Tmin, and sampling interval Δt, and set B = (Tmax - Tmin) / Δt.

[0079] Step A03: Calculate the gas concentration accumulation rate C based on the current gas concentration Cgas(t), the previous gas concentration Cgas(t-Δt), and the sampling interval Δt, and set C=[Cgas(t)-Cgas(t-Δt)] / Δt;

[0080] Step A04: The voltage drop rate, temperature rise gradient, and gas concentration accumulation rate are taken as target battery characteristics.

[0081] Specifically, the three-level filtering strategy refers to processing the voltage signal sequentially through mean filtering, median filtering, and low-pass filtering to remove noise. The processed voltage signal refers to stable and accurate voltage time-series data obtained after noise reduction and smoothing by the three-level filtering strategy. The processed temperature signal refers to temperature data closer to the actual physical value obtained after optimal estimation and noise reduction by Kalman filtering. The processed gas signal refers to gas concentration change trend data obtained after wavelet denoising to remove background noise and sensor drift interference. The current voltage value refers to the processed voltage signal acquired at the current sampling time point. The previous voltage value refers to the processed voltage signal acquired at a time point one sampling period Δt before the current sampling time point. The sampling interval refers to the time between two consecutive samples taken by the battery data acquisition module. The interval length, such as 1 second, refers to the maximum temperature of the battery pack, which is the maximum value of the processed temperature signal measured among all battery cells in the battery pack at the current sampling time point. The minimum temperature of the battery pack is the minimum value of the processed temperature signal measured among all battery cells in the battery pack at the current sampling time point. The gas concentration at the current moment refers to the processed gas signal at the current sampling time point. The gas concentration at the previous moment refers to the processed gas signal collected at a time point with an interval of one sampling period Δt before the current sampling time point. The voltage drop rate refers to the magnitude of the voltage drop of a battery cell per unit time. Under normal operating conditions, A is close to 0. The voltage drop rate increases significantly during short-circuit faults. The temperature rise gradient refers to the rate of temperature rise of a battery cell per unit time. During short-circuit faults, the temperature rise gradient increases significantly due to the presence of local hot spots. The gas concentration accumulation rate refers to the cumulative change in the concentration of the characteristic gas per unit time.

[0082] Specifically, the battery signal processing module 102 performs signal processing and feature extraction to remove background noise from the original signal, thereby improving data purity and reliability. It also transforms the processed battery signal into indicators with clear physical meaning, so as to quantify the severity of internal battery anomalies and thus improve the accuracy of short circuit detection.

[0083] Specifically, the single-cell short-circuit monitoring module 103 obtains the comprehensive short-circuit characteristic index based on the characteristics of the target battery, and normalizes the voltage drop rate, temperature rise gradient, and gas concentration accumulation rate to obtain the voltage drop rate value A1, temperature rise gradient value B1, and gas concentration accumulation rate value C1.

[0084] The comprehensive short-circuit characteristic index SSY is calculated based on the voltage drop rate value A1, temperature rise gradient value B1, gas concentration accumulation rate value C1, voltage drop rate weight w1, temperature rise gradient value weight w2, and gas concentration accumulation rate weight w3. SSY is set as A1×w1+B1×w2+C1×w3, and w1+w2+w3=1.

[0085] The real-time operating condition quantization value Gk is obtained based on the battery physical data, and the short-circuit feature adaptive threshold is obtained based on the target battery characteristics and the real-time operating condition quantization value. The short-circuit feature adaptive threshold includes a first adaptive threshold SSY1 and a second adaptive threshold SSY2.

[0086] The voltage drop rate value A1 and the temperature rise gradient value B1 are fused to obtain the feature fusion index Tz. Based on the real-time operating condition quantization value Gk, the feature fusion index Tz, the basic threshold weight set Wb={wb1,wb2}, and the operating condition sensitivity coefficient set Ks={ks1,ks2}, the first adaptive threshold SSY1 and the second adaptive threshold SSY2 are calculated. The following settings are made:

[0087] SSY1=wb1×[1+ks1×Gk]×[1+Tz], SSY2=wb2×[1+ks2×Gk]×[1+Tz], and wb1<wb2, ks1<ks2, SSY1<SSY2;

[0088] The comprehensive short-circuit characteristic index SSY is compared with the first adaptive threshold SSY1 and the second adaptive threshold SSY2. Based on the comparison result, the state of the comprehensive short-circuit characteristic index is determined, and a short-circuit graded early warning is output based on the determination result. Wherein:

[0089] When SSY≤SSY1, the single-unit short-circuit monitoring module 103 determines that the state of the comprehensive short-circuit characteristic index is low and outputs the primary warning as a short-circuit graded warning.

[0090] When SSY1<SSY≤SSY2, the single-unit short-circuit monitoring module 103 determines the state of the comprehensive short-circuit characteristic index to be moderate and outputs the intermediate warning as a short-circuit graded warning.

[0091] When SSY > SSY2, the single-unit short-circuit monitoring module 103 determines the state of the comprehensive short-circuit characteristic index to be high and outputs the advanced warning as a short-circuit graded warning.

[0092] Specifically, the normalization process refers to mapping the voltage drop rate, temperature rise gradient, and gas concentration accumulation rate to the [0,1] interval using the max-min normalization method. The temperature rise gradient value weight is a coefficient that measures the importance of the temperature rise gradient value in the comprehensive short-circuit characteristic index, and the gas concentration accumulation rate value weight is a coefficient that measures the importance of the gas concentration accumulation rate value in the comprehensive short-circuit characteristic index. In this embodiment, by using a step size of 0.05, all combinations of w1, w2, and w3 are traversed, and the comprehensive short-circuit characteristic index under each combination is obtained. The true positive rate and false positive rate are calculated, and an ROC curve is plotted. The weight combination that maximizes the area under the curve is selected as the value of w1, w2, and w3, thus obtaining w1. =0.5, w2=0.3, w3=0.2. The comprehensive short-circuit characteristic index refers to the comprehensive risk level of an internal short circuit in a battery cell. The real-time operating condition quantification value refers to the numerical value that quantifies the intensity of the operating conditions currently experienced by the battery cell. In this embodiment, the process of obtaining the real-time operating condition quantification value Gk is as follows: quantification calculation is performed based on vehicle speed Vc, current L, and gear Gp. Gk is set as Gk=m1×(Vc / Vmax)+m2×(|L| / Lmax)+m3×(Gp / Gmax), where Vmax represents the maximum vehicle speed, Lmax represents the maximum allowable current of the battery, Gmax represents the highest gear, m1 is the vehicle speed weight, m2 is the current weight, and m3 is the gear weight, and m1+m2+m3 are the same. =1, where the vehicle's maximum speed is entered through the vehicle specifications, the maximum allowable current of the battery is obtained through the cell specifications of the new energy vehicle, and the highest gear refers to the number of forward gears of the vehicle, obtained through the transmission specifications. For example, if the maximum number of forward gears is 8, then Gmax = 8 gears. In this embodiment, m1 = 0.4, m2 = 0.4, and m3 = 0.2. In this embodiment, m1, m2, and m3 are assigned values ​​using an expert experience method. The expert experience method refers to the process by which domain experts assign values ​​to m1, m2, and m3 based on past experience and physical meaning. The physical meaning includes: the current directly determines the polarization internal resistance voltage drop of the battery, which is the main factor in voltage drop; vehicle speed affects the average power and Heat dissipation and speed adjustment are auxiliary corrections. The feature fusion calculation refers to the process of multiplying and coupling the voltage drop rate value A1 and the temperature rise gradient value B1 to obtain the feature fusion index Tz, such as Tz=A1×B1. The basic threshold weight set includes a first basic threshold wb1 and a second basic threshold wb2. The first basic threshold refers to the basic reference value of the first adaptive threshold under the ideal operating condition when Gk=0 and Tz=0. The second basic threshold refers to the basic reference value of the second adaptive threshold under the ideal operating condition when Gk=0 and Tz=0. In this embodiment, the fluctuation range of the comprehensive short-circuit characteristic index is 0.3-0.7 obtained by statistically distributing a large number of fault-free batteries under standard operating conditions. Therefore, wb1 is set to 0.3, wb2=0.7, the set of operating condition sensitivity coefficients includes a first operating condition sensitivity coefficient ks1 and a second operating condition sensitivity coefficient ks2. The first operating condition sensitivity coefficient is a coefficient that measures the influence of the real-time operating condition quantization value on the first adaptive threshold, and the second operating condition sensitivity coefficient is a coefficient that measures the influence of the real-time operating condition quantization value on the second adaptive threshold. In this embodiment, the first operating condition sensitivity coefficient and the second operating condition sensitivity coefficient are valued through a data experiment method. The data experiment method refers to placing the battery cell on a test bench, simulating different real-time operating condition quantization values, and recording the comprehensive short-circuit characteristic index corresponding to each real-time operating condition quantization value. The 85th percentile is used as the low-risk boundary, and the 99th percentile is used as the high-risk boundary. The comprehensive short-circuit characteristic index obtained from the low-risk boundary and high-risk boundary is fitted into low-risk curves and high-risk curves, respectively. The slope of the low-risk curve is found to be 0.5, and the slope of the high-risk curve is 0.8, using the least squares method. Therefore, ks1 = 0.5 and ks2 = 0.8 are set. The state of the comprehensive short-circuit characteristic index refers to its degree of severity, including low, medium, and high. In this embodiment, when the short-circuit warning is a primary warning, the warning is only recorded in the system operation log without triggering an alarm. When the short-circuit warning is a medium warning, a maintenance schedule is pushed to the vehicle management interface via Bluetooth as a reminder. When the short-circuit warning is a high warning, the high voltage is immediately cut off, and the personnel on board are notified to evacuate.

[0093] Specifically, the single-unit short-circuit monitoring module 103 acquires the adaptive threshold of short-circuit characteristics and judges the state of the comprehensive short-circuit characteristic index to construct a dual-threshold adaptive mechanism based on operating condition quantification and feature fusion, so as to achieve accurate classification and differentiated response of short-circuit risk, thereby improving the accuracy, robustness and overall vehicle safety of the early warning system.

[0094] Specifically, the short-circuit trend prediction module 104 acquires the short-circuit situation based on the battery physicochemical data, compares the gas proportion factor Qt with the gas composition fingerprint spectrum F={f1,f2,...,fn} to obtain the similarity set S={s1,s2,...,sn}, where n represents the order, n=1,2,...,n;

[0095] The maximum similarity max(S) is obtained based on the similarity set S. The maximum similarity max(S) is then compared with a preset maximum similarity max(S0). Based on the comparison result, the state of the maximum similarity is determined, and the short-circuit condition is output according to the determination result. Where:

[0096] When max(S)≤max(S0), the short-circuit trend prediction module 104 determines that the state with the maximum similarity is dissimilar and outputs the absence of a short-circuit cause as a short-circuit situation.

[0097] When max(S)>max(S0), the short-circuit trend prediction module 104 determines the state of maximum similarity as similar and outputs the existence of a short-circuit cause as a short-circuit situation.

[0098] Specifically, the gas composition fingerprint spectrum refers to a pre-established standard database of characteristic gas concentration distributions under battery failure modes. Here, f1 represents the first fingerprint feature, representing the gas proportion factor distribution template corresponding to the first preset battery failure mode; f2 represents the second fingerprint feature, representing the gas proportion factor distribution template corresponding to the second preset battery failure mode; fn represents the nth fingerprint feature, representing the gas proportion factor distribution template corresponding to the nth preset battery failure mode. The maximum similarity refers to the set of cosine similarities between the gas proportion factor and all fingerprint features in the gas composition fingerprint spectrum. Here, s1 represents the similarity between the gas proportion factor and the first fingerprint feature; s2 represents the similarity between the gas proportion factor and the second fingerprint feature; and sn represents the similarity between the gas proportion factor and the nth fingerprint feature. The maximum similarity refers to the set of cosine similarities between the gas proportion factor and all fingerprint features in the gas composition fingerprint spectrum. The maximum similarity value in the similarity set refers to the preset maximum similarity value used to judge the state of the maximum similarity. In this embodiment, based on the balance experiment, 0.75≤max(S0)≤0.90 is set. The balance experiment includes: collecting normal samples by simulating real vehicle working conditions, obtaining fault samples by simulating short circuits such as needle penetration and compression, comparing all samples with the preset gas composition fingerprint spectrum, calculating the maximum similarity value of each sample, obtaining the normal sample similarity set and the fault sample similarity set, plotting the distribution of the two sets of similarity data on the same coordinate system, and observing that the overlapping area of ​​the normal interval and the fault interval is [0.75,0.90]. Therefore, 0.75≤max(S0)≤0.90 is set. The state of the maximum similarity refers to the degree of similarity represented by the maximum similarity, including dissimilarity and similarity.

[0099] Specifically, the short-circuit trend prediction module 104 judges the state with the highest similarity so as to accurately identify the specific mode of short-circuit fault from the gas characteristic level, effectively eliminate false alarms caused by environmental interference and sensor drift, thereby improving the accuracy and reliability of short-circuit cause diagnosis.

[0100] Figure 3 This is a schematic diagram illustrating the situation comparison process in the short-circuit detection system for a new energy vehicle battery cell provided in an embodiment of the present invention. Figure 3 As shown, the short-circuit trend prediction module 104 compares the short-circuit situation with the comprehensive short-circuit characteristic index, specifically as follows:

[0101] When a short circuit occurs because there is a cause for the short circuit:

[0102] If SSY≤SSY1, the comparison result is determined to be conflicting, and the feature extraction process is windowed.

[0103] Otherwise, if the comparison result is determined to be without conflict, no window correction will be performed on the feature extraction process;

[0104] When the short circuit condition is that there is no cause for the short circuit:

[0105] If SSY≤SSY1, the comparison result is determined to be that there is no conflict, and no window correction is performed in the feature extraction process;

[0106] Otherwise, if the comparison result indicates a conflict, a window correction is performed on the feature extraction process.

[0107] The process of window correction for feature extraction is as follows: extend the time window of feature extraction, extend and correct the sampling interval Δt according to the window correction factor kc1 to obtain the extended sampling interval Δt1, set Δt1=Δt×(1+kc1), and 0.2≤kc1≤0.5, and use the extended sampling interval Δt1 as the sampling interval Δt to re-extract the target battery features.

[0108] Specifically, the window correction factor refers to a preset coefficient for extending and correcting the sampling interval Δt. In this embodiment, the range of values ​​for the window correction factor is obtained through a traversal optimization method. The traversal optimization method includes: creating a scenario where the gas signal and the electrothermal signal are inconsistent, such as simulating a micro short circuit where the gas has responded but the electrothermal signal is weak; taking values ​​for the window correction factor kc1 in steps of 0.05; calculating the extended sampling interval Δt1; and observing its impact on the subsequent judgment results. The fluctuation range of the window correction factor kc1 with the highest conflict resolution rate and the most significant improvement in signal-to-noise ratio is found to be [0.2, 0.5]. Therefore, 0.2 ≤ kc1 ≤ 0.5 is set.

[0109] Specifically, the short-circuit trend prediction module 104 compares the short-circuit situation with the comprehensive short-circuit characteristic index to proactively trigger adaptive correction of feature extraction parameters when evidence is inconsistent, re-examine the data to eliminate contradictions, thereby improving the robustness of information fusion diagnosis and ensuring the reliability of short-circuit detection results.

[0110] Specifically, the short-circuit trend prediction module 104 performs short-circuit trend prediction on the comprehensive short-circuit characteristic index: the target battery characteristics and real-time operating condition quantification values ​​are input into the short-circuit trend prediction model to obtain the trend prediction result Ypv output by the short-circuit trend prediction model, where v represents the order of the values ​​in the trend prediction result, and v=1, 2, 3, ..., v.

[0111] Specifically, the risk window control module 105 compares the comprehensive short-circuit characteristic index SSY with the trend prediction result Ypv, judges the state of the comprehensive short-circuit characteristic index based on the comparison result, and obtains the intervention window period based on the judgment result, wherein:

[0112] When SSY≤Ypv, the risk window control module 105 determines that the state of the comprehensive short-circuit characteristic index is safe and does not acquire the intervention window period;

[0113] When SSY > Ypv, the risk window control module 105 determines that the state of the comprehensive short-circuit characteristic index is unsafe and obtains the intervention window period: the intervention window period CM is calculated from the trend prediction time starting point t1 to the time point tv corresponding to Ypv, and CM is set as tv-t1.

[0114] Specifically, the short-circuit trend prediction model refers to a long short-term memory neural network model that takes target battery features and real-time operating condition quantization values ​​as input data and trend prediction results as output data. The construction process of the short-circuit trend prediction model is as follows: Four input nodes are set in the input layer to receive target battery features and real-time operating condition quantization values ​​respectively. Two layers of long short-term memory networks are used as core feature extraction units, with 32 memory units in each layer. Dropout regularization is introduced after each layer of long short-term memory networks, with a dropout rate set to 0.2. The output layer has 15 fully connected output nodes, corresponding to the trend prediction results for the next 15 consecutive time steps, with a time step size of 6 seconds. Each output value is accompanied by a timestamp, resulting in a pre-built model. Historical operating condition data is divided into a 70% training set, a 20% validation set, and a 10% test set. The pre-built model is input into the input set and trained using the Adam adaptive moment estimator optimizer. The initial learning rate is set to 0.001, and the mean squared error is used as the loss function. Early stopping is used to prevent overfitting, resulting in a trained model. The validation set is then input into the trained model, and training is terminated when the validation set loss no longer decreases after 10 consecutive rounds, resulting in a validated model. The test set is then input into the validated model for testing. The validated model with an accuracy of 92% is used as the short-circuit trend prediction model. The historical operating data refers to a database storing historical target battery characteristics, real-time operating condition quantification values, and their corresponding actual comprehensive short-circuit characteristic indices. The trend prediction result refers to the comprehensive short-circuit characteristic index sequence for multiple consecutive time steps in the future obtained based on the short-circuit trend prediction model. Each Ypv represents the short-circuit trend prediction model's prediction of the v-th time step in the future. The quantitative estimation of the risk level at each time point, the state of the comprehensive short-circuit characteristic index refers to the safety level of the comprehensive short-circuit characteristic index, including safe and unsafe, the trend prediction time starting point t1 refers to the starting time point when the short-circuit trend prediction model makes a prediction, and the time point tv corresponding to Ypv refers to the time position corresponding to the v-th time step predicted by the short-circuit trend prediction model, which is obtained through the timestamp attached to the output value.

[0115] Specifically, the risk window control module 105 obtains future multi-step risk prediction values ​​and makes a forward-looking comparison with the current comprehensive short-circuit characteristic index, so as to transform qualitative risk judgment into quantitative intervention window period, accurately predict the risk evolution speed and remaining disposal time, thereby improving the proactive early warning capability of battery cell detection and the timeliness of system response.

[0116] Specifically, the risk window control module 105 compares the intervention window period CM with the first preset intervention window period CM1 and the second preset intervention window period CM2, where CM1 < CM2. Based on the comparison result, it judges the situation of the intervention window period and outputs a tiered control strategy based on the judgment result, wherein:

[0117] When CM≤CM1, the risk window control module 105 determines that the situation during the intervention window period is highly urgent and outputs the first response strategy as a graded control strategy.

[0118] When CM1<CM≤CM2, the risk window control module 105 determines that the situation during the intervention window period is moderately urgent and outputs the second response strategy as a graded control strategy.

[0119] When CM > CM2, the risk window control module 105 determines that the situation during the intervention window period is of low urgency and outputs the third response strategy as a graded control strategy.

[0120] Specifically, the first preset intervention window period refers to the preset lower limit value for judging the situation during the intervention window period, and the second preset intervention window period refers to the preset upper limit value for judging the situation during the intervention window period. In this embodiment, a whole vehicle thermal runaway simulation model is built in the cloud, and the execution time required for different intervention strategies is simulated. Based on the required execution time, the time division points for executing the first response strategy, the second response strategy, and the third response strategy are 30 seconds and 80 seconds, respectively. Therefore, CM1 = 30 seconds and CM2 = 80 seconds are set. The situation during the intervention window period refers to the urgency of the short circuit event reflected by the intervention window period, including high urgency, medium urgency, and low urgency. The first response strategy includes: immediately disconnecting the battery main circuit, activating the automatic fire extinguishing system, sending an emergency rescue request through the vehicle terminal, and turning on the hazard warning lights. The second response strategy includes: limiting the charging and discharging power to below 20%, activating the battery active cooling system, prompting the driver to go to the nearest repair shop immediately through the vehicle display screen, and recording the fault code. The third response strategy includes: recording the fault information to the battery management system log and suggesting that the driver schedule an inspection through the vehicle system.

[0121] Specifically, the risk window control module 105 divides the urgency level according to the intervention window period and matches differentiated response strategies according to different urgency levels, so as to transform abstract time prediction into specific and executable safety handling actions, thereby improving the timeliness and precision of response to battery short circuit events.

[0122] Specifically, the risk window control module 105 acquires the electro-mechanical coupling value by: normalizing the shell deformation rate D to obtain the shell deformation rate value D1; and calculating the electro-mechanical coupling value MH based on the voltage drop rate value A1, the shell deformation rate value D1, the second weight w4 of the voltage drop rate value, and the weight w5 of the shell deformation rate value, setting MH = A1 × w4 + D1 × w5, and w4 + w5 = 1.

[0123] The electro-force coupling value MH is compared with the preset electro-force coupling value MH0. Based on the comparison result, the state of the electro-force coupling value is judged, and the output process of the graded control strategy is adjusted in stages according to the judgment result.

[0124] When MH≤MH0, the risk window control module 105 determines that the state of the electro-mechanical coupling value is not serious and does not perform graded adjustment on the output process of the graded control strategy;

[0125] When MH > MH0, the risk window control module 105 determines that the state of the electro-mechanical coupling value is severe and adjusts the output process of the graded control strategy in a graded manner.

[0126] Specifically, the second weight of the voltage drop rate value refers to a coefficient that measures the importance of the voltage drop rate value in the electro-mechanical coupling value, and the weight of the casing deformation rate value refers to a coefficient that measures the importance of the casing deformation rate value in the electro-mechanical coupling value. In this embodiment, principal component analysis is performed on the voltage drop rate value A1 and the casing deformation rate value D1, and the loading coefficient of the first principal component is taken as the basis for weight allocation. It is found that the loadings of the two features on the first principal component are similar, so w4=0.5 and w5=0.5 are set. The electro-mechanical coupling value refers to a value that characterizes the coupling strength between electrical abnormalities and mechanical abnormalities inside the battery. The preset electro-mechanical coupling value refers to a preset value for judging the state of the electro-mechanical coupling value. In this embodiment, MH0 is set to 0.65. This is achieved by collecting data under normal operating conditions, such as obtaining the battery's electrical-force coupling values ​​under conditions of rest, constant speed, rapid acceleration, and charging, forming an electrical-force coupling value sequence. This sequence is then sorted in ascending order to obtain a sorted electrical-force coupling value sequence. The 95th percentile of this sorted sequence is used as the preset electrical-force coupling value, resulting in MH0 = 0.65. The state of the electrical-force coupling value refers to the severity of the battery cell short circuit reflected by the electrical-force coupling value, including severe and non-severe. The graded adjustment refers to the process of upgrading the current graded control strategy by one level for output, such as upgrading low emergency to medium emergency, and medium emergency to high emergency.

[0127] Specifically, the risk window control module 105 judges the state of the electro-mechanical coupling value to achieve dual verification of electrical and mechanical characteristics, effectively eliminating false risks caused by interference from a single signal source, thereby improving the reliability of battery short circuit detection.

[0128] Based on the same inventive concept, such as Figure 4 As shown in the figure, this embodiment of the invention also provides a method for a short-circuit detection system for a single battery cell in a new energy vehicle, including:

[0129] Step S1: Collect the target battery signal and battery physical and chemical data;

[0130] Step S2: Perform signal processing on the target battery signal to obtain the processed battery signal, and extract features from the processed battery signal to obtain the target battery features.

[0131] Step S3: Obtain the comprehensive short-circuit characteristic index based on the target battery characteristics, obtain the real-time operating condition quantization value based on the battery physical data, obtain the short-circuit characteristic adaptive threshold based on the target battery characteristics and the real-time operating condition quantization value, and output the short-circuit graded early warning based on the comprehensive short-circuit characteristic index and the short-circuit characteristic adaptive threshold.

[0132] Step S4: Obtain the short circuit condition based on the battery physical data, compare the short circuit condition with the comprehensive short circuit characteristic index to obtain the comparison result, perform window correction on the feature extraction process based on the comparison result, and predict the short circuit trend of the comprehensive short circuit characteristic index to obtain the trend prediction result.

[0133] Step S5: Obtain the intervention window period based on the trend prediction results, output the graded control strategy based on the intervention window period, obtain the electro-mechanical coupling value, and adjust the output process of the graded control strategy in a graded manner based on the electro-mechanical coupling value.

[0134] The technical solution of the present invention has been described above with reference to the preferred embodiments shown in the accompanying drawings. However, it will be readily understood by those skilled in the art that the scope of protection of the present invention is obviously not limited to these specific embodiments. Without departing from the principles of the present invention, those skilled in the art can make equivalent changes or substitutions to the relevant technical features, and the technical solutions after these changes or substitutions will all fall within the scope of protection of the present invention.

Claims

1. A short-circuit detection system for a single battery cell in a new energy vehicle, characterized in that, include: The battery data acquisition module, battery signal processing module, single-cell short-circuit monitoring module, short-circuit trend prediction module, and risk window control module include: The short-circuit trend prediction module is used to acquire short-circuit conditions based on battery physical data, compare the short-circuit conditions with the comprehensive short-circuit characteristic index to obtain the comparison results, and also to perform window correction on the feature extraction process based on the comparison results, and to predict the short-circuit trend of the comprehensive short-circuit characteristic index to obtain the trend prediction results. The risk window control module is used to obtain the intervention window period based on the trend prediction results, and output the graded control strategy according to the intervention window period. It is also used to obtain the electro-mechanical coupling value and adjust the output process of the graded control strategy according to the electro-mechanical coupling value.

2. The new energy vehicle battery cell short-circuit detection system according to claim 1, characterized in that, The battery data acquisition module is used to acquire target battery signals and battery physical and chemical data; The battery signal processing module is used to process the target battery signal to obtain the processed battery signal, and also to extract features from the processed battery signal using feature extraction methods to obtain the target battery features. The single-cell short-circuit monitoring module is used to obtain the comprehensive short-circuit characteristic index based on the target battery characteristics, obtain the real-time operating condition quantification value based on the battery physical data, obtain the short-circuit characteristic adaptive threshold based on the target battery characteristics and the real-time operating condition quantification value, and output the short-circuit graded early warning based on the comprehensive short-circuit characteristic index and the short-circuit characteristic adaptive threshold.

3. The new energy vehicle battery cell short-circuit detection system according to claim 2, characterized in that, The battery signal processing module performs signal processing on the target battery signal, specifically: it processes the voltage signal through a three-level filtering strategy to obtain a processed voltage signal, processes the temperature signal through Kalman filtering to obtain a processed temperature signal, and processes the gas signal through wavelet denoising to obtain a processed gas signal. The processed voltage signal, processed temperature signal, and processed gas signal are then used as the processed battery signal. The battery signal processing module extracts features from the processed battery signal using a feature extraction method, which includes: Step A01: Calculate the voltage drop rate A based on the current voltage value u(t), the previous voltage value u(t-Δt), and the sampling interval Δt, and set A=[u(t)-u(t-Δt)] / Δt; Step A02: Calculate the temperature rise gradient B based on the battery pack's highest temperature Tmax, lowest temperature Tmin, and sampling interval Δt, and set B = (Tmax - Tmin) / Δt. Step A03: Calculate the gas concentration accumulation rate C based on the current gas concentration Cgas(t), the previous gas concentration Cgas(t-Δt), and the sampling interval Δt, and set C=[Cgas(t)-Cgas(t-Δt)] / Δt; Step A04: The voltage drop rate, temperature rise gradient, and gas concentration accumulation rate are taken as target battery characteristics.

4. The new energy vehicle battery cell short-circuit detection system according to claim 3, characterized in that, The single-cell short-circuit monitoring module obtains the comprehensive short-circuit characteristic index based on the characteristics of the target battery, and normalizes the voltage drop rate, temperature rise gradient, and gas concentration accumulation rate to obtain the voltage drop rate value A1, temperature rise gradient value B1, and gas concentration accumulation rate value C1. The comprehensive short-circuit characteristic index SSY is calculated based on the voltage drop rate value A1, temperature rise gradient value B1, gas concentration accumulation rate value C1, voltage drop rate weight w1, temperature rise gradient value weight w2, and gas concentration accumulation rate weight w3. SSY is set as A1×w1+B1×w2+C1×w3, and w1+w2+w3=1. The real-time operating condition quantization value Gk is obtained based on the battery physical data, and the short-circuit feature adaptive threshold is obtained based on the target battery characteristics and the real-time operating condition quantization value. The short-circuit feature adaptive threshold includes a first adaptive threshold SSY1 and a second adaptive threshold SSY2. The voltage drop rate value A1 and the temperature rise gradient value B1 are fused to obtain the feature fusion index Tz. Based on the real-time operating condition quantization value Gk, the feature fusion index Tz, the basic threshold weight set Wb={wb1,wb2}, and the operating condition sensitivity coefficient set Ks={ks1,ks2}, the first adaptive threshold SSY1 and the second adaptive threshold SSY2 are calculated. The following settings are made: SSY1=wb1×[1+ks1×Gk]×[1+Tz], SSY2=wb2×[1+ks2×Gk]×[1+Tz], and wb1<wb2, ks1<ks2, SSY1<SSY2; The comprehensive short-circuit characteristic index SSY is compared with the first adaptive threshold SSY1 and the second adaptive threshold SSY2. Based on the comparison result, the state of the comprehensive short-circuit characteristic index is determined, and a short-circuit graded early warning is output based on the determination result. Wherein: When SSY≤SSY1, the single-unit short-circuit monitoring module determines the state of the comprehensive short-circuit characteristic index to be low and outputs the primary warning as a short-circuit graded warning. When SSY1<SSY≤SSY2, the single-unit short-circuit monitoring module determines the state of the comprehensive short-circuit characteristic index to be moderate and outputs the intermediate warning as a short-circuit graded warning. When SSY > SSY2, the single-unit short-circuit monitoring module determines the state of the comprehensive short-circuit characteristic index to be high and outputs the advanced warning as a short-circuit graded warning.

5. The new energy vehicle battery cell short-circuit detection system according to claim 4, characterized in that, The short-circuit trend prediction module acquires the short-circuit situation based on the battery physicochemical data, compares the gas proportion factor Qt with the gas composition fingerprint spectrum F={f1,f2,...,fn} to obtain the similarity set S={s1,s2,...,sn}, where n represents the order, n=1,2,...,n; The maximum similarity max(S) is obtained based on the similarity set S. The maximum similarity max(S) is then compared with a preset maximum similarity max(S0). Based on the comparison result, the state of the maximum similarity is determined, and the short-circuit condition is output according to the determination result. Where: When max(S)≤max(S0), the short-circuit trend prediction module determines the state of maximum similarity as dissimilar and outputs the absence of short-circuit cause as a short-circuit situation; When max(S)>max(S0), the short-circuit trend prediction module determines the state of maximum similarity as similar and outputs the existence of a short-circuit cause as a short-circuit situation.

6. The new energy vehicle battery cell short-circuit detection system according to claim 5, characterized in that, The short-circuit trend prediction module compares the short-circuit situation with the comprehensive short-circuit characteristic index, specifically as follows: When a short circuit occurs because there is a cause for the short circuit: If SSY≤SSY1, the comparison result is determined to be conflicting, and the feature extraction process is windowed. Otherwise, if the comparison result is determined to be without conflict, no window correction will be performed on the feature extraction process; When the short circuit condition is that there is no cause for the short circuit: If SSY≤SSY1, the comparison result is determined to be that there is no conflict, and no window correction is performed in the feature extraction process; Otherwise, if the comparison result indicates a conflict, a window correction is performed on the feature extraction process. The process of window correction for feature extraction is as follows: extend the time window of feature extraction, extend and correct the sampling interval Δt according to the window correction factor kc1 to obtain the extended sampling interval Δt1, set Δt1=Δt×(1+kc1), and 0.2≤kc1≤0.5, and use the extended sampling interval Δt1 as the sampling interval Δt to re-extract the target battery features.

7. The new energy vehicle battery cell short-circuit detection system according to claim 6, characterized in that, The short-circuit trend prediction module performs short-circuit trend prediction on the comprehensive short-circuit characteristic index: the target battery characteristics and real-time operating condition quantification values ​​are input into the short-circuit trend prediction model to obtain the trend prediction result Ypv output by the short-circuit trend prediction model, where v represents the order of the values ​​in the trend prediction result, and v=1, 2, 3, ..., v. The comprehensive short-circuit characteristic index SSY is compared with the trend prediction result Ypv. Based on the comparison result, the state of the comprehensive short-circuit characteristic index is judged, and the intervention window period is obtained based on the judgment result. When SSY≤Ypv, the risk window control module determines that the state of the comprehensive short-circuit characteristic index is safe and does not acquire the intervention window period; When SSY > Ypv, the risk window control module determines that the state of the comprehensive short-circuit characteristic index is unsafe and obtains the intervention window period: the intervention window period CM is calculated from the trend prediction time starting point t1 to the time point tv corresponding to Ypv, and CM is set as tv-t1.

8. The new energy vehicle battery cell short-circuit detection system according to claim 7, characterized in that, The risk window control module compares the intervention window period CM with the first preset intervention window period CM1 and the second preset intervention window period CM2, where CM1 < CM2. Based on the comparison result, it judges the situation of the intervention window period and outputs a tiered control strategy based on the judgment result, wherein: When CM≤CM1, the risk window control module determines that the situation during the intervention window period is highly urgent and outputs the first response strategy as a graded control strategy. When CM1<CM≤CM2, the risk window control module determines that the situation during the intervention window period is moderately urgent and outputs the second response strategy as a graded control strategy. When CM > CM2, the risk window control module determines that the situation during the intervention window period is of low urgency and outputs the third response strategy as a graded control strategy.

9. The new energy vehicle battery cell short-circuit detection system according to claim 8, characterized in that, The risk window control module acquires the electro-mechanical coupling value by: normalizing the shell deformation rate D to obtain the shell deformation rate value D1; and calculating the electro-mechanical coupling value MH based on the voltage drop rate value A1, the shell deformation rate value D1, the second weight w4 of the voltage drop rate value, and the weight w5 of the shell deformation rate value, setting MH = A1 × w4 + D1 × w5, and w4 + w5 = 1. The electro-force coupling value MH is compared with the preset electro-force coupling value MH0. Based on the comparison result, the state of the electro-force coupling value is judged, and the output process of the graded control strategy is adjusted in stages according to the judgment result. When MH≤MH0, the risk window control module determines that the state of the electro-mechanical coupling value is not serious and does not perform graded adjustment on the output process of the graded control strategy; When MH > MH0, the risk window control module determines that the state of the electro-mechanical coupling value is severe and adjusts the output process of the graded control strategy in a graded manner.

10. A method for applying a short-circuit detection system for a new energy vehicle battery cell as described in any one of claims 1-9, comprising: Step S1: Collect the target battery signal and battery physical and chemical data; Step S2: Perform signal processing on the target battery signal to obtain the processed battery signal, and extract features from the processed battery signal to obtain the target battery features. Step S3: Obtain the comprehensive short-circuit characteristic index based on the target battery characteristics, obtain the real-time operating condition quantization value based on the battery physical data, obtain the short-circuit characteristic adaptive threshold based on the target battery characteristics and the real-time operating condition quantization value, and output the short-circuit graded early warning based on the comprehensive short-circuit characteristic index and the short-circuit characteristic adaptive threshold. Step S4: Obtain the short circuit condition based on the battery physical data, compare the short circuit condition with the comprehensive short circuit characteristic index to obtain the comparison result, perform window correction on the feature extraction process based on the comparison result, and predict the short circuit trend of the comprehensive short circuit characteristic index to obtain the trend prediction result. Step S5: Obtain the intervention window period based on the trend prediction results, output the graded control strategy based on the intervention window period, obtain the electro-mechanical coupling value, and adjust the output process of the graded control strategy in a graded manner based on the electro-mechanical coupling value.

Images