A battery thermal runaway early warning method and device

Abstract

The application provides a battery thermal runaway early warning method and device, comprising: collecting original acoustic signals released by a target battery pack by using an acoustic sensor; obtaining working condition information of the target battery pack, performing dynamic noise reduction processing on the original acoustic signals based on the working condition information, suppressing multi-source interference noise, obtaining de-noised acoustic signals, and extracting acoustic features of the de-noised acoustic signals in a preset frequency band; constructing a thermal runaway acoustic trigger index based on an energy threshold, a combined feature change threshold and a continuous trigger time, and generating an initial early warning signal according to the thermal runaway acoustic trigger index and the acoustic features; obtaining state information of the target battery pack, and performing multi-modal fusion analysis based on the state information and the initial early warning signal to generate a thermal runaway verification signal; and performing thermal runaway graded early warning on the target battery pack based on the thermal runaway verification signal and the initial early warning signal. The application can improve the accuracy of thermal runaway early warning.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of batteries, and particularly relates to a battery thermal runaway early warning method and device. BACKGROUND

[0002] With the large-scale application of lithium ion batteries in the fields of electric vehicles and energy storage systems, the thermal runaway safety problem is increasingly prominent. Traditional monitoring methods (such as temperature and voltage sensors) can only trigger an alarm when thermal runaway has occurred, with a serious response lag, which is difficult to meet the early warning needs. Therefore, a monitoring technology that can identify thermal runaway in the early stage or even the incubation period is urgently needed.

[0003] Sound detection technology can issue an early warning several minutes to tens of minutes before thermal runaway occurs by capturing acoustic signals (such as electrolyte vaporization, separator rupture, lithium dendrite fracture, etc.) generated by micro-failure inside the battery, thus gaining critical time for safety protection. Studies have shown that the battery will release acoustic signals of specific frequency bands before thermal runaway: the early stage (20-100 kHz ultrasonic wave) corresponds to SEI film decomposition and micro-short circuit, and the middle and late stages (1-20 kHz audible sound) correspond to large-scale separator collapse and electrolyte boiling. Compared with traditional sensors, acoustic monitoring has the advantages of high sensitivity, fast response speed, and is not affected by external environmental temperature, and can directly reflect the internal state of the battery.

[0004] In the prior art, the application of acoustic monitoring still faces problems such as strong environmental noise interference and high false alarm rate. SUMMARY

[0005] The technical problem to be solved by the present application is to provide a battery thermal runaway early warning method and device to improve the accuracy of thermal runaway early warning.

[0006] In a first aspect, the present application provides a battery thermal runaway early warning method, which comprises the following steps:

[0007] An acoustic sensor is used to collect original acoustic signals released by a target battery pack; the acoustic sensor is pre-arranged or arranged in real time inside or on the surface of the target battery pack;

[0008] Working condition information of the target battery pack is obtained, the original acoustic signals are dynamically denoised based on the working condition information to suppress multi-source interference noise, denoised acoustic signals are obtained, and acoustic features of the denoised acoustic signals in a preset frequency band are extracted; the multi-source interference noise includes cooling system noise, mechanical vibration noise and electromagnetic interference noise;

[0009] A thermal runaway acoustic trigger index is constructed based on an energy threshold, a combination feature change threshold and a continuous trigger time, and an initial early warning signal is generated according to the thermal runaway acoustic trigger index and the acoustic features;

[0010] Obtain state information of the target battery pack, and perform multi-modal fusion analysis based on the state information and the initial early warning signal to generate a thermal runaway verification signal; the thermal runaway verification signal is used to verify whether a causal time sequence relationship corresponding to the initial early warning signal conforms to a physical law;

[0011] Based on the thermal runaway verification signal and the initial early warning signal, a thermal runaway graded early warning is performed on the target battery pack.

[0012] Optionally, the dynamic noise reduction processing includes:

[0013] Based on an adaptive filtering algorithm, a reference noise signal collected from the vicinity of the noise source is used to dynamically cancel the cooling system noise in the original acoustic signal to obtain a first acoustic signal;

[0014] The first acoustic signal is subjected to energy-weighted wavelet packet decomposition, and high-energy subbands are screened and reconstructed to suppress mechanical vibration noise to obtain a second acoustic signal;

[0015] Based on a high-frequency current ripple signal of the target battery pack, electromagnetic interference noise in the second acoustic signal is identified and filtered out to obtain a third acoustic signal;

[0016] A pre-trained U-Net neural network is used to reduce noise of a time-frequency spectrum of the third acoustic signal to obtain a de-noised acoustic signal.

[0017] Optionally, the preset frequency band is 3KHZ to 8KHZ.

[0018] Optionally, an expression of the thermal runaway acoustic trigger index is:

[0019]

[0020]

[0021]

[0022]

[0023]

[0024] wherein, the thermal runaway acoustic trigger index is used to indicate whether to generate the initial early warning signal, and when the value is 1, the initial early warning signal is generated, and when the value is 0, the initial early warning signal is not generated, the thermal runaway acoustic early warning instantaneous trigger index is, the thermal runaway acoustic early warning continuous trigger index is, the time-frequency spectrum of the de-noised acoustic signal is, the background noise energy baseline is, a decibel scale corresponding to an acoustic feature of the denoised acoustic signal in a preset frequency band, , represents a weight, represents a zero-crossing rate of the denoised acoustic signal in the 3-8 KHZ frequency band, representing the frequency component and stationarity of the denoised acoustic signal, represents a short-time energy of the denoised acoustic signal in the 3-8 KHZ frequency band, representing the intensity of the denoised acoustic signal, represents a spectral flux of the denoised acoustic signal in the 3-8 KHZ frequency band, represents a sustained trigger time of the acoustic trigger early warning, represents a Fourier transform, represents an expression of the denoised acoustic signal.

[0025] Optionally, the state information of the target battery pack includes cell temperature information, cell voltage information, and state information of auxiliary equipment; the auxiliary equipment includes a cooling system and a power conversion system.

[0026] Optionally, multi-modal fusion analysis is performed based on the state information and the initial early warning signal to generate a thermal runaway verification signal, including:

[0027] adjusting the confidence weight of the initial early warning signal and / or the battery pack state information according to the state information of the auxiliary equipment;

[0028] constructing a timing verification function; the timing verification function is used to verify whether the occurrence sequence of the initial early warning signal, the cell temperature sudden rise signal, the cell voltage sudden drop signal, and the fire trigger signal conforms to the causal timing chain of thermal runaway;

[0029] integrating and calculating the initial early warning signal, the cell temperature change rate, and the cell voltage sudden drop amplitude according to the confidence weight and the timing verification function to generate a thermal runaway verification signal.

[0030] Optionally, adjusting the confidence weight of the initial early warning signal and / or the battery pack state information according to the state information of the auxiliary equipment includes:

[0031] by a calculation formula

[0032]

[0033]

[0034]

[0035]

[0036]

[0037] obtaining a confidence weight ; wherein, represents a cell temperature weight, represents a cold state, represents a fault, represents a shutdown, represents normal operation, represents a water pump speed, represents a water pump minimum speed, represents a confidence of a power conversion system PCS, is used to indicate the operating state of the power conversion system, is used to indicate the fire fighting trigger state, represents a gain coefficient.

[0038] Optionally, the expression of the timing verification function is:

[0039]

[0040] wherein, represents the first trigger time of the initial early warning signal, specifically, the starting time when the value changes from 0 to 1 and the value is 1 for five consecutive frames, represents the time when the temperature rises sharply and the temperature change rate is greater than a preset temperature change rate threshold, represents the time when the voltage drops sharply and the voltage change rate is greater than a preset voltage change rate threshold, represents the rising edge trigger time of the dry node signal of the fire fighting system, represents a temperature response delay, represents a short circuit formation speed, represents a heat spread to fire fighting trigger time, represents a full chain reaction window.

[0041] Optionally, the expression of the thermal runaway verification signal is:

[0042]

[0043] wherein, represents the thermal runaway verification signal value, represents an acoustic trigger signal, which is a Boolean quantity, and is determined by an acoustic trigger early warning value , represents a temperature change rate, represents a voltage drop amplitude, represents a confidence weight, represents a Sigmoid activation function, H(t) is a Heaviside function, and the voltage drop caused by thermal runaway is a sudden event, which is a step change in nature and is suitable for being described by the Heaviside function.

[0044] In a second aspect, the present application provides a battery thermal runaway early warning device, comprising:

[0045] The acquisition module is configured to acquire original acoustic signals released by the target battery pack by using an acoustic sensor; the acoustic sensor is arranged in advance or in real time inside or on the surface of the target battery pack;

[0046] The denoising module is configured to obtain working condition information of the target battery pack, perform dynamic denoising processing on the original acoustic signals based on the working condition information, suppress multi-source interference noise, obtain denoised acoustic signals, and extract acoustic features of the denoised acoustic signals in a preset frequency band; the multi-source interference noise includes cooling system noise, mechanical vibration noise, and electromagnetic interference noise.

[0047] The first early warning module is configured to construct a thermal runaway acoustic trigger index based on an energy threshold, a combined feature change threshold, and a continuous trigger time, and generate an initial early warning signal according to the thermal runaway acoustic trigger index and the acoustic features;

[0048] The verification module is configured to obtain state information of the target battery pack, and perform multi-modal fusion analysis based on the state information and the initial early warning signal to generate a thermal runaway verification signal; the thermal runaway verification signal is used to verify whether a causal time sequence relationship corresponding to the initial early warning signal conforms to a physical law.

[0049] The second early warning module is configured to perform thermal runaway hierarchical early warning on the target battery pack based on the thermal runaway verification signal and the initial early warning signal.

[0050] The present application has at least the following beneficial effects:

[0051] The working condition information is used to perform dynamic denoising processing on the original acoustic signals to suppress the cooling system noise, the mechanical vibration noise, and the electromagnetic interference noise, so that multi-source noise separation can be realized, and the accuracy of subsequent battery thermal runaway early warning can be improved; the state information of the target battery pack is obtained, and multi-modal fusion analysis is performed based on the state information and the initial early warning signal, instead of relying on the acoustic signal alone, but innovatively linking the acoustic signal with temperature, voltage, and cooling system, PCS, fire extinguishing and other auxiliary equipment states provided by the BMS to verify whether a causal time sequence relationship corresponding to the initial early warning signal conforms to a physical law, so as to solve the problem of multi-modal fusion misassociation and improve the accuracy of subsequent battery thermal runaway early warning. BRIEF DESCRIPTION OF DRAWINGS

[0052] The accompanying drawings are used to provide a further understanding of the technical solutions of the present application, and constitute a part of the specification, and are used to explain the technical solutions of the present application together with the embodiments of the present application, and do not constitute a limitation on the technical solutions of the present application.

[0053] Figure 1 A flowchart of a battery thermal runaway early warning method in one of the embodiments of the present application;

[0054] Figure 2 A structural diagram of a battery thermal runaway early warning system in another of the embodiments of the present application;

[0055] Figure 3 A structural diagram of a battery thermal runaway early warning device in one of the embodiments of the present application. DETAILED DESCRIPTION

[0056] The technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, but not all the embodiments of the present application. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative work fall within the scope of protection of the present application.

[0057] In view of the problems of strong environmental noise interference and high false alarm rate of the traditional battery thermal runaway early warning method, the present application provides a battery thermal runaway early warning method and device. The original acoustic signal is dynamically denoised based on working condition information, the cooling system noise, mechanical vibration noise and electromagnetic interference noise are suppressed, multi-source noise separation can be achieved, which is conducive to improving the accuracy of subsequent battery thermal runaway early warning. The state information of the target battery pack is obtained, and multi-modal fusion analysis is performed based on the state information and the initial early warning signal. Instead of relying on acoustic signals alone, the temperature, voltage and state of auxiliary equipment such as cooling system, PCS and fire extinguishing provided by BMS are innovatively analyzed in linkage, to verify whether the causal time sequence relationship corresponding to the initial early warning signal conforms to the physical law, which is conducive to solving the problem of multi-modal fusion misassociation and improving the accuracy of subsequent battery thermal runaway early warning.

[0058] Embodiment 1

[0059] As shown in the figure, the battery thermal runaway early warning method includes the following steps: Figure 1

[0060] Step 11, using an acoustic sensor to collect the original acoustic signal released by the target battery pack.

[0061] In the embodiments of the present application, the acoustic sensor is designed to cover the acoustic characteristics of the whole thermal runaway stage, and the acoustic sensor is prearranged or arranged in real time inside or on the surface of the target battery pack. ​

[0062] Due to the existence of noise interference, the original acoustic signal released by the target battery pack is a mixed acoustic signal of multi-source interference signals such as cooling system, mechanical vibration, electromagnetic interference and the acoustic signal of the opening of the battery pack safety valve, and the original acoustic signal can be expressed as: , wherein, represents the acoustic signal of the opening of the battery pack safety valve, represents the cooling system noise, represents the mechanical vibration noise, represents the electromagnetic interference noise.

[0063] Step 12, obtaining the working condition information of the target battery pack, performing dynamic noise reduction processing on the original acoustic signal based on the working condition information, suppressing the multi-source interference noise, obtaining the de-noised acoustic signal, and extracting the acoustic features of the de-noised acoustic signal in the preset frequency band.

[0064] In the embodiment of the present application, the multi-source interference noise includes cooling system noise, mechanical vibration noise and electromagnetic interference noise.

[0065] For example, the sound sources and characteristics of the multi-source interference noise are shown in Table 1.

[0066] Table 1

[0067] .

[0068] It should be noted that in the embodiment of the present application, the de-noised acoustic signal is specifically the opening sound signal of the battery safety valve. The opening of the battery safety valve is a landmark physical event in the thermal runaway process. When thermal runaway occurs inside the battery, the electrolyte boils to produce a large amount of gas, causing the internal pressure of the battery to rise sharply. When the pressure reaches the opening threshold of the safety valve (usually 1-2 MPa), the safety valve will mechanically open to release the internal pressure. This process inevitably produces a characteristic acoustic signal. Specifically, the opening of the battery safety valve is a physical process that is completed instantaneously, producing a burst acoustic signal. The opening of the safety valve releases high-pressure gas, producing a high-energy acoustic signal (usually greater than 30 dB). The frequency of the opening of the battery safety valve is concentrated in the 3-8 kHz characteristic frequency band, which is significantly different from the background noise. Moreover, the opening sound of the battery safety valve occurs in the middle and late stages of the thermal runaway process, but is earlier than the complete outbreak of thermal runaway and the fire stage. This time window is crucial for taking emergency measures (such as power failure and starting the fire extinguishing system), and can issue a warning several minutes to several tens of minutes before thermal runaway occurs, which is crucial for safety protection, effectively preventing the spread of thermal runaway and fire.

[0069] It should be noted that the acoustic characteristics of the extracted de-noised acoustic signal in the preset frequency band are specifically: extracting the acoustic combination characteristics of the de-noised acoustic signal (battery safety valve opening sound) in the 3-8 kHz frequency band. Among them, the acoustic combination characteristics include burstiness, high energy, and spectral mutation characteristics. Specifically, the expression of the acoustic combination characteristics is , represents the weight, represents the zero-crossing rate of the de-noised acoustic signal in the 3-8 kHz frequency band, representing the frequency component and stationarity of the de-noised acoustic signal, represents the short-time energy of the de-noised acoustic signal in the 3-8 kHz frequency band, representing the intensity of the de-noised acoustic signal, represents the spectral flux of the de-noised acoustic signal in the 3-8 kHz frequency band.

[0070] In the embodiment of the present application, the dynamic noise reduction process includes steps 12.1.1 to 12.1.4.

[0071] Step 12.1.1, based on an adaptive filtering algorithm, using a reference noise signal collected from the vicinity of the noise source to dynamically cancel the cooling system noise in the original acoustic signal, to obtain a first acoustic signal.

[0072] Specifically, in combination with the noise template signal collected separately from the vicinity of the noise source (such as cooling fan, water pump) and the measured data of the cooling fan or water pump impeller speed, the cooling system noise is dynamically canceled.

[0073] In a feasible implementation, the expression of the first acoustic signal is as follows:

[0074]

[0075]

[0076]

[0077]

[0078] wherein, represents the cooling system noise estimate, represents the reference noise signal, which is a noise template signal collected separately from the vicinity of the noise source (such as cooling fan, water pump), and usually does not contain the opening sound of the battery safety valve, represents the original acoustic signal and the cooling system noise estimate , representing the residual signal after noise cancellation, represents the weight, represents the step factor, represents the leakage factor, represents the initial number of rotations per minute of the cooling fan or water pump impeller, represents the initial number of rotations per minute of the cooling fan or water pump impeller, represents the initial step size.

[0079] Step 12.1.2, energy-weighted wavelet packet decomposition is performed on the first acoustic signal, high-energy subbands are screened and reconstructed to suppress mechanical vibration noise, and a second acoustic signal is obtained.

[0080] It should be noted that mechanical vibration noise is usually periodic and relatively dispersed narrowband signal, and the opening valve sound has the characteristics of burst and high energy, and the energy is concentrated in a certain period of time. By retaining high-energy subbands, the characteristics of the opening valve sound can be effectively retained, and mechanical vibration noise generated by compressor, relay switch or support resonance can be effectively filtered out, thereby greatly reducing the false alarm rate of the system.

[0081] Specifically, by calculating the energy proportion of each wavelet packet subband, high-energy subbands are dynamically screened and weighted reconstructed, and low-energy noise is suppressed while retaining effective signals such as opening valve sound, thereby realizing rapid noise reduction of mechanical vibration noise.

[0082] In a feasible implementation, the expression of the second acoustic signal is as follows:

[0083]

[0084]

[0085] wherein, represents the second acoustic signal, represents wavelet packet decomposition, represents the decomposition level, represents the subband index set of the Top50% energy, represents the energy of the th wavelet packet subband.

[0086] Step 12.1.3, based on the high-frequency current ripple signal of the target battery pack, electromagnetic interference noise in the second acoustic signal is identified and filtered out, and a third acoustic signal is obtained.

[0087] Specifically, by detecting high-frequency electromagnetic noise (such as pulse interference generated by the inverter / PCS) unrelated to the opening valve sound, the strong correlation between the high-frequency electromagnetic noise and the current ripple is used for identification and filtering.

[0088] In a feasible implementation, the expression of the third acoustic signal is as follows:

[0089]

[0090] wherein, represents the third acoustic signal, represents the battery pack high-frequency current ripple, which can be obtained by a Hall sensor + filtering, represents the time delay range.

[0091] Step 12.1.4, denoising the time-frequency spectrum of the third acoustic signal by using a pre-trained U-Net neural network to obtain a denoised acoustic signal.

[0092] Specifically, the U-Net time-frequency denoising learns the mapping relationship between the noisy time-frequency spectrum and the pure time-frequency spectrum through an encoding-decoding architecture, and uses a skip connection to preserve high-frequency details.

[0093] In a feasible implementation, the expression of the denoised acoustic signal is:

[0094]

[0095] wherein, represents the inverse short-time Fourier transform, represents the short-time Fourier transform, represents a convolutional neural network algorithm of an encoder-decoder structure.

[0096] Step 13, constructing a thermal runaway acoustic trigger index based on an energy threshold, a combined feature change threshold, and a continuous trigger time, and generating an initial early warning signal according to the thermal runaway acoustic trigger index and the acoustic features.

[0097] Specifically, the triple criteria of the energy threshold (for example, 30 dB), the combined feature change rate threshold (for example, 1000 dB / s), and the continuous trigger time (for example, 50 ms to 200 ms) trigger the warning, and accurately identify the burst signal of the opening valve sound in a complex noise environment. The triple criteria are associated with the logical AND (AND) of the time window, which can filter long-time interference (> 200 ms) and short-time pulses (< 50 ms), and meet the real-time energy threshold (> 30 dB), the real-time combined feature change rate threshold (> 1000 dB / s).

[0098] In a feasible implementation, the expression of the thermal runaway acoustic trigger index is:

[0099]

[0100]

[0101]

[0102]

[0103]

[0104] wherein, represents a thermal runaway acoustic trigger index for indicating whether to generate an initial warning signal, and the initial warning signal is generated when the value is 1, and the initial warning signal is not generated when the value is 0, is a thermal runaway acoustic warning instantaneous trigger index, is a thermal runaway acoustic warning continuous trigger index, represents a time-frequency spectrum of a denoised acoustic signal, represents a background noise energy baseline, represents a continuous trigger time of an acoustic trigger warning, represents a Fourier transform, represents an expression of the denoised acoustic signal.

[0105] Step 14, obtaining state information of the target battery pack, and performing multi-modal fusion analysis based on the state information and the initial warning signal to generate a thermal runaway verification signal.

[0106] The thermal runaway verification signal is used to verify whether a causal time sequence relationship corresponding to the initial warning signal conforms to a physical law.

[0107] In the embodiment of the application, the state information of the target battery pack includes cell temperature information, cell voltage information, and state information of auxiliary equipment; the auxiliary equipment includes a cooling system and a power conversion system.

[0108] Specifically, the multi-modal fusion analysis based on the state information and the initial warning signal to generate the thermal runaway verification signal includes steps 14.1 to 14.3.

[0109] Step 14.1, adjusting the initial warning signal and / or the confidence weight of the battery pack state information according to the state information of the auxiliary equipment.

[0110] Specifically, a cooling system failure (cold machine failure, water pump speed too low) causes a sudden temperature change similar to thermal runaway, at which time the temperature weight should be reduced; the electromagnetic noise will cover the real valve opening sound when the PCS fails, at which time the power conversion system confidence should be reduced; the fire extinguishing is the end of the chain reaction of thermal runaway, and the fire extinguishing trigger should be given greater weight as a strong confirmation signal.

[0111] In a feasible implementation manner, the thermal runaway verification signal is calculated by the following formula

[0112]

[0113]

[0114]

[0115]

[0116]

[0117] Obtain confidence weights ;in, Indicates cell temperature weighting. Indicates the cold state of the machine. Indicates a fault. Indicates that it is closed. This indicates that the system is operating normally. Indicates the water pump speed. Indicates the minimum speed of the water pump. This represents the confidence level of the power conversion system PCS. Used to indicate the operating status of the power conversion system Used to indicate the fire alarm trigger status. This represents the gain coefficient, which is set to 0.5 based on the CATL thermal runaway experimental dataset.

[0118] Step 14.2: Construct the timing verification function.

[0119] The timing verification function is used to verify whether the order of occurrence of the initial warning signal, the cell temperature surge signal, the cell voltage drop signal, and the fire trigger signal conforms to the causal timing chain of thermal runaway.

[0120] In one feasible implementation, the expression for the timing verification function is:

[0121]

[0122] in, This indicates the time when the initial warning signal is first triggered, specifically... The starting moment when the value changes from 0 to 1 and remains 1 for five consecutive frames. This indicates the moment when the temperature rises sharply and the rate of temperature change exceeds a preset temperature change rate threshold. This indicates the moment when the voltage drops sharply and the rate of voltage change exceeds a preset voltage change rate threshold. Indicates the trigger time of the rising edge of the dry contact signal in the fire protection system. Indicates temperature response delay. Indicates the speed at which a short circuit forms. This indicates the time it takes for heat to spread to trigger a fire alarm. This indicates the full-chain reaction window.

[0123] In one feasible implementation, the temperature response delay is set to 500ms, the short circuit formation speed is set to 300ms, the time for heat propagation to fire alarm triggering is set to 800ms, and the full-chain reaction window is set to 2000ms.

[0124] Step 14.3, according to the confidence weight and the time sequence verification function, the initial early warning signal, the temperature change rate of the battery cell, and the voltage drop amplitude are integrated to generate a thermal runaway verification signal.

[0125] Specifically, the expression of the thermal runaway verification signal is:

[0126]

[0127] represents the value of the thermal runaway verification signal, represents the acoustic trigger signal, which is a Boolean quantity, determined by the acoustic trigger early warning value represents the temperature change rate, represents the voltage drop amplitude, represents the confidence weight, represents the Sigmoid activation function, represents the Heaviside function, since the voltage drop caused by thermal runaway is a sudden event, its physical nature is a step change (such as membrane collapse → internal short circuit → voltage drop), which is suitable for describing with Heaviside function.

[0128] Step 15, based on the thermal runaway verification signal and the initial early warning signal, the target battery pack is given a thermal runaway early warning.

[0129] Specifically, it includes first to third level early warning:

[0130] The first level is the millisecond-level acoustic early warning, which is completed independently based on the initial early warning signal;

[0131] The second level is the second-level response fusion verification early warning, which confirms the thermal runaway risk based on the thermal runaway verification signal;

[0132] The third level is the protection execution level, which confirms the thermal runaway risk and executes the emergency shutdown operation by the battery management system.

[0133] In a feasible implementation, by capturing the 3-8 kHz valve opening sound characteristics (energy > 30 dB, combined feature change rate > 1000 / s, duration 50-200 ms) in real time, the first level early warning (millisecond level) is triggered; combined with acoustic signal, temperature change rate (> 5℃ / s) and voltage drop (> 0.5V / 100ms) and auxiliary equipment state, the interference is excluded by causal time sequence verification, the thermal runaway risk is confirmed, and the second level early warning (second level) is triggered; after confirming the thermal runaway risk, the emergency shutdown operation is executed by the battery management system, and the third level early warning (second level) is triggered.

[0134] ​​It is worth mentioning that based on the working condition information, the original acoustic signal is dynamically denoised, the cooling system noise, mechanical vibration noise and electromagnetic interference noise are suppressed, multi-source noise separation can be realized, and the accuracy of subsequent battery thermal runaway early warning can be improved; the state information of the target battery pack is obtained, and multi-modal fusion analysis is carried out based on the state information and the initial early warning signal, not single dependent on the acoustic signal, but innovatively linking the temperature, voltage and cooling system, PCS, fire fighting and other auxiliary equipment state provided by the BMS for linkage analysis, verifying whether the causal time sequence relationship corresponding to the initial early warning signal conforms to the physical law, which is helpful to solve the problem of multi-modal fusion misassociation, and improve the accuracy of subsequent battery thermal runaway early warning.

[0135] Embodiment 2

[0136] In the embodiments of the present application, a battery thermal runaway early warning system is also provided, in which the modules cooperate with each other to execute the battery thermal runaway early warning method in embodiment 1.

[0137] Specifically, as shown in Figure 2 The battery thermal runaway early warning system is composed of a sound detection device, an early thermal runaway early warning host, a BMS and a battery pack, adopts a two-level sound detection system design, and the sound detection device with a specific array distribution completes the collection of specific frequency band acoustic signals released by each battery pack; the early thermal runaway early warning host completes the collection and processing of the thermal runaway early warning information reported by each sound detection device, and synchronously receives the cell temperature, cell voltage information and auxiliary equipment state information of each battery pack provided by the BMS to realize thermal runaway early warning auxiliary judgment; the BMS receives the thermal runaway early warning signal sent by the early thermal runaway early warning host and executes emergency shutdown to realize the protection of the battery.

[0138] Specifically, the sound detection device realizes early thermal runaway early warning and hierarchical safety response by monitoring the abnormal acoustic signals (such as electrolyte boiling, diaphragm rupture, valve opening, etc.) in the battery pack in real time and combining multi-module cooperative processing.

[0139] The main modules of the sound detection device include:

[0140] Sound acquisition module: adopt a microphone to cover the acoustic characteristics of the whole thermal runaway stage.

[0141] Sound signal processing module: realize signal processing and dynamic noise reduction, realize acoustic feature extraction through a specific algorithm, and output acoustic trigger early warning signal.

[0142] Communication module: realize data interaction with the early thermal runaway early warning host, and can realize the uploading of acoustic signal and thermal runaway preliminary early warning.

[0143] The early thermal runaway early warning host machine collects the thermal runaway early warning original data and the hierarchical early warning signal reported by each sound detection device in real time, collects the battery cell temperature, the cell voltage information and the auxiliary equipment state information reported by the BMS, and realizes the positioning of the thermal runaway battery pack by a specific early thermal runaway early warning algorithm, and finally uploads the BMS and executes the emergency shutdown by the BMS to realize the protection of the battery. The main modules of the early thermal runaway early warning host machine include:

[0144] The thermal runaway early warning processing module obtains the thermal runaway early warning original data and the hierarchical early warning signal, the battery cell temperature and voltage information and the auxiliary equipment state information from the communication module, and realizes the positioning information output of the thermal runaway battery pack by a specific algorithm and uploads the BMS through the communication module.

[0145] The communication module realizes the data interaction with the sound detection device and the BMS, can realize the collection of acoustic signals and thermal runaway preliminary early warning, and can realize the uploading of thermal runaway early warning.

[0146] Embodiment 3

[0147] In the embodiments of the present application, the present application also provides a battery thermal runaway early warning device, as shown in the figure, the battery thermal runaway early warning device 300 includes: Figure 3 The acquisition module 301 is used for acquiring the original acoustic signal released by the target battery pack by using an acoustic sensor; the acoustic sensor is arranged in the interior or on the surface of the target battery pack in advance or in real time;

[0148] The denoising module 302 is used for obtaining the working condition information of the target battery pack, performing dynamic noise reduction processing on the original acoustic signal based on the working condition information, suppressing multi-source interference noise to obtain a denoised acoustic signal, and extracting acoustic features of the denoised acoustic signal in a preset frequency band; the multi-source interference noise includes cooling system noise, mechanical vibration noise and electromagnetic interference noise;

[0149] The first early warning module 303 is used for constructing a thermal runaway acoustic trigger index based on an energy threshold, a combined feature change threshold and a continuous trigger time, and generating an initial early warning signal according to the thermal runaway acoustic trigger index and the acoustic features;

[0150] The verification module 304 is used for obtaining the state information of the target battery pack, and performing multi-modal fusion analysis based on the state information and the initial early warning signal to generate a thermal runaway verification signal; the thermal runaway verification signal is used to verify whether the causal time sequence relationship corresponding to the initial early warning signal conforms to the physical law;

[0151] The second early warning module 305 is used for performing thermal runaway hierarchical early warning on the target battery pack based on the thermal runaway verification signal and the initial early warning signal.

[0152]

[0153] ​It should be noted that the information interaction, execution process and the like between the above apparatuses / units are based on the same concept as the method embodiments of the present application, and the specific functions and the brought technical effects can be referred to the method embodiments part. The skilled in the art can clearly understand that, for the convenience and brevity of description, only the above-mentioned division of each functional unit / module is taken as an example for illustration, and in actual application, the above-mentioned functions can be completed by different functional units / modules according to needs, that is, the internal structure of the apparatus is divided into different functional units or modules to complete all or part of the functions described above. Each functional unit / module in the embodiments can be integrated in one processing unit, or each unit can exist physically, or two or more units can be integrated in one unit. The integrated unit can be realized in the form of hardware or in the form of software functional unit. In addition, the specific names of each functional unit / module are only for easy distinction, and do not limit the protection scope of the present application. The specific working process of the unit / module in the system can refer to the corresponding process in the foregoing method embodiments, which will not be repeated here.

[0154] Those of ordinary skill in the art will understand that the above discussion of any embodiment is merely exemplary and is not intended to suggest that the protection scope of the present application is limited to these examples; under the idea of the present application, the technical features in the above embodiments or different embodiments can also be combined, the steps can be implemented in any order, and there are many other changes of different aspects of one or more embodiments of the present application as described above. In order to be brief, they are not provided in details.

[0155] One or more embodiments of the present application are intended to cover all such alternatives, modifications and variations falling within the broad scope of the present application. Therefore, any omission, modification, equivalent replacement, improvement and the like made in the spirit and principles of one or more embodiments of the present application should be included in the protection scope of the present application.

Claims

1. A battery thermal runaway pre-warning method, characterized in that, The method comprises the following steps: acquiring original acoustic signals released by a target battery pack by using an acoustic sensor; the acoustic sensor is arranged in the interior or on the surface of the target battery pack in advance or in real time; acquiring working condition information of the target battery pack, performing dynamic noise reduction processing on the original acoustic signals based on the working condition information, suppressing multi-source interference noise, obtaining de-noised acoustic signals, and extracting acoustic characteristics of the de-noised acoustic signals in a preset frequency band; the multi-source interference noise includes cooling system noise, mechanical vibration noise, and electromagnetic interference noise; constructing a thermal runaway acoustic trigger index based on an energy threshold, a combined feature change threshold, and a continuous trigger time, and generating an initial warning signal according to the thermal runaway acoustic trigger index and the acoustic characteristics; acquiring state information of the target battery pack, and performing multi-modal fusion analysis based on the state information and the initial warning signal to generate a thermal runaway verification signal; the thermal runaway verification signal is used to verify whether a cause-effect time sequence relationship corresponding to the initial warning signal conforms to a physical law; based on the thermal runaway verification signal and the initial warning signal, performing a thermal runaway grading warning on the target battery pack.

2. The battery thermal runaway pre-alarm method of claim 1, wherein, The dynamic noise reduction processing comprises the following steps: based on an adaptive filtering algorithm, using a reference noise signal collected near a noise source to dynamically cancel the cooling system noise in the original acoustic signals to obtain a first acoustic signal; performing energy-weighted wavelet packet decomposition on the first acoustic signal, screening and reconstructing high-energy subbands to suppress mechanical vibration noise to obtain a second acoustic signal; based on a high-frequency current ripple signal of the target battery pack, identifying and filtering electromagnetic interference noise in the second acoustic signal to obtain a third acoustic signal; using a pre-trained U-Net neural network to de-noise a time-frequency spectrum of the third acoustic signal to obtain the de-noised acoustic signals.

3. The battery thermal runaway pre-alarm method of claim 2, wherein, The preset frequency band is 3KHZ to 8KHZ.

4. The battery thermal runaway pre-warning method of claim 3, wherein, The expression of the thermal runaway acoustic trigger index is: wherein, represents the thermal runaway acoustic trigger indicator, used to indicate whether to generate an initial early warning signal, and takes a value of 1 to generate the initial early warning signal, and takes a value of 0 to not generate the initial early warning signal, is a thermal runaway acoustic early warning instantaneous trigger indicator, is a thermal runaway acoustic early warning continuous trigger indicator, represents a time-frequency spectrum of the denoised acoustic signal, represents a background noise energy baseline, represents a decibel scale corresponding to an acoustic feature of the denoised acoustic signal in a preset frequency band, , represents a weight, represents a zero-crossing rate of the denoised acoustic signal in the frequency band, representing the frequency component and stationarity of the denoised acoustic signal, represents a short-time energy of the denoised acoustic signal in the frequency band, representing the intensity of the denoised acoustic signal, represents a spectral flux of the denoised acoustic signal in the frequency band, represents a continuous trigger time of the acoustic trigger early warning, represents a Fourier transform, represents an expression of the denoised acoustic signal.

5. The battery thermal runaway pre-warning method of claim 4, wherein, The state information of the target battery pack includes cell temperature information, cell voltage information, and state information of auxiliary equipment; the auxiliary equipment includes a cooling system and a power conversion system.

6. The battery thermal runaway pre-alarm method of claim 5, wherein, The multi-modal fusion analysis based on the state information and the initial warning signal to generate a thermal runaway verification signal comprises the following steps: adjusting confidence weight of the initial warning signal and / or battery pack state information according to the state information of the auxiliary equipment; constructing a time sequence verification function; the time sequence verification function is used to verify whether an occurrence sequence among the initial warning signal, a cell temperature sudden rise signal, a cell voltage sudden drop signal, and a fire fighting trigger signal conforms to a cause-effect time sequence chain of thermal runaway; integrally calculating the initial warning signal, a cell temperature change rate, and a cell voltage sudden drop amplitude according to the confidence weight and the time sequence verification function to generate the thermal runaway verification signal.

7. The battery thermal runaway pre-warning method of claim 6, wherein, The adjusting confidence weight of the initial warning signal and / or battery pack state information according to the state information of the auxiliary equipment comprises the following steps: by a calculation formula obtaining the confidence weight ; wherein represents a cell temperature weight, represents a cold state, represents a fault, represents a shutdown, represents a normal operation, represents a water pump speed, represents a water pump minimum speed, represents a confidence of a power conversion system (PCS), is used to indicate an operating state of the power conversion system, is used to indicate a fire protection trigger state, represents a gain coefficient.

8. The battery thermal runaway pre-warning method of claim 7, wherein, The expression of the time sequence verification function is: wherein, represents the first time when the initial early warning signal is triggered, specifically represents the starting time when the value changes from 0 to 1 and the value is 1 for five consecutive frames, represents the time when the temperature rises sharply and the temperature change rate is greater than the preset temperature change rate threshold, represents the time when the voltage drops sharply and the voltage change rate is greater than the preset voltage change rate threshold, represents the time when the dry node signal of the fire extinguishing system rises, represents the temperature response delay, represents the short circuit formation speed, represents the time when the heat spreads to the fire extinguishing trigger, represents the full chain reaction window.

9. The battery thermal runaway pre-warning method of claim 8, wherein, The expression of the thermal runaway verification signal is: wherein, represents the thermal runaway verification signal value, represents an acoustic trigger signal, whose data type is Boolean, triggered by an acoustic trigger warning value determined, represents the temperature change rate, represents the voltage drop amplitude, represents the confidence weight, represents the Sigmoid activation function, represents the Heaviside function, since the voltage drop caused by thermal runaway is a sudden event, its physical nature is a step change, which is suitable for describing with the Heaviside function.

10. A battery thermal runaway pre-warning device, characterized in that, ​ The acquisition module is configured to acquire, by using an acoustic sensor, an original acoustic signal released by a target battery pack; The acoustic sensor is arranged in the target battery pack in advance or in real time; The de-noising module is configured to obtain working condition information of the target battery pack, perform dynamic de-noising processing on the original acoustic signal based on the working condition information, suppress multi-source interference noise, obtain a de-noised acoustic signal, and extract an acoustic feature of the de-noised acoustic signal in a preset frequency band; The multi-source interference noise includes cooling system noise, mechanical vibration noise, and electromagnetic interference noise; The first early warning module is configured to construct a thermal runaway acoustic trigger index based on an energy threshold, a combined feature change threshold, and a continuous trigger time, and generate an initial early warning signal according to the thermal runaway acoustic trigger index and the acoustic feature; The verification module is configured to obtain state information of the target battery pack, and perform multi-modal fusion analysis based on the state information and the initial early warning signal to generate a thermal runaway verification signal; the thermal runaway verification signal is used to verify whether a cause-and-effect time sequence relationship corresponding to the initial early warning signal conforms to a physical law; The second early warning module is configured to perform thermal runaway hierarchical early warning on the target battery pack based on the thermal runaway verification signal and the initial early warning signal.

Images