Early warning method, device, system, equipment, medium and product for battery thermal runaway

Abstract

The invention discloses a battery thermal runaway early warning method, device, system, equipment, medium and product, and relates to the technical field of energy storage safety. The method comprises the following steps: firstly, processing a safety valve pressure relief and exhaust sound signal, a PACK box vibration signal and a thermal runaway significant signal which are obtained by monitoring in a latest unit time period so as to obtain thermal runaway identification probabilities corresponding to the sound vibration signal and the significant signal respectively; then, according to the probabilities, calculating to obtain a posterior probability under each thermal runaway judgment condition combination, and finally, taking a certain thermal runaway judgment condition combination corresponding to the maximum posterior probability as a battery thermal runaway early warning decision basis for the target battery and outputting the battery thermal runaway early warning decision basis. In this way, it can be ensured that the finally obtained battery thermal runaway posterior probability is obviously larger than the battery thermal runaway posterior probability obtained based on any single monitoring signal, and the battery early-stage thermal runaway detection capacity is greatly improved.

Description

Technical Field

[0001] The present invention belongs to the field of energy storage safety technology, and specifically relates to a battery thermal runaway early warning method, device, system, equipment, medium and product. Background Art

[0002] As a representative of new energy power batteries, lithium-ion batteries have caused more people to think about their potential and possibilities in energy storage. With the continuous breakthroughs in lithium-ion battery technology, the development of lithium-ion batteries in energy storage is very rapid, and related applications have entered the commercial operation stage.

[0003] As lithium battery explosion accidents happen from time to time, people's concerns about its safety have always been a major problem for its rapid development and large-scale application. The fires in energy storage systems that have occurred in recent years have aroused widespread concern about lithium battery energy storage systems, making energy storage safety highly valued in the industry. However, due to the exponential growth of the energy storage market, a large amount of capital and enterprises have entered across industries, resulting in uneven quality of energy storage products and increasing potential safety hazards. Energy storage safety is a systematic project. From the equipment level, battery cells, inverters, temperature control and fire protection systems are all related to energy storage safety; from the perspective of the entire life cycle of energy storage, design, manufacturing, transportation, installation, commissioning and operation and maintenance are all critical. Problems in any link will bury hidden dangers to power station safety. Energy storage safety requires the industry to innovate technology and improve quality to achieve inherent safety at the product level, and also requires regulatory agencies to further improve rules and standards and build a regulatory system covering the entire life cycle of energy storage without blind spots. In addition, the existing industry guidelines for the development of the energy electronics industry also point out the need to strengthen the development and utilization of multi-dimensional safety testing technology for energy storage batteries, thermal runaway safety warning technology, and evaluation systems. It is also clear that around the intelligent warning and security of energy storage systems, we must develop intelligent safety warning technologies based on multiple physical parameters such as sound, heat, force, electricity, and gas, as well as efficient and clean fire-fighting technologies, establish a safety grading assessment system for energy storage systems, and develop dynamic intelligent assessment technologies for the safety status of energy storage systems based on operating data-driven and advanced artificial intelligence algorithms.

[0004] In summary, in view of the characteristics of electrochemical energy storage systems, such as high energy density, great fire hazards, rapid spread between equipment, and great difficulty in extinguishing fires, we should take "early detection and early disposal" as the basic principle to study the very early detection method of the initial stage of thermal runaway of lithium-ion batteries, and extinguish the hidden dangers of fire in the bud. This is of great significance for both energy storage system manufacturers and energy storage system users, as it can effectively reduce the property losses caused by fires.

[0005] At present, using battery voltage / resistance, temperature, gas, sound, smoke and flame as diagnostic signals for fault identification during thermal runaway is a necessary and feasible solution to achieve internal thermal runaway identification and early warning of batteries and maintain safe and stable battery operation. However, in existing battery thermal runaway monitoring solutions, most of them are based on a single sensor to detect whether thermal runaway occurs, which will inevitably have the following defects:

[0006] (1) For the electrical signal monitoring and early warning method: On the one hand, compared with the diffusion speed of battery thermal runaway, the electrical signals such as the voltage and current of the single cell have the characteristic of slow change, which makes the timeliness problem of using them to characterize battery failures; on the other hand, these electrical signal parameters are related to the properties of the battery itself, and their values ​​are affected by the battery type, capacity and internal conditions, which can easily lead to misjudgment of thermal runaway early warning. In addition, for voltage signals, different thermal runaway initiation methods will lead to different voltage drop processes. For example, mechanical abuse such as needle puncture will cause the battery voltage to drop to 0V, while the process of voltage drop to 0V caused by thermal abuse or electrical abuse is slow and complicated. For resistance, there is currently an electrochemical impedance spectroscopy (EIS) thermal runaway prediction method. Although EIS can detect temperature changes inside the battery, external disturbances or poor internal contact of the battery will also cause changes in internal resistance, making this early warning method inaccurate. In addition, using EIS for battery thermal runaway early warning also faces cost challenges.

[0007] (2) Regarding the temperature monitoring and early warning method: Due to the heat conduction of the battery, there is a certain error in the temperature inside and outside the battery. The method of using thermocouples or temperature sensors to measure the battery surface temperature to predict thermal runaway has a time delay, resulting in battery thermal runaway before the set warning temperature is reached, ultimately leading to warning failure.

[0008] (3) Regarding pressure monitoring and early warning methods: Current fiber optic pressure sensors face complex temperature-pressure decoupling problems and device cost issues, making it difficult to apply them in large-scale energy storage.

[0009] (4) Gas signal monitoring and early warning method: During thermal runaway, the battery will produce characteristic gases such as CO, H2, O2, and alkanes and alkenes, causing the concentration of the gas in the surrounding environment to increase rapidly. Therefore, the gas signal can also be used as a characteristic signal of battery thermal runaway. However, the gas concentration of the battery module is related to the placement of the sensor in the battery box, and the real-time response of detection using this method is poor. In addition, the life of the gas sensor is usually 3-5 years, which is much lower than the life of the energy storage system, which undoubtedly increases the cost of the system.

[0010] (5) Sound signal detection and warning method: When the battery thermal runaway causes the safety valve to open, a specific sound signal will be generated. At this time, the thermal runaway is not severe. By identifying this signal, an early warning of thermal runaway can be achieved. However, this method is easily affected by environmental noise. When the fault data sample is limited, it is easy to miss or misjudge. The recognition accuracy needs to be improved.

[0011] It can be seen from this that it is difficult to achieve accurate early warning of thermal runaway in the battery system with only a single monitoring parameter. It is necessary to try to combine multiple sensor signals and develop multi-parameter coupling early warning technology to accurately identify and early warn of battery thermal runaway situations, thereby improving the sensitivity and accuracy of the battery thermal runaway warning system.

[0012] In addition, the existing technical solutions for multi-sensor fusion detection of battery thermal runaway also have some inherent defects. For example, the existing invention patent "Method and device for predicting thermal runaway of lithium batteries in energy storage power stations based on multi-sensor data fusion" (application number 202311438280.X) mentions the use of four sensors, temperature, voltage, combustible gas and smoke, to fuse and predict thermal runaway of lithium batteries. This solution only achieves thermal runaway warning, not early warning; its disadvantage is that it does not consider the changes in internal pressure and acoustic characteristics when the battery cell is in thermal runaway, especially the changes in acoustic characteristics, which is particularly important for early warning, because when the lithium battery is in thermal runaway, the internal pressure often increases first, and then the safety valve explodes and makes a sound, and then the gas is released, and the temperature and electrical signal parameters change accordingly. The existing invention patent "Energy Storage Power Station Fire Alarm Method and System Based on Gas and Sound Information Fusion" (application number 202210322191.8) mentions the use of hydrogen and acoustic sensor fusion to predict power station fires, but it does not actually mention how the data or features of the two sensors are fused through algorithms. It only mentions how to make simple logical judgments based on the information of the two sensors for early warning. The existing invention patent "A Battery Thermal Runaway Early Warning Device and Method Based on Explosion-proof Valve Monitoring" (application number 202310862635.1) mentions the use of three types of sensors, pressure, sound and gas, for thermal runaway early warning. It only sets three abnormal warnings based on the data of each sensor, and does not mention how to use these sensor data for fusion warning. The existing invention patent "A Lithium-ion Battery Thermal Runaway Early Warning System and Method" (authorization number CN109786872B) mentions the use of four types of sensors, namely ultrasonic, temperature, voltage and smoke, to achieve thermal runaway early warning, but lacks pressure and audible sound sensors, cannot achieve early warning, and does not mention how to fuse. Summary of the invention

[0013] The purpose of the present invention is to provide a battery thermal runaway early warning method, device, system, computer equipment, computer readable storage medium and computer program product to solve the problems of insufficient ability of existing battery thermal runaway monitoring solutions in early battery thermal runaway detection and low warning accuracy.

[0014] In order to achieve the above object, the present invention adopts the following technical solutions:

[0015] In a first aspect, a battery thermal runaway early warning method is provided, comprising:

[0016] Acquire the safety valve pressure relief exhaust sound signal, the PACK box vibration signal and K significant thermal runaway signals obtained by monitoring the target battery in the most recent unit time period, wherein K represents a positive integer, and the significant thermal runaway signal refers to a battery monitoring signal that changes flatly over time when the battery is operating normally but changes dramatically over time when the battery is in thermal runaway;

[0017] The safety valve pressure relief exhaust sound signal, the PACK box vibration signal and the K significant thermal runaway signals are synchronously sampled and processed respectively to obtain a sound signal sequence corresponding to the safety valve pressure relief exhaust sound signal, a vibration signal sequence corresponding to the PACK box vibration signal and K significant signal sequences corresponding to the K significant thermal runaway signals one by one;

[0018] Performing cross-correlation fusion processing on the acoustic signal sequence and the vibration signal sequence to obtain acoustic and vibration signal feature data;

[0019] Importing the acoustic vibration signal characteristic data into a first thermal runaway classification and recognition model pre-trained based on a first artificial intelligence algorithm, and outputting a thermal runaway recognition probability corresponding to the acoustic vibration signal, wherein the thermal runaway recognition probability refers to the probability of classifying and identifying a battery thermal runaway event occurring in the target battery in the most recent unit time period;

[0020] For each thermal runaway significant signal among the K thermal runaway significant signals, importing the corresponding significant signal sequence into a corresponding second thermal runaway classification and recognition model obtained by pre-training based on a second artificial intelligence algorithm, and outputting the corresponding thermal runaway recognition probability;

[0021] According to the acoustic vibration signal and the K significant thermal runaway signals, 2 K+1A thermal runaway determination situation combination, wherein the thermal runaway determination situation combination includes each thermal runaway determination flag bit corresponding to each signal in the acoustic vibration signal and the K thermal runaway significant signals, the thermal runaway determination flag bit is a valid bit, indicating that a battery thermal runaway event occurs in the most recent unit time period based on the corresponding signal alone, and the thermal runaway judgment flag bit is an invalid bit, indicating that a battery thermal runaway event does not occur in the most recent unit time period based on the corresponding signal alone;

[0022] In the 2 K+1 For each thermal runaway judgment situation combination in the thermal runaway judgment situation combination, the posterior probability under the corresponding combination is calculated according to the following formula:

[0023]

[0024] In the formula, c represents a positive integer, P h,c represents the posterior probability under the cth thermal runaway judgment situation combination, and the cth thermal runaway judgment situation combination belongs to the 2 K+1 Thermal runaway judgment combination, P c,purpose represents the probability of the cth thermal runaway judgment situation combination under the condition of a thermal runaway event, k represents a positive integer, p k represents the thermal runaway recognition probability corresponding to the kth signal among the acoustic vibration signal and the K thermal runaway significant signals, B c,k represents the thermal runaway determination flag corresponding to the kth signal in the cth thermal runaway determination situation combination, P c,Npurpose represents the probability of the cth thermal runaway judgment situation combination under the condition that no thermal runaway event occurs, p c,k,x represents the thermal runaway false alarm rate corresponding to the kth signal, P purpose Indicates the empirical probability of a battery thermal runaway event occurring in the target battery;

[0025] In the 2 K+1 A certain thermal runaway judgment situation combination among the thermal runaway judgment situation combinations and corresponding to the maximum a posteriori probability is used as a battery thermal runaway early warning decision basis for the target battery and is output.

[0026] Based on the above invention content, a new solution based on multi-monitoring signal fusion to achieve the purpose of early warning of battery thermal runaway is provided, that is, firstly, the safety valve pressure relief exhaust sound signal, the PACK box vibration signal and the thermal runaway significant signal monitored in the most recent unit time period are synchronously sampled and processed to obtain a sound signal sequence, a vibration signal sequence and a significant signal sequence, and then the sound signal sequence and the vibration signal sequence are cross-correlatedly fused to obtain the acoustic and vibration signal feature data and import them into the pre-trained first thermal runaway classification and recognition model to obtain the thermal runaway recognition probability corresponding to the acoustic and vibration signal, and also the significant signal is integrated into the thermal runaway classification and recognition model. The signal sequence is imported into the pre-trained second thermal runaway classification and recognition model to obtain the thermal runaway recognition probability corresponding to the significant signal, and then the posterior probability under each thermal runaway judgment situation combination is calculated according to the above probability. Finally, a thermal runaway judgment situation combination corresponding to the maximum posterior probability is used as the basis for the battery thermal runaway early warning decision for the target battery and is output. This ensures that the final posterior probability of battery thermal runaway will be significantly greater than the posterior probability of battery thermal runaway based on any single monitoring signal, which greatly improves the detection capability of early battery thermal runaway and is convenient for practical application and promotion.

[0027] In a possible design, the K significant thermal runaway signals include a concentration signal of flammable and explosive gas in a PACK box, an air pressure signal in a PACK box, a voltage signal of a battery cell in a PACK box, and / or a temperature signal of a battery cell in a PACK box.

[0028] In a possible design, when the K significant thermal runaway signals include a cell voltage signal in a PACK box or a cell temperature signal in a PACK box and the target battery includes a plurality of cells, for each significant thermal runaway signal in the K significant thermal runaway signals, the corresponding significant signal sequence is introduced into a corresponding second thermal runaway classification and recognition model obtained by pre-training based on a second artificial intelligence algorithm, and the corresponding thermal runaway recognition probability is output, including:

[0029] For the voltage signal of the battery cells in the PACK box or the temperature signal of the battery cells in the PACK box among the K significant thermal runaway signals, firstly import the corresponding significant signal sequence of each battery cell into the second thermal runaway classification and recognition model obtained by pre-training based on the second artificial intelligence algorithm, output the corresponding thermal runaway recognition probability of each battery cell, and then take the corresponding maximum value of the thermal runaway recognition probability as the corresponding final value of the thermal runaway recognition probability, wherein each battery cell belongs to the multiple battery cells.

[0030] In a possible design, cross-correlation fusion processing is performed on the acoustic signal sequence and the vibration signal sequence to obtain acoustic and vibration signal feature data, including:

[0031] Performing windowing and framing processing on the acoustic signal sequence and the vibration signal sequence respectively to obtain M acoustic signal frames corresponding to the acoustic signal sequence and M vibration signal frames corresponding to the vibration signal sequence, wherein M represents a positive integer greater than or equal to 2;

[0032] For each of the M sound signal frames and the M vibration signal frames, the corresponding short-time energy is calculated respectively, and the corresponding spectrum centroid is obtained by short-time Fourier transform;

[0033] For each pair of sound signal frames and vibration signal frames in the M sound signal frames and the M vibration signal frames and having the same sampling period, cross-correlation calculation is performed on all sampling values ​​of the corresponding two frames to obtain a corresponding linear cross-correlation sequence;

[0034] Calculating the root mean square value of the linear cross-correlation sequence of each pair of sound signal frames and vibration signal frames respectively, and summarizing them to obtain a linear cross-correlation root mean square value sequence;

[0035] Performing the cross-correlation calculation on all short-time energies of the M sound signal frames and the M vibration signal frames to obtain a short-time energy cross-correlation sequence, and further performing the cross-correlation calculation on all frequency spectrum centroids of the M sound signal frames and the M vibration signal frames to obtain a frequency spectrum centroid cross-correlation sequence;

[0036] The linear cross-correlation root mean square value sequence, the short-time energy cross-correlation sequence and / or the spectrum centroid cross-correlation sequence are combined into acoustic vibration signal characteristic data.

[0037] In one possible design, the first artificial intelligence algorithm or the second artificial intelligence algorithm adopts a support vector machine algorithm.

[0038] In the second aspect, a battery thermal runaway early warning device is provided, including a monitoring signal acquisition unit, a signal sequence sampling unit, a cross-correlation fusion processing unit, a thermal runaway identification processing unit, a thermal runaway situation construction unit, a posterior probability calculation unit and a warning decision output unit;

[0039] The monitoring signal acquisition unit is used to acquire the safety valve pressure relief exhaust sound signal, the PACK box vibration signal and K significant thermal runaway signals obtained by monitoring the target battery in the most recent unit time period, wherein K represents a positive integer, and the significant thermal runaway signal refers to a battery monitoring signal that changes flatly over time when the battery is operating normally but changes dramatically over time when the battery is in thermal runaway;

[0040] The signal sequence sampling unit is communicatively connected to the monitoring signal acquisition unit, and is used to synchronously sample and process the safety valve pressure relief exhaust sound signal, the PACK box vibration signal, and the K significant thermal runaway signals, respectively, to obtain a sound signal sequence corresponding to the safety valve pressure relief exhaust sound signal, a vibration signal sequence corresponding to the PACK box vibration signal, and K significant signal sequences corresponding one-to-one to the K significant thermal runaway signals;

[0041] The cross-correlation fusion processing unit is communicatively connected to the signal sequence sampling unit and is used to perform cross-correlation fusion processing on the acoustic signal sequence and the vibration signal sequence to obtain acoustic and vibration signal feature data;

[0042] The thermal runaway identification processing unit is communicatively connected to the cross-correlation fusion processing unit and the signal sequence sampling unit, respectively, and is used to import the acoustic vibration signal feature data into a first thermal runaway classification and identification model obtained by pre-training based on a first artificial intelligence algorithm, and output a thermal runaway identification probability corresponding to the acoustic vibration signal, and also for each thermal runaway significant signal among the K thermal runaway significant signals, import the corresponding significant signal sequence into a corresponding second thermal runaway classification and identification model obtained by pre-training based on a second artificial intelligence algorithm, and output the corresponding thermal runaway identification probability, wherein the thermal runaway identification probability refers to the probability of classifying and identifying that a battery thermal runaway event occurs in the target battery in the most recent unit time period;

[0043] The thermal runaway situation constructing unit is communicatively connected to the monitoring signal acquiring unit and is used to construct and obtain 2 according to the acoustic vibration signal and the K thermal runaway significant signals. K+1 A thermal runaway determination situation combination, wherein the thermal runaway determination situation combination includes each thermal runaway determination flag bit corresponding to each signal in the acoustic vibration signal and the K thermal runaway significant signals, the thermal runaway determination flag bit is a valid bit, indicating that a battery thermal runaway event occurs in the most recent unit time period based on the corresponding signal alone, and the thermal runaway judgment flag bit is an invalid bit, indicating that a battery thermal runaway event does not occur in the most recent unit time period based on the corresponding signal alone;

[0044] The posterior probability calculation unit is respectively connected to the thermal runaway identification processing unit and the thermal runaway situation construction unit for K+1 For each thermal runaway judgment situation combination in the thermal runaway judgment situation combination, the posterior probability under the corresponding combination is calculated according to the following formula:

[0045]

[0046] In the formula, c represents a positive integer, P h,crepresents the posterior probability under the cth thermal runaway judgment situation combination, and the cth thermal runaway judgment situation combination belongs to the 2 K+1 Thermal runaway judgment combination, P c,purpose represents the probability of the cth thermal runaway judgment situation combination under the condition of a thermal runaway event, k represents a positive integer, p k represents the thermal runaway recognition probability corresponding to the kth signal among the acoustic vibration signal and the K thermal runaway significant signals, B c,k represents the thermal runaway determination flag corresponding to the kth signal in the cth thermal runaway determination situation combination, P c,Npurpose represents the probability of the cth thermal runaway judgment situation combination under the condition that no thermal runaway event occurs, p c,k,x represents the thermal runaway false alarm rate corresponding to the kth signal, P purpose Indicates the empirical probability of a battery thermal runaway event occurring in the target battery;

[0047] The early warning decision output unit is communicatively connected to the posterior probability calculation unit, and is used to calculate the K+1 A certain thermal runaway judgment situation combination among the thermal runaway judgment situation combinations and corresponding to the maximum a posteriori probability is used as a battery thermal runaway early warning decision basis for the target battery and is output.

[0048] In a third aspect, the present invention provides a battery thermal runaway early warning system, comprising a microphone, a triaxial acceleration sensor, a thermal runaway significant signal acquisition module, a processing device and a fire control mainboard, wherein the microphone, the triaxial acceleration sensor and the thermal runaway significant signal acquisition module are respectively arranged on a target battery;

[0049] The microphone is communicatively connected to the processing device, and is used to collect the pressure relief exhaust sound signal of the safety valve of the target battery in real time, and transmit the collected result to the processing device in real time;

[0050] The three-axis acceleration sensor is communicatively connected to the processing device, and is used to collect the vibration signal of the PACK box of the target battery in real time, and transmit the collected result to the processing device in real time;

[0051] The thermal runaway significant signal acquisition module is communicatively connected to the processing device, and is used to acquire K thermal runaway significant signals of the target battery in real time, and transmit the acquisition results to the processing device in real time, wherein K represents a positive integer, and the thermal runaway significant signal refers to a battery monitoring signal that changes flatly over time when the battery is operating normally but changes dramatically over time when the battery is in thermal runaway;

[0052] The processing device is communicatively connected to the fire control mainboard, and is used to execute the battery thermal runaway early warning method as described in the first aspect or any possible design of the first aspect, and transmit the obtained battery thermal runaway early warning decision basis for the target battery to the fire control mainboard;

[0053] The fire control mainboard is used to determine and execute a corresponding battery thermal runaway early warning decision according to the battery thermal runaway early warning decision basis.

[0054] In a fourth aspect, the present invention provides a computer device comprising a memory, a processor and a transceiver which are communicatively connected in sequence, wherein the memory is used to store a computer program, the transceiver is used to send and receive messages, and the processor is used to read the computer program to execute the battery thermal runaway early warning method as described in the first aspect or any possible design of the first aspect.

[0055] In a fifth aspect, the present invention provides a computer-readable storage medium having instructions stored thereon. When the instructions are executed on a computer, the battery thermal runaway early warning method as described in the first aspect or any possible design of the first aspect is executed.

[0056] In a sixth aspect, the present invention provides a computer program product, including a computer program or instructions, which, when executed by a computer, implements the battery thermal runaway early warning method as described in the first aspect or any possible design of the first aspect.

[0057] Beneficial effects of the above scheme:

[0058] (1) The present invention creatively provides a new solution for achieving the purpose of early warning of battery thermal runaway based on multi-monitoring signal fusion, that is, firstly synchronously sampling and processing the safety valve pressure relief exhaust sound signal, the PACK box vibration signal and the thermal runaway significant signal obtained by monitoring in the most recent unit time period to obtain a sound signal sequence, a vibration signal sequence and a significant signal sequence, then cross-correlation fusion processing is performed on the sound signal sequence and the vibration signal sequence to obtain the acoustic and vibration signal feature data and import it into the pre-trained first thermal runaway classification and recognition model to obtain the thermal runaway recognition probability corresponding to the acoustic and vibration signal, and also import the significant signal sequence into the pre-trained second thermal runaway classification and recognition model to obtain the thermal runaway recognition probability corresponding to the significant signal, and then calculate the posterior probability under each thermal runaway judgment situation combination according to the above-mentioned probability, and finally use a certain thermal runaway judgment situation combination corresponding to the maximum posterior probability as the battery thermal runaway early warning decision basis for the target battery and output it, so as to ensure that the final battery thermal runaway posterior probability will be significantly greater than the battery thermal runaway posterior probability obtained based on any single monitoring signal, thereby greatly improving the detection capability of early battery thermal runaway;

[0059] (2) It also has the advantages of low requirements for hardware computing power, good real-time performance, higher accuracy than using a single monitoring signal, and low misjudgment rate and missed judgment rate, which facilitates practical application and promotion. BRIEF DESCRIPTION OF THE DRAWINGS

[0060] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0061] Figure 1 A schematic diagram of the process flow of a battery thermal runaway early warning method provided in an embodiment of the present application.

[0062] Figure 2 A schematic diagram of the structure of a battery thermal runaway early warning device provided in an embodiment of the present application.

[0063] Figure 3 A schematic diagram of the structure of a battery thermal runaway early warning system provided in an embodiment of the present application.

[0064] Figure 4 A schematic diagram of the structure of a computer device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0065] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the present invention will be briefly introduced below in combination with the drawings and the description of the embodiments or the prior art. Obviously, the following description of the structures of the drawings is only some embodiments of the present invention. For ordinary technicians in this field, other embodiments can be obtained based on these embodiments without creative work. It should be noted that the description of these embodiments is used to help understand the present invention, but does not constitute a limitation of the present invention.

[0066] It should be understood that although the terms first and second, etc. may be used herein to describe various objects, these objects should not be limited by these terms. These terms are only used to distinguish one object from another object. For example, a first object can be referred to as a second object, and similarly, a second object can be referred to as a first object without departing from the scope of the exemplary embodiments of the present invention.

[0067] It should be understood that the term "and / or" that may appear in this document is merely a description of the association relationship between associated objects, indicating that there may be three relationships. For example, A and / or B can indicate three situations: A exists alone, B exists alone, or A and B exist at the same time. For another example, A, B and / or C can indicate the existence of any one of A, B and C or any combination of them. The term " / and" that may appear in this document describes another type of association object relationship, indicating that there may be two relationships. For example, A / and B can indicate two situations: A exists alone or A and B exist at the same time. In addition, the character " / " that may appear in this document generally indicates that the previous and next associated objects are in an "or" relationship.

[0068] Example

[0069] like Figure 1 As shown, the battery thermal runaway early warning method provided in the first aspect of this embodiment can be, but is not limited to, executed by a computer device having certain computing resources and communicating with a number of signal acquisition modules (such as microphones, three-axis acceleration sensors, BMS modules, gas sensors, and air pressure sensors, etc.), such as an energy storage system control motherboard, a personal computer (Personal Computer, PC, refers to a multi-purpose computer with a size, price, and performance suitable for personal use; desktops, laptops, small laptops, tablets, and ultrabooks are all personal computers), smart phones, personal digital assistants (Personal Digital Assistant, PDA) or wearable devices and other electronic devices. Figure 1 As shown, the battery thermal runaway early warning method may include but is not limited to the following steps S1 to S8.

[0070] S1. Obtain the safety valve pressure relief exhaust sound signal, PACK box vibration signal and K significant thermal runaway signals obtained by monitoring the target battery in the most recent unit time period, where K represents a positive integer, and the significant thermal runaway signal refers to a battery monitoring signal that changes smoothly over time when the battery is operating normally but changes dramatically over time when the battery is in thermal runaway.

[0071] In the step S1, the unit time period may be specifically but not limited to a few seconds (e.g., 3 seconds or 5 seconds, etc., that is, the most recent unit time period is within the most recent 3 seconds or within the most recent 5 seconds, etc.). The target battery may be, but not limited to, a lithium-ion battery, which is specifically a battery pack module (PACK in English) consisting of a number of lithium battery cells installed in a sealed box, and is conventionally configured with a safety valve, a BMS (Battery Management System) module, a microphone, a three-axis acceleration sensor, a gas sensor, and an air pressure sensor, etc. The safety valve pressure relief exhaust sound signal can be specifically acquired and transmitted in real time by the microphone. At the same time, since the safety valve pressure relief exhaust sound signal will show a strong correlation when the safety valve pressure relief is opened, it needs to be applied in this embodiment. The PACK box vibration signal can be specifically acquired and transmitted in real time by the three-axis acceleration sensor. At the same time, since the PACK box vibration signal will also show a strong correlation when the safety valve pressure relief is opened, it also needs to be applied in this embodiment. Since the significant thermal runaway signal has the characteristics of changing smoothly with time when the battery is operating normally but changing dramatically with time when the battery is in thermal runaway, it is also necessary to apply it in the present embodiment; specifically, the K significant thermal runaway signals include but are not limited to the concentration signal of flammable and explosive gas in the PACK box (the flammable and explosive gas in the PACK box here can be, but is not limited to, characteristic gases such as carbon monoxide, hydrogen, methane and / or alkanes), the air pressure signal in the PACK box, the voltage signal of the battery cell in the PACK box and / or the temperature signal of the battery cell in the PACK box, etc., wherein the concentration signal of flammable and explosive gas in the PACK box can be specifically acquired and transmitted in real time by the gas sensor, the air pressure signal in the PACK box can be specifically acquired and transmitted in real time by the air pressure sensor, the voltage signal of the battery cell in the PACK box and / or the temperature signal of the battery cell in the PACK box can be specifically acquired in real time by the BMS module (when the target battery includes multiple battery cells, the voltage signal and temperature signal of each battery cell can be acquired and transmitted by the BMS model in real time). In addition, considering that smoke and flames are late manifestations of thermal runaway and are meaningless for early warning, this embodiment does not consider these two types of signals; and the microphone, the three-axis acceleration sensor, the gas sensor, the air pressure sensor, etc. can be implemented using existing analog related products or existing digital related products.

[0072] S2. Perform synchronous sampling processing on the safety valve pressure relief exhaust sound signal, the PACK box vibration signal and the K significant thermal runaway signals respectively to obtain a sound signal sequence corresponding to the safety valve pressure relief exhaust sound signal, a vibration signal sequence corresponding to the PACK box vibration signal and K significant signal sequences corresponding one by one to the K significant thermal runaway signals.

[0073] In step S2, the synchronous sampling process refers to synchronous periodic sampling process of different signals (for example, different signals are sampled once every 1 millisecond), so that the acoustic signal sequence, the vibration signal sequence and the significant signal sequence with the same number of sampling points can be obtained. In addition, the acoustic signal sequence, the vibration signal sequence and the significant signal sequence each contain a plurality of sampling values, and are time series data obtained by sequentially arranging all the sampling values ​​in order from early to late according to the sampling time.

[0074] S3. Perform cross-correlation fusion processing on the acoustic signal sequence and the vibration signal sequence to obtain acoustic-vibration signal feature data.

[0075] In step S3, since the safety valve exhaust sound signal and the PACK box vibration signal both show strong correlation when the safety valve is venting and releasing pressure, it is necessary to perform feature-level fusion processing on these two signals in order to obtain a comprehensive basis for judging whether the safety valve is open (i.e., the acoustic and vibration signal characteristic data). Specifically, the acoustic signal sequence and the vibration signal sequence are cross-correlatedly fused to obtain the acoustic and vibration signal characteristic data, including but not limited to the following steps S31 to S36.

[0076] S31. Perform windowing and framing processing on the acoustic signal sequence and the vibration signal sequence respectively to obtain M acoustic signal frames corresponding to the acoustic signal sequence and M vibration signal frames corresponding to the vibration signal sequence, where M represents a positive integer greater than or equal to 2.

[0077] In step S31, the windowing and framing process is an existing signal processing method for dividing a continuous digital signal into a series of overlapping frames so that each frame contains a certain length of sampled data. Specifically, the windowing and framing process includes, but is not limited to, using a Hamming window or a Hanning window, making the frame length 10 to 20 milliseconds, and making the frame shift half of the frame length. In addition, in order to filter out the noise in the safety valve pressure relief exhaust sound signal and the PACK box vibration signal, it is also necessary to perform digital low-pass filtering on the sound signal sequence and the vibration signal sequence respectively before the windowing and framing process.

[0078] S32. For each frame in the M sound signal frames and the M vibration signal frames, the corresponding short-time energy is calculated respectively, and the corresponding spectrum centroid is obtained by short-time Fourier transform.

[0079] In step S32, the short-time energy refers to the calculation of the energy of the signal frame to represent the signal amplitude of the frame; the specific calculation method is to perform amplitude square operation on the windowed signal frame, and accumulate the results to obtain the energy value of the frame. The short-time Fourier transform (STFT, short-time Fourier transform, or short-term Fourier transform) is a mathematical transformation related to Fourier transform, which is used to determine the frequency and phase of the local sinusoidal wave of the time-varying signal. The spectrum centroid is the weighted average frequency of the spectrum power distribution. Therefore, the spectrum of each frame can be obtained by the short-time Fourier transform first, and then the spectrum centroid of each frame can be obtained conventionally.

[0080] S33. For each pair of sound signal frames and vibration signal frames in the M sound signal frames and the M vibration signal frames and having the same sampling period, cross-correlation calculation is performed on all sampling values ​​of the corresponding two frames to obtain a corresponding linear cross-correlation sequence.

[0081] In the step S33, since the safety valve pressure relief exhaust sound signal and the PACK box vibration signal are both monitored within the most recent unit time period, and are both subjected to the synchronous sampling process and the windowing and framing process to obtain the M sound signal frames and the M vibration signal frames, there will be M pairs of sound signal frames and vibration signal frames, and in each pair of sound signal frames and vibration signal frames, the corresponding two frames can have the same number of sampling values, and then the following formula can be used for cross-correlation calculation to obtain each value in the linear cross-correlation sequence:

[0082]

[0083] Wherein, n and n″ represent positive integers less than or equal to N, n′ represents a positive integer greater than or equal to n and less than or equal to N, N represents the number of sampling values ​​in the sound signal frame or the vibration signal frame, and r av,n represents the nth value in the linear cross-correlation sequence, S a,n′ represents the n'th sampling value in the acoustic signal frame, S v,n′+1-n represents the n′+1-nth sampling value in the vibration signal frame, S a,n″ represents the n″th sampling value in the acoustic signal frame, S v,n″ represents the n″th sampling value in the vibration signal frame. In addition, the linear cross-correlation sequence is also a time series data.

[0084] S34. Calculate the root mean square value of the linear cross-correlation sequence of each pair of sound signal frames and vibration signal frames respectively, and summarize them to obtain a linear cross-correlation root mean square value sequence.

[0085] In step S34, the root mean square value refers to the square root of the average value of the sum of squares of a set of values, so M root mean square values ​​corresponding to the M pairs of sound signal frames and vibration signal frames can be conventionally calculated to obtain the linear cross-correlation root mean square value sequence containing the M root mean square values. In addition, the linear cross-correlation root mean square value sequence is also a time series data.

[0086] S35. Perform the cross-correlation calculation on all short-time energies of the M sound signal frames and the M vibration signal frames to obtain a short-time energy cross-correlation sequence, and also perform the cross-correlation calculation on all frequency spectrum centroids of the M sound signal frames and the M vibration signal frames to obtain a frequency spectrum centroid cross-correlation sequence.

[0087] In step S35, the specific formula for performing the cross-correlation calculation on all the short-time energies or all the spectrum centroids can be obtained by referring to the conventional derivation of step S33, which will not be repeated here. In addition, the short-time energy cross-correlation sequence and the spectrum centroid cross-correlation sequence are time series data containing M values ​​respectively.

[0088] S36. Combine the linear cross-correlation root mean square value sequence, the short-time energy cross-correlation sequence and / or the spectrum centroid cross-correlation sequence into acoustic vibration signal feature data.

[0089] In the step S36, the linear cross-correlation root mean square value sequence, the short-time energy cross-correlation sequence and the spectrum centroid cross-correlation sequence may be preferably combined into a vector containing 3×M values ​​to serve as the acoustic vibration signal characteristic data.

[0090] S4. Import the acoustic vibration signal characteristic data into the first thermal runaway classification and recognition model pre-trained based on the first artificial intelligence algorithm, and output a thermal runaway recognition probability corresponding to the acoustic vibration signal, wherein the thermal runaway recognition probability refers to the probability of classifying and identifying that a battery thermal runaway event occurs in the target battery in the most recent unit time period.

[0091] In step S4, the artificial intelligence algorithm is a core artificial intelligence algorithm that specifically studies how computers simulate or implement human learning behavior to acquire new knowledge or skills and reorganize existing knowledge structures to continuously improve their own performance. It is the fundamental way to make computers intelligent. In order to reduce the demand for front-end computing power and the number of samples required for model training, the first artificial intelligence algorithm preferably uses, but is not limited to, a support vector machine algorithm. Based on a certain amount of sample data (i.e., based on the historical exhaust sound signal of the safety valve pressure relief and the historical vibration signal of the PACK box collected when the safety valve pressure relief is opened multiple times, the multiple historical sound and vibration signal feature data obtained through the aforementioned steps S2 to S3 and corresponding to the multiple safety valve pressure relief openings) and label data (i.e., multiple battery thermal runaway effective values ​​corresponding to the multiple historical sound and vibration signal feature data, i.e., the multiple historical sound and vibration signal feature data are used as learning target data), the first thermal runaway classification and identification model can be trained through conventional calibration verification modeling methods (the specific process includes the calibration process and verification process of the model, i.e., first by comparing the model simulation results with the measured data, and then adjusting the model parameters according to the comparison results, so that the simulation results are consistent with the actual process). At the same time, in the calibration verification modeling process of the first thermal runaway classification and identification model, the model parameters can be specifically tuned by using a tree-structured Bayesian optimization algorithm. In addition, the sound and vibration signal is the abbreviation of the safety valve pressure relief exhaust sound signal and the PACK box vibration signal.

[0092] S5. For each thermal runaway significant signal among the K thermal runaway significant signals, the corresponding significant signal sequence is imported into the corresponding second thermal runaway classification and recognition model obtained by pre-training based on the second artificial intelligence algorithm, and the corresponding thermal runaway recognition probability is output.

[0093] In step S5, in order to reduce the demand for front-end computing power and the number of samples required for model training, the second artificial intelligence algorithm preferably but not limited to adopts a support vector machine algorithm. Similarly, based on a certain amount of sample data (i.e., based on historical thermal runaway significant signals collected when the safety valve is depressurized and opened multiple times, multiple historical significant signal sequences obtained by the aforementioned step S2 and corresponding one-to-one to the multiple safety valves being depressurized and opened) and label data (i.e., multiple battery thermal runaway effective values ​​corresponding one-to-one to the multiple historical significant signal sequences, i.e., the multiple historical significant signal sequences are used as another learning target data), the second thermal runaway classification and identification model can also be trained through conventional calibration verification modeling methods. In addition, in the calibration verification modeling process of the second thermal runaway classification and identification model, the model parameters can also be tuned using a tree-structured Bayesian optimization algorithm.

[0094] In the step S5, when the K significant thermal runaway signals include a cell voltage signal in a PACK box or a cell temperature signal in a PACK box and the target battery includes a plurality of cells, there will be a plurality of cell voltage signals or a plurality of cell temperature signals corresponding one-to-one to the plurality of cells for the cell voltage signal in the PACK box or the cell temperature signal in the PACK box, and then a plurality of thermal runaway recognition probabilities corresponding one-to-one to the plurality of cells will be obtained. Therefore, in order to obtain the thermal runaway recognition probability for the global cell voltage signal or the global cell temperature signal, preferably, when the K significant thermal runaway signals include a cell voltage signal in a PACK box or a cell temperature signal in a PACK box and the target battery includes a plurality of cells, for For each of the K significant thermal runaway signals, the corresponding significant signal sequence is imported into the corresponding second thermal runaway classification and recognition model pre-trained based on the second artificial intelligence algorithm, and the corresponding thermal runaway recognition probability is output, including but not limited to: for the battery cell voltage signal in the PACK box or the battery cell temperature signal in the PACK box among the K significant thermal runaway signals, the corresponding significant signal sequence of each battery cell is first imported into the corresponding second thermal runaway classification and recognition model pre-trained based on the second artificial intelligence algorithm, and the corresponding thermal runaway recognition probability of each battery cell is output, and then the corresponding maximum value of the thermal runaway recognition probability is used as the corresponding final value of the thermal runaway recognition probability, wherein each battery cell belongs to the multiple battery cells. For example, if the multiple battery cells include battery cell A, battery cell B and battery cell C, then based on the battery cell voltage signal of battery cell A, the significant signal sequence of battery cell A is obtained through the aforementioned step S2, and the significant signal sequence is imported into the second thermal runaway classification and recognition model corresponding to the battery cell voltage signal in the PACK box, and the thermal runaway recognition probability corresponding to the battery cell A is output; and based on the battery cell voltage signal of battery cell B, the significant signal sequence of battery cell B is obtained through the aforementioned step S2, and the significant signal sequence is imported into the second thermal runaway classification and recognition model corresponding to the battery cell voltage signal in the PACK box, and the thermal runaway recognition probability corresponding to the battery cell B is output. The thermal runaway recognition probability corresponding to the battery cell A; and based on the battery cell voltage signal of the battery cell C, the significant signal sequence of the battery cell C is obtained through the aforementioned step S2, and the significant signal sequence is imported into the second thermal runaway classification and recognition model corresponding to the battery cell voltage signal in the PACK box, and the thermal runaway recognition probability corresponding to the battery cell C is output; finally, if the thermal runaway recognition probability corresponding to the battery cell C is the maximum value among the three probabilities (i.e., the thermal runaway recognition probabilities corresponding to the battery cells A, B and C), the maximum value is used as the final value of the thermal runaway recognition probability corresponding to the battery cell voltage signal in the PACK box.

[0095] S6. According to the acoustic vibration signal and the K significant thermal runaway signals, construct 2 K+1 A thermal runaway determination situation combination, wherein the thermal runaway determination situation combination includes thermal runaway determination flags corresponding to each signal in the acoustic vibration signal and the K thermal runaway significant signals, the thermal runaway determination flag is a valid bit, indicating that a battery thermal runaway event occurs in the most recent unit time period based on a separate determination based on the corresponding signal, and the thermal runaway determination flag is an invalid bit, indicating that a battery thermal runaway event does not occur in the most recent unit time period based on a separate determination based on the corresponding signal.

[0096] In step S6, since each signal corresponds to two possible separate detection results (i.e., determining whether a battery thermal runaway event occurs or does not occur in the most recent unit time period), 2 K+1 For example, if the valid bit is represented by "1", the invalid bit is represented by "0", and the K significant thermal runaway signals include the flammable and explosive gas concentration signal in the PACK box, the air pressure signal in the PACK box, the battery cell voltage signal in the PACK box, and the battery cell temperature signal in the PACK box, and the individual signals are also the acoustic vibration signal, the flammable and explosive gas concentration signal in the PACK box, the air pressure signal in the PACK box, the battery cell voltage signal in the PACK box, and the battery cell temperature signal in the PACK box, then K=4, the 2 K+1 The combinations of thermal runaway judgment situations can be represented one by one by the following 32 character strings: 00000~11111. For further example, the character string 10101 can represent such a combination of thermal runaway judgment situations: based on the acoustic vibration signal, it is determined that a battery thermal runaway event has occurred in the most recent unit period; based on the flammable and explosive gas concentration signal in the PACK box, it is determined that no battery thermal runaway event has occurred in the most recent unit period; based on the air pressure signal in the PACK box, it is determined that a battery thermal runaway event has occurred in the most recent unit period; based on the battery cell voltage signal in the PACK box, it is determined that no battery thermal runaway event has occurred in the most recent unit period; based on the battery cell temperature signal in the PACK box, it is determined that a battery thermal runaway event has occurred in the most recent unit period.

[0097] S7. For the K+1 For each thermal runaway judgment situation combination in the thermal runaway judgment situation combination, the posterior probability under the corresponding combination is calculated according to the following formula:

[0098]

[0099] In the formula, c represents a positive integer, P h,crepresents the posterior probability under the cth thermal runaway judgment situation combination, and the cth thermal runaway judgment situation combination belongs to the 2 K+1 Thermal runaway judgment combination, P c,purpose represents the probability of the cth thermal runaway judgment situation combination under the condition of a thermal runaway event, k represents a positive integer, p k represents the thermal runaway recognition probability corresponding to the kth signal among the acoustic vibration signal and the K thermal runaway significant signals, B c,k represents the thermal runaway determination flag corresponding to the kth signal in the cth thermal runaway determination situation combination, P c,Npurpose represents the probability of the cth thermal runaway judgment situation combination under the condition that no thermal runaway event occurs, p c,k,x represents the thermal runaway false alarm rate corresponding to the kth signal, P purpose Indicates the empirical probability of a battery thermal runaway event occurring in the target battery.

[0100] In step S7, the derivation process of the posterior probability can be conventionally derived based on existing probabilistic knowledge, which will not be repeated here. The thermal runaway false alarm rate is also called the false alarm rate. It is usually set to a given value based on the reliability requirements of different scenarios, such as 1%, which means that 1 false alarm is allowed in 100 times; constant false alarm rate detection is a very important method in sensor detection, and the detection performance can be quickly verified by Monte Carlo simulation and other means. The empirical probability is an empirical value obtained through actual experiments. It is not an assumed value, but also a given value.

[0101] S8. In the 2 K+1 A certain thermal runaway judgment situation combination among the thermal runaway judgment situation combinations and corresponding to the maximum a posteriori probability is used as a battery thermal runaway early warning decision basis for the target battery and is output.

[0102] Therefore, based on the battery thermal runaway early warning method described in the aforementioned steps S1 to S8, a new solution based on the fusion of multiple monitoring signals to achieve the purpose of battery thermal runaway early warning is provided, that is, firstly, the safety valve pressure relief exhaust sound signal, the PACK box vibration signal and the thermal runaway significant signal obtained by monitoring in the most recent unit time period are synchronously sampled and processed to obtain a sound signal sequence, a vibration signal sequence and a significant signal sequence, and then the sound signal sequence and the vibration signal sequence are cross-correlatedly fused to obtain the acoustic and vibration signal feature data and import them into the pre-trained first thermal runaway classification and recognition model to obtain the thermal runaway recognition probability corresponding to the acoustic and vibration signal, and also the significant signal sequence is imported into the pre-trained second thermal runaway classification and recognition model. In the out-of-control classification and identification model, the thermal runaway identification probability corresponding to the significant signal is obtained, and then the posterior probability under each thermal runaway judgment situation combination is calculated based on the above-mentioned probability. Finally, a thermal runaway judgment situation combination corresponding to the maximum posterior probability is used as the basis for the battery thermal runaway early warning decision for the target battery and is output. This ensures that the final posterior probability of battery thermal runaway will be significantly greater than the posterior probability of battery thermal runaway based on any single monitoring signal, greatly improving the detection capability of early thermal runaway of the battery. It has the advantages of low hardware computing power requirements, good real-time performance, higher accuracy than using a single monitoring signal, and low misjudgment rate and missed judgment rate, which is convenient for practical application and promotion.

[0103] like Figure 2 As shown, the second aspect of this embodiment provides a virtual device for implementing the battery thermal runaway early warning method described in the first aspect, including a monitoring signal acquisition unit, a signal sequence sampling unit, a cross-correlation fusion processing unit, a thermal runaway identification processing unit, a thermal runaway situation construction unit, a posterior probability calculation unit and a warning decision output unit;

[0104] The monitoring signal acquisition unit is used to acquire the safety valve pressure relief exhaust sound signal, the PACK box vibration signal and K significant thermal runaway signals obtained by monitoring the target battery in the most recent unit time period, wherein K represents a positive integer, and the significant thermal runaway signal refers to a battery monitoring signal that changes flatly over time when the battery is operating normally but changes dramatically over time when the battery is in thermal runaway;

[0105] The signal sequence sampling unit is communicatively connected to the monitoring signal acquisition unit, and is used to synchronously sample and process the safety valve pressure relief exhaust sound signal, the PACK box vibration signal, and the K significant thermal runaway signals, respectively, to obtain a sound signal sequence corresponding to the safety valve pressure relief exhaust sound signal, a vibration signal sequence corresponding to the PACK box vibration signal, and K significant signal sequences corresponding one-to-one to the K significant thermal runaway signals;

[0106] The cross-correlation fusion processing unit is communicatively connected to the signal sequence sampling unit and is used to perform cross-correlation fusion processing on the acoustic signal sequence and the vibration signal sequence to obtain acoustic and vibration signal feature data;

[0107] The thermal runaway identification processing unit is communicatively connected to the cross-correlation fusion processing unit and the signal sequence sampling unit, respectively, and is used to import the acoustic vibration signal feature data into a first thermal runaway classification and identification model obtained by pre-training based on a first artificial intelligence algorithm, and output a thermal runaway identification probability corresponding to the acoustic vibration signal, and also for each thermal runaway significant signal among the K thermal runaway significant signals, import the corresponding significant signal sequence into a corresponding second thermal runaway classification and identification model obtained by pre-training based on a second artificial intelligence algorithm, and output the corresponding thermal runaway identification probability, wherein the thermal runaway identification probability refers to the probability of classifying and identifying that a battery thermal runaway event occurs in the target battery in the most recent unit time period;

[0108] The thermal runaway situation constructing unit is communicatively connected to the monitoring signal acquiring unit and is used to construct and obtain 2 according to the acoustic vibration signal and the K thermal runaway significant signals. K+1 A thermal runaway determination situation combination, wherein the thermal runaway determination situation combination includes each thermal runaway determination flag bit corresponding to each signal in the acoustic vibration signal and the K thermal runaway significant signals, the thermal runaway determination flag bit is a valid bit, indicating that a battery thermal runaway event occurs in the most recent unit time period based on the corresponding signal alone, and the thermal runaway judgment flag bit is an invalid bit, indicating that a battery thermal runaway event does not occur in the most recent unit time period based on the corresponding signal alone;

[0109] The posterior probability calculation unit is respectively connected to the thermal runaway identification processing unit and the thermal runaway situation construction unit for K+1 For each thermal runaway judgment situation combination in the thermal runaway judgment situation combination, the posterior probability under the corresponding combination is calculated according to the following formula:

[0110]

[0111] In the formula, c represents a positive integer, P h,c represents the posterior probability under the cth thermal runaway judgment situation combination, and the cth thermal runaway judgment situation combination belongs to the 2 K+1 Thermal runaway judgment combination, P c,purpose represents the probability of the cth thermal runaway judgment situation combination under the condition of a thermal runaway event, k represents a positive integer, p k represents the thermal runaway recognition probability corresponding to the kth signal among the acoustic vibration signal and the K thermal runaway significant signals, B c,krepresents the thermal runaway determination flag corresponding to the kth signal in the cth thermal runaway determination situation combination, P c,Npurpose represents the probability of the cth thermal runaway judgment situation combination under the condition that no thermal runaway event occurs, p c,k,x represents the thermal runaway false alarm rate corresponding to the kth signal, P purpose Indicates the empirical probability of a battery thermal runaway event occurring in the target battery;

[0112] The early warning decision output unit is communicatively connected to the posterior probability calculation unit, and is used to calculate the K+1 A certain thermal runaway judgment situation combination among the thermal runaway judgment situation combinations and corresponding to the maximum a posteriori probability is used as a battery thermal runaway early warning decision basis for the target battery and is output.

[0113] The working process, working details and technical effects of the aforementioned device provided in the second aspect of this embodiment can be referred to the battery thermal runaway early warning method described in the first aspect, and will not be described in detail here.

[0114] like Figure 3 As shown, the third aspect of this embodiment provides a physical system for implementing the battery thermal runaway early warning method described in the first aspect, comprising a microphone, a triaxial acceleration sensor, a thermal runaway significant signal acquisition module, a processing device and a fire control mainboard, wherein the microphone, the triaxial acceleration sensor and the thermal runaway significant signal acquisition module are respectively arranged on the target battery;

[0115] The microphone is communicatively connected to the processing device, and is used to collect the pressure relief exhaust sound signal of the safety valve of the target battery in real time, and transmit the collected result to the processing device in real time;

[0116] The three-axis acceleration sensor is communicatively connected to the processing device, and is used to collect the vibration signal of the PACK box of the target battery in real time, and transmit the collected result to the processing device in real time;

[0117] The thermal runaway significant signal acquisition module is communicatively connected to the processing device, and is used to acquire K thermal runaway significant signals of the target battery in real time, and transmit the acquisition results to the processing device in real time, wherein K represents a positive integer, and the thermal runaway significant signal refers to a battery monitoring signal that changes flatly over time when the battery is operating normally but changes dramatically over time when the battery is in thermal runaway;

[0118] The processing device is communicatively connected to the fire control mainboard, and is used to execute the battery thermal runaway early warning method as described in the first aspect, and transmit the obtained battery thermal runaway early warning decision basis for the target battery to the fire control mainboard;

[0119] The fire control mainboard is used to determine and execute a corresponding battery thermal runaway early warning decision according to the battery thermal runaway early warning decision basis.

[0120] In one possible design, the thermal runaway significant signal acquisition module includes but is not limited to a gas sensor for real-time acquisition of flammable and explosive gas concentration signals in a PACK box, a pressure sensor for real-time acquisition of air pressure signals in a PACK box, and / or a BMS module for real-time acquisition of battery cell voltage signals in a PACK box and / or battery cell temperature signals in a PACK box, etc., wherein the flammable and explosive gas concentration signal in the PACK box, the air pressure signal in the PACK box, the battery cell voltage signal in the PACK box and / or the battery cell temperature signal in the PACK box are used as the thermal runaway significant signal.

[0121] The working process, working details and technical effects of the aforementioned system provided in the third aspect of this embodiment can be referred to the battery thermal runaway early warning method described in the first aspect, and will not be described in detail here.

[0122] like Figure 4 As shown, the fourth aspect of this embodiment provides a computer device for executing the battery thermal runaway early warning method as described in the first aspect, including a memory, a processor and a transceiver that are sequentially connected in communication, wherein the memory is used to store a computer program, the transceiver is used to send and receive messages, and the processor is used to read the computer program to execute the battery thermal runaway early warning method as described in the first aspect. For example, the memory may include, but is not limited to, a random access memory (Random-Access Memory, RAM), a read-only memory (Read-Only Memory, ROM), a flash memory (Flash Memory), a first-in-first-out memory (First Input First Output, FIFO) and / or a first-in-last-out memory (First Input Last Output, FILO), etc.; the processor may include, but is not limited to, a microprocessor of the STM32F105 series. In addition, the computer device may also include, but is not limited to, a power module, a display screen and other necessary components.

[0123] The working process, working details and technical effects of the aforementioned computer device provided in the fourth aspect of this embodiment can be referred to the battery thermal runaway early warning method described in the first aspect, and will not be described in detail here.

[0124] In a fifth aspect of this embodiment, there is provided a computer-readable storage medium storing instructions including the battery thermal runaway early warning method as described in the first aspect, that is, the computer-readable storage medium stores instructions, and when the instructions are executed on a computer, the battery thermal runaway early warning method as described in the first aspect is executed. The computer-readable storage medium refers to a carrier for storing data, which may include but is not limited to computer-readable storage media such as floppy disks, optical disks, hard disks, flash memories, USB flash drives, and / or memory sticks, and the computer may be a general-purpose computer, a special-purpose computer, a computer network, or other programmable devices.

[0125] The working process, working details and technical effects of the aforementioned computer-readable storage medium provided in the fifth aspect of this embodiment can be referred to the battery thermal runaway early warning method described in the first aspect, and will not be repeated here.

[0126] A sixth aspect of this embodiment provides a computer program product, including a computer program or an instruction, wherein when the computer program or the instruction is executed by a computer, the battery thermal runaway early warning method as described in the first aspect is implemented. The computer may be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device.

[0127] Finally, it should be noted that the above description is only a preferred embodiment of the present invention and is not intended to limit the protection scope of the present invention. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A battery thermal runaway early warning method, characterized in that: include: Acquire the safety valve pressure relief exhaust sound signal, the PACK box vibration signal and K significant thermal runaway signals obtained by monitoring the target battery in the most recent unit time period, wherein K represents a positive integer, and the significant thermal runaway signal refers to a battery monitoring signal that changes flatly over time when the battery is operating normally but changes dramatically over time when the battery is in thermal runaway; The safety valve pressure relief exhaust sound signal, the PACK box vibration signal and the K significant thermal runaway signals are synchronously sampled and processed respectively to obtain a sound signal sequence corresponding to the safety valve pressure relief exhaust sound signal, a vibration signal sequence corresponding to the PACK box vibration signal and K significant signal sequences corresponding to the K significant thermal runaway signals one by one; Performing cross-correlation fusion processing on the acoustic signal sequence and the vibration signal sequence to obtain acoustic and vibration signal feature data; Importing the acoustic vibration signal characteristic data into a first thermal runaway classification and recognition model pre-trained based on a first artificial intelligence algorithm, and outputting a thermal runaway recognition probability corresponding to the acoustic vibration signal, wherein the thermal runaway recognition probability refers to the probability of classifying and identifying a battery thermal runaway event occurring in the target battery in the most recent unit time period; For each thermal runaway significant signal among the K thermal runaway significant signals, importing the corresponding significant signal sequence into a corresponding second thermal runaway classification and recognition model obtained by pre-training based on a second artificial intelligence algorithm, and outputting the corresponding thermal runaway recognition probability; According to the acoustic vibration signal and the K significant thermal runaway signals, 2 K+1 A thermal runaway determination situation combination, wherein the thermal runaway determination situation combination includes each thermal runaway determination flag bit corresponding to each signal in the acoustic vibration signal and the K thermal runaway significant signals, the thermal runaway determination flag bit is a valid bit, indicating that a battery thermal runaway event occurs in the most recent unit time period based on the corresponding signal alone, and the thermal runaway judgment flag bit is an invalid bit, indicating that a battery thermal runaway event does not occur in the most recent unit time period based on the corresponding signal alone; In the 2 K+1 For each thermal runaway judgment situation combination in the thermal runaway judgment situation combination, the posterior probability under the corresponding combination is calculated according to the following formula: In the formula, c represents a positive integer, P h,c represents the posterior probability under the cth thermal runaway judgment situation combination, and the cth thermal runaway judgment situation combination belongs to the 2 K+1 Thermal runaway judgment combination, P c,purpose represents the probability of the cth thermal runaway judgment situation combination under the condition of a thermal runaway event, k represents a positive integer, p k represents the thermal runaway recognition probability corresponding to the kth signal among the acoustic vibration signal and the K thermal runaway significant signals, B c,k represents the thermal runaway determination flag corresponding to the kth signal in the cth thermal runaway determination situation combination, P c,Npurpose represents the probability of the cth thermal runaway judgment situation combination under the condition that no thermal runaway event occurs, p c,k,x represents the thermal runaway false alarm rate corresponding to the kth signal, P purpose Indicates the empirical probability of a battery thermal runaway event occurring in the target battery; In the 2 K+1 A certain thermal runaway judgment situation combination among the thermal runaway judgment situation combinations and corresponding to the maximum a posteriori probability is used as a battery thermal runaway early warning decision basis for the target battery and is output.

2. The battery thermal runaway early warning method according to claim 1, characterized in that: The K significant thermal runaway signals include a concentration signal of inflammable and explosive gas in the PACK box, an air pressure signal in the PACK box, a voltage signal of a battery cell in the PACK box, and / or a temperature signal of a battery cell in the PACK box.

3. The battery thermal runaway early warning method according to claim 2, characterized in that: When the K significant thermal runaway signals include a cell voltage signal in a PACK box or a cell temperature signal in a PACK box and the target battery includes a plurality of cells, for each significant thermal runaway signal in the K significant thermal runaway signals, the corresponding significant signal sequence is introduced into a corresponding second thermal runaway classification and recognition model obtained by pre-training based on a second artificial intelligence algorithm, and the corresponding thermal runaway recognition probability is output, including: For the voltage signal of the battery cells in the PACK box or the temperature signal of the battery cells in the PACK box among the K significant thermal runaway signals, firstly import the corresponding significant signal sequence of each battery cell into the second thermal runaway classification and recognition model obtained by pre-training based on the second artificial intelligence algorithm, output the corresponding thermal runaway recognition probability of each battery cell, and then take the corresponding maximum value of the thermal runaway recognition probability as the corresponding final value of the thermal runaway recognition probability, wherein each battery cell belongs to the multiple battery cells.

4. The battery thermal runaway early warning method according to claim 1, characterized in that: The acoustic signal sequence and the vibration signal sequence are cross-correlatedly fused to obtain acoustic and vibration signal feature data, including: Performing windowing and framing processing on the acoustic signal sequence and the vibration signal sequence respectively to obtain M acoustic signal frames corresponding to the acoustic signal sequence and M vibration signal frames corresponding to the vibration signal sequence, wherein M represents a positive integer greater than or equal to 2; For each of the M sound signal frames and the M vibration signal frames, the corresponding short-time energy is calculated respectively, and the corresponding spectrum centroid is obtained by short-time Fourier transform; For each pair of sound signal frames and vibration signal frames in the M sound signal frames and the M vibration signal frames and having the same sampling period, cross-correlation calculation is performed on all sampling values ​​of the corresponding two frames to obtain a corresponding linear cross-correlation sequence; Calculating the root mean square value of the linear cross-correlation sequence of each pair of sound signal frames and vibration signal frames respectively, and summarizing them to obtain a linear cross-correlation root mean square value sequence; Performing the cross-correlation calculation on all short-time energies of the M sound signal frames and the M vibration signal frames to obtain a short-time energy cross-correlation sequence, and further performing the cross-correlation calculation on all frequency spectrum centroids of the M sound signal frames and the M vibration signal frames to obtain a frequency spectrum centroid cross-correlation sequence; The linear cross-correlation root mean square value sequence, the short-time energy cross-correlation sequence and / or the spectrum centroid cross-correlation sequence are combined into acoustic vibration signal characteristic data.

5. The battery thermal runaway early warning method according to claim 1, characterized in that: The first artificial intelligence algorithm or the second artificial intelligence algorithm adopts a support vector machine algorithm.

6. A battery thermal runaway early warning device, characterized in that: It includes a monitoring signal acquisition unit, a signal sequence sampling unit, a cross-correlation fusion processing unit, a thermal runaway identification processing unit, a thermal runaway situation construction unit, a posterior probability calculation unit and a warning decision output unit; The monitoring signal acquisition unit is used to acquire the safety valve pressure relief exhaust sound signal, the PACK box vibration signal and K significant thermal runaway signals obtained by monitoring the target battery in the most recent unit time period, wherein K represents a positive integer, and the significant thermal runaway signal refers to a battery monitoring signal that changes flatly over time when the battery is operating normally but changes dramatically over time when the battery is in thermal runaway; The signal sequence sampling unit is communicatively connected to the monitoring signal acquisition unit, and is used to synchronously sample and process the safety valve pressure relief exhaust sound signal, the PACK box vibration signal, and the K significant thermal runaway signals, respectively, to obtain a sound signal sequence corresponding to the safety valve pressure relief exhaust sound signal, a vibration signal sequence corresponding to the PACK box vibration signal, and K significant signal sequences corresponding one-to-one to the K significant thermal runaway signals; The cross-correlation fusion processing unit is communicatively connected to the signal sequence sampling unit and is used to perform cross-correlation fusion processing on the acoustic signal sequence and the vibration signal sequence to obtain acoustic and vibration signal feature data; The thermal runaway identification processing unit is communicatively connected to the cross-correlation fusion processing unit and the signal sequence sampling unit, respectively, and is used to import the acoustic vibration signal feature data into a first thermal runaway classification and identification model obtained by pre-training based on a first artificial intelligence algorithm, and output a thermal runaway identification probability corresponding to the acoustic vibration signal, and also for each thermal runaway significant signal among the K thermal runaway significant signals, import the corresponding significant signal sequence into a corresponding second thermal runaway classification and identification model obtained by pre-training based on a second artificial intelligence algorithm, and output the corresponding thermal runaway identification probability, wherein the thermal runaway identification probability refers to the probability of classifying and identifying that a battery thermal runaway event occurs in the target battery in the most recent unit time period; The thermal runaway situation constructing unit is communicatively connected to the monitoring signal acquiring unit and is used to construct and obtain 2 according to the acoustic vibration signal and the K thermal runaway significant signals. K+1 A thermal runaway determination situation combination, wherein the thermal runaway determination situation combination includes each thermal runaway determination flag bit corresponding to each signal in the acoustic vibration signal and the K thermal runaway significant signals, the thermal runaway determination flag bit is a valid bit, indicating that a battery thermal runaway event occurs in the most recent unit time period based on the corresponding signal alone, and the thermal runaway judgment flag bit is an invalid bit, indicating that a battery thermal runaway event does not occur in the most recent unit time period based on the corresponding signal alone; The posterior probability calculation unit is respectively connected to the thermal runaway identification processing unit and the thermal runaway situation construction unit for K+1 For each thermal runaway judgment situation combination in the thermal runaway judgment situation combination, the posterior probability under the corresponding combination is calculated according to the following formula: In the formula, c represents a positive integer, P h,c represents the posterior probability under the cth thermal runaway judgment situation combination, and the cth thermal runaway judgment situation combination belongs to the 2 K+1 Thermal runaway judgment combination, P c,purpose represents the probability of the cth thermal runaway judgment situation combination under the condition of a thermal runaway event, k represents a positive integer, p k represents the thermal runaway recognition probability corresponding to the kth signal among the acoustic vibration signal and the K thermal runaway significant signals, B c,k represents the thermal runaway determination flag corresponding to the kth signal in the cth thermal runaway determination situation combination, P c,Npurpose represents the probability of the cth thermal runaway judgment situation combination under the condition that no thermal runaway event occurs, p c,k,x represents the thermal runaway false alarm rate corresponding to the kth signal, P purpose Indicates the empirical probability of a battery thermal runaway event occurring in the target battery; The early warning decision output unit is communicatively connected to the posterior probability calculation unit, and is used to calculate the K+1 A certain thermal runaway judgment situation combination among the thermal runaway judgment situation combinations and corresponding to the maximum a posteriori probability is used as a battery thermal runaway early warning decision basis for the target battery and is output.

7. A battery thermal runaway early warning system, characterized in that: It includes a microphone, a three-axis acceleration sensor, a thermal runaway significant signal acquisition module, a processing device and a fire control mainboard, wherein the microphone, the three-axis acceleration sensor and the thermal runaway significant signal acquisition module are respectively arranged on the target battery; The microphone is communicatively connected to the processing device, and is used to collect the pressure relief exhaust sound signal of the safety valve of the target battery in real time, and transmit the collected result to the processing device in real time; The three-axis acceleration sensor is communicatively connected to the processing device, and is used to collect the vibration signal of the PACK box of the target battery in real time, and transmit the collected result to the processing device in real time; The thermal runaway significant signal acquisition module is communicatively connected to the processing device, and is used to acquire K thermal runaway significant signals of the target battery in real time, and transmit the acquisition results to the processing device in real time, wherein K represents a positive integer, and the thermal runaway significant signal refers to a battery monitoring signal that changes flatly over time when the battery is operating normally but changes dramatically over time when the battery is in thermal runaway; The processing device is communicatively connected to the fire control mainboard, and is used to execute the battery thermal runaway early warning method according to any one of claims 1 to 5, and transmit the obtained battery thermal runaway early warning decision basis for the target battery to the fire control mainboard; The fire control mainboard is used to determine and execute a corresponding battery thermal runaway early warning decision according to the battery thermal runaway early warning decision basis.

8. A computer device, characterized in that: The invention comprises a memory, a processor and a transceiver which are communicatively connected in sequence, wherein the memory is used to store a computer program, the transceiver is used to send and receive messages, and the processor is used to read the computer program to execute the battery thermal runaway early warning method as claimed in any one of claims 1 to 5.

9. A computer-readable storage medium, characterized in that The computer-readable storage medium stores instructions, and when the instructions are executed on the computer, the battery thermal runaway early warning method as described in any one of claims 1 to 5 is executed.

10. A computer program product comprising a computer program or instructions, characterized in that When the computer program or the instruction is executed by a computer, the battery thermal runaway early warning method according to any one of claims 1 to 5 is implemented.

Images