New energy automobile lithium battery thermal runaway monitoring system and method

Abstract

The invention discloses a new energy automobile lithium battery thermal runaway monitoring system and method. The system comprises a vehicle-mounted lithium battery, a multi-source data acquisition module, a multi-source data fusion module, a hidden Markov model monitoring module, a state identification module, a data analysis processing module, an early warning module, an emergency processing module, a wireless communication module, a cloud platform and a battery management system. The multi-source data acquisition module is responsible for acquiring multi-dimensional data such as voltage, temperature and internal resistance of a lithium battery, and the data fusion module processes the data and then transmits the data to the hidden Markov model monitoring module for state monitoring and prediction. According to the system, a high-order hidden Markov model and a hidden semi-Markov model are adopted, state retention time modeling and a self-adaptive online learning mechanism are combined, and conversion of a battery from a normal state to an abnormal state and a thermal runaway state can be accurately recognized. The system remarkably improves the safety and stability of the lithium battery of the new energy automobile, and is suitable for power battery management of the new energy automobile.

Description

Technical Field

[0001] The present invention belongs to the technical field of lithium battery management for new energy vehicles, and specifically relates to a thermal runaway monitoring system and method for lithium batteries for new energy vehicles. Background Art

[0002] With the widespread use of new energy vehicles, lithium batteries, as the core power source, have a direct impact on the operation of vehicles. Lithium batteries are prone to thermal runaway under conditions of high temperature, overcharge or mechanical damage, which can lead to serious safety accidents such as fire or explosion, posing a high safety risk.

[0003] At present, traditional battery management systems (BMS) usually monitor battery status by setting fixed thresholds such as temperature (such as 60°C) and voltage (such as 4.2V). When the battery temperature exceeds 60°C or the voltage exceeds 4.2V, the system will issue an alarm. However, this static early warning method has limited effect when dealing with rapid fluctuations in battery temperature and voltage, and cannot predict potential failure risks such as thermal runaway. Especially under complex working conditions, static thresholds are difficult to adapt to the dynamic changes in battery status, resulting in the inability to issue early warnings in a timely manner.

[0004] In actual use, lithium batteries have a high risk of thermal runaway due to the high energy demand of electric vehicles, especially in the case of overcharging and abnormal temperature, which may quickly lead to uncontrollable heat diffusion and cause serious safety accidents. In order to accurately grasp the internal state of the battery pack, it is necessary to rely on multiple sensors for real-time monitoring to capture the signals before thermal runaway in time.

[0005] Existing BMS systems may fail under high temperature conditions and fail to output valid data, resulting in missed fault reports or false alarms. In addition, the BMS system cannot fully record the key data generated by the battery during thermal runaway, which is not conducive to post-accident analysis and optimization of battery pack design. At the same time, some existing detection methods lack air pressure sensors, have slow response speeds, and cannot meet high timeliness requirements. Although pressure sensors can provide early warnings by detecting battery cell deformation, they have large errors and low accuracy, making it difficult to provide reliable real-time monitoring.

[0006] Therefore, an improved method is urgently needed to improve the accuracy and timeliness of thermal runaway detection of lithium batteries. Hidden Markov Model (HMM) is a common time series analysis tool that can predict battery status through real-time monitoring data. However, the existing HMM has the problem of insufficient accuracy when processing multidimensional data, especially when dealing with complex working conditions of lithium batteries. To this end, an improved hidden Markov model is proposed, which is specifically used for lithium battery monitoring systems of new energy vehicles. It will significantly improve the real-time and accuracy of battery status monitoring and solve the problem of insufficient accuracy of current detection methods under complex working conditions. Summary of the invention

[0007] The purpose of the present invention is to overcome the defects of the above-mentioned prior art and provide a new energy vehicle lithium battery thermal runaway monitoring system and method.

[0008] The purpose of the present invention can be achieved by the following technical solutions:

[0009] On one hand, the present invention provides a new energy vehicle lithium battery thermal runaway monitoring system, including a vehicle-mounted lithium battery, the system comprising: a multi-source data acquisition module, a multi-source data fusion module, a hidden Markov model monitoring module, a state recognition module, a data analysis and processing module, an early warning module, an emergency processing module, a wireless communication module, a cloud platform and a battery management system;

[0010] The multi-source data acquisition module is used to collect multi-dimensional data of the vehicle-mounted lithium battery, and the output end of the multi-source data acquisition module is connected to the multi-source data fusion module;

[0011] The multi-source data fusion module is used to fuse the collected multi-dimensional data and transmit the fused multi-dimensional data to the hidden Markov model monitoring module;

[0012] The hidden Markov model monitoring module includes a pre-trained hidden Markov model and a hidden semi-Markov model, which is used to perform real-time state analysis on the vehicle-mounted lithium battery through the hidden Markov model according to multi-dimensional data, output the state information of the vehicle-mounted lithium battery, transmit the state information of the vehicle-mounted lithium battery to the state recognition module, predict the duration of the working state of the lithium battery according to the hidden semi-Markov model, and transmit the predicted duration to the data analysis and processing module;

[0013] The state recognition module is used to determine the current working state of the vehicle-mounted lithium battery according to the state information of the vehicle-mounted lithium battery, and send the working state of the vehicle-mounted lithium battery to the data analysis and processing module;

[0014] The data analysis and processing module is used to receive the working status and predicted duration of the vehicle-mounted lithium battery, analyze the working status and predicted duration of the vehicle-mounted lithium battery, and perform corresponding emergency processing;

[0015] The warning module is used to control the sound and light alarm to send out an alarm signal, and send the warning information to the driver terminal through wireless communication;

[0016] The emergency processing module is connected to the battery management system and is used to automatically cut off the vehicle-mounted lithium battery power supply and start the vehicle-mounted cooling system.

[0017] Furthermore, the multi-source data acquisition module includes a voltage sensor, a current sensor, a temperature sensor and an internal resistance sensor;

[0018] The multi-dimensional data of the vehicle-mounted lithium battery includes the operating voltage of the lithium battery, the charge and discharge current of the lithium battery, the temperature of the lithium battery and the internal resistance of the lithium battery.

[0019] Furthermore, the voltage sensor is used to monitor the working voltage of the lithium battery and is installed at the positive and negative output terminals of the lithium battery;

[0020] The current sensor is used to monitor the charging and discharging current of the lithium battery in real time and is installed in the lithium battery circuit;

[0021] The temperature sensor is used to monitor the temperature of the lithium battery in real time and is installed on the surface and surrounding of the lithium battery;

[0022] The internal resistance sensor is used to detect the change of the internal resistance of the lithium battery.

[0023] Furthermore, the working state of the vehicle-mounted lithium battery includes a normal state, a slightly abnormal state, a seriously abnormal state and a thermal runaway state.

[0024] Furthermore, the data analysis and processing module is controlled and connected with the early warning module and the emergency processing module.

[0025] Furthermore, the wireless communication module is connected to the multi-source data acquisition module, the early warning module, and the cloud platform, and is used to upload the multi-dimensional data of the vehicle lithium battery collected in real time by the multi-source data acquisition module and the early warning information of the early warning module to the cloud platform.

[0026] Another aspect of the present invention provides a method for monitoring thermal runaway of a lithium battery in a new energy vehicle, comprising the following steps:

[0027] Step S1: collecting multidimensional data of the vehicle-mounted lithium battery in real time through a multi-source data collection module, and processing the collected multidimensional data through a multi-source data fusion module to obtain pre-processed multidimensional data;

[0028] Step S2: The pre-processed multi-dimensional data is subjected to state prediction by a pre-trained hidden Markov model of a hidden Markov model monitoring module to obtain the probability of transition of each working state of the lithium battery, and the probability of transition of each working state of the lithium battery is transmitted to the state recognition module, and the state recognition module predicts the working state of the lithium battery according to the probability of transition of each working state of the lithium battery;

[0029] Step S3: if the predicted working state of the lithium battery changes, the predicted working state of the lithium battery is analyzed by a pre-trained hidden semi-Markov model of the hidden Markov model monitoring module to obtain a predicted duration of the predicted working state of the lithium battery;

[0030] Step S4: transmitting the predicted working state and duration of the lithium battery to the data analysis and processing module, the data analysis and processing module analyzes the predicted working state and duration of the lithium battery, and performs corresponding emergency processing through the early warning module and the emergency processing module;

[0031] Step S5: The multi-dimensional data of the vehicle lithium battery collected in real time by the multi-source data collection module and the warning information of the warning module are uploaded to the cloud platform through the wireless communication module.

[0032] Furthermore, the collected multidimensional data is processed by a multi-source data fusion module, specifically including the following steps:

[0033] The Kalman filter is used to process the noise in the multi-dimensional data of the vehicle lithium battery. The update formula of the Kalman filter is as follows:

[0034] x t =x t-1 +K t (z t -Hx t-1 )

[0035] Among them, x t is the estimated value of the current state, z t is the observed data, H is the observation matrix, K t is the Kalman gain;

[0036] Linear interpolation is used to fill in the missing values ​​in the multidimensional data of vehicle lithium batteries. The linear interpolation formula is:

[0037]

[0038] The multi-dimensional data of the on-board lithium battery is standardized, and the standardization formula is:

[0039]

[0040] Among them, μ is the mean of the data and σ is the standard deviation of the data.

[0041] Furthermore, the step S4 specifically includes the following steps:

[0042] If the predicted working state of the lithium battery is normal, no processing will be done;

[0043] If the predicted working state of the lithium battery is a slightly abnormal state, and the predicted duration is less than the first preset time, control information is sent to the early warning module, and the early warning module sends a primary early warning to the user terminal;

[0044] If the predicted working state of the lithium battery is a slightly abnormal state, and the predicted duration is greater than or equal to the first preset time, control information is sent to the early warning module, and the early warning module sends an intermediate early warning to the user terminal;

[0045] If the predicted working state of the lithium battery is a serious abnormal state, and the predicted duration is less than the second preset time, control information is sent to the early warning module, and the early warning module sends an intermediate early warning to the user terminal;

[0046] If the predicted working state of the lithium battery is a serious abnormal state, and the predicted duration is greater than or equal to the second preset time, a control message is sent to the early warning module, the early warning module sends a high-level early warning to the user terminal, and sends a control signal to the emergency processing module, and the emergency processing module automatically cuts off the power supply of the vehicle lithium battery and starts the vehicle cooling system;

[0047] If the predicted working state of the lithium battery is a thermal runaway state, control information is sent to the early warning module, the early warning module sends an advanced early warning to the user terminal, and sends a control signal to the emergency processing module, which automatically cuts off the power supply of the on-board lithium battery and starts the on-board cooling system.

[0048] Furthermore, the hidden Markov model and the hidden semi-Markov model are periodically updated according to the data stored in the cloud platform through an adaptive online learning mechanism.

[0049] Compared with the prior art, the present invention has the following advantages:

[0050] (1) The present invention uses a high-order hidden Markov model (HMM) combined with historical state information to predict state transitions, and introduces a hidden semi-Markov model (HSMM) to model the state residence time. This model improvement can not only improve the prediction accuracy of battery state changes, but also accurately capture the change process of the battery from normal state to abnormal and thermal runaway state, thereby improving the overall intelligence and predictive ability of the monitoring system.

[0051] (2) The present invention acquires multi-dimensional data such as voltage, current, temperature, internal resistance, etc. in real time through a multi-source data acquisition module, and combines data fusion technology to filter noise, compensate for missing values, and standardize the data to ensure data accuracy. At the same time, the system has an adaptive online learning mechanism that can dynamically update model parameters according to battery aging and changes in the external environment to maintain monitoring accuracy during long-term operation. This improvement improves the intelligence and adaptability of the system, ensuring that the system can accurately perform status monitoring and fault prediction under different working conditions.

[0052] (3) Based on the battery status identification, the present invention designs a multi-level early warning mechanism. According to the battery status change, the system can issue early warning signals of different levels, from mild abnormality to severe abnormality or thermal runaway. At the same time, the system links the battery management system (BMS) and automatically performs emergency operations, such as cutting off the power supply and starting the cooling system, to ensure safety when a thermal runaway state is detected. This improvement greatly improves the response speed and processing reliability of the battery during thermal runaway, effectively reducing the risk of accidents.

[0053] (4) The present invention proposes a thermal runaway monitoring system and method for lithium batteries of new energy vehicles based on an improved hidden Markov model. Multi-dimensional data of the battery is collected in real time through multi-source sensors, and battery status monitoring and fault prediction are performed in combination with high-order hidden Markov models (HMM) and hidden semi-Markov models (HSMM). The system has an adaptive online learning mechanism that can dynamically adjust model parameters according to battery aging and external environmental changes to ensure monitoring accuracy. At the same time, through a multi-level early warning and emergency handling mechanism, the system can issue early warnings in a timely manner and execute automatic emergency measures, effectively improving the speed and reliability of battery safety monitoring and ensuring the safety and stability of lithium batteries of new energy vehicles. BRIEF DESCRIPTION OF THE DRAWINGS

[0054] Figure 1 is a system structure diagram of the present invention;

[0055] Figure 2 It is a flowchart of the system's multi-level early warning and emergency response;

[0056] Figure 3 It is a flow chart of the steps of the present invention. DETAILED DESCRIPTION

[0057] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of the present invention.

[0058] This embodiment proposes a new energy vehicle lithium battery thermal runaway monitoring system based on an improved hidden Markov model, which can quickly and accurately detect thermal runaway faults of battery packs. The system adopts an in-vehicle bus control mode (such as CAN bus) to achieve real-time data transmission and efficient processing of multiple sensors (voltage, current, temperature, etc.). At the same time, the system also transmits data to the cloud platform through a wireless communication module to achieve remote monitoring. Combined with the improved hidden Markov model (HMM), the system can accurately predict the battery status and identify abnormalities and thermal runaway conditions.

[0059] A new energy vehicle lithium battery thermal runaway monitoring system based on an improved hidden Markov model is characterized by comprising a vehicle-mounted lithium battery, a multi-source data acquisition module, a multi-source data fusion module, a hidden Markov model monitoring module, a state recognition module, a data analysis and processing module, an early warning module, an emergency processing module, a wireless communication module, a cloud platform and a battery management system (BMS); the multi-source data acquisition module is used to collect multi-dimensional data of the vehicle-mounted lithium battery through multiple sensors, the data including real-time parameters such as battery voltage, current, temperature, internal resistance, etc., and the output end of the multi-source data acquisition module is connected to the multi-source data fusion module;

[0060] The multi-source data fusion module is used to perform fusion processing on the collected multi-dimensional data, including noise filtering, missing value compensation and standardization processing on the data, and the output after the data fusion is connected to the hidden Markov model monitoring module;

[0061] The hidden Markov model monitoring module is used to perform real-time state analysis on the fused multi-dimensional data, and use the hidden Markov model to model and monitor the state of the lithium battery. The output end of the monitoring module is connected to the state recognition module; the duration of the working state of the lithium battery is predicted according to the hidden semi-Markov model;

[0062] Hidden Markov Model (HMM) is used to analyze the transition process of lithium battery state in real time. HMM can build a battery state transition probability model based on multi-dimensional data (voltage, current, temperature, internal resistance, etc.). By analyzing the observed data at the current moment, HMM can identify different battery states (such as normal state, mild abnormality, severe abnormality and thermal runaway) and calculate the probability of battery state transition.

[0063] HMM is suitable for predicting changes in battery status in a short period of time. For example, during the monitoring process, the battery status transitions from normal to slightly abnormal, or continues to maintain the current status. Since HMM does not have the ability to directly model the state residence time, it mainly focuses on the state transition process.

[0064] Hidden Semi-Markov Model (HSMM) is used to model the residence time of each state. Based on HMM, HSMM adds modeling and prediction of the length of time the battery state stays in each state. By calculating the residence time distribution of each state, HSMM can determine whether the battery stays in a certain state (such as a mild abnormal state) for too long and predict when it may change to a more serious state (such as a serious abnormality or thermal runaway). This is particularly important in battery failure prediction and early warning.

[0065] Real-time state recognition and transition prediction (completed by HMM): In the system, the hidden Markov model is first used to analyze multidimensional data in real time to identify the current battery state and calculate the probability of state transition. HMM can quickly monitor short-term changes in battery state and identify whether the battery has changed from a normal state to an abnormal state.

[0066] State residence time modeling (done by HSMM): Once HMM identifies that the battery state has changed or entered a certain state, HSMM will take over and start modeling the residence time of the state. By combining historical data with current monitoring data, HSMM analyzes the distribution of the battery's residence time in a certain state, predicts how long the state will last, and determines whether it is necessary to issue an early warning signal or take emergency measures.

[0067] HMM provides real-time transition prediction of battery status, ensuring that the system can quickly identify status changes.

[0068] HSMM enhances the ability to predict long-term trends in battery status by modeling state residence time, especially providing more accurate early warning during the transition of the battery from mild abnormality to severe abnormality or thermal runaway.

[0069] HMM detects battery status: When HMM detects that the battery enters a slightly abnormal state, it outputs the transition probability of the state, and the system recognizes that the battery is abnormal.

[0070] HSMM modeling residence time: HSMM then starts to calculate the residence time of the battery in the abnormal state, determines whether the mild abnormal state will last too long, and predicts whether it will transfer to a serious abnormal state or thermal runaway state.

[0071] Early warning and emergency response: Through the residence time analysis of HSMM, the system can decide when to trigger the early warning or emergency response mechanism to avoid serious faults such as thermal runaway.

[0072] The state recognition module recognizes various states of the battery according to the state information output by the monitoring module, including a normal state, a slightly abnormal state, a seriously abnormal state, and a thermal runaway state;

[0073] The data analysis and processing module is used to further analyze and evaluate the status recognition results. When the lithium battery status is abnormal, the data analysis and processing module will send the abnormal status information to the early warning module and the emergency processing module;

[0074] The warning module is connected to the sound and light alarm and the driver terminal. When the battery status is abnormal, the warning module controls the sound and light alarm to send an alarm signal, and at the same time sends the warning information to the driver terminal through the wireless communication module to ensure that the driver can receive the alarm in time and take countermeasures;

[0075] The emergency processing module is linked with the battery management system (BMS) and can perform emergency processing when the battery enters a serious abnormal state or thermal runaway state, including automatically cutting off the power supply and starting the vehicle cooling system to prevent accidents;

[0076] The wireless communication module (such as LTE / 5G, LoRaWAN, NB-IoT or Zigbee) is used to transmit battery status data and alarm information to the cloud platform. The cloud platform has data storage, analysis and remote monitoring functions. Users can remotely view the operating status of the lithium battery and receive warning information through the user terminal. It should be noted that the wireless communication module has different functions from the bus system (such as CAN bus) inside the new energy vehicle. The former is responsible for data transmission outside the vehicle, while the bus system is used for high-speed communication and real-time data interaction between devices inside the vehicle. The two work together to ensure the overall data transmission efficiency and monitoring capabilities of the system.

[0077] Furthermore, the multi-source data fusion module collects data such as voltage, current, temperature, internal resistance, etc. in real time through sensors, and pre-processes the data, specifically including:

[0078] The voltage sensor is used to monitor the operating voltage of the lithium battery to ensure that the battery operates normally within the operating range. The voltage sensor is installed at the positive and negative output terminals of the battery to detect battery voltage changes in real time to prevent damage to the battery due to overvoltage or undervoltage. Since current usually becomes abnormal earlier than voltage, timely monitoring of current changes is crucial to prevent failures;

[0079] The current sensor is used to monitor the charge and discharge current of the lithium battery in real time to ensure that the current is always maintained within a safe range. The current sensor is installed in the battery circuit and can detect abnormal conditions such as overcurrent or short circuit early, thereby providing early warning for the system and further reducing the safety hazards caused by voltage abnormalities;

[0080] The temperature sensor is used to monitor the temperature of the lithium battery in real time to ensure that the battery temperature remains within a safe range to avoid overheating or thermal runaway of the battery due to high temperature. The temperature sensor is usually installed on the surface of the battery module and key parts around it;

[0081] The internal resistance sensor is used to detect the change of the internal resistance of the lithium battery to determine the aging and health status of the battery. The internal resistance sensor can obtain the impedance change of the battery in real time and evaluate the service life and performance degradation of the battery;

[0082] The noise filtering module is used to filter the noise of the data collected by the sensor, eliminate the data errors caused by the external environment and electromagnetic interference, and ensure the accuracy of the data;

[0083] The standardization processing module is used to perform standardization processing on the collected data of different physical quantities so that data of different dimensions can be effectively compared and analyzed in the same model.

[0084] Furthermore, the hidden Markov model adopts a high-order hidden Markov hypothesis, combines historical state information to predict state transition, and models the state residence time by introducing a hidden semi-Markov model.

[0085] Furthermore, the hidden Markov model and hidden semi-Markov model adopt an adaptive online learning mechanism, which can update model parameters in real time according to battery aging and external environmental changes, thereby improving the accuracy and sensitivity of monitoring, including:

[0086] 1. Real-time data update: The online learning mechanism can continuously receive real-time operating data of lithium batteries (such as voltage, current, temperature, internal resistance, etc.), and update the state transition probability and residence time model of the model based on these data. Through such dynamic data stream input, the model can continuously adjust itself to reduce the prediction error caused by static models.

[0087] 2. Adaptive parameter adjustment: As lithium batteries age and external conditions (such as temperature, humidity, and load) change, the performance of the battery may gradually decay, which will affect the probability of its state transition. The online learning mechanism automatically adjusts the model parameters of HMM and HSMM through an adaptive algorithm, allowing the monitoring system to adapt to changes in battery performance. For example, when the internal resistance of the battery increases, the model will adjust its threshold and transition probability to more accurately predict future state changes.

[0088] 3. Dynamic model reconstruction: The online learning mechanism can not only update the model parameters, but also dynamically reconstruct the entire model structure when necessary. For example, when the battery ages faster than expected or the environment changes dramatically, the system will remodel the battery state transfer process based on the latest data to maintain high-precision monitoring. This reconstruction ensures that the system can still work effectively even in more extreme or uncommon situations.

[0089] How online learning mechanisms are applied:

[0090] 1. Continuous input and update of data streams: Sensors in the system (such as voltage, current, temperature, and internal resistance sensors) continuously collect working data of lithium batteries and input these data into the hidden Markov model (HMM) and hidden semi-Markov model (HSMM). The online learning mechanism compares these data with the current parameters of the model through an algorithm, calculates the error, and updates the model weights.

[0091] 2. Adaptive adjustment of model parameters: The online learning mechanism can calculate the probability of current state transition and the distribution of residence time based on each updated data, and adjust the model parameters in real time. For example, as the battery ages, certain abnormal states may become more frequent, and the system will automatically adjust the transition probability of these states to improve prediction accuracy.

[0092] 3. Model evaluation and feedback loop: The system periodically evaluates the accuracy of the model. By comparing the actual observed data with the data predicted by the model, the system can calculate the model's error and continuously optimize the model based on the feedback loop. This process ensures that the system can maintain accurate state predictions even if the battery state changes gradually.

[0093] Furthermore, the multi-level warning mechanism can send out a corresponding warning signal when the battery state changes from a normal state to a slightly abnormal state or a seriously abnormal state, and trigger an emergency processing mechanism when the battery enters a thermal runaway state.

[0094] Furthermore, the emergency processing module is linked with the battery management system (BMS) and can automatically perform emergency operations such as cutting off the power supply and starting the cooling system when a thermal runaway state is detected.

[0095] Furthermore, the cloud platform can also provide a remote monitoring function, and users can view the operating status of the lithium battery in real time through the cloud platform, and perform remote maintenance and management of the battery based on the analysis results provided by the data statistical analysis module.

[0096] A new energy vehicle lithium battery thermal runaway monitoring system and method based on an improved hidden Markov model further comprises the following steps:

[0097] Step S1: collect multidimensional data of the vehicle-mounted lithium battery in real time through the multi-source data acquisition module, and process the collected multidimensional data through the multi-source data fusion module to obtain pre-processed multidimensional data; specifically, the following steps are included:

[0098] The Kalman filter is used to process the noise in the multi-dimensional data of the vehicle lithium battery. The update formula of the Kalman filter is as follows:

[0099] x t =x t-1 +K t (z t -Hx t-1 )

[0100] Among them, x t is the estimated value of the current state, z t is the observed data, H is the observation matrix, K t is the Kalman gain;

[0101] Linear interpolation is used to fill in the missing values ​​in the multidimensional data of vehicle lithium batteries. The linear interpolation formula is:

[0102]

[0103] The multi-dimensional data of the on-board lithium battery is standardized, and the standardization formula is:

[0104]

[0105] Among them, μ is the mean of the data and σ is the standard deviation of the data.

[0106] Step S2: The pre-processed multi-dimensional data is subjected to state prediction by a pre-trained hidden Markov model of a hidden Markov model monitoring module to obtain the probability of transition of each working state of the lithium battery, and the probability of transition of each working state of the lithium battery is transmitted to the state recognition module, and the state recognition module predicts the working state of the lithium battery according to the probability of transition of each working state of the lithium battery;

[0107] Step S3: if the predicted working state of the lithium battery changes, the predicted working state of the lithium battery is analyzed by a pre-trained hidden semi-Markov model of the hidden Markov model monitoring module to obtain a predicted duration of the predicted working state of the lithium battery;

[0108] Step S4: transmitting the predicted working state and duration of the lithium battery to the data analysis and processing module, the data analysis and processing module analyzes the predicted working state and duration of the lithium battery, and performs corresponding emergency processing through the early warning module and the emergency processing module;

[0109] Step S5: The multi-dimensional data of the vehicle lithium battery collected in real time by the multi-source data collection module and the warning information of the warning module are uploaded to the cloud platform through the wireless communication module.

[0110] The step S4 specifically comprises the following steps:

[0111] If the predicted working state of the lithium battery is normal, no processing will be done;

[0112] If the predicted working state of the lithium battery is a slightly abnormal state, and the predicted duration is less than the first preset time, control information is sent to the early warning module, and the early warning module sends a primary early warning to the user terminal;

[0113] If the predicted working state of the lithium battery is a slightly abnormal state, and the predicted duration is greater than or equal to the first preset time, control information is sent to the early warning module, and the early warning module sends an intermediate early warning to the user terminal;

[0114] If the predicted working state of the lithium battery is a serious abnormal state, and the predicted duration is less than the second preset time, control information is sent to the early warning module, and the early warning module sends an intermediate early warning to the user terminal;

[0115] If the predicted working state of the lithium battery is a serious abnormal state, and the predicted duration is greater than or equal to the second preset time, a control message is sent to the early warning module, the early warning module sends a high-level early warning to the user terminal, and sends a control signal to the emergency processing module, and the emergency processing module automatically cuts off the power supply of the vehicle lithium battery and starts the vehicle cooling system;

[0116] If the predicted working state of the lithium battery is a thermal runaway state, control information is sent to the early warning module, the early warning module sends an advanced early warning to the user terminal, and sends a control signal to the emergency processing module, which automatically cuts off the power supply of the on-board lithium battery and starts the on-board cooling system.

[0117] The state set of the hidden Markov model is S={S1, S2, S3, S4}, which respectively correspond to a normal state, a slightly abnormal state, a seriously abnormal state and a thermal runaway state.

[0118] The state transfer matrix A is in the form of:

[0119]

[0120] Among them, a ij Indicates that from state S i Transfer to state S j The model calculates the transition probability of the battery state by processing the observation data at the current moment and the historical moment.

[0121] In one embodiment, the system introduces a hidden semi-Markov model (HSMM) to model the residence time of the battery state. The hidden semi-Markov model can describe the duration of the battery in a certain state, and adopts the following residence time modeling formula:

[0122]

[0123] Among them, μ i For state S i The average residence time, σ i is the standard deviation of the state. Through this model, the system can more accurately describe the battery's residence time in different states, thereby improving the accuracy of fault prediction.

[0124] In one embodiment, the hidden Markov model uses a forward-backward algorithm to predict the state. The forward algorithm is used to calculate the observed sequence O = {O1, O2, ..., O t}, the battery is in state S iThe forward probability α t The calculation formula for (i) is:

[0125]

[0126] α t (i): Forward probability, indicating that the system is in state S at time t i The probability of ji State transition probability, indicating the transition from state S j Transfer to state S i The probability of i (O t ):Observation probability, indicating that in state S i When O t The probability of this observation.

[0127] The backward algorithm is used to calculate the probability of the sequence observed from time t to the end time, the backward probability β t The calculation formula for (i) is:

[0128]

[0129] β t (i): Backward probability, indicating that starting from time t, the system changes from state S i The probability of transferring to the subsequent time t+1, t+2, ..., T.

[0130] Through the forward-backward algorithm, the system can predict the posterior probability of the current battery state and predict and evaluate the battery state based on the probability, including:

[0131] The system monitors the battery's operating status in real time through a variety of sensors (such as voltage, current, temperature, etc.). At each moment, the system obtains battery status information, such as the current voltage, whether the temperature is too high, whether the internal resistance is normal, etc.

[0132] These real-time observation data reflect the current state of the battery, and the model will use these data to determine whether the battery is in a normal state or in some abnormal state.

[0133] In addition to the current data, HMM also considers past observations, that is, the battery status information at previous times. These historical data can help the model understand the trends and changes in the battery status, such as whether the battery has been in a stable state or whether it has gradually become abnormal.

[0134] By introducing historical data, HMM can better grasp the dynamic trend of battery state changes instead of relying solely on the current instantaneous state.

[0135] The hidden Markov model defines the probability of transitioning from one state (such as a normal state) to another state (such as a slightly abnormal state) through a state transition matrix (i.e., matrix A). These transition probabilities reflect the possibility of the battery transitioning from one state to another.

[0136] For example, assuming that the battery is currently in a normal state, the element a in the transfer matrix A 12 Indicates the probability of the battery transitioning from a normal state to a slightly abnormal state. If this probability is large, the model predicts that the battery is likely to enter a slightly abnormal state in the short term.

[0137] HMM combines the observation data at the current moment (i.e. the current state) with the observation data at the historical moment (i.e. the previous state) to comprehensively analyze the state change trend of the battery.

[0138] Through this data processing, the model can make predictions about future states, such as whether the battery is about to move from a normal state to an abnormal state, or whether the battery has a high probability of entering thermal runaway.

[0139] By processing these data, the model calculates the specific probability of transitioning from one state to another. These transition probabilities help the system predict future state changes of the battery.

[0140] For example, if the current status data indicates that the battery temperature is rising, and the historical data also shows that the battery temperature has continued to rise over the past few minutes, the model may conclude that there is a high probability that the battery has moved from a mild abnormal state to a severe abnormal state.

[0141] If the above functions are implemented in the form of software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, or the part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium, including several instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk, etc., which can store program code.

[0142] The above is only a specific embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any technician familiar with the technical field can easily think of various equivalent modifications or replacements within the technical scope disclosed by the present invention, and these modifications or replacements should be included in the protection scope of the present invention. Therefore, the protection scope of the present invention shall be based on the protection scope of the claims.

Claims

1. A new energy vehicle lithium battery thermal runaway monitoring system, comprising a vehicle-mounted lithium battery, characterized in that: The system includes: a multi-source data acquisition module, a multi-source data fusion module, a hidden Markov model monitoring module, a state recognition module, a data analysis and processing module, an early warning module, an emergency processing module, a wireless communication module, a cloud platform and a battery management system; The multi-source data acquisition module is used to collect multi-dimensional data of the vehicle-mounted lithium battery, and the output end of the multi-source data acquisition module is connected to the multi-source data fusion module; The multi-source data fusion module is used to fuse the collected multi-dimensional data and transmit the fused multi-dimensional data to the hidden Markov model monitoring module; The hidden Markov model monitoring module includes a pre-trained hidden Markov model and a hidden semi-Markov model, which is used to perform real-time state analysis on the vehicle-mounted lithium battery through the hidden Markov model according to multi-dimensional data, output the state information of the vehicle-mounted lithium battery, transmit the state information of the vehicle-mounted lithium battery to the state recognition module, predict the duration of the working state of the lithium battery according to the hidden semi-Markov model, and transmit the predicted duration to the data analysis and processing module; The state recognition module is used to determine the current working state of the vehicle-mounted lithium battery according to the state information of the vehicle-mounted lithium battery, and send the working state of the vehicle-mounted lithium battery to the data analysis and processing module; The data analysis and processing module is used to receive the working status and predicted duration of the vehicle-mounted lithium battery, analyze the working status and predicted duration of the vehicle-mounted lithium battery, and perform corresponding emergency processing; The warning module is used to control the sound and light alarm to send out an alarm signal, and send the warning information to the driver terminal through wireless communication; The emergency processing module is connected to the battery management system and is used to automatically cut off the vehicle-mounted lithium battery power supply and start the vehicle-mounted cooling system.

2. A new energy vehicle lithium battery thermal runaway monitoring system according to claim 1, characterized in that: The multi-source data acquisition module includes a voltage sensor, a current sensor, a temperature sensor and an internal resistance sensor; The multi-dimensional data of the vehicle-mounted lithium battery includes the operating voltage of the lithium battery, the charge and discharge current of the lithium battery, the temperature of the lithium battery and the internal resistance of the lithium battery.

3. A new energy vehicle lithium battery thermal runaway monitoring system according to claim 2, characterized in that: The voltage sensor is used to monitor the working voltage of the lithium battery and is installed at the positive and negative output terminals of the lithium battery; The current sensor is used to monitor the charging and discharging current of the lithium battery in real time and is installed in the lithium battery circuit; The temperature sensor is used to monitor the temperature of the lithium battery in real time and is installed on the surface and surrounding of the lithium battery; The internal resistance sensor is used to detect the change of the internal resistance of the lithium battery.

4. A new energy vehicle lithium battery thermal runaway monitoring system according to claim 1, characterized in that: The working states of the vehicle-mounted lithium battery include a normal state, a slightly abnormal state, a seriously abnormal state and a thermal runaway state.

5. A new energy vehicle lithium battery thermal runaway monitoring system according to claim 1, characterized in that: The data analysis and processing module is control-connected with the early warning module and the emergency processing module.

6. A new energy vehicle lithium battery thermal runaway monitoring system according to claim 1, characterized in that: The wireless communication module is connected to the multi-source data acquisition module, the early warning module and the cloud platform, and is used to upload the multi-dimensional data of the vehicle lithium battery collected in real time by the multi-source data acquisition module and the early warning information of the early warning module to the cloud platform.

7. A method for monitoring thermal runaway of lithium batteries for new energy vehicles, characterized in that: The following steps are involved: Step S1: collecting multidimensional data of the vehicle-mounted lithium battery in real time through a multi-source data collection module, and processing the collected multidimensional data through a multi-source data fusion module to obtain pre-processed multidimensional data; Step S2: The pre-processed multi-dimensional data is subjected to state prediction by a pre-trained hidden Markov model of a hidden Markov model monitoring module to obtain the probability of transition of each working state of the lithium battery, and the probability of transition of each working state of the lithium battery is transmitted to the state recognition module, and the state recognition module predicts the working state of the lithium battery according to the probability of transition of each working state of the lithium battery; Step S3: if the predicted working state of the lithium battery changes, the predicted working state of the lithium battery is analyzed by a pre-trained hidden semi-Markov model of the hidden Markov model monitoring module to obtain a predicted duration of the predicted working state of the lithium battery; Step S4: transmitting the predicted working state and duration of the lithium battery to the data analysis and processing module, the data analysis and processing module analyzes the predicted working state and duration of the lithium battery, and performs corresponding emergency processing through the early warning module and the emergency processing module; Step S5: The multi-dimensional data of the vehicle lithium battery collected in real time by the multi-source data collection module and the warning information of the warning module are uploaded to the cloud platform through the wireless communication module.

8. A method for monitoring thermal runaway of a lithium battery for a new energy vehicle according to claim 7, characterized in that: The collected multi-dimensional data is processed by a multi-source data fusion module, specifically comprising the following steps: The Kalman filter is used to process the noise in the multi-dimensional data of the vehicle lithium battery. The update formula of the Kalman filter is as follows: x t =x t-1 +K t (z t -Hx t-1 ) Among them, x t is the estimated value of the current state, x t-1 is the estimated value of the state at the previous moment, z t is the observed data, H is the observation matrix, K t is the Kalman gain; Linear interpolation is used to fill in the missing values ​​in the multidimensional data of vehicle lithium batteries. The linear interpolation formula is: x i is the current missing value, x i-1 is the data before the missing value, x i+1 The data after the missing value; The multi-dimensional data of the on-board lithium battery is standardized, and the standardization formula is: Among them, x′ is the data after standardization, x is the data before standardization, μ is the mean of the data, and σ is the standard deviation of the data.

9. A method for monitoring thermal runaway of a lithium battery for a new energy vehicle according to claim 7, characterized in that: The step S4 specifically comprises the following steps: If the predicted working state of the lithium battery is normal, no processing will be done; If the predicted working state of the lithium battery is a slightly abnormal state, and the predicted duration is less than the first preset time, control information is sent to the early warning module, and the early warning module sends a primary early warning to the user terminal; If the predicted working state of the lithium battery is a slightly abnormal state, and the predicted duration is greater than or equal to the first preset time, control information is sent to the early warning module, and the early warning module sends an intermediate early warning to the user terminal; If the predicted working state of the lithium battery is a serious abnormal state, and the predicted duration is less than the second preset time, control information is sent to the early warning module, and the early warning module sends an intermediate early warning to the user terminal; If the predicted working state of the lithium battery is a serious abnormal state, and the predicted duration is greater than or equal to the second preset time, a control message is sent to the early warning module, the early warning module sends a high-level early warning to the user terminal, and sends a control signal to the emergency processing module, and the emergency processing module automatically cuts off the power supply of the vehicle lithium battery and starts the vehicle cooling system; If the predicted working state of the lithium battery is a thermal runaway state, control information is sent to the early warning module, the early warning module sends an advanced early warning to the user terminal, and sends a control signal to the emergency processing module, which automatically cuts off the power supply of the on-board lithium battery and starts the on-board cooling system.

10. A method for monitoring thermal runaway of a lithium battery for a new energy vehicle according to claim 7, characterized in that: The hidden Markov model and the hidden semi-Markov model are periodically updated according to the data stored in the cloud platform through an adaptive online learning mechanism.

Images