Power battery thermal management method and system based on aging state adaptation

Abstract

This invention provides a power battery thermal management method and system based on aging state adaptation, belonging to the field of battery thermal management technology. The method includes: Step 1: Machine learning of discharge records of different power batteries to train a battery discharge prediction model; Step 2: Inputting the expected data of the target power battery into the battery discharge prediction model to obtain the predicted temperature curve of the target power battery; Step 3: Determining the future thermal management window based on the position of the predicted temperature in the predicted temperature curve within the ideal operating temperature range; Step 4: Determining and applying the thermal management strategy based on the heat generation power distribution within the future thermal management window. This invention's power battery thermal management method and system based on aging state adaptation improves the prediction accuracy of future battery output parameters, dynamically determines the timing and intensity of different thermal management methods based on the predicted heat load characteristics of discharge stages that do not meet thermal management expectations, and reduces the risk of thermal runaway in aging batteries.

Description

Technical Field

[0001] This invention relates to the field of battery thermal management technology, and in particular to a power battery thermal management method and system based on aging condition adaptation. Background Technology

[0002] With the rapid development of new energy vehicles, energy storage systems and portable electronic devices, the safety, reliability and service life of power batteries, as core energy storage units, are receiving increasing attention.

[0003] During actual operation, power batteries inevitably experience performance degradation, or "aging," due to various factors such as charge-discharge cycles, ambient temperature, and load conditions. During battery aging, their voltage characteristics, internal resistance changes, and heat generation mechanisms exhibit high nonlinearity and time-varying characteristics. Traditional power battery thermal management systems typically rely on fixed thresholds or empirical rules for thermal regulation; for example, activating cooling devices when the battery temperature exceeds a certain upper limit and heating devices when it falls below a lower limit. Such methods do not adequately consider individual battery differences and their aging states over time, making it difficult to implement precise and dynamic thermal management for batteries at different stages of aging. Especially for aged batteries, their heat generation characteristics, heat conduction paths, and thermal response speeds differ significantly from new batteries. Using a uniform thermal management strategy can easily lead to inappropriate thermal management, thereby increasing the risk of thermal runaway.

[0004] In view of this, there is an urgent need for a power battery thermal management method and system that is adaptive to aging conditions, in order to at least address the above-mentioned shortcomings. Summary of the Invention

[0005] One objective of this invention is to provide a power battery thermal management method and system based on aging state adaptation. It utilizes complete discharge records of the power battery under different aging states and operating conditions to train a battery discharge prediction model. The model learns the complex nonlinear relationship between input and output parameters. When used for discharge prediction later, it can better combine multiple factors such as battery aging state to predict battery temperature, battery voltage, and DC internal resistance, improving the prediction accuracy of future battery output parameters. The future thermal management window is determined based on the predicted temperature curve and the introduced ideal operating temperature range. The timing and intensity of different thermal management methods are dynamically determined based on the predicted heat load characteristics of discharge stages that do not meet thermal management expectations, effectively reducing the risk of thermal runaway in aging batteries and extending battery life.

[0006] The power battery thermal management method based on aging state adaptation provided in this embodiment of the invention includes:

[0007] Step 1: Machine learning of discharge records of different power batteries to train a battery discharge prediction model; the discharge records include: input parameters and output parameters. Input parameter types include: aging state, battery specifications, discharge rate, ambient temperature and depth of discharge; output parameter types include: battery temperature, battery voltage and DC internal resistance.

[0008] Step 2: Input the expected data corresponding to the input parameter type of the target power battery into the battery discharge prediction model to obtain the predicted temperature curve of the target power battery;

[0009] Step 3: Determine the future thermal management window based on the position of the predicted temperature in the ideal operating temperature range according to the predicted temperature curve;

[0010] Step 4: Determine and apply the thermal management strategy based on the heat generation power distribution within the future thermal management window.

[0011] Preferably, step 1: Machine learning of discharge records from different power batteries to train a battery discharge prediction model, including:

[0012] The input parameters in the discharge record are characterized, and an input feature vector is constructed;

[0013] The output parameters in the discharge record are characterized, and an output feature vector is constructed;

[0014] The battery discharge prediction model is trained by using the input feature vector as the input of the preset machine learning model and the output feature vector as the output of the machine learning model.

[0015] Preferably, in step 2, the expected discharge rate data of the target power battery is determined based on its application conditions.

[0016] Preferably, the future thermal management window is determined, including:

[0017] Analyze the interval position;

[0018] If the predicted temperature is higher than the upper limit of the ideal operating temperature range, the corresponding depth of discharge at the predicted temperature is collected and used as the first window.

[0019] If the predicted temperature is lower than the lower limit of the ideal operating temperature range, the corresponding depth of discharge at the predicted temperature is collected and used as a second window.

[0020] The first and second windows will be used together as the future thermal management window.

[0021] Preferably, the methods for obtaining the heat generation power distribution within the future thermal management window include:

[0022] Based on the experimental measurement records of the target power battery, determine the entropy change coefficient of the target power battery;

[0023] Determine the predicted internal resistance corresponding to the expected discharge depth within the future thermal management window;

[0024] Based on the discharge rate of the target battery, the battery specifications of the target power battery, and the predicted internal resistance, determine the irreversible heat generation power corresponding to the expected depth of discharge.

[0025] The reversible heat generation power is determined based on the target battery's discharge rate, the predicted temperature corresponding to the expected depth of discharge, and the entropy change coefficient.

[0026] The irreversible heat generation power and reversible heat generation power are each associated with their corresponding expected discharge depth. Once the expected discharge depth within the future thermal management window has been associated, the heat generation power distribution is obtained.

[0027] Preferably, the entropy change coefficient of the target power battery is determined based on the experimental measurement records of the target power battery, including:

[0028] Determine the battery aging model based on the aging state of the tested battery;

[0029] Based on the battery aging model, predict the location of the nonlinear inflection point;

[0030] The set temperature points of the constant temperature chamber are dynamically set according to the preset number of temperature points and the position of the nonlinear inflection point based on the aging state of the test battery.

[0031] The experimental measurement records of the test battery are obtained based on the set temperature points;

[0032] Obtain sub-measurement data from experimental measurement records under different known aging states;

[0033] By fitting the open-circuit voltage value versus temperature curve in the same SOC state of the sub-measurement data, the entropy change coefficient curve of the known aging state can be obtained.

[0034] By comparing the entropy change coefficient change curves under different known aging states, the target entropy change coefficient change curve of the target power battery can be deduced.

[0035] The entropy change coefficient is determined based on the target entropy change coefficient curve and the future thermal management window.

[0036] Preferably, step 4: Determine and apply a thermal management strategy based on the heat generation power distribution within the future thermal management window, including:

[0037] Based on the window type of the continuous windows in the future thermal management window, the risk types are determined; the risk types include: low temperature risk and high temperature risk.

[0038] Determine the thermal management model based on the type of risk;

[0039] Based on the heat generation power distribution within the continuous window, the total heat generation power curve and the dominant heat type pattern are determined; the dominant heat type pattern is the trend of the proportion of the dominant heat type in the total heat generation power.

[0040] The mode intensity curve is determined based on the total power curve of the discharge heat generation;

[0041] The mode intensity curve is modified according to the dominant law of thermal type to obtain the target mode intensity curve;

[0042] The thermal management strategy is determined based on the thermal management mode and target mode intensity curve associated with the continuous window.

[0043] Preferably, the model intensity curve is corrected according to the dominant law of thermal type to obtain the target model intensity curve, including:

[0044] Based on the dominant laws of heat type, the trend slope, entropy thermal response delay coefficient, and trend switching point are extracted, and a thermal behavior feature vector is constructed.

[0045] A neural fuzzy logic network is constructed using fuzzy rules. The thermal behavior feature vector is input into the neural fuzzy logic network to obtain the correction factor.

[0046] The target mode intensity curve is obtained by correcting the mode intensity curve using the correction factor.

[0047] The aging-adaptive power battery thermal management method provided in this embodiment of the invention further includes:

[0048] After each discharge, the fuzzy membership function parameters and rule weights are dynamically optimized based on the error feedback between the actual temperature and the predicted temperature during the most recent historical discharge.

[0049] The aging-adaptive power battery thermal management system provided in this embodiment of the invention includes:

[0050] The training module is used to learn from the discharge records of different power batteries and train the battery discharge prediction model. The discharge records include input parameters and output parameters. The types of input parameters include: aging state, battery specifications, discharge rate, ambient temperature and depth of discharge. The types of output parameters include: battery temperature, battery voltage and DC internal resistance.

[0051] The prediction module is used to input the expected data of the target power battery corresponding to the input parameter type into the battery discharge prediction model to obtain the predicted temperature curve of the target power battery.

[0052] The management window determination module is used to determine the future thermal management window based on the position of the predicted temperature in the predicted temperature curve within the ideal operating temperature range.

[0053] The management strategy application module is used to determine and apply thermal management strategies based on the heat generation power distribution within the future thermal management window.

[0054] The beneficial effects of this invention are as follows:

[0055] This invention utilizes complete discharge records of power batteries under different aging states and operating conditions to train a battery discharge prediction model. The model learns the complex nonlinear relationship between input and output parameters. When used for subsequent discharge prediction, it can better combine multiple factors such as battery aging state to predict battery temperature, battery voltage, and DC internal resistance, improving the prediction accuracy of future battery output parameters. Based on the predicted temperature curve and the introduced ideal operating temperature range, the future thermal management window is determined. The timing and intensity of different thermal management methods are dynamically determined based on the predicted thermal load characteristics of discharge stages that do not meet thermal management expectations, effectively reducing the risk of thermal runaway in aging batteries and extending battery life.

[0056] Other features and advantages of the invention will be set forth in the description which follows, and will be apparent in part from the description, or may be learned by practicing the invention. The objects and other advantages of the invention may be realized and obtained by means of the structures particularly pointed out in this application.

[0057] The technical solution of the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. Attached Figure Description

[0058] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used in conjunction with embodiments of the invention to explain the invention and do not constitute a limitation thereof. In the drawings:

[0059] Figure 1 This is a schematic diagram of a power battery thermal management method based on aging state adaptation in an embodiment of the present invention;

[0060] Figure 2 This is a schematic diagram of a power battery thermal management system based on aging state adaptation in an embodiment of the present invention. Detailed Implementation

[0061] The preferred embodiments of the present invention will be described below with reference to the accompanying drawings. It should be understood that the preferred embodiments described herein are for illustration and explanation only and are not intended to limit the present invention.

[0062] This invention provides a power battery thermal management method based on aging state adaptation, such as... Figure 1 As shown, it includes:

[0063] Step 1: Machine learning of discharge records of different power batteries to train a battery discharge prediction model; the discharge records include: input parameters and output parameters. The types of input parameters include: aging state, battery specifications, discharge rate, ambient temperature and depth of discharge; the types of output parameters include: battery temperature, battery voltage and DC internal resistance.

[0064] In this embodiment, the discharge record refers to historical data recorded by the power battery within a complete discharge cycle, including: aging state (capacity decay), battery specifications (rated capacity, size, weight, and chemical system), discharge rate, ambient temperature, depth of discharge (percentage of discharged capacity relative to the total usable capacity of the battery), battery temperature, battery voltage, and DC internal resistance. The battery discharge prediction model is a mathematical model built based on machine learning. By learning the relationship between input and output parameters in the discharge record, it can predict future changes in the battery's output parameters based on new input conditions. Step 1 specifically includes:

[0065] Step 11: Characterize the input parameters in the discharge record and construct the input feature vector.

[0066] In this embodiment, parameter characterization refers to converting different parameter types in the discharge record into numerical forms (feature values) that are more suitable for machine learning models to process. The above feature values ​​are used to construct vectors according to preset vector construction rules to obtain input feature vectors. For example, [85%, 480, 5C, 26, 30%] indicates that the SOH of the power battery is 85%, the battery size is 480kg, the discharge rate is 5C, the ambient temperature is 26 degrees Celsius, and the depth of discharge is 30%.

[0067] Step 12: Characterize the output parameters in the discharge record and construct the output feature vector.

[0068] In this embodiment, the output feature vector is constructed in the same way as the input feature vector. For example, the input feature vector is [29, 389.8, 0.375], which means that the battery temperature is 29 degrees Celsius, the battery voltage is 389.8V, and the battery internal resistance is 0.375mΩ when the depth of discharge is 30%.

[0069] Step 13: Use the input feature vector as the input of the preset machine learning model and the output feature vector as the output of the machine learning model to train the battery discharge prediction model.

[0070] In this embodiment, the preset machine learning models include: linear regression, random forest, and neural networks.

[0071] Step 2: Input the expected data corresponding to the input parameter type of the target power battery into the battery discharge prediction model to obtain the predicted temperature curve of the target power battery.

[0072] In this embodiment, the target power battery is the power battery for which thermal management decisions are currently required. The expected data corresponding to the input parameter type includes: the aging state of the target power battery, battery specifications, discharge rate, ambient temperature, and expected depth of discharge from 0% to 100%. When the expected data is input into the battery discharge prediction model, it is vectorized according to the feature vectorization rules of the input parameters in the machine learning samples before being input into the battery discharge prediction model. The predicted temperature curve of the target power battery is a curve fitted from coordinate points in a two-dimensional Cartesian coordinate system where the horizontal axis represents the expected depth of discharge and the vertical axis represents the predicted temperature.

[0073] Step 3: Determine the future thermal management window based on the position of the predicted temperature in the ideal operating temperature range in the predicted temperature curve.

[0074] In this embodiment, the ideal operating temperature range is set according to the battery type. For example, for lithium-ion power batteries, the ideal operating temperature range is 15℃-35℃. The range position refers to the relative position of the predicted temperature within the ideal operating temperature range. The future thermal management window is the expected depth of discharge range requiring thermal management, for example, when the depth of discharge is between 0-10% and 80%-100%. Step 3 specifically includes:

[0075] Step 31: Analyze the interval position.

[0076] In this embodiment, the analytical interval position is used to determine the upper limit temperature and the lower limit temperature of the ideal operating temperature range.

[0077] Step 32: If the predicted temperature is higher than the upper limit of the ideal operating temperature range, collect the corresponding depth of discharge for the predicted temperature and use it as the first window.

[0078] In this embodiment, the first window represents the expected depth of discharge range where the temperature is too high.

[0079] Step 33: If the predicted temperature is lower than the lower limit of the ideal operating temperature range, collect the corresponding depth of discharge for the predicted temperature and use it as the second window.

[0080] In this embodiment, the second window represents the expected depth of discharge range where the temperature is too low.

[0081] Step 34: Use both the first and second windows as the future thermal management windows.

[0082] Step 4: Determine and apply the thermal management strategy based on the heat generation power distribution within the future thermal management window.

[0083] In this embodiment, the heat generation power distribution is: the total heat generation power corresponding to different expected discharge depths within the future thermal management window, and the respective proportions of irreversible and reversible heat generation power in the total heat generation power. The thermal management strategy is: the timing and intensity of starting and stopping the thermal management system (active cooling and heating). Specifically, the method for obtaining the heat generation power distribution within the future thermal management window in step 4 includes:

[0084] Step 4a: Determine the entropy change coefficient of the target power battery based on the experimental measurement records of the target power battery.

[0085] In this embodiment, the experimental measurement record is as follows: the target battery is placed in a constant temperature chamber and left to stand at different temperatures. The open-circuit voltage of the target battery at different SOC states is measured, and the process record of fitting the curve of the battery open-circuit voltage changing with temperature is obtained. For example, the target battery is placed in a constant temperature chamber and left to stand for 3 hours at temperatures of 25℃, 30℃, 35℃, and 40℃ respectively. The open-circuit voltage values ​​of the target battery at an SOC of 40% are measured to be 3.65175V, 3.6528V, 3.6545V, and 3.6553V respectively. The curve of the battery open-circuit voltage changing with temperature is fitted, and the slope of this curve is the entropy change coefficient when the predicted depth of discharge is 60%.

[0086] Step 4b: Determine the predicted internal resistance corresponding to the expected discharge depth within the future thermal management window.

[0087] In this embodiment, the predicted internal resistance is output along with the predicted temperature in step 2.

[0088] Step 4c: Determine the irreversible heat generation power corresponding to the expected depth of discharge based on the discharge rate of the target battery, the battery specifications of the target power battery, and the predicted internal resistance.

[0089] In this embodiment, the battery operating current is determined based on the discharge rate of the target battery. Determine the battery quality based on the battery specifications of the target power battery. According to the battery operating current Battery quality and predicting internal resistance Calculate the irreversible heat production power as follows: .

[0090] Step 4d: Determine the reversible heat generation power based on the target battery's discharge rate, the predicted temperature corresponding to the expected depth of discharge, and the entropy change coefficient.

[0091] In this embodiment, based on the battery operating current Predicted temperature corresponding to expected discharge depth and entropy change coefficient Calculate the reversible heat production power as follows: .

[0092] Step 4e: Associate the irreversible heat generation power and the reversible heat generation power with the corresponding expected discharge depth. Once the expected discharge depth within the future thermal management window has been associated, the heat generation power distribution is obtained.

[0093] The working principle and beneficial effects of the above technical solution are as follows:

[0094] This invention utilizes complete discharge records of power batteries under different aging states and operating conditions to train a battery discharge prediction model. The model learns the complex nonlinear relationship between input and output parameters. When used for subsequent discharge prediction, it can better combine multiple factors such as battery aging state to predict battery temperature, battery voltage, and DC internal resistance, improving the prediction accuracy of future battery output parameters. Based on the predicted temperature curve and the introduced ideal operating temperature range, the future thermal management window is determined. The timing and intensity of different thermal management methods are dynamically determined based on the predicted thermal load characteristics of discharge stages that do not meet thermal management expectations, effectively reducing the risk of thermal runaway in aging batteries and extending battery life.

[0095] In one embodiment, step 4a: determining the entropy change coefficient of the target power battery based on experimental measurement records, including:

[0096] The battery aging model is determined based on the aging state of the tested battery.

[0097] In this embodiment, the battery aging model is as follows:

[0098] ;

[0099] Wherein, SOH represents the aging state of the test battery, and the test battery is a battery of the same specification as the target power battery. The entropy change coefficient of the new battery. This is the attenuation limit value. This is the degradation coefficient. For example, the entropy change coefficient of a new LFP battery is approximately -0.0015 V / K, the degradation limit of an LFP battery is approximately -0.0005 V / K, and the degradation coefficient of an LFP battery, obtained by fitting historical data, is 0.8, indicating an aging state of 75%.

[0100] Based on the battery aging model, the location of the nonlinear inflection point is predicted.

[0101] In this embodiment, when predicting the location of the nonlinear inflection point, the zero point is found by calculating the second derivative of the battery aging model using the numerical differentiation method. Continuing with the above example, the predicted location of the nonlinear inflection point is 25 degrees Celsius.

[0102] The set temperature points of the constant temperature chamber are dynamically set according to the preset number of temperature points and the position of the nonlinear inflection point based on the aging state of the test battery.

[0103] In this embodiment, the number of preset temperature points for the aging state is manually set. For example, the number of temperature points corresponding to SOH≥90% is 3, the number of temperature points corresponding to 80%≤SOH<90% is 4, and the number of temperature points corresponding to SOH<80% is 5. When dynamically setting the temperature points of the constant temperature chamber, in addition to meeting the requirement of the number of temperature points, the set temperature points need to include the temperature at the nonlinear inflection point, the temperature above the nonlinear inflection point, and the temperature below the nonlinear inflection point. For example, the set temperature points for SOH≥90% are 25 degrees Celsius, 35 degrees Celsius, and 45 degrees Celsius; the set temperature points for SOH<80% are 5 degrees Celsius, 15 degrees Celsius, 25 degrees Celsius, 35 degrees Celsius, and 45 degrees Celsius.

[0104] The experimental measurement records of the test battery are obtained based on the set temperature points.

[0105] Obtain sub-measurement data under different known aging states from the experimental measurement records.

[0106] In this embodiment, the known aging state refers to the known aging state of the test battery. The sub-measurement data includes the open-circuit voltage value under different SOC states in the known aging state of the corresponding test battery as a function of temperature.

[0107] By fitting the open-circuit voltage value versus temperature curve in the same SOC state from the sub-measurement data, the entropy change coefficient curve of the known aging state can be obtained.

[0108] In this embodiment, the entropy change coefficient of the corresponding SOC state under the known aging state can be obtained by fitting the curve of the open circuit voltage value under the same SOC state as the temperature change in the sub-measurement data. Based on the SOC state, the DoD is inferred. With DoD as the horizontal axis and the entropy change coefficient as the vertical axis, the curve of the entropy change coefficient changing with DoD under the known aging state is plotted (entropy change coefficient change curve).

[0109] By comparing the entropy change coefficient curves of different known aging states, the target entropy change coefficient curve of the target power battery can be deduced.

[0110] In this embodiment, during the deduction, the entropy coefficient change curves of different known aging states are compared. If the deviation of the curve is less than the preset deviation threshold, the entropy coefficient change curve of the known aging state is fitted to obtain the target entropy coefficient change curve.

[0111] The entropy change coefficient is determined based on the target entropy change coefficient curve and the future thermal management window.

[0112] In this embodiment, when determining the entropy change coefficient based on the target entropy change coefficient change curve and the future thermal management window, the entropy change coefficient of the target power battery can be read by comparing the expected discharge depth corresponding to the future thermal management window with the target entropy change coefficient change curve.

[0113] The working principle and beneficial effects of the above technical solution are as follows:

[0114] This invention acquires experimental measurement records, determines the number of temperature points to be measured based on different aging states of the test battery, and introduces a battery aging model to predict the location of nonlinear inflection points. Based on the number of temperature points and the location of nonlinear inflection points, the set temperature points of the constant temperature chamber are dynamically set to achieve adaptive sampling, avoid sampling blind spots, and improve the accuracy of obtaining the entropy change coefficient curve. By comparing the entropy change coefficient curves of different known aging states, the target entropy change coefficient curve of the target power battery is deduced, overcoming the problems of limited experimental measurement data and difficulty in determining the entropy change coefficient, improving the rationality of the target entropy change coefficient curve deduction, and increasing the accuracy of entropy change coefficient determination.

[0115] In one embodiment, step 4: determining and applying a thermal management strategy based on the heat generation power distribution within the future thermal management window, including:

[0116] Step 41: Determine the risk type based on the window type of the continuous windows in the future thermal management window; the risk types include: low temperature risk and high temperature risk.

[0117] In this embodiment, the window type includes a first window and a second window. When determining the risk type based on the window type, if the window type is the first window, the risk type is high temperature risk; if the window type is the second window, the risk type is low temperature risk.

[0118] Step 42: Determine the thermal management model based on the type of risk.

[0119] In this embodiment, the thermal management mode includes active cooling (e.g., liquid cooling) and active heating (e.g., PTC heating). When the risk type is high temperature risk, the determined thermal management mode is active cooling; when the risk type is low temperature risk, the determined thermal management mode is active heating.

[0120] Step 43: Based on the heat generation power distribution within the continuous window, determine the total heat generation power curve and the dominant heat type law; the dominant heat type law is the trend of the proportion of the dominant heat type in the total heat generation power of the discharge.

[0121] In this embodiment, a continuous window refers to a continuous expected discharge depth. When the depth difference between the collected expected discharge depths is less than a certain percentage, the corresponding expected discharge depths constitute a continuous window. The total discharge heat generation power curve is obtained by analyzing the heat generation power distribution within the continuous window. The horizontal axis of the coordinate system containing the total discharge heat generation power curve represents the expected discharge depth, the vertical axis represents the total discharge heat generation power, and the range of the horizontal axis is the range of expected discharge depths within the continuous window. The dominant heat generation types include irreversible heat generation power and reversible heat generation power. The proportion trends include: gradually decreasing proportion, gradually increasing proportion, decreasing first and then increasing proportion, and increasing first and then decreasing proportion.

[0122] Step 44: Determine the mode intensity curve based on the total power curve of the discharge heat generation.

[0123] In this embodiment, the horizontal axis of the coordinate system containing the mode intensity curve is the expected discharge depth, and the vertical axis is the mode intensity. The mode intensity value is the intensity quantification value of the thermal management mode that overcomes the total power value of the discharge heat generated at the corresponding expected discharge depth, such as the cooling power of the liquid cooling system.

[0124] Step 45: Correct the mode intensity curve according to the dominant law of thermal type to obtain the target mode intensity curve.

[0125] In this embodiment, the mode intensity curve is modified according to the dominant thermal type law, taking into account the Joule thermal runaway and entropy thermal response hysteresis factors. The intensity is increased when the irreversible thermal dominance trend strengthens, and the intensity is reduced or the mode is initiated earlier when the reversible thermal dominance trend strengthens, thus obtaining a more accurate mode intensity curve (target mode intensity curve). Step 45 specifically includes:

[0126] Step 451: Based on the dominant law of heat type, extract the trend slope, entropy thermal response delay coefficient and trend switching point, and construct thermal behavior feature vector.

[0127] In this embodiment, the trend slope is the result of differentiating the proportion of the dominant thermal type in the total discharge heat generation power with respect to the expected discharge depth. The entropy thermal response delay coefficient is the percentage delay of the DoD of the entropy thermal peak compared to the DoD of the total heat generation power peak. The trend switching point is the inflection point where the Joule thermal dominant trend changes from non-dominant to dominant. The constructed thermal behavior feature vector is: [trend slope, entropy thermal response delay coefficient, trend switching point, entropy thermal activity], where the trend slope... Entropy thermal response delay coefficient Trend switching point Entropy thermal activity This represents the integral percentage of reversible heat generation power within a continuous window. .

[0128] Step 452: Construct a neural fuzzy logic network using fuzzy rules, input the thermal behavior feature vector into the neural fuzzy logic network, and obtain the correction factor.

[0129] In this embodiment, the fuzzy rules are obtained through data mining. The mining target is historical records of experts correcting initial curves based on heat production power distribution. The initial curves are constructed in the same way as the mode intensity curves. For example, if... For high and If it is large, then the correction factor is very high; if for low and If the value is high, then the correction factor is low; where high, low, large, and small are defined by Gaussian membership functions, for example, for the trend slope. The membership function parameters are: , The parameters of the neural fuzzy logic network are dynamically optimized after each discharge, including: after each discharge, based on the error feedback between the actual temperature and the predicted temperature during the most recent historical discharge, the fuzzy membership function parameters and rule weights are dynamically optimized.

[0130] Step 453: Correct the mode intensity curve according to the correction factor to obtain the target mode intensity curve.

[0131] In this embodiment, the correction factor and the corresponding mode intensity curve are multiplied together to obtain the target mode intensity curve, where the corresponding curve refers to the expected discharge depth.

[0132] Step 46: Determine the thermal management strategy based on the thermal management mode and target mode intensity curve associated with the continuous window.

[0133] In this embodiment, the thermal management strategy is determined based on the thermal management mode and target mode intensity curve associated with the continuous window: establishing a one-to-one correspondence between the thermal management mode, target mode intensity, and expected discharge depth, and executing the thermal management mode with the corresponding target mode intensity at the expected discharge depth as the thermal management strategy.

[0134] The working principle and beneficial effects of the above technical solution are as follows:

[0135] This invention, taking into account factors such as the susceptibility of Joule heating to runaway and the lag in entropy-thermal response, enhances the intensity of cooling when the irreversible thermal dominance trend strengthens, and reduces the intensity or initiates cooling earlier when the reversible thermal dominance trend strengthens, thereby obtaining a more accurate mode intensity curve (target mode intensity curve) and further improving the control precision of subsequent thermal management. Specifically, the dominant thermal type is characterized, and a thermal behavior feature vector is constructed based on the trend slope, entropy-thermal response delay coefficient, trend switching point, and entropy-thermal activity. A neural fuzzy logic network is constructed using fuzzy rules, and the thermal behavior feature vector is input into the neural fuzzy logic network to obtain correction factors. Cooling is initiated earlier by using the entropy-thermal response delay coefficient to avoid temperature overshoot; intervention before the thermal runaway critical point is achieved by using the trend switching point to avoid the risk of thermal runaway; heating intensity is automatically reduced when there is a low temperature risk and entropy-thermal activity to avoid overheating; and strong cooling throughout the process is avoided when there is a high temperature risk but Joule heating is slow, saving energy. After each correction, the neural fuzzy logic network performs independent feedback optimization, resulting in stronger adaptability.

[0136] This invention provides a power battery thermal management system based on aging condition adaptation, such as... Figure 2 As shown, it includes:

[0137] Training module 1 is used for machine learning of discharge records of different power batteries to train a battery discharge prediction model. The discharge records include input parameters and output parameters. The types of input parameters include: aging state, battery specifications, discharge rate, ambient temperature and depth of discharge. The types of output parameters include: battery temperature, battery voltage and DC internal resistance.

[0138] Prediction module 2 is used to input the expected data of the target power battery corresponding to the input parameter type into the battery discharge prediction model to obtain the predicted temperature curve of the target power battery.

[0139] The management window determination module 3 is used to determine the future thermal management window based on the position of the predicted temperature in the ideal operating temperature range in the predicted temperature curve.

[0140] The management strategy application module 4 is used to determine and apply the thermal management strategy based on the heat generation power distribution within the future thermal management window.

[0141] The management strategy application module 4 performs the following operations:

[0142] Based on the window type of the continuous windows in the future thermal management window, the risk types are determined; the risk types include: low temperature risk and high temperature risk.

[0143] Determine the thermal management model based on the type of risk;

[0144] Based on the heat generation power distribution within the continuous window, the total heat generation power curve and the dominant heat type pattern are determined; the dominant heat type pattern is the trend of the proportion of the dominant heat type in the total heat generation power.

[0145] The mode intensity curve is determined based on the total power curve of the discharge heat generation;

[0146] The mode intensity curve is modified according to the dominant law of thermal type to obtain the target mode intensity curve;

[0147] The thermal management strategy is determined based on the thermal management mode and target mode intensity curve associated with the continuous window.

[0148] Among them, the model intensity curve is corrected according to the dominant law of thermal type to obtain the target model intensity curve, including:

[0149] Based on the dominant laws of heat type, the trend slope, entropy thermal response delay coefficient, and trend switching point are extracted, and a thermal behavior feature vector is constructed.

[0150] A neural fuzzy logic network is constructed using fuzzy rules. The thermal behavior feature vector is input into the neural fuzzy logic network to obtain the correction factor.

[0151] The target mode intensity curve is obtained by correcting the mode intensity curve using the correction factor.

[0152] After each discharge, the fuzzy membership function parameters and rule weights are dynamically optimized based on the error feedback between the actual temperature and the predicted temperature during the most recent historical discharge.

[0153] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of the claims of this invention and their equivalents, this invention also intends to include these modifications and variations.

Claims

1. A power battery thermal management method based on aging state adaptation, characterized in that, include: Step 1: Machine learning is used to record the discharge data of different power batteries to train a battery discharge prediction model; The discharge record includes input parameters and output parameters. Input parameter types include: aging status, battery specifications, discharge rate, ambient temperature, and depth of discharge. Output parameter types include: battery temperature, battery voltage, and DC internal resistance. Step 2: Input the expected data corresponding to the input parameter type of the target power battery into the battery discharge prediction model to obtain the predicted temperature curve of the target power battery; Step 3: Based on the position of the predicted temperature within the ideal operating temperature range in the predicted temperature curve, determine the future thermal management window, including: Analyze the interval position; If the predicted temperature is higher than the upper limit of the ideal operating temperature range, the corresponding depth of discharge at the predicted temperature is collected and used as the first window. If the predicted temperature is lower than the lower limit of the ideal operating temperature range, the corresponding depth of discharge at the predicted temperature is collected and used as a second window. The first and second windows will be used together as the future thermal management window; Step 4: Determine and apply the thermal management strategy based on the heat generation power distribution within the future thermal management window; The methods for obtaining the heat generation power distribution within the future thermal management window include: Based on the experimental measurement records of the target power battery, determine the entropy change coefficient of the target power battery; Determine the predicted internal resistance corresponding to the expected discharge depth within the future thermal management window; Based on the discharge rate of the target battery, the battery specifications of the target power battery, and the predicted internal resistance, determine the irreversible heat generation power corresponding to the expected depth of discharge. The reversible heat generation power is determined based on the target battery's discharge rate, the predicted temperature corresponding to the expected depth of discharge, and the entropy change coefficient. The irreversible heat generation power and the reversible heat generation power are each associated with the corresponding expected discharge depth. Once the expected discharge depth within the future thermal management window is associated, the heat generation power distribution is obtained. Step 4: Based on the heat generation power distribution within the future thermal management window, determine and apply the thermal management strategy, including: Based on the window type of the continuous windows in the future thermal management window, the risk types are determined; the risk types include: low temperature risk and high temperature risk. Determine the thermal management model based on the type of risk; Based on the heat generation power distribution within the continuous window, the total heat generation power curve and the dominant heat type pattern are determined; the dominant heat type pattern is the trend of the proportion of the dominant heat type in the total heat generation power. The mode intensity curve is determined based on the total power curve of the discharge heat generation; The mode intensity curve is modified according to the dominant law of thermal type to obtain the target mode intensity curve; The thermal management strategy is determined based on the thermal management mode and target mode intensity curve associated with the continuous window.

2. The power battery thermal management method based on aging state adaptation as described in claim 1, characterized in that, Step 1: Machine learning of discharge records from different power batteries to train a battery discharge prediction model, including: The input parameters in the discharge record are characterized, and an input feature vector is constructed; The output parameters in the discharge record are characterized, and an output feature vector is constructed; The battery discharge prediction model is trained by using the input feature vector as the input of the preset machine learning model and the output feature vector as the output of the machine learning model.

3. The power battery thermal management method based on aging state adaptation as described in claim 1, characterized in that, In step 2, the expected discharge rate of the target power battery is determined based on its application conditions.

4. The power battery thermal management method based on aging state adaptation as described in claim 1, characterized in that, Based on the experimental measurement records of the target power battery, determine the entropy change coefficient of the target power battery, including: Determine the battery aging model based on the aging state of the tested battery; Based on the battery aging model, predict the location of the nonlinear inflection point; The set temperature points of the constant temperature chamber are dynamically set according to the preset number of temperature points and the position of the nonlinear inflection point based on the aging state of the test battery. The experimental measurement records of the test battery are obtained based on the set temperature points; Obtain sub-measurement data from experimental measurement records under different known aging states; By fitting the open-circuit voltage value versus temperature curve in the same SOC state of the sub-measurement data, the entropy change coefficient curve of the known aging state can be obtained. By comparing the entropy change coefficient change curves under different known aging states, the target entropy change coefficient change curve of the target power battery can be deduced. The entropy change coefficient is determined based on the target entropy change coefficient curve and the future thermal management window.

5. The power battery thermal management method based on aging state adaptation as described in claim 1, characterized in that, The model intensity curve is corrected based on the dominant thermal type to obtain the target model intensity curve, including: Based on the dominant laws of heat type, the trend slope, entropy thermal response delay coefficient, and trend switching point are extracted, and a thermal behavior feature vector is constructed. A neural fuzzy logic network is constructed using fuzzy rules. The thermal behavior feature vector is input into the neural fuzzy logic network to obtain the correction factor. The target mode intensity curve is obtained by correcting the mode intensity curve using the correction factor.

6. The power battery thermal management method based on aging state adaptation as described in claim 5, characterized in that, Also includes: After each discharge, the fuzzy membership function parameters and rule weights are dynamically optimized based on the error feedback between the actual temperature and the predicted temperature during the most recent historical discharge.

7. A power battery thermal management system based on aging condition self-adaptation, characterized in that, include: The training module is used to learn the discharge records of different power batteries and train the battery discharge prediction model. The discharge record includes input parameters and output parameters. Input parameter types include: aging status, battery specifications, discharge rate, ambient temperature, and depth of discharge. Output parameter types include: battery temperature, battery voltage, and DC internal resistance. The prediction module is used to input the expected data of the target power battery corresponding to the input parameter type into the battery discharge prediction model to obtain the predicted temperature curve of the target power battery. The management window determination module is used to determine the future thermal management window based on the position of the predicted temperature in the predicted temperature curve within the ideal operating temperature range. This includes: Analyze the interval position; If the predicted temperature is higher than the upper limit of the ideal operating temperature range, the corresponding depth of discharge at the predicted temperature is collected and used as the first window. If the predicted temperature is lower than the lower limit of the ideal operating temperature range, the corresponding depth of discharge at the predicted temperature is collected and used as a second window. The first and second windows will be used together as the future thermal management window; The management strategy application module is used to determine and apply the thermal management strategy based on the heat generation power distribution within the future thermal management window. The methods for obtaining the heat generation power distribution within the future thermal management window include: Based on the experimental measurement records of the target power battery, determine the entropy change coefficient of the target power battery; Determine the predicted internal resistance corresponding to the expected discharge depth within the future thermal management window; Based on the discharge rate of the target battery, the battery specifications of the target power battery, and the predicted internal resistance, determine the irreversible heat generation power corresponding to the expected depth of discharge. The reversible heat generation power is determined based on the target battery's discharge rate, the predicted temperature corresponding to the expected depth of discharge, and the entropy change coefficient. The irreversible heat generation power and the reversible heat generation power are each associated with the corresponding expected discharge depth. Once the expected discharge depth within the future thermal management window is associated, the heat generation power distribution is obtained. The management strategy application module performs the following operations: Based on the window type of the continuous windows in the future thermal management window, the risk types are determined; the risk types include: low temperature risk and high temperature risk. Determine the thermal management model based on the type of risk; Based on the heat generation power distribution within the continuous window, the total heat generation power curve and the dominant heat type pattern are determined; the dominant heat type pattern is the trend of the proportion of the dominant heat type in the total heat generation power. The mode intensity curve is determined based on the total power curve of the discharge heat generation; The mode intensity curve is modified according to the dominant law of thermal type to obtain the target mode intensity curve; The thermal management strategy is determined based on the thermal management mode and target mode intensity curve associated with the continuous window.

Images