A lithium battery thermal safety boundary determination method based on fuzzy comprehensive evaluation

Abstract

The present application belongs to the technical field of lithium battery thermal safety management, and relates to a lithium battery thermal safety boundary determination method based on fuzzy comprehensive evaluation, comprising: obtaining a factor set, determining the evaluation corresponding to the factor set and the weight vector corresponding to the factor set; constructing a fuzzy relation matrix using the evaluation corresponding to the factor set; synthesizing the membership function of the fuzzy relation matrix and the weight vector based on fuzzy synthesis operation to generate a fuzzy comprehensive evaluation system of the lithium battery safety boundary. The present application can reduce the thermal runaway prediction time error.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of lithium battery thermal safety management, in particular to a lithium battery thermal safety boundary determination method based on fuzzy comprehensive evaluation. BACKGROUND

[0002] The existing lithium battery safety evaluation mostly adopts a single parameter threshold method (such as temperature or voltage limit), which is difficult to describe the nonlinear coupling of temperature, SOC, pressure, heat dissipation, rate, aging and other factors; the traditional method relies on a large number of experimental data to establish an empirical model, which has problems such as long modeling cycle and poor adaptability; the existing technical solutions propose safety test requirements, but lack a dynamic safety boundary quantification method, and the existing evaluation system has insufficient description accuracy for the nonlinear thermal runaway evolution process. SUMMARY

[0003] In order to solve the problems existing in the prior art, the purpose of the present application is to provide a lithium battery thermal safety boundary determination method based on fuzzy comprehensive evaluation, which reduces the thermal runaway prediction time error compared with the traditional method.

[0004] To achieve the above purpose, the present application provides the following solutions:

[0005] A lithium battery thermal safety boundary determination method based on fuzzy comprehensive evaluation, comprising:

[0006] obtaining a factor set, determining the evaluation remarks corresponding to the factor set and the weight vector corresponding to the factor set;

[0007] constructing a fuzzy relation matrix using the evaluation remarks corresponding to the factor set;

[0008] synthesizing the membership function and the weight vector of the fuzzy relation matrix based on fuzzy synthesis operation to generate a fuzzy comprehensive evaluation system of the lithium battery safety boundary.

[0009] Optionally, the factor set includes battery type, SOC value, pressure value, temperature value, cooling condition, charge / discharge rate and cycle number.

[0010] Optionally, determining the evaluation remarks corresponding to the factor set includes dividing the factors in the factor set into safe, relatively safe, general, relatively dangerous and dangerous.

[0011] Optionally, determining the weight vector corresponding to the factor set includes:

[0012] scaling and assigning the factors in the factor set, establishing a judgment matrix, calculating the characteristic vector, square root method weight and consistency index using the elements in the judgment matrix, and performing multi-source weight fusion to obtain the subjective fusion weight:

[0013]

[0014] wherein, is the number of experts, is the normalized vector, is the vector normalization operator, is the subjective fusion weight, CF k is the comprehensive weight of the kth expert, determined by the consistency and experience weight, is the subjective fusion weight given by the kth expert, is the normalization operator;

[0015] The normalized index direction is used to calculate the information entropy, i.e., the objective entropy weight:

[0016]

[0017] wherein, is the information entropy of the jth index, is the number of samples, is the normalized proportion of the ith sample in the jth index, is the difference degree of the jth index, is the entropy weight of the jth index, is the sum of the difference degrees of all indexes;

[0018] The composite weight is calculated according to the subjective fusion weight and the objective entropy weight:

[0019]

[0020] wherein, is the fusion coefficient, is the time or iteration step, is the transition midpoint, is the control steepness.

[0021] Optionally, constructing the fuzzy relation matrix comprises:

[0022]

[0023] wherein, is the membership degree of the nth factor to the mth comment.

[0024] Optionally, the membership degree function comprises:

[0025] The thermal stability coefficients of the factors of different types of batteries are converted into membership degrees:

[0026] ;

[0027] wherein, is safe, is safer, is general, is a battery type membership function;

[0028] converts the SOC value in the factor set into membership:

[0029] ;

[0030] wherein, is a membership function of SOC, =0.15 calibrated by constant current charge-discharge experiment, is the central value of the jth comment on the SOC axis, is the standard deviation of the jth comment;

[0031] converts the pressure value in the factor set into membership:

[0032] ;

[0033] wherein, is a membership function of pressure value, is the pressure threshold value of mechanical strength test under the condition of shell rupture pressure≥200kPa.

[0034] Optionally, determining the membership function further comprises:

[0035] converts the temperature value in the factor set into membership:

[0036] ;

[0037] wherein, is a membership function of temperature value, is a slope factor, is a test temperature, is a fitted inflection temperature;

[0038] converts the cooling condition in the factor set into membership:

[0039] ;

[0040] wherein, is the membership of cooling condition to grade j, is safer, is general, is more dangerous;

[0041] converts the charge-discharge rate in the factor set into membership:

[0042] ;

[0043] wherein, is a membership function of the cycle number, is a C-rate threshold value;

[0044] convert the cycle number in the factor set into a membership degree:

[0045]

[0046] wherein, is a membership function of the cycle number, is a cycle number, is a safe cycle number.

[0047] Optionally, the fuzzy comprehensive evaluation system of the lithium battery safety boundary comprises:

[0048] a weighted average synthesis is performed on the membership function and the weight vector of the fuzzy relation matrix, and a time function and a scene gating vector are introduced to obtain a dynamic synthesis result;

[0049] a loss function is constructed by using a labeled historical set or an online weak supervision signal, and recursive least squares or small step gradient descent is adopted to further defuzzify the score vector to obtain a heat runaway probability that can be mapped;

[0050] the heat runaway probability that can be mapped is combined with a thermal safety boundary criterion to generate the fuzzy comprehensive evaluation system of the lithium battery safety boundary.

[0051] The present application has the following beneficial effects:

[0052] In view of the problem that the single parameter threshold method ignores the coupling of multiple parameters and lacks a dynamic safety boundary quantification method in the prior art lithium battery safety evaluation, the present application realizes a breakthrough improvement through the following technical means and produces a significant technical effect:

[0053] The present application establishes a multi-dimensional factor set U containing battery type (u1), SOC value (u2), pressure value (u3), temperature value (u4), cooling condition (u5), charge and discharge rate (u6), and cycle number (u7), quantifies the interaction effect between parameters through a fuzzy relation matrix R, and integrates a weight vector W by using fuzzy synthesis operation, thereby breaking through the limitation of a single parameter, reducing the heat runaway false alarm rate, improving the system robustness, and realizing quantitative calculation of the battery thermal safety boundary. BRIEF DESCRIPTION OF DRAWINGS

[0054] ​In order to make the technical solutions in the embodiments of the present application or the prior art clearer, the accompanying drawings needed in the embodiments will be briefly introduced below. Obviously, the accompanying drawings in the following description are only some embodiments of the present application, and for those skilled in the art, other drawings can be obtained from these drawings without creative labor.

[0055] Figure 1 A flow chart of a lithium battery thermal safety boundary determination method based on fuzzy comprehensive evaluation for an embodiment of the present application;

[0056] Figure 2 A schematic diagram of the relationship between battery evaluation parameters and membership degrees for an embodiment of the present application. DETAILED DESCRIPTION

[0057] The technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the accompanying drawings in the embodiments of the present application. Obviously, the described embodiments are only some embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor fall within the scope of the present application.

[0058] In order to make the above-mentioned purposes, features and advantages of the present application more obvious and easy to understand, the present application will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0059] As shown in Figure 1 The present embodiment discloses a lithium battery thermal safety boundary determination method based on fuzzy comprehensive evaluation, which comprises: acquiring a factor set, determining the evaluation remarks corresponding to the factor set and the weight vector corresponding to the factor set; constructing a fuzzy relation matrix by using the evaluation remarks corresponding to the factor set; synthesizing the membership degree function of the fuzzy relation matrix and the weight vector based on fuzzy synthesis operation to generate a fuzzy comprehensive evaluation system of the lithium battery safety boundary.

[0060] Specifically, the embodiment discloses a lithium battery thermal safety boundary determination method based on fuzzy comprehensive evaluation, which comprises the following steps: according to the characteristics and safety requirements of the lithium battery, selecting the battery type, SOC value, pressure, temperature, cooling condition and other parameters as key indexes. Based on the fuzzy mathematics theory, a fuzzy comprehensive evaluation model of the lithium battery safety boundary is established. The model constructs a fuzzy comprehensive evaluation matrix by determining the factor set (i.e. the evaluation index set) and the comment set (such as “safe”, “relatively safe”, “general”, “relatively dangerous” and “dangerous”). The elements in the matrix reflect the membership degree of each evaluation index to the lithium battery safety boundary, i.e. the possibility distribution of each index at different safety levels. The expert system method, the analytic hierarchy process and other methods are used to combine the knowledge and experience to allocate the weight of each index. Through fuzzy synthesis operation, the fuzzy membership degree vector of each evaluation index and the weight vector are synthesized to obtain the fuzzy comprehensive evaluation system of the lithium battery safety boundary.

[0061] A lithium battery safety boundary model based on fuzzy comprehensive evaluation is established: ;

[0062] Among them, the factor set is , represents the selected battery type, represents the SOC value, represents the pressure value, represents the temperature value, represents the cooling condition, represents the charge and discharge rate, represents the cycle number; represents the comment domain of qualitative evaluation , respectively represent safety (S), relatively safe (SS), general (N), relatively dangerous (SD) and dangerous (D); represents the weight vector; represents the membership function mapping; represents the fuzzy relation matrix.

[0063] More specifically, the parameter quantification method comprises:

[0064] The dimension / gaging standard of the battery type (u1) is shown in Table 1.

[0065] Table 1

[0066]

[0067] The dimension / gaging standard of the SOC value (u2) is shown in Table 2.

[0068] Table 2

[0069]

[0070] The parameter calibration method of the SOC value (u2) is:

[0071] DVA: , and constant current charge and discharge cycles are carried out according to GB / T 31486-2015;

[0072] wherein, is a voltage value, is a battery capacity, is a function relationship of voltage change with SOC.

[0073] The dimension / gaging standard of the pressure value (u3) is shown in Table 3.

[0074] Table 3

[0075]

[0076] The parameter calibration method of the pressure value (u3) is:

[0077] The pressure deformation constitutive equation is: , and the breaking pressure is greater than or equal to 200 kPa;

[0078] wherein, is a stress, is an elastic modulus, is a strain, is a viscosity coefficient, is a time.

[0079] The dimension / gaging standard of the temperature value (u4) is shown in Table 4.

[0080] Table 4

[0081]

[0082] The parameter calibration method of the temperature value (u4) is:

[0083] Three-dimensional thermal simulation and measured data fusion: ;

[0084] wherein, is a volume heat source intensity, is a core temperature, is an outer surface temperature, is a characteristic radius, is an equivalent thermal conductivity.

[0085] The dimension / gaging standard of the cooling condition (u5) is: {1-natural convection, 2-forced air cooling, 3-liquid cooling}.

[0086] The dimension / gaging standard of the charge and discharge rate (u6) is shown in Table 5.

[0087] Table 5

[0088]

[0089] The parameter calibration method of the charge and discharge rate (u6) is as follows:

[0090] Perform 1C 5C step charge and discharge:

[0091] wherein, is the instantaneous heat power of the battery cell during constant current charging / discharging, is the C-rate, is the constant term of zero deviation.

[0092] The dimension / classification standard of the cycle number (u7) is shown in Table 6.

[0093] Table 6

[0094]

[0095] Further, the weight vector corresponding to the factor set comprises: scaling the factors in the factor set, establishing a judgment matrix, calculating a characteristic vector, a square root method weight and a consistency index using elements in the judgment matrix, performing multi-source weight fusion to obtain a subjective fusion weight, normalizing an index direction, calculating information entropy, i.e., an objective entropy weight, using the normalized index direction, and calculating a composite weight according to the subjective fusion weight and the objective entropy weight.

[0096] Specifically, the judgment matrix is constructed as follows: wherein the factors are valued according to a 1-9 scaling method, representing the value j corresponding to the factor i:

[0097]

[0098] The characteristic vector is calculated as follows: wherein, is the maximum eigenvalue, which is calculated using a square root method:

[0099]

[0100] wherein, is the relative weight of the index , and is the element in the kth row and jth column of the judgment matrix.

[0101] The consistency index is calculated as follows: when the random consistency ratio , the matrix is determined to be valid; wherein, ​​​​For random consistency index.

[0102] Dynamic weight correction, establish expert knowledge base , using evidence theory in AHP based on multi-source weight fusion: AHP multi-expert fusion of subjective layer, set the first expert based on AHP subjective weight vector , its expert confidence is . Then the subjective weight fusion is:

[0103]

[0104] Where, is the number of experts, is the normalized vector, is the vector normalization operator.

[0105] Entropy weight of objective layer, data matrix represents the observation of the first sample in the first index. According to the index direction:

[0106] Positive index (the larger the value, the safer, such as cooling capacity level):

[0107]

[0108] Reverse index (the larger the value, the more dangerous, such as temperature, pressure, C-rate risk):

[0109]

[0110] Interval optimal index (such as SOC exists "most safe interval"): can be converted into "distance to optimal interval" and then normalized, or directly use membership function into .

[0111] After normalization, calculate .

[0112] First, according to the index attribute to do normalization to get :

[0113]

[0114] Information entropy:

[0115] Where, is the information entropy of the first index, is the number of samples, is the normalized proportion of sample s in the first index, is the difference degree of the first index, is the entropy weight of the ith index.

[0116] Final composite weight:

[0117]

[0118] wherein, is the fusion coefficient, is the time / iteration step, is the transition midpoint, control steepness, dimensionless In implementation, scene gating can be introduced Element-level scaling is performed on

[0119] Further, the membership function includes: converting the thermal stability coefficient of different types of batteries in the factor set into membership, converting the SOC value in the factor set into membership, converting the pressure value in the factor set into membership, converting the temperature value in the factor set into membership, converting the cooling condition in the factor set into membership, converting the charging and discharging rate in the factor set into membership, and converting the cycle number in the factor set into membership.

[0120] Specifically, the membership function is designed as:

[0121] The battery type (u1) adopts a classification correction coefficient, which affects the threshold values of other parameters:

[0122] ;

[0123] wherein, is the battery type correction coefficient, used to adjust the critical value of other parameters, and the NCM battery temperature threshold needs to be reduced by 10%, is a lithium iron phosphate battery, is a lithium titanate battery.

[0124] SOC value: fitting a Gaussian distribution based on the DVA test curve:

[0125] ;

[0126] ;

[0127] wherein, the bandwidth is determined by fitting the DVA test curve, and the test refers to the IEC 62660-2 standard.

[0128] Pressure value (u3): , the threshold refers to the shell mechanical strength test data.

[0129] Temperature value (u4): general (N): ​The S-shaped function describes the temperature transition zone characteristics, where the temperature function slope factor of 0.3 is obtained by fitting the ARC test curve.

[0130] Cooling condition (u5): safer (SS): ;

[0131] Charge-discharge rate (u6): safe (S): ;

[0132] Cycle number (u7): general (N): ;

[0133] Among them, the cycle number transition interval 200 times is obtained by capacity attenuation inflection point analysis.

[0134] Membership function calculation:

[0135] Battery type, different types of battery thermal stability coefficient (lithium iron phosphate 0.92, ternary NCM 0.75) is converted to membership: ;

[0136] SOC value: Gaussian distribution based on DVA test curve fitting: ; Wherein The mean value is calibrated by constant current charge and discharge experiment Taken from GB / T 31486-2015 test data;

[0137] Pressure value: ; Threshold according to: shell rupture pressure ≥200kPa mechanical strength test data;

[0138] Temperature value: Slope factor k=0.3 by ARC test curve fitting inflection point temperature T0=45℃ taken from three-dimensional thermal simulation data;

[0139] Cooling condition: discrete assignment method: ;

[0140] Charge-discharge rate: step membership: , Boundary condition: according to the safety C-rate threshold of battery type;

[0141] Cycle number: linear decay model: , Transition interval: 200 times transition zone determined by capacity attenuation inflection point analysis, is the cycle number, is the reference safety cycle number.

[0142] As shown in Figure 2 , the fuzzy evaluation matrix is composed of the membership of 7 evaluation factors to 5 comment levels:

[0143] wherein represents the membership degree of the ith factor to the jth comment.

[0144] Further, the fuzzy comprehensive evaluation system of the lithium battery safety boundary is generated by: weighting and averaging the membership degree functions of the fuzzy relation matrix and the weight vector, introducing a time function and a scene gating vector to obtain a dynamic synthesis result; constructing a loss function with a labeled history set or an online weakly supervised signal, and using recursive least squares or small step gradient descent, further defuzzifying with a score vector to obtain a heat runaway probability that can be mapped; and using the heat runaway probability that can be mapped in combination with a thermal safety boundary criterion to generate the fuzzy comprehensive evaluation system of the lithium battery safety boundary.

[0145] Specifically, the multi-level fuzzy synthesis operation includes:

[0146] Multi-source membership aggregation: construct a fuzzy evaluation matrix i , the ith row of which is the membership degree vector of the factor u to the comment set V={S, SS, N, SD, D}. Weighted average synthesis is performed by weighting the membership degree functions of the fuzzy relation matrix and the weight vector according to the subjective and objective fusion weights .

[0147] wherein, is the final fusion weight vector, is the weight component of the factor , is the membership degree of the factor to the comment .

[0148] Scene-adaptive dynamic correction mechanism: to depict the dynamic influence of the system / working condition / aging on the judgment, a time (or iteration step) function and a scene gating vector are introduced:

[0149]

[0150] wherein, is the AHP weight, is the entropy weight, is the Hadamard product, is to make the sum of the elements 1, is a scene variable (such as cooling type, ambient temperature, module position, etc.). At the same time, the membership degree function parameters of temperature / SOC / pressure, etc. are fine-tuned (for example, .

[0151] Accordingly, the dynamic synthesis is obtained as follows:

[0152] ;

[0153] wherein, is the membership center of the jth comment, is the scaling term, is the bandwidth, is the time-dependent fuzzy comment matrix.

[0154] wherein is not the final value, which is input to the evidence theory correction module as the basic probability assignment of : let the prior distribution be , then according to the Dempster combination:

[0155]

[0156] the grade vector after prior fusion is obtained . Then the defuzzification is performed with the score vector .

[0157] The real-time working condition is injected into the synthesis process to realize the linkage correction of "weight and membership".

[0158] Online learning parameters (closed loop update , etc.);

[0159] Loss function is constructed with labeled history set or online weak supervision signal (for example):

[0160]

[0161] wherein, is the safety boundary threshold, is the "true" safety degree obtained by post-mortem (or soft score of event label), is the discrete grade label, contains and other parameters, is the current moment safety index, is the grade membership vector after correction of evidence theory, is the supervision signal, is the weighting coefficient, is the parameter vector. Recursive least squares (RLS) or small step gradient descent is adopted:

[0162]

[0163] wherein, is the gradient of the parameter vector;

[0164] Defuzzification and hot safety boundary output (result definition);

[0165] Defuzzification and hot safety boundary output (result definition); De-fuzzification:

[0166]

[0167] where, is the transpose of the score vector.

[0168] The thermal safety margin criterion is given by:

[0169]

[0170] and can map the thermal runaway probability:

[0171]

[0172] where, is the thermal runaway probability estimate, is the logistic function, is the shift parameter.

[0173] is used for hierarchical alarm and energy / thermal management linkage control. Thus, the executable output of the fuzzy comprehensive evaluation system for generating the thermal safety margin of lithium batteries is obtained .

[0174] The above-described embodiments are merely intended to describe the preferred modes of the present application, and are not intended to limit the scope of the present application. Various modifications and improvements to the technical solutions of the present application made by those of ordinary skill in the art without departing from the design spirit of the present application shall fall within the scope of protection of the present application as defined by the claims.

Claims

1.A method for determining thermal safety boundary of lithium battery based on fuzzy comprehensive evaluation, characterized in that, The method comprises the following steps: acquiring a factor set, determining evaluation corresponding to the factor set and weight vector corresponding to the factor set; the factor set comprises battery type, SOC value, pressure value, temperature value, cooling condition, charge and discharge rate and cycle number; constructing a fuzzy relation matrix by using the evaluation corresponding to the factor set; based on fuzzy composition operation, synthesizing membership function and weight vector of the fuzzy relation matrix to generate a fuzzy comprehensive evaluation system of lithium battery safety boundary; the membership function comprises: converting thermal stability coefficient of different types of batteries in the factor set into membership degree: ； wherein, is safe, is safer, is general, is a battery type membership function; converting SOC value in the factor set into membership degree: ； wherein, is a membership function of SOC, = 0.15 calibrated by constant current charge-discharge experiment, is a center value of the jth comment on the SOC axis, is a standard deviation of the jth comment; converting pressure value in the factor set into membership degree: ； wherein a membership function of pressure values, is the pressure threshold for mechanical strength testing at a shell burst pressure ≥ 200 kPa. 2.The method for determining the thermal safety boundary of lithium battery based on fuzzy comprehensive evaluation according to claim 1, characterized in that, determining the evaluation corresponding to the factor set comprises dividing the factors in the factor set into safe, safer, general, more dangerous and dangerous. 3.The method for determining the thermal safety boundary of lithium battery based on fuzzy comprehensive evaluation according to claim 1, characterized in that, determining the weight vector corresponding to the factor set comprises: scaling the factors in the factor set, establishing a judgment matrix, calculating eigenvector, square root method weight and consistency index by using elements in the judgment matrix, and carrying out multi-source weight fusion to obtain subjective fusion weight: wherein, is the number of experts, is the normalized vector, is the vector normalization operator, is the subjective fusion weight, CF k is the comprehensive weight of the kth expert, determined by the consistency and experience weights, is the subjective fusion weight given by the kth expert, is the normalization operator; normalizing index direction, calculating information entropy, i.e. objective entropy weight, by using normalized index direction: wherein, is the information entropy of the jth index, is the number of samples, is the normalized proportion of the ith sample in the jth index, is the difference degree of the jth index, is the entropy weight of the jth index, is the sum of the difference degrees of all indexes; calculating compound weight according to the subjective fusion weight and the objective entropy weight: wherein, is a fusion coefficient, is a time or iteration step, is a transition midpoint, is a control steepness. 4.The method for determining the thermal safety boundary of lithium battery based on fuzzy comprehensive evaluation according to claim 1, wherein, constructing the fuzzy relation matrix comprises: wherein, is the membership of the nth factor to the mth comment. 5.The method for determining the thermal safety boundary of lithium battery based on fuzzy comprehensive evaluation according to claim 1, wherein, determining the membership function further comprises: converting temperature value in the factor set into membership degree: ； wherein, is a membership function of temperature value, is a slope factor, is a test temperature, is a fitted inflection temperature; converting cooling condition in the factor set into membership degree: ； wherein, is the degree of membership of the cooling condition to the class j, is safer, is average, is more dangerous; converting charge and discharge rate in the factor set into membership degree: ； wherein, is a membership function of the charge and discharge rate, is a C-rate threshold value; converting cycle number in the factor set into membership degree: ； wherein, is a membership function of the number of cycles, is the number of cycles, is the number of safe cycles. 6.The method for determining thermal safety boundary of lithium battery based on fuzzy comprehensive evaluation according to claim 1, characterized in that, generating the fuzzy comprehensive evaluation system of the lithium battery safety boundary comprises: performing weighted average synthesis on the membership function and the weight vector of the fuzzy relation matrix, introducing time function and scene gate vector to obtain dynamic synthesis result; constructing loss function by using labeled history set or online weak supervision signal, and further fuzzing by using score vector by using recursive least square or small step gradient descent to obtain a heat runaway probability which can be mapped; using the heat runaway probability which can be mapped in combination with thermal safety boundary criterion to generate the fuzzy comprehensive evaluation system of the lithium battery safety boundary.

Images