Method and device for predicting remaining life of lithium-ion battery considering switching state

Abstract

The present invention relates to a method and device for predicting the remaining service life of a lithium-ion battery taking into account a switching state provided by an embodiment of the present invention, and belongs to the technical field of lithium-ion battery health management. The method and device acquire degradation data and historical state data of the lithium-ion battery, and according to a pre-constructed lithium-ion battery degradation model and a lithium-ion battery operating state switching model, thereby considering the influence of state switching on lithium-ion battery degradation, and effectively solves the problem of difficulty in predicting the RUL of lithium-ion batteries under actual operating conditions; at the same time, compared with other methods, the pre-constructed lithium-ion battery degradation model and the lithium-ion battery operating state switching model are updated through Bayesian posteriori, and the method has the ability of real-time prediction; and the present invention pre-constructs a lithium-ion battery remaining probability density function model, which is more conducive to online calculation, and provides a basis and ideas for the embedded and cloud platform research and development of lithium-ion battery RUL prediction and health management system.

Description

Technical Field

[0001] The present invention relates to the technical field of lithium-ion battery health management, and in particular to a method and device for predicting the remaining life of a lithium-ion battery taking into account a switching state. Background Art

[0002] New energy vehicles have been widely recognized in the automotive industry for their improved performance and efficiency as well as their contribution to solving environmental problems such as greenhouse gas emissions and global warming. As the energy source of new energy vehicles, lithium-ion batteries have the characteristics of high energy density, no memory, and long life cycle, and are the core components of new energy vehicles. The life of lithium-ion batteries is directly related to the service life and safety of electric vehicles. Therefore, accurate remaining useful life (RUL) prediction can improve the battery management level, reduce battery operation risks, and extend the battery life cycle based on the battery management system (BMS), which is of great significance to improving the availability and reliability of electric vehicles. At the same time, RUL can be used as an indicator to further guide battery operation and maintenance, reasonably predict the evolution of battery status, and improve energy utilization efficiency, which is of great significance to reducing the economic cost and sustainable development of electric vehicles.

[0003] In the related art, the RUL of lithium-ion batteries is usually predicted by physical model-based, data-driven or fusion methods. Among them, the physical model only has a good characterization effect on the discharge and charging behaviors of lithium-ion batteries. The more external factors are considered in the long-term RUL prediction, the more components in the model, resulting in a decrease in prediction accuracy; the data-driven method, its prediction effect depends on the scale of historical data, requires a large amount of data support, and is difficult to use in BMS due to its high complexity; the fusion method has good prediction performance for RUL, and combining two or more physical model-based or data-driven methods can solve the single limitation of the above methods. However, the actual working state of the battery is constantly switching between the two modes of storage and work, and there is uncertainty in the duration, switching rules, and residence time in each state. The three methods of the related art are basically only for batteries that are continuously running or assume that the battery follows the same degradation mechanism in different states, ignoring the interaction between storage and working states. These factors lead to different degradation paths of lithium-ion batteries, making it difficult for BMS to accurately predict the remaining life of lithium-ion batteries on a calendar time scale. Summary of the invention

[0004] In view of this, the purpose of the present invention is to provide a method and device for predicting the remaining life of a lithium-ion battery taking into account the switching state, so as to overcome the problem that the current degradation paths of lithium-ion batteries are different, making it difficult for the BMS to accurately predict the remaining life of lithium-ion batteries on a calendar time scale.

[0005] To achieve the above objectives, the present invention adopts the following technical solutions:

[0006] On the one hand, a method for predicting the remaining life of a lithium-ion battery considering a switching state includes:

[0007] Obtaining degradation data and status monitoring information of the lithium-ion battery to be predicted; the degradation data includes historical degradation data at the current moment; the online status monitoring information includes historical status data at the current moment;

[0008] Performing hyperparameter estimation on a pre-constructed lithium-ion battery degradation model using maximum likelihood estimation according to the historical degradation data of the lithium-ion battery to be predicted;

[0009] Performing hyperparameter estimation on a pre-constructed lithium-ion battery operating state switching model according to the historical state data of the lithium-ion battery to be predicted;

[0010] Based on the Bayesian theory, updating the posterior parameters of the lithium-ion battery degradation model and the posterior parameters of the lithium-ion battery operation state switching model according to the degradation data and the state monitoring information;

[0011] Inputting the degradation data and the state monitoring information into a pre-constructed lithium-ion battery residual probability density function model to obtain a lithium-ion battery residual probability density function of the lithium-ion battery to be predicted;

[0012] The remaining life of the lithium-ion battery to be predicted is obtained according to the remaining probability density function of the lithium-ion battery.

[0013] Optionally, also include:

[0014] Acquiring a life prediction index of a lithium-ion battery, wherein the life prediction index includes a degradation characterization amount of the lithium-ion battery;

[0015] A lithium-ion battery degradation model of a nonlinear drift Wiener process is established according to the life prediction index; a lithium-ion battery operating state switching model is established according to the Markov chain and the lithium-ion battery operating state;

[0016] The lithium-ion battery degradation model and the lithium-ion battery operation state switching model are integrated by using a fuzzy system to determine the lithium-ion battery residual probability density function model.

[0017] Optionally, establishing a lithium-ion battery degradation model of a nonlinear drift Wiener process according to the life prediction index includes:

[0018] According to the degradation characteristics of lithium-ion batteries, the nonlinear drift Wiener process is expressed as:

[0019]

[0020] Among them, X(t) represents the degradation amount of the lithium-ion battery at time t, and X(0) represents the degradation amount at time 0, which is 0; is a random parameter that characterizes the individual differences of lithium-ion batteries. Follows normal distribution; is nonlinear drift degradation; σ is the diffusion parameter, σ>0; B(t) is the standard Brownian motion, which characterizes the random dynamic characteristics of the degraded device;

[0021] The method of performing hyperparameter estimation on a pre-constructed lithium-ion battery degradation model by using maximum likelihood estimation according to the historical degradation data of the lithium-ion battery to be predicted includes:

[0022] If there are M lithium batteries with relevant degradation history data, let the number of degradation data of the i-th lithium battery be N i (0≤i≤M), and the measurement time of the i-th battery is t i，j (1≤j≤N i ), where j represents the jth measurement, X(t i，j ) represents the degradation of the ith battery at the jth measurement, and we get:

[0023]

[0024] Where: X i，j represents X(t i，j ); X(0) is the degradation amount at time 0. The degradation amount of the lithium battery at time 0 is 0, that is, X(0) = 0; is a random parameter The prior, that is

[0025] The M degradation data obey the multivariate Gaussian distribution. Let the degradation data of the i-th lithium battery be Get the mean and covariance:

[0026]

[0027] Where:

[0028]

[0029] The log-likelihood function generated by the degradation data from all M degraded devices:

[0030]

[0031] Select a general and more applicable drift function, that is, ξ(τ; η) = ηt η-1, maximize the above formula, that is, find the first-order partial derivative to be 0, and the model hyperparameter You can get it.

[0032] Optionally, establishing a lithium-ion battery operating state switching model according to the Markov chain and the lithium-ion battery operating state includes:

[0033] According to the Markov chain, the lithium-ion battery state space S = {0, 1} is defined, where the 0 state represents that the lithium-ion battery is in the working state, and the 1 state represents that the lithium-ion battery is in the storage state;

[0034] The first working time of the lithium battery is recorded as T1, and T1 is assumed to be an exponential distribution with parameter λ and expectation 1 / λ, and each subsequent working time is independent and identically distributed with T1; and the first storage time of the lithium-ion battery is recorded as T2, and T2 is assumed to be an exponential distribution with parameter μ and expectation 1 / μ, and each subsequent storage time is independent and identically distributed with T2; the state of the lithium-ion battery at time t is recorded as X(t); the transition probability function is proved by the Kolmogorov forward equation:

[0035]

[0036] At any time, the probability distribution of the lithium battery working and storing in the future can be determined according to the values ​​of the parameters λ and μ. In order to simplify the symbolic representation, κ(λ, μ) is called the set of parameters λ and μ.

[0037] The step of estimating hyperparameters of a pre-built lithium-ion battery operating state switching model according to the historical state data of the lithium-ion battery to be predicted includes:

[0038] The historical data are directly fitted using the GAMFIT function of MATLAB to obtain the shape parameter and scale parameter estimates of the gamma distribution, thereby obtaining the value of κ(λ,μ).

[0039] Optionally, the step of fusing the lithium-ion battery degradation model and the lithium-ion battery operation state switching model using a fuzzy system to determine a lithium-ion battery residual probability density function model includes:

[0040] According to the concept of first arrival time, determine T = inf{t: X(t)≥ω|X(0)<ω}, where T is the first arrival time;

[0041] Let the service life of the lithium-ion battery be the time when it first reaches the retirement threshold ω; let the current time be t k , remaining life l k , according to the definition of equipment life: l k =tt k , get kThe remaining life Lk of the equipment at the moment is defined as:

[0042] L k = inf{l k >0:X(t k +l k )≥ω}

[0043] If there is no moment to cross the boundary before reaching the boundary precisely at a certain time t, then the remaining probability density function of the lithium-ion battery at the first arrival time of X(t), t≥0, which is the fixed threshold ω, is:

[0044]

[0045] In the above formula, f X(t) =(ω, t) is the residual probability density function of the lithium-ion battery at the time of first reaching the fixed threshold ω; S B (t) is the time-dependent boundary function corresponding to the standard Brownian motion,

[0046] The total probability formula is used to obtain the residual probability density function of the lithium-ion battery that first reaches the failure threshold:

[0047]

[0048]

[0049] Where w0 and w1 are the probability of the working and storage probabilities affecting the capacity degradation of lithium-ion batteries, p0 and p1 are the distribution probabilities of working and storing in the future. Assuming that the initial probability distribution of battery working and storing is π0(1, 0), it means that the probability of the system working at the beginning of operation is 100%, and the probability of the lithium-ion battery being in working and storing in the future can be obtained through the E state transition probability matrix (p0, p1);

[0050] The degradation of lithium-ion batteries is divided into three stages: initial stage, middle stage, and final stage. The degradation is accelerated in the initial stage, stable in the middle stage, and accelerated again after reaching the final stage. The fuzzy control rules are set as follows based on the actual situation of lithium-ion batteries:

[0051] If the remaining capacity is x1, that is, ω- xk In the early stage, the effect of working probability x2 decreases and the effect of early recovery probability x3 is larger;

[0052] If the remaining capacity x1 is in the medium term, the effect x2 of the working probability is natural weight and the effect x3 of the recovery probability is moderate;

[0053] If the remaining capacity x1 is at the end, the effect x2 of the working probability will be larger and the effect x3 of the recovery probability will be slightly larger.

[0054] Optionally, the updating of the posterior parameters of the lithium-ion battery degradation model and the posterior parameters of the lithium-ion battery operation state switching model based on the Bayesian theory and the degradation data and the state monitoring information includes:

[0055] In the Bayesian framework, the random parameters of the degradation model are The posterior distribution of will follow Bayesian theory, given by The definition distribution of and the Brownian motion properties give the degenerate data likelihood function:

[0056]

[0057] Get Updated Mean and variance of the parameters:

[0058]

[0059]

[0060] Assume that the prior distribution of p(κ) is a gamma distribution, with parameters λ~gama(α0, β0), μ~gama(α1, β1), where α0, α1 are shape parameters, and β0, β1 are scale parameters; the posterior distribution lithium-ion battery residual probability density function is:

[0061]

[0062] where n 0→1 (t k ) is the lithium-ion battery in [0, t k ] The number of times the working state is converted to the storage state within a certain period of time, n 1→0 (t k ) is the lithium battery in [0, t k ] is the number of times the storage state is converted to the working state within a certain period of time, that is, n(t k ) satisfies n(t k )=n 0-1 (tk)+n 1→0 (t k );Let d 0→1 (t k ) and d 1→0 (t k ) is the lithium-ion battery in the interval [0, t k ] is the cumulative stay time in the working and storage states, then t k Satisfy t k =d 0→1 (t k )+d 1→0 (t k). On the other hand, a lithium-ion battery remaining life prediction device considering a switching state includes a processor and a memory, wherein the processor is connected to the memory:

[0063] Wherein, the processor is used to call and execute the program stored in the memory;

[0064] The memory is used to store the program, and the program is used to execute at least any of the above-mentioned methods for predicting the remaining life of a lithium-ion battery taking into account the switching state.

[0065] The technical solution provided by the present invention includes at least the following beneficial effects:

[0066] The method and device for predicting the remaining life of a lithium-ion battery taking into account the switching state provided in the embodiment of the present invention obtains the degradation data and historical state data of the lithium-ion battery, and considers the influence of state switching on the degradation of the lithium-ion battery according to the pre-constructed lithium-ion battery degradation model and the lithium-ion battery operation state switching model, thereby effectively solving the problem of the difficulty in predicting the RUL of lithium-ion batteries under actual operation states, so as to overcome the problem that the current degradation paths of lithium-ion batteries are different, making it difficult for BMS to accurately predict the remaining life of lithium-ion batteries on a calendar time scale; at the same time, compared with other methods, the pre-constructed lithium-ion battery degradation model and the lithium-ion battery operation state switching model are updated through Bayesian posteriori, and the method has the ability of real-time prediction; and the present invention pre-constructs a lithium-ion battery residual probability density function model, which is more conducive to online calculation, and provides a basis and ideas for the embedded and cloud platform research and development of lithium-ion battery RUL prediction and health management system. BRIEF DESCRIPTION OF THE DRAWINGS

[0067] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0068] Figure 1 A schematic flow chart of a method for predicting the remaining life of a lithium-ion battery taking into account a switching state provided by an embodiment of the present invention;

[0069] Figure 2 A schematic flow chart of a method for predicting the remaining life of a lithium-ion battery taking into account a switching state provided in a specific embodiment of the present invention;

[0070] Figure 3 A lithium-ion battery degradation path diagram provided by an embodiment of the present invention;

[0071] Figure 4A schematic diagram of PDF and RUL prediction results of a lithium-ion battery provided in an embodiment of the present invention; wherein (a) is a schematic diagram of PDF prediction results of a lithium-ion battery, and (b) is a schematic diagram of RUL prediction results of a lithium-ion battery;

[0072] Figure 5 A schematic diagram of the structure of a lithium-ion battery remaining life prediction device taking switching states into consideration provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0073] To make the purpose, technical solution and advantages of the present invention clearer, the technical solution of the present invention will be described in detail below. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other implementation methods obtained by ordinary technicians in this field without creative work belong to the scope of protection of the present invention.

[0074] Figure 1 A schematic diagram of a flow chart of a method for predicting the remaining life of a lithium-ion battery considering a switching state provided by an embodiment of the present invention is provided in Figure 1 , this embodiment may include the following steps:

[0075] Step S11, obtaining degradation data and status monitoring information of the lithium-ion battery to be predicted; the degradation data includes historical degradation data at the current moment; and the online status monitoring information includes historical status data at the current moment;

[0076] Step S12, performing hyperparameter estimation on a pre-constructed lithium-ion battery degradation model using maximum likelihood estimation according to historical degradation data of the lithium-ion battery to be predicted;

[0077] Step S13, performing hyperparameter estimation on a pre-constructed lithium-ion battery operating state switching model according to historical state data of the lithium-ion battery to be predicted;

[0078] Step S14: Based on the Bayesian theory, the posterior parameters of the lithium-ion battery degradation model and the posterior parameters of the lithium-ion battery operation state switching model are updated according to the degradation data and the state monitoring information;

[0079] Step S15, inputting the degradation data and the state monitoring information into a pre-constructed lithium-ion battery residual probability density function model to obtain a lithium-ion battery residual probability density function of the lithium-ion battery to be predicted;

[0080] Step S16: Obtain the remaining life of the lithium-ion battery to be predicted according to the remaining probability density function of the lithium-ion battery.

[0081] It can be understood that the technical solution provided by the embodiment of the present invention can effectively solve the problem of difficult RUL prediction of lithium-ion batteries under actual operating conditions by acquiring historical degradation data and historical state data of lithium-ion batteries, thereby considering the impact of state switching on lithium-ion battery degradation; at the same time, compared with other methods, the pre-constructed lithium-ion battery degradation model and lithium-ion battery operating state switching model are updated through Bayesian posteriori, which has the ability of real-time prediction; and the present invention pre-constructs a lithium-ion battery residual probability density function model, which is more conducive to online calculation, and provides a basis and ideas for the embedded and cloud platform development of lithium-ion battery RUL prediction and health management system.

[0082] Figure 2 A flowchart of a method for predicting the remaining life of a lithium-ion battery taking into account the switching state is provided in accordance with a specific embodiment of the present invention. Figure 1-Figure 2 , the remaining life prediction method of a lithium-ion battery taking into account the switching state of the present application can be described.

[0083] In some embodiments, specifically, before predicting the remaining life of a lithium-ion battery taking into account the switching state, a model construction step is also included:

[0084] Step S21, obtaining a life prediction index of a lithium-ion battery, wherein the life prediction index includes a degradation characterization value of the lithium-ion battery;

[0085] Step S22, establishing a lithium-ion battery degradation model of a nonlinear drift Wiener process according to the life prediction index;

[0086] Step S23, establishing a lithium-ion battery operating state switching model according to the Markov chain and the lithium-ion battery operating state;

[0087] Step S24: using a fuzzy system to fuse the lithium-ion battery degradation model and the lithium-ion battery operation state switching model to determine a lithium-ion battery residual probability density function model.

[0088] The degradation characterization quantity of the lithium-ion battery may include: capacity, voltage, internal resistance or temperature, etc., or other quantities. In the embodiment of the present invention, taking capacitance as an example, the remaining life prediction method of the lithium-ion battery considering the switching state and the process of constructing the model are described:

[0089] Step S22 and step S12 may include the following steps:

[0090] A11: According to the degradation characteristics of lithium-ion batteries, the nonlinear drift Wiener process is expressed as follows:

[0091]

[0092] Among them, X(t) represents the degradation amount of the lithium-ion battery at time t, and X(0) represents the degradation amount at time 0, which is generally considered to be 0; is a random parameter that characterizes the individual differences of lithium-ion batteries. Follows normal distribution; is nonlinear drift degradation; σ is the diffusion parameter, σ>0; B(t) is the standard Brownian motion, which characterizes the random dynamic characteristics of the degraded device.

[0093] A12: Assume that there are M lithium batteries with relevant degradation history data. Let the number of degradation data of the i-th lithium battery be Ni (0≤i≤M), and the measurement time of the i-th battery be t i，j (1≤j≤N i ), where j represents the jth measurement, so X(t i，j ) represents the degradation of the ith battery at the jth measurement, and we get:

[0094]

[0095] Where: X i，j represents X(t i，j ); X(0) is the degradation amount at time 0. Without loss of generality, it is assumed that the degradation amount of the lithium battery at time 0 is 0, that is, X(0) = 0; is a random parameter The prior, that is

[0096] According to the first formula and the independent increment property of the standard Brownian motion process, we know that the M degradation data obey the multivariate Gaussian distribution. Let the degradation data of the i-th lithium battery be From this we can get the mean and covariance:

[0097]

[0098] Where:

[0099]

[0100] A13: So there is a log-likelihood function generated by the degradation data from all M degraded devices:

[0101]

[0102] Select a general and more applicable drift function, that is, ξ(τ; η) = ηt η-1 , maximize the above formula, that is, find the first-order partial derivative to be 0, and the model hyperparameter You can get it.

[0103] In some embodiments, step S23 may include the following steps:

[0104] A21: Lithium-ion battery equipment has a working state and a storage state during operation. Therefore, according to the definition of Markov chain, the lithium-ion battery state space S={0, 1}, where state 0 represents that the lithium-ion battery is in a working state, and state 1 represents that the lithium-ion battery is in a storage state at this time.

[0105] A22: Without loss of generality, let the first working time of the lithium battery be T1, and let T1 be an exponential distribution with parameter λ and expectation 1 / λ, and each subsequent working time is independent and identically distributed with T1; similarly, let the first storage time of the lithium-ion battery be T2, and let T2 be an exponential distribution with parameter μ and expectation 1 / μ, and each subsequent storage time is independent and identically distributed with T2; let the state of the lithium-ion battery at time t be X(t). It is easy to prove the transition probability function from the Kolmogorov forward equation:

[0106]

[0107] Therefore, at any time, the probability distribution of the lithium battery's operation and storage in the future can be determined according to the values ​​of the parameters λ and μ. In order to simplify the symbolic representation, κ(λ, μ) is called the set of parameters λ and μ.

[0108] A23: At the same time, in order to obtain the parameter value of k(λ, μ), the GAMFIT function of MATLAB is used to directly fit the historical data (exponential distribution is a special gamma distribution), and the shape parameter and scale parameter estimates of the gamma distribution can be obtained, thereby obtaining the value of κ(λ, μ).

[0109] In some embodiments, step S24 includes the following specific steps:

[0110] The residual probability density function of lithium-ion batteries can be expressed as Probability Density Function, abbreviated as PDF.

[0111] A31: Based on the concept of first arrival time, we have the following formula:

[0112] T=inf{t:X(t)≥ω|X(0)<ω}

[0113] A32: Let the service life of the lithium-ion battery be the time when it first reaches the retirement threshold ω. In order to solve its RUL at different times, let the current time be t k , remaining life l k , according to the definition of equipment life: l k =tt k , get k The remaining life of the equipment at the time L k definition:

[0114] L k = inf{l k >0:X(t k +l k )≥ω}

[0115] A33: Assuming that there is no crossing of the boundary before reaching the boundary precisely at a certain time t, then the PDF of the first arrival time of X(t), t≥0 to reach the fixed threshold ω is:

[0116]

[0117] In the above formula, f X(t) =(ω, t) is the residual probability density function of the lithium-ion battery at the time of first reaching the fixed threshold ω; S B (t) is the time-dependent boundary function corresponding to the standard Brownian motion,

[0118]

[0119] A34: Using the total probability formula for the above equation, we can get the PDF of the first failure threshold of lithium-ion batteries:

[0120]

[0121]

[0122] Where w0 and w1 are the probabilities of the influence of working and storage probabilities on the capacity degradation of lithium-ion batteries, p0 and p1 are the distribution probabilities of working and storage in the future. Assuming that the initial probability distribution of battery working and storage is π0(1, 0), it means that the probability of the system working at the beginning of operation is 100%, and the probability of the lithium-ion battery being in working and storage in the future can be obtained through the E-state transition probability matrix (p0, p1).

[0123] A35: Lithium-ion battery degradation is roughly divided into three stages: accelerated degradation in the early stage, stable in the middle stage, and accelerated degradation again after reaching the end stage. Therefore, a single-input dual-output fuzzy controller is designed. The capacity is divided into three categories: initial, middle, and final stages and used as the only input of the fuzzy system. The above-calculated future working probability and storage probability affecting lithium-ion degradation are used as outputs. Under the working state, there are two domains: natural degradation and degradation affected by capacity recovery. Under the storage state, there are three degradation impact domains: small, medium, and large, with capacity recovery phenomena existing at different times.

[0124] A36: The fuzzy control rules are set as follows based on the actual situation of lithium-ion batteries:

[0125] ① If the remaining capacity is x1, that is, ω-x in the above formulak In the initial stage, the effect of the working probability x2 will be slightly reduced (to prevent the degradation rate from being too fast, resulting in a small analytical remaining capacity and a small predicted RUL) and the effect of the early recovery probability x3 will be relatively large.

[0126] ② If the remaining capacity x1 is in the medium term, then the impact of the working probability x2 will be a natural weight (because the medium term is relatively stable and there is no phenomenon of capacity degradation too fast) and the recovery probability impact x3 is moderate.

[0127] ③ If the remaining capacity x1 is in the later stage, the influence of the working probability x2 will be too large (if the influence is reduced as in the early stage, the RUL predicted by the analytical formula will be too large, which will reduce the safety performance in the final stage. Therefore, the probability should be chosen to be too large to ensure the reliability in the final stage) and the recovery probability influence x3 will be slightly too large.

[0128] Table 1 provides the fuzzy membership of lithium-ion batteries, that is, the specific representation form mentioned above, and its coordinate value depends on the specific performance of the battery.

[0129] Table 1. Fuzzy membership of lithium-ion batteries

[0130]

[0131] In the table, x1 is the input of the system, which is the remaining capacity of the lithium-ion battery at the current moment, and μ1 is the membership function corresponding to the remaining capacity, which is used to identify the period of the lithium-ion battery. x2 and x3 are the degradation probabilities of lithium-ion batteries affected by work and storage, with a maximum value of 1. μ2 and μ3 are the corresponding membership functions, that is, input the remaining capacity, and output different degradation impact rates in different periods. The a coordinate is specifically defined according to the battery model and historical data.

[0132] In some embodiments, step S14 may include the following specific steps:

[0133] A41: Random parameters of degradation models in a Bayesian framework The posterior distribution of will follow Bayesian theory, given by The definition distribution of and the properties of Brownian motion give the degenerate data likelihood function:

[0134]

[0135] A42: Update available Mean and variance of the parameters:

[0136]

[0137]

[0138] A43: At the same time, in order to be able to update the parameters of the switching model under the Bayesian system, it is assumed that the prior distribution of p(κ) is a gamma distribution, with parameters λ~gama(α0, β0), μ~gama(α1, β1), where α0, α1 are shape parameters, β0, β1 are scale parameters; the posterior distribution PDF is thus:

[0139]

[0140] where n 0→1 (t k ) is the lithium-ion battery in [0, t k ] The number of times the working state is converted to the storage state within a certain period of time, n 1→0 (t k ) is the lithium battery in [0, t k ] is the number of times the storage state is converted to the working state within a certain period of time, that is, n(t k ) satisfies n(t k )=n 0→1 (t k )+n 1→0 (t k );Let d 0→1 (t k ) and d 1→0 (t k ) is the lithium-ion battery in the interval [0, t k ] is the cumulative stay time in the working and storage states, then t k Satisfy t k =d 0-1 (t k )+d 1-0 (t k ).

[0141] In some embodiments, online degradation data of the lithium-ion battery to be predicted and online status monitoring information may be obtained to perform real-time RUL prediction. Figure 3 A lithium-ion battery degradation path diagram provided by an embodiment of the present invention; Figure 4 A schematic diagram of a lithium-ion battery PDF and RUL prediction result provided by an embodiment of the present invention; wherein (a) is a schematic diagram of a lithium-ion battery PDF prediction result, and (b) is a schematic diagram of a lithium-ion battery RUL prediction result; see Figure 3-Figure 4 , the specific steps are as follows:

[0142] Combined with steps S12 and S13, the historical degradation data and status data of lithium-ion batteries B0005, B0006, and B0007 are used to obtain hyperparameter estimates of the model.

[0143] Once the degradation data and status data of the lithium-ion battery are obtained at a certain moment, the model parameters are updated in step S14, and then brought into the PDF of step S15 to obtain the expected RUL of the lithium-ion battery, such as Figure 4 shown.

[0144] The remaining life prediction device for lithium-ion batteries taking switching states into consideration provided by an embodiment of the present invention acquires degradation data and historical state data of lithium-ion batteries, and according to a pre-constructed lithium-ion battery degradation model and a lithium-ion battery operating state switching model, thereby taking into consideration the influence of state switching on lithium-ion battery degradation, and effectively solves the problem of difficulty in predicting the RUL of lithium-ion batteries under actual operating states; at the same time, compared with other methods, the pre-constructed lithium-ion battery degradation model and the lithium-ion battery operating state switching model are updated through Bayesian posteriori, and the device has the ability of real-time prediction; and the present invention pre-constructs a lithium-ion battery remaining probability density function model, which is more conducive to online calculation, and provides a basis and ideas for the embedded and cloud platform research and development of lithium-ion battery RUL prediction and health management systems.

[0145] Based on a general inventive concept, an embodiment of the present invention further provides a device for predicting the remaining life of a lithium-ion battery taking the switching state into consideration.

[0146] Figure 5 A schematic diagram of a lithium-ion battery remaining life prediction device considering switching states provided by an embodiment of the present invention, see Figure 5 The device provided in the embodiment of the present invention may include: a processor 51 and a memory 52, wherein the processor 51 is connected to the memory 52. ​​The processor 51 is used to call and execute the program stored in the memory 52; the memory 52 is used to store the program, and the program is at least used to execute the remaining life prediction method of the lithium-ion battery considering the switching state in the above embodiment.

[0147] It can be understood that the same or similar parts of the above embodiments can be referenced to each other, and the contents not described in detail in some embodiments can refer to the same or similar contents in other embodiments.

[0148] It should be noted that, in the description of the present invention, the terms "first", "second", etc. are only used for descriptive purposes and cannot be understood as indicating or implying relative importance. In addition, in the description of the present invention, unless otherwise specified, the meaning of "plurality" refers to at least two.

[0149] Any process or method description in a flowchart or otherwise described herein may be understood to represent a module, segment or portion of code that includes one or more executable instructions for implementing the steps of a specific logical function or process, and the scope of the preferred embodiments of the present invention includes alternative implementations in which functions may not be performed in the order shown or discussed, including performing functions in a substantially simultaneous manner or in the reverse order depending on the functions involved, which should be understood by those skilled in the art to which the embodiments of the present invention belong.

[0150] It should be understood that the various parts of the present invention can be implemented by hardware, software, firmware or a combination thereof. In the above-mentioned embodiments, a plurality of steps or methods can be implemented by software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented by hardware, as in another embodiment, it can be implemented by any one of the following technologies known in the art or their combination: a discrete logic circuit having a logic gate circuit for implementing a logic function for a data signal, a dedicated integrated circuit having a suitable combination of logic gate circuits, a programmable gate array (PGA), a field programmable gate array (FPGA), etc.

[0151] A person skilled in the art may understand that all or part of the steps in the method for implementing the above-mentioned embodiment may be completed by instructing related hardware through a program, and the program may be stored in a computer-readable storage medium, which, when executed, includes one or a combination of the steps of the method embodiment.

[0152] In addition, each functional unit in each embodiment of the present invention may be integrated into a processing module, or each unit may exist physically separately, or two or more units may be integrated into one module. The above-mentioned integrated module may be implemented in the form of hardware or in the form of a software functional module. If the integrated module is implemented in the form of a software functional module and sold or used as an independent product, it may also be stored in a computer-readable storage medium.

[0153] The storage medium mentioned above can be a read-only memory, a magnetic disk or an optical disk, etc.

[0154] In the description of this specification, the description with reference to the terms "one embodiment", "some embodiments", "examples", "specific examples", or "some examples" means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic representation of the above terms does not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described may be combined in any one or more embodiments or examples in a suitable manner.

[0155] Although the embodiments of the present invention have been shown and described above, it is to be understood that the above embodiments are exemplary and are not to be construed as limitations of the present invention. A person skilled in the art may change, modify, replace and vary the above embodiments within the scope of the present invention.

Claims

1. A method for predicting the remaining life of a lithium-ion battery considering a switching state, characterized in that: include: Obtaining degradation data and condition monitoring information of lithium-ion batteries to be predicted; The degradation data includes historical degradation data at the current moment; The state monitoring information includes historical state data at the current moment; Performing hyperparameter estimation on a pre-constructed lithium-ion battery degradation model using maximum likelihood estimation according to the historical degradation data of the lithium-ion battery to be predicted; Performing hyperparameter estimation on a pre-constructed lithium-ion battery operating state switching model according to the historical state data of the lithium-ion battery to be predicted; Based on the Bayesian theory, updating the posterior parameters of the lithium-ion battery degradation model and the posterior parameters of the lithium-ion battery operation state switching model according to the degradation data and the state monitoring information; Inputting the degradation data and the state monitoring information into a pre-constructed lithium-ion battery residual probability density function model to obtain a lithium-ion battery residual probability density function of the lithium-ion battery to be predicted; Obtaining the remaining life of the lithium-ion battery to be predicted according to the remaining probability density function of the lithium-ion battery; According to the Markov chain and the operating state of the lithium-ion battery, a lithium-ion battery operating state switching model is established, including defining a lithium-ion battery state space S={0, 1} according to the Markov chain, wherein the 0 state represents that the lithium-ion battery is in a working state, and the 1 state represents that the lithium-ion battery is in a storage state; The first working time of the lithium-ion battery is recorded as T1. Assume that T1 is an exponential distribution with parameter λ and expectation 1 / λ, and each subsequent working time is independent and identically distributed with T1. Also, the first storage time of the lithium-ion battery is recorded as T2, and T2 is assumed to be an exponential distribution with parameter μ and expectation 1 / μ, and each subsequent storage time is independent and identically distributed with T2; the state of the lithium-ion battery at time t is recorded as X(t); the transition probability function is proved by the Kolmogorov forward equation: At any time, the probability distribution of the lithium battery's operation and storage in the future can be determined according to the values ​​of the parameters λ and μ. κ(λ, μ) is called the set of parameters λ and μ. The step of estimating hyperparameters of a pre-built lithium-ion battery operating state switching model according to the historical state data of the lithium-ion battery to be predicted includes: The historical data is directly fitted using the GAMFIT function of MATLAB to obtain the shape parameter and scale parameter estimates of the gamma distribution, thereby obtaining the value of κ(λ, μ); The lithium-ion battery degradation model and the lithium-ion battery operation state switching model are integrated by using a fuzzy system to determine the residual probability density function model of the lithium-ion battery, including determining T=inf{t:X(t)≥ω|X(0)<ω} according to the concept of first arrival time; wherein T is the first arrival time; Let the service life of the lithium-ion battery be the time when it first reaches the retirement threshold ω; let the current time be t k , remaining life l k , according to the definition of equipment life: k =tt k , get k The remaining life Lk of the equipment at the moment is defined as: L k inf{l k >0:X(t k +l k )≥ω} If there is no moment to cross the boundary before reaching the boundary precisely at a certain time t, then the remaining probability density function of the lithium-ion battery at the first arrival time of X(t), t≥0, which first reaches the fixed threshold ω, is: In the above formula, f x(t) =(ω,t) is the residual probability density function of the lithium-ion battery at the time of first reaching the fixed threshold ω; S B (t) is the time-dependent boundary function corresponding to the standard Brownian motion, The total probability formula is used to obtain the residual probability density function of the lithium-ion battery that first reaches the failure threshold: Where w0 and w1 are the probability of the working and storage probabilities affecting the capacity degradation of lithium-ion batteries, p0 and p1 are the distribution probabilities of working and storing in the future. Assuming that the initial probability distribution of battery working and storing is π0(1, 0), it means that the probability of the system working at the beginning of operation is 100%, and the probability of the lithium-ion battery being in working and storing in the future can be obtained through the E state transition probability matrix (p0, p1); The degradation of lithium-ion batteries is divided into three stages: initial stage, middle stage, and final stage. The degradation is accelerated in the initial stage, stable in the middle stage, and accelerated again after reaching the final stage. The fuzzy control rules are set as follows based on the actual situation of lithium-ion batteries: If the remaining capacity is x1, that is, ω-x k In the early stage, the effect of working probability x2 decreases and the effect of early recovery probability x3 is larger; If the remaining capacity x1 is in the medium term, the effect x2 of the working probability is natural weight and the effect x3 of the recovery probability is moderate; If the remaining capacity x1 is at the end, the effect of the working probability x2 will be larger and the effect of the recovery probability x3 will be slightly larger.

2. The method according to claim 1, characterized in that Also includes: Acquiring a life prediction index of a lithium-ion battery, wherein the life prediction index includes a degradation characterization amount of the lithium-ion battery; A lithium-ion battery degradation model of a nonlinear drift Wiener process is established according to the life prediction index.

3. The method according to claim 2, characterized in that The lithium-ion battery degradation model of the nonlinear drift Wiener process is established according to the life prediction index, comprising: According to the degradation characteristics of lithium-ion batteries, the nonlinear drift Wiener process is expressed as: Among them, X(t) represents the degradation amount of the lithium-ion battery at time t, and X(0) represents the degradation amount at time 0, which takes the value of 0; is a random parameter that characterizes the individual differences of lithium-ion batteries. Follows normal distribution; is nonlinear drift degradation; σ is the diffusion parameter, σ>0; B(t) is the standard Brownian motion, which characterizes the random dynamic characteristics of the degraded device; The method of performing hyperparameter estimation on a pre-built lithium-ion battery degradation model using maximum likelihood estimation according to the historical degradation data of the lithium-ion battery to be predicted includes: If there are M lithium batteries with relevant historical degradation data; the number of degradation data of the i-th lithium battery is N i (0≤i≤M), and the measurement time of the i-th battery is t i，j (1≤j≤N i ), where j represents the jth measurement, X(t i,j ) represents the degradation of the ith battery at the jth measurement, and we get: Where: X i,j represents X(t i,j ); X(0) is the degradation amount at time 0. The degradation amount of the lithium battery at time 0 is 0, that is, X(0) = 0; is a random parameter The prior, that is The M degradation data obey the multivariate Gaussian distribution. Let the degradation data of the i-th lithium battery be Get the mean and covariance: Where: The log-likelihood function generated by the degradation data from all M degraded devices: Select a general and more applicable drift function, that is, ξ(τ; η) = ηt η-1 , maximize the above formula, that is, find the first-order partial derivative to be 0, and the model hyperparameter You can get it.

4. The method according to claim 2, characterized in that: The updating of the a posteriori parameters of the lithium-ion battery degradation model and the a posteriori parameters of the lithium-ion battery operation state switching model based on the Bayesian theory according to the degradation data and the state monitoring information includes: In the Bayesian framework, the random parameters of the degradation model are The posterior distribution of will follow Bayesian theory, given by The definition distribution of and the Brownian motion properties give the degenerate data likelihood function: Get Updated Mean and variance of the parameters: Assume that the prior distribution of p(κ) is a gamma distribution, with parameters λ~gama(α0, β0), μ~gama(α1, β1), where α0, α1 are shape parameters, and β0, β1 are scale parameters; the posterior distribution lithium-ion battery residual probability density function is: Where n 0→1 (t k ) is the lithium-ion battery in [0, t k ] The number of times the working state is converted to the storage state within a certain period of time, n 1→0 (t k ) is the lithium battery in [0, t k ] is the number of times the storage state is converted to the working state within a certain period of time, that is, n(t k ) satisfies n(t k )=n 0→1 (t k )+n 1→0 (t k ); let d0(t k ) and d1(t k ) is the lithium-ion battery in the interval [0, t k ] is the cumulative stay time in the working and storage states, then t k Satisfy t k =d0(t k )+d1(t k ).

5. A lithium-ion battery remaining life prediction device considering switching states, characterized in that: The invention comprises a processor and a memory, wherein the processor is connected to the memory: Wherein, the processor is used to call and execute the program stored in the memory; The memory is used to store the program, and the program is at least used to execute the method for predicting the remaining life of a lithium-ion battery taking into account the switching state as described in any one of claims 1-4.

Images