Battery aging evaluation method based on multi-source multi-scale high-dimensional state space modeling

Abstract

The present invention relates to the field of new energy power system energy storage. Disclosed is a battery aging evaluation method based on multi-source multi-scale high-dimensional state space modeling. The method comprises: acquiring time series of a sample battery in each discharge process of preset discharge counts; on the basis of discharge parameters corresponding to the time series, determining a first state transition path and a second state transition path; determining a working state transition path reference; calculating a plurality of sample distances between the second state transition path and the working state transition path reference; using each sample distance as an input and using a corresponding target SOH value as an output to train a battery aging evaluation model; calculating a target distance between the state transition paths of a target battery to be predicted and the working state transition path reference; and inputting the target distance into the battery aging evaluation model to obtain a SOH value of said target battery. The present invention can improve the precision of battery aging prediction.

Description

Battery aging assessment method based on multi-source multi-scale high-dimensional state space modeling TECHNICAL FIELD

[0001] The present application relates to the field of new energy power system energy storage, in particular to a battery aging assessment method based on multi-source multi-scale high-dimensional state space modeling. BACKGROUND

[0002] Battery aging prediction refers to predicting the performance degradation of batteries by establishing mathematical models and using experimental data. This involves multiple aspects, including capacity degradation, internal resistance increase, cycle life, etc. of batteries. The following are some common battery aging prediction models and related industry progress:

[0003] Physics-based models: Physics-based models use electrochemical principles and physical equations to describe the chemical reactions and transport processes occurring in batteries to predict battery aging behavior. Such models can provide an in-depth understanding of the internal processes of batteries, but usually require complex mathematical expressions and a large number of parameters.

[0004] Statistical-based models: Statistical-based models establish statistical relationships to predict battery aging behavior by analyzing a large amount of experimental data. Such models are usually data-driven and do not require an in-depth understanding of the internal processes of batteries, but may have limited predictive ability for complex aging mechanisms.

[0005] Deep learning models: In recent years, deep learning technology has also made some progress in battery aging prediction. Deep learning models can learn complex nonlinear relationships from large amounts of data and have certain potential for battery aging prediction. For example, models such as recurrent neural networks (LSTM) can be used for time series data prediction.

[0006] Multi-physical coupling models: In recent years, more and more research has begun to explore multi-physical coupling models, combining electrochemical models, thermal models, mechanical models, etc. to more comprehensively describe the complex interactions within batteries and make more accurate aging predictions.

[0007] The above physics-based, multi-physical coupling models require a large number of difficult-to-measure parameters for modeling and have complex calculation processes and high computational costs, making it difficult to apply to actual environments. Machine learning models based on deep learning use black-box modeling methods, which have poor interpretability and transferability. At the same time, due to the differences in characteristics of each battery and working conditions, it is also difficult to obtain battery failure samples, so the method of training neural networks with large data samples and prediction often results in large errors. The method based on statistical models often only uses mathematical methods for statistical analysis without considering the physical mechanism of battery aging.

[0008] From the perspective of the model, the characteristic model of the circuit model does not consider the influence factors of various external environmental changes; the SOX index system belongs to basic performance indicators and cannot further depict the evolution process in the whole life cycle in a fine-grained manner, and there is a deficiency in analyzing the battery aging, state mutation and other fault modes; it cannot be directly and dynamically connected with the reconfigurable working mode; only time domain data is relied on, and features in other domains such as frequency domain are not considered. Although the electrochemical model simulates the battery operation process by using a system of differential equations, the model parameters are difficult to measure and update, and it is difficult to be directly applied in the actual environment. SUMMARY

[0009] The purpose of the present application is to provide a battery aging evaluation method based on multi-source multi-scale high-dimensional state space modeling, which can improve the battery aging prediction accuracy.

[0010] To achieve the above purpose, the present application provides the following scheme:

[0011] The battery aging evaluation method based on multi-source multi-scale high-dimensional state space modeling comprises:

[0012] Obtain the time sequence of each discharge process of the sample battery in the preset discharge times; the time sequence includes the sequence of voltage change over time and the sequence of current change over time; the preset discharge times include the first preset discharge times and the second preset discharge times after the first preset discharge times are completed;

[0013] According to the discharge parameters corresponding to the time sequence, determine the state transition path of the battery working state of the sample battery in each discharge process of the preset discharge times; the state transition path includes the first state transition path of the battery working state of the sample battery in each discharge process of the first preset discharge times and the second state transition path of the battery working state of the sample battery in each discharge process of the second preset discharge times; the state transition path is the motion trajectory of the battery working state of the battery in each discharge process;

[0014] According to the first state transition path of the battery working state of the sample battery in each discharge process of the first preset discharge times, determine the working state transition path reference;

[0015] Calculate the distance between the second state transition path of the battery working state of the sample battery in each discharge process of the second preset discharge times and the working state transition path reference, to obtain a plurality of sample distances;

[0016] Obtain the target SOH value of the sample battery in each discharge process of the second preset discharge times;

[0017] The neural network is trained by taking each sample distance as input and taking the corresponding target SOH value as output, to obtain a battery aging assessment model;

[0018] A distance between a state transition path of the target battery after a discharge process and the working state transition path benchmark is calculated to obtain a target distance;

[0019] The target distance is input into the battery aging assessment model to obtain an SOH value of the target battery after completing the preset target number of discharge processes.

[0020] Optionally, according to the discharge parameters corresponding to the time sequence, a state transition path of the battery working state of the sample battery in each discharge process of the preset number of discharges is determined, specifically including:

[0021] The time sequence is segmented according to a preset rule to obtain a plurality of time sequence segments; each time sequence segment includes a voltage time sequence segment and a current time sequence segment;

[0022] Each time sequence segment is discretized to obtain a plurality of discrete sequence data; the discrete sequence data includes voltage discrete sequence data and current discrete sequence data;

[0023] The temperature, SoC and discharge rate corresponding to each time sequence segment are obtained;

[0024] According to the discharge parameters corresponding to each time sequence segment arranged in chronological order according to each discharge process, a first state transition path of the battery working state of the sample battery in each discharge process of the first preset number of discharges and a second state transition path of the battery working state of the sample battery in each discharge process of the second preset number of discharges are determined; the input feature of the battery working state is the discharge parameter; the output feature of the battery working state is the voltage discrete sequence data; the discharge parameter includes the number of time sequence segments, temperature, SoC, discharge rate and current discrete sequence data.

[0025] Optionally, each time sequence segment is discretized by applying discrete Fourier transform.

[0026] Optionally, according to the first state transition path of the battery working state of the sample battery in each discharge process of the first preset number of discharges, a working state transition path benchmark is determined, specifically including:

[0027] The battery state transition probability of the first state transition path is calculated, and a working state transition path benchmark is determined according to the maximum value of the battery state transition probability; the working state transition path benchmark is composed of state transition paths corresponding to the maximum values of the battery state transition probabilities of the respective state transition paths.

[0028] Optionally, the neural network is an RNN neural network.

[0029] Optionally, the target distance and the sample distance are calculated by applying a frechet distance formula.

[0030] Optionally, the frechet distance formula is as follows:

[0031] wherein F(A, B) is the frechet distance between a state set A(alpha(i)) corresponding to a state transition path and a state set B(beta(j)) corresponding to another state transition path; d(A(alpha(i)), B(beta(j))) is the Euclidean or cosine distance between the state set A(alpha(i)) corresponding to a state transition path and the state set B(beta(j)) corresponding to another state transition path; A(alpha(i)) is a set composed of a state alpha(i) corresponding to a state transition path; B(beta(j)) is a set composed of a state beta(j) corresponding to another state transition path; alpha(i) is the i-th state in the state set A(alpha(i)) corresponding to a state transition path; beta(j) is the j-th state in the state set B(beta(j)) corresponding to another state transition path; inf is the lower limit; and N is the total number of states in the state transition path.

[0032] A computer device, comprising: a memory, a processor to store a computer program on the memory and executable on the processor, and the processor executes the computer program to implement the battery aging evaluation method based on multi-source multi-scale high-dimensional state space modeling according to any one of the above.

[0033] A computer readable storage medium having a computer program stored thereon, the computer program being executed by a processor to implement the battery aging evaluation method based on multi-source multi-scale high-dimensional state space modeling according to any one of the above.

[0034] A computer program product comprising a computer program, the computer program being executed by a processor to implement the battery aging evaluation method based on multi-source multi-scale high-dimensional state space modeling according to any one of the above.

[0035] According to the embodiments of the present application, the following technical effects are provided:

[0036] The application can model specific states of the battery, and the fine-grained model itself has the properties of invariance, timeliness, universality and completeness, so that the modeling data of other batteries can also be used as available samples of specific batteries, thereby accelerating the prediction progress and improving the prediction SOH accuracy, and meeting the battery aging evaluation and prediction under various working conditions. BRIEF DESCRIPTION OF DRAWINGS

[0037] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the drawings needed in the embodiments will be briefly introduced below. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative labor.

[0038] Fig. 1 is a flowchart of a battery aging evaluation method based on multi-source multi-scale high-dimensional state space modeling according to an embodiment of the present application.

[0039] Fig. 2 is a first-order Thevenin equivalent circuit model of a battery.

[0040] Fig. 3 is a fine-grained model state constructed according to a discharge process voltage time series.

[0041] Fig. 4 is a battery fine-grained model considering the input and output physical quantities of the working condition.

[0042] Fig. 5 is a state transition process of a high-dimensional state space.

[0043] Fig. 6 is a battery aging and modeling analysis step diagram.

[0044] Fig. 7 describes the structure of an RNN neural network for predicting SOH.

[0045] Fig. 8 is an internal structure diagram of a computer device. DETAILED DESCRIPTION

[0046] The technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only some embodiments of the present application, not all embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor are within the scope of protection of the present application.

[0047] The purpose of the present application is to provide a battery aging evaluation method based on multi-source multi-scale high-dimensional state space modeling, aiming to improve the accuracy of battery aging prediction.

[0048] In view of the foregoing problems of the current SOX index system, such as coarse granularity, no environmental interaction, and limited feature dimension, a fine-grained modeling level is added, a new model unit is introduced, and the invariance is met, that is, the change of external incentive factors does not affect the internal composition of the model unit, which can support the evaluation demand of dynamic reconfigurable mode, and the timeliness is met, that is, the state change of the model unit can be mapped to the aging process of the battery system, which can describe and predict the phase change and aging process of the battery, the model should also have universality, that is, the model unit can describe different battery types, batch numbers and working states, and can meet various application scenarios such as gradient utilization and battery mixed insertion, and completeness, the basic features of the battery are comprehensively described through multi-domain feature extraction, and the influence factors at different scales are comprehensively considered to comprehensively evaluate the state and performance of the battery.

[0049] A multi-source multi-scale high-dimensional state space is constructed through a new fine-grained modeling level, each measurement in a time period in a discharge sequence of the battery is defined as an input feature and an output feature, and the input feature and the output feature are quantified to define a state vector of the battery. Thus, each discharge process of the battery can be defined as a motion trajectory of a state space formed by all state vectors, and the SOH state of the battery is evaluated by analyzing the change of the state migration trajectory of the battery in the discharge process.

[0050] 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 drawings and specific embodiments.

[0051] Embodiment 1

[0052] As shown in FIG. 1, the battery aging evaluation method based on multi-source multi-scale high-dimensional state space modeling in the embodiment includes:

[0053] Step S101: acquiring a time sequence of each discharge process of a sample battery in a preset discharge number; the time sequence includes a sequence of voltage change over time and a sequence of current change over time; the preset discharge number includes a first preset discharge number and a second preset discharge number performed after the first preset discharge number is completed.

[0054] Step S102: determining a state migration path of the battery working state of the sample battery in each discharge process in the preset discharge number according to the discharge parameters corresponding to the time sequence; the state migration path includes a first state migration path of the battery working state of the sample battery in each discharge process of the first preset discharge number and a second state migration path of the battery working state of the sample battery in each discharge process of the second preset discharge number; the state migration path is a motion trajectory of the battery working state of the battery in each discharge process.

[0055] S102 specifically comprises:

[0056] Step S1021: segmenting the time series according to a preset rule to obtain a plurality of time series segments; each of the time series segments comprises a voltage time series segment and a current time series segment.

[0057] Step S1022: discretizing each of the time series segments to obtain a plurality of discrete sequence data; the discrete sequence data comprises voltage discrete sequence data and current discrete sequence data.

[0058] As a specific embodiment, the Fourier transform is applied to discretize each of the time series segments.

[0059] Step S1023: obtaining the temperature, SoC and discharge rate corresponding to each of the time series segments.

[0060] Step S1024: determining the first state transition path of the battery operating state of the sample battery in each discharge process of the first preset number of discharges and the second state transition path of the battery operating state of the sample battery in each discharge process of the second preset number of discharges according to the discharge parameters corresponding to each of the time series segments arranged in chronological order; the input feature of the battery operating state is the discharge parameter; the output feature of the battery operating state is the voltage discrete sequence data; the discharge parameter includes the number of time series segments, temperature, SoC, discharge rate and current discrete sequence data.

[0061] Step S103: determining the working state transition path reference according to the first state transition path of the battery operating state of the sample battery in each discharge process of the first preset number of discharges.

[0062] Specifically, the battery state transition probability of the first state transition path is calculated, and the working state transition path reference is determined according to the maximum value of the battery state transition probability; the working state transition path reference is composed of the state transition path corresponding to the maximum value of the battery state transition probability of each state transition path.

[0063] Step S104: calculating the distance between the second state transition path of the battery operating state of the sample battery in each discharge process of the second preset number of discharges and the working state transition path reference to obtain a plurality of sample distances.

[0064] As a specific embodiment, the frechet distance formula is applied to calculate the sample distance.

[0065] Step S105: obtaining the target SOH value of the sample battery in each discharge process of the second preset number of discharges.

[0066] Step S106: training the neural network with each sample distance as input and the corresponding target SOH value as output to obtain a battery aging assessment model.

[0067] As a specific embodiment, the neural network is an RNN neural network.

[0068] Step S107: calculating the distance between the state transition path after the target battery discharge process and the working state transition path benchmark to obtain a target distance.

[0069] As a specific embodiment, the target distance is calculated by applying a frechet distance formula.

[0070] The frechet distance formula is:

[0071] wherein d is a Euclidean or cosine distance, α is a state in a state set A corresponding to one state transition path, and β is a state in a state set B corresponding to another state transition path.

[0072] Step S108: inputting the target distance into the battery aging assessment model to obtain the SOH value of the target battery after completing the preset target number of discharge processes.

[0073] As shown in FIGS. 2-7, FIG. 2 describes the first-order Thevenin equivalent circuit model of the battery and the voltage and current changes of the battery after being connected to the circuit. FIG. 3 describes the decomposition process of taking a discharge time sequence segment in a time period in a first discharge sequence of the battery as modeling input data. The present application first segments the time sequence of the first discharge process of the battery (voltage and current time-varying sequence) according to time to form voltage and current time sequence segments as the basic data for modeling, and calculates the discrete Fourier spectrum sequence data of voltage and current based on the voltage and current time sequence. At the same time, the discrete Fourier spectrum sequence data of voltage and current, the temperature, SOC, and discharge rate measured during the time sequence segment are taken as data samples, the multi-source and multi-scale battery working state is defined according to the data samples, the input features of the state include the time sequence segment number, the discrete Fourier spectrum sequence of the current time sequence, the temperature, the SOC, the discharge rate, and the time sequence segment number, the output features of the state report the discrete Fourier spectrum sequence of the voltage time sequence, and the migration probability of the battery state is calculated according to the order of the time sequence segments generated in the previous G discharge sequences, that is, the migration from the state corresponding to a time sequence segment to the state corresponding to the next time sequence segment is calculated as a migration, according to the migration times of each state, the migration probability of each state migrating to other states is calculated, thereby constructing the dynamics process of the high-dimensional state space, calculating the maximum migration probability from the minimum time sequence segment number to the maximum time sequence segment number according to the time sequence segment number, and thereby obtaining the state migration path corresponding to the maximum migration probability as the main working path of the battery state migration. For the discharge sequence after G+1 times, the state migration path corresponding to G+1 times is generated according to the time sequence segments generated by the discharge sequence, and the distance between the state migration path corresponding to G+1 times and the main working path is calculated according to the frechet distance formula. The sample set of the corresponding relationship between the distance and the cycle number g (i.e., taking the corresponding relationship between the distance and the cycle number as a sample, and collecting multiple such corresponding relationships to form a sample set) is taken as the input of the RNN neural network; at the same time, the sample set of the corresponding relationship between the remaining capacity C and the cycle number g (wherein the SOH is calculated according to the ratio of the capacity measured after each discharge to the initial rated capacity) is taken as the output of the RNN neural network, thereby recursively predicting the remaining capacity of the battery. Wherein, g represents the cycle number variable, G represents the number of discharge sequences required to calculate the battery state migration probability, and one discharge process corresponds to one cycle.

[0074] Step one: segment the voltage (current) sequence of the first discharge process of the battery in the time dimension, the number of segments is N, the total number of elements of the entire discharge sequence is N*p, and the segmented voltage (current) time sequence segments can be numbered in time order as V0,...,V Nand C0,...,C N , and each element in each segmented sequence is p, taking the V0 time sequence segment as an example, the element contained is V 0,0 ,...,V 0,p-1 , and the temperature, SOC, and discharge rate corresponding to each time sequence segment are TM0,...,TM N , SOC0,...,SOC N , and D0,...,D N , and the segmentation process is shown in FIG. 2.

[0075] Step two: calculate the discrete Fourier spectrum of the voltage and current time sequence based on the discharge time sequence segment, and the formula is as follows:

[0076] wherein, W p = e (-2πi) / p is one of the p roots of 1, n∈[0,N], i,j∈[0,p-1].

[0077] Step three: define the multi-source multi-scale battery working state s n , the input features include: time sequence segment number n, discrete Fourier spectrum sequence of the current time sequence ( Because the segmented sequence is relatively stable, it mainly contains low-frequency components, so a value much smaller than p can be taken), temperature sequence TM0,...,TM N , SOC sequence SOC0,...,SOC N , and discharge rate sequence D0,...,D N , and the output feature is the discrete Fourier spectrum of the voltage time sequence ( Because the segmented sequence is relatively stable, it mainly contains low-frequency components, so a value much smaller than p can be taken):

[0078] Step four: define all states s n,m (n∈[0,N], m∈[0,M]) in the M discharge processes of the battery as a state space Φ, and the N+1 subspaces of the state space Φ are Ω n (n∈[0,N]), Ω n ={s n,0 ,s n,1 ,...,s n,M}, and calculate the battery state Ω i →Ω i+1The transition probability of (i∈[0,N-1]) is used to construct the dynamic process of the high-dimensional state space. Figure 4 describes the basic structure of the multi-source multi-scale battery fine-grained model considering the working condition proposed by the present application. Figure 5 describes the state transition process between the high-dimensional state space constituted by the corresponding working state of the battery fine-grained model, and the thick line describes the state transition process of the main working path. As shown in Figures 4 and 5.

[0079] For Ω i , the transition probability of a state s i,u to each state s i+1 in Ω i+1,v is defined as:

[0080] Where |Ω i+1 | is the total number of states in the Ω i+1 subspace, and σ(s i,u ,s i+1,v ) is the total number of times s i,u transitions to s i+1,v in M discharging processes, i.e., s i,u and s i+1,v are two adjacent states in the same discharging process.

[0081] Step five: the state transition path that satisfies the maximum probability condition given by formula (5) is the main working path. The main working path can be statistically calibrated according to sample data within a certain period of time of system normal operation, and used as the reference of the working state transition path of the battery. As the number of discharging times of the battery increases, the working state transition path of the battery gradually deviates from the main working path, which is manifested as the increasing distance from the main working path. The change in distance and the change in battery SOH have a mapping relationship.

[0082] Step six: the state set of the path corresponding to p max is A, and the state set of the other path is B. The frechet distance of the two paths is defined as:

[0083] Where d is the Euclidean or cosine distance, and α(i) and β(j) are the states in A and B, respectively.

[0084] Step seven: calculate the path state set B t corresponding to each discharging time t to obtain the F(A,B t ) sequence as the input of the RNN neural network; at the same time, calculate the battery SOH value Q t obtained by each discharging time t as the output of the RNN neural network, thereby recursively predicting the battery SOH value.

[0085] The neural network structure is shown in Fig. 6, wherein the input vector X t = F(A, B t ), the encoding dimension is 1, and the output vector y t = Q t , the encoding dimension is 1, and the hidden layer dimension is 20.

[0086] As a specific embodiment, the present application calibrates the state set A of the main working path by using the sample data obtained from the first 1000 discharging times (the sample number 1000 can be determined by the user, that is, the first preset discharging times) of the sample battery, trains the RNN neural network by using the sample data of the sample battery from the 1001th discharging time after 1000 discharging times to the 3000th discharging time (3000 is determined according to the actual training effect, and the training loss value of the neural network is required to be less than a given threshold value, for example, a loss threshold value of 0.01, that is, the second preset discharging times is 2000 (that is, 3000-1000=2000)), and after the neural network training is completed (that is, the training loss value of the neural network is less than the given threshold), the SOH value of the target battery to be predicted can be predicted based on the neural network input F(A, B t ) sequence.

[0087] In actual application, the trained neural network can be used for SOH prediction in the battery operation process. First, Xt is calculated and input into the neural network model, and the predicted SOH value can be output after the calculation by the neural network model.

[0088] In addition, in the present application, the sample battery includes a plurality of training batteries, and the second preset discharging times corresponding to each battery are set according to the needs, that is, each training battery can obtain a plurality of training data with different values of the second preset discharging times, and the training data of the plurality of training batteries are used as the training data of the neural network. For example, the training data of each training battery can be the distance between the state path and the main working path obtained from each discharging process starting from the value of the second preset discharging times of 1001 to the value of the second preset discharging times of 3000, and 2000 training data of each training battery can be obtained; or one training data can be obtained at intervals, so that 1000 training data of each training battery can be obtained.

[0089] The present application proposes a battery state model for constructing a multi-source multi-scale high-dimensional state space through a new fine-grained modeling level, and defines each discharging process of the battery as a motion trajectory of the state space formed by all state vectors, evaluates the SOH state of the battery by analyzing the change of the state migration trajectory of the battery in the discharging process, and overcomes the problems of not considering the working condition of the battery, few fault samples, insufficient prediction accuracy, poor interpretability and poor transferability existing in current various prediction algorithms.

[0090] Embodiment 2

[0091] A computer device, comprising: a memory, a processor to store a computer program on the memory and executable on the processor, the processor executes the computer program to implement the battery aging assessment method based on multi-source multi-scale high-dimensional state space modeling in embodiment 1.

[0092] Embodiment 3

[0093] A computer readable storage medium, having stored thereon a computer program, the computer program being executed by a processor to implement the battery aging assessment method based on multi-source multi-scale high-dimensional state space modeling in embodiment 1.

[0094] Embodiment 4

[0095] A computer program product, comprising a computer program, the computer program being executed by a processor to implement the battery aging assessment method based on multi-source multi-scale high-dimensional state space modeling in embodiment 1.

[0096] Embodiment 5

[0097] A computer device, which can be a database, and an internal structure diagram thereof can be as shown in FIG. 8. The computer device includes a processor, a memory, an input / output interface (I / O) and a communication interface. The processor, the memory and the input / output interface are connected through a system bus, and the communication interface is connected to the system bus through the input / output interface. The processor of the computer device is configured to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operating system and the computer program in the non-volatile storage medium to run. The database of the computer device is configured to store transactions to be processed. The input / output interface of the computer device is configured to exchange information between the processor and external devices. The communication interface of the computer device is configured to communicate with external terminals through a network connection. The computer program is executed by the processor to implement the battery aging assessment method based on multi-source multi-scale high-dimensional state space modeling in embodiment 1.

[0098] It should be noted that the object information (including but not limited to object device information, object personal information, etc.) and data (including but not limited to data for analysis, stored data, displayed data, etc.) involved in the present application are all information and data authorized by the object or fully authorized by all parties, and the collection, use and processing of related data need to comply with relevant laws, regulations and standards of relevant countries and regions.

[0099] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program. The computer program can be stored in a non-volatile computer readable storage medium, and when the computer program is executed, the processes of the above-mentioned embodiments of the methods can be included. Any reference to memory, database or other medium used in the embodiments provided by the present application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical storage, high-density embedded non-volatile memory, resistive memory (ReRAM), magnetoresistive random access memory (MRAM), ferroelectric memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. As an illustration but not limitation, RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM), etc. The database involved in the embodiments provided by the present application can include at least one of a relational database and a non-relational database. The non-relational database can include a distributed database based on a block chain, etc., without being limited thereto. The processor involved in the embodiments provided by the present application can be a general-purpose processor, a central processing unit, a graphics processing unit, a digital signal processor, a programmable logic device, a data processing logic device based on quantum computing, etc., without being limited thereto.

[0100] Any combination of the technical features of the above embodiments can be made. In order to make the description simple, all possible combinations of the technical features in the above embodiments are not described, however, as long as the combination of the technical features does not exist, it should be considered as the scope of the present application.

[0101] The principles and implementation modes of the present application are described by using specific examples. The above description of the embodiments is only used to help understand the method of the present application and its core idea; at the same time, for those skilled in the art, according to the idea of the present application, the specific implementation mode and application range will be changed. In conclusion, the content of the present application should not be understood as a limitation.

Claims

1. A battery aging assessment method based on multi-source multi-scale high-dimensional state space modeling, characterized in that, The method comprises: obtaining a time sequence of each discharge process of a preset number of discharges of a sample battery; the time sequence comprises a sequence of voltage change over time and a sequence of current change over time; the preset number of discharges comprises a first preset number of discharges and a second preset number of discharges after the first preset number of discharges is completed; determining a state transition path of a battery operating state of the sample battery in each discharge process of the preset number of discharges according to a discharge parameter corresponding to the time sequence; the state transition path comprises a first state transition path of the battery operating state of the sample battery in each discharge process of the first preset number of discharges and a second state transition path of the battery operating state of the sample battery in each discharge process of the second preset number of discharges; the state transition path is a motion trajectory of the battery operating state of the battery in each discharge process; determining a working state transition path reference according to the first state transition path of the battery operating state of the sample battery in each discharge process of the first preset number of discharges; calculating a distance between the second state transition path of the battery operating state of the sample battery in each discharge process of the second preset number of discharges and the working state transition path reference to obtain a plurality of sample distances; obtaining a target SOH value in each discharge process of the second preset number of discharges of the sample battery; training a neural network with each sample distance as input and the corresponding target SOH value as output to obtain a battery aging evaluation model; calculating a distance between a state transition path of a target battery after a discharge process and the working state transition path reference to obtain a target distance; inputting the target distance into the battery aging evaluation model to obtain an SOH value of the target battery after a preset target number of discharge processes; determining a state transition path of a battery operating state of the sample battery in each discharge process of the preset number of discharges according to a discharge parameter corresponding to the time sequence, specifically comprising: segmenting the time sequence according to a preset rule to obtain a plurality of time sequence segments; each time sequence segment comprises a voltage time sequence segment and a current time sequence segment; discretizing each time sequence segment to obtain a plurality of discrete sequence data; the discrete sequence data comprises voltage discrete sequence data and current discrete sequence data; obtaining temperature, SoC and discharge rate corresponding to each time sequence segment; determining the first state transition path of the battery operating state of the sample battery in each discharge process of the first preset number of discharges and the second state transition path of the battery operating state of the sample battery in each discharge process of the second preset number of discharges according to the discharge parameters corresponding to each time sequence segment arranged in chronological order according to each discharge process; the input feature of the battery operating state is the discharge parameter; the output feature of the battery operating state is the voltage discrete sequence data; the discharge parameter comprises the number of time sequence segments, temperature, SoC, discharge rate and current discrete sequence data; According to a first state transition path of the sample battery in each discharge process of a first preset number of times of discharges, a working state transition path benchmark is determined, and specifically includes: A battery state transition probability of the first state transition path is calculated, and according to a maximum value of the battery state transition probability, a working state transition path benchmark is determined; the working state transition path benchmark is composed of a state transition path corresponding to the maximum value of the battery state transition probability of each state transition path.

2. The battery aging assessment method based on multi-source multi-scale high-dimensional state space modeling according to claim 1, characterized in that, Each of the time series segments is discretized by using a discrete Fourier transform.

3. The battery aging assessment method based on multi-source multi-scale high-dimensional state space modeling according to claim 1, characterized in that, The neural network is an RNN neural network.

4. The battery aging assessment method based on multi-source multi-scale high-dimensional state space modeling according to claim 1, characterized in that, The target distance and the sample distance are calculated by using a frechet distance formula.

5. The battery aging assessment method based on multi-source multi-scale high-dimensional state space modeling according to claim 4, characterized in that, The Frechet distance formula is: Wherein, F(A, B) is the frechet distance between the state set A(α(i)) corresponding to one state transition path and the state set B(β(j)) corresponding to another state transition path; d(A(α(i)), B(β(j))) is the Euclidean or cosine distance between the state set A(α(i)) corresponding to one state transition path and the state set B(β(j)) corresponding to another state transition path; A(α(i)) is the set composed of the state α(i) corresponding to one state transition path; B(β(j)) is the set composed of the state β(j) corresponding to another state transition path; α(i) is the i-th state in the state set A(α(i)) corresponding to one state transition path; β(j) is the j-th state in the state set B(β(j)) corresponding to another state transition path; inf is the lower limit; N is the number of all states in the state transition path.

6. A computer apparatus comprising: The memory and the processor store a computer program which is stored on the memory and can run on the processor, and the processor executes the computer program to implement the battery aging evaluation method based on multi-source multi-scale high-dimensional state space modeling according to any one of claims 1-5.

7. A computer-readable storage medium having stored thereon a computer program, characterized in that, The computer program is executed by the processor to implement the battery aging evaluation method based on multi-source multi-scale high-dimensional state space modeling according to any one of claims 1-5.

8. A computer program product comprising a computer program, characterized in that, The computer program is executed by the processor to implement the battery aging evaluation method based on multi-source multi-scale high-dimensional state space modeling according to any one of claims 1-5. The computer program is executed by the processor to implement the battery aging evaluation method based on multi-source multi-scale high-dimensional state space modeling according to any one of claims 1-5.

Images