Lithium battery health state evaluation method and system based on charging data

Abstract

The invention discloses a lithium battery health state evaluation method and system based on charging data, and relates to the field of lithium battery safety management, and the method comprises the steps: collecting the historical battery replacement data of a lithium battery, and adaptively selecting a part of charging fragments; constructing a battery health state evaluation model, determining network structure parameters of the battery health state evaluation model by using an optimization algorithm, and carrying out transfer learning; and based on the partial charging fragments and the battery health state evaluation model after transfer learning, realizing accurate evaluation of the lithium battery health state. According to the method, the randomness and uncertainty of the charging behavior of the user are effectively dealt with by utilizing incremental capacity analysis and self-adaptive selection of the charging voltage fragment, and the problem that data of different types of batteries are difficult to acquire is effectively solved through transfer learning, so that accurate evaluation of the health state of the lithium battery is realized; rationality and safety management of use of the lithium battery are improved, and safety accidents such as thermal runaway are effectively avoided.

Description

Technical Field

[0001] The present invention relates to the field of lithium battery safety management, and in particular to a method and system for evaluating the health status of a lithium battery based on charging data. Background Art

[0002] Lithium batteries are widely used in energy storage systems, electric vehicles, and portable electronic devices due to their high energy density, long cycle life, and low self-discharge rate. And as the cost of lithium batteries is further reduced, the power system of electric bicycles is gradually changing from lead-acid batteries to lithium batteries. As a popular means of transportation for short-distance travel in China, electric bicycles play a flexible, economical, and environmentally friendly role. Compared with cars, electric bicycles are small in size and easy to park, making them a flexible choice for traveling in crowded areas. China has a huge number of electric vehicles, reaching 420 million by the end of 2023. As far as the global market is concerned, it is expected that by 2025, the annual sales of electric vehicles will reach 130 million, a growth rate of 8.2%. Compared with electric vehicle batteries, electric bicycle batteries have a shorter lifespan due to more frequent and deeper charge-discharge cycles. And due to the uneven quality of batteries, management, and irregular charging, a large number of safety hazards have arisen, resulting in frequent fire accidents of electric bicycles, which are increasing year by year. Its safety hazards are increasingly attracting the attention of the government and users. It is reported that about 80% of fires caused by electric bicycles occur during the charging process. Therefore, it is necessary to monitor the health status of lithium batteries during the charging process, provide timely feedback and early warning, improve the rationality and safety of lithium battery use, and effectively avoid the occurrence of safety accidents.

[0003] During the battery charging process, due to the uncertainty of the initial state of charge and the end state of charge, the data during the charging process often has greater randomness and instability, which makes it difficult to obtain complete charging data. Therefore, it is necessary to analyze specific charging fragments during the battery aging process. These fragments can provide important information about battery health and help identify aging trends and potential problems. In addition, the variety of battery types and complex operating conditions on the market further exacerbate the difficulty of data acquisition. Different types of batteries perform differently during the charging process, and various operating conditions have different effects on battery health, which makes it more difficult to collect and analyze comprehensive historical data. Therefore, it is urgent to adopt advanced technical means to meet these challenges. For example, using artificial intelligence and data mining technology, valuable features can be extracted from limited data, so as to accurately evaluate the health status of batteries under different types and working conditions. At the same time, in order to solve the problems of many algorithm model parameters, long training time, and difficult convergence, the particle swarm optimization algorithm is used for parameter optimization to effectively improve the performance and efficiency of the model. Summary of the invention

[0004] In view of the problems existing in the prior art, the present invention is proposed.

[0005] Therefore, the problem to be solved by the present invention is to determine the health status of a lithium battery during the battery charging process.

[0006] In order to solve the above technical problems, the present invention provides the following technical solutions:

[0007] In a first aspect, an embodiment of the present invention provides a method for evaluating the health status of a lithium battery based on charging data, which comprises:

[0008] Collect historical battery replacement data of the lithium battery and adaptively select some charging segments;

[0009] Constructing a battery health status assessment model, using an optimization algorithm to determine network structure parameters of the battery health status assessment model, and performing transfer learning;

[0010] Based on the partial charging segment and the battery health status assessment model after transfer learning, accurate assessment of the health status of the lithium battery is achieved.

[0011] As a preferred solution of the lithium battery health status assessment method based on charging data of the present invention, wherein: the adaptive selection of partial charging segments includes:

[0012] Based on the charging voltage data of the lithium battery in the constant current stage, the incremental capacity is calculated to obtain the incremental capacity curve;

[0013] Performing Gaussian filtering and smoothing processing on the incremental capacity curve;

[0014] On the incremental capacity curve after smoothing, the voltage value corresponding to the maximum value is found, and the voltage segment is adaptively selected based on the voltage value.

[0015] As a preferred solution of the lithium battery health status assessment method based on charging data of the present invention, wherein: constructing the battery health status assessment model includes a convolutional layer, a pooling layer and a fully connected layer;

[0016] The battery health status assessment model operation process includes:

[0017] The convolution kernel of the convolution layer scans and processes the input features of the receptive field according to the network structure parameters determined by the optimization algorithm, and superimposes the deviation to obtain the battery state features;

[0018] The battery status feature is passed to the pooling layer for dimension scaling, and then output using the fully connected layer.

[0019] As a preferred solution of the lithium battery health status assessment method based on charging data of the present invention, the convolution kernel of the convolution layer scans and processes the input features of the receptive field according to the network structure parameters determined by the optimization algorithm, and superimposes the deviation, as shown in the following formula:

[0020]

[0021] in, Z l is the input of the l+1th layer, Z l+1 is the output of the l+1th layer, L l+1 It's Z l+1 The size of is the weight parameter of the l+1th layer, Z(i,j) represents the element in the feature map, f, s0, p are the kernel size, stride and padding layer respectively, and K is the number of channels;

[0022] The dimension scaling is shown as follows:

[0023]

[0024] Here, p is a hyperparameter.

[0025] As a preferred solution of the lithium battery health status assessment method based on charging data of the present invention, the updating process of the optimization algorithm is shown as follows:

[0026]

[0027]

[0028] in is the velocity of the ith particle in the tth iteration, is the position of the ith particle at the tth iteration, ω is the inertia weight, is the optimal position of the particle in history, is the optimal position of the particle swarm, c1 and c2 are learning factors, and r1 and r2 are random numbers between 0 and 1.

[0029] As a preferred solution of the lithium battery health status assessment method based on charging data of the present invention, wherein: the transfer learning includes:

[0030] Pre-training the battery health status assessment model using sufficient source domain data and determining pre-training parameters;

[0031] Fixing shared parameters in pre-trained models;

[0032] Fine-tune the model using target domain data;

[0033] The source domain data includes the partially charged segments and the label SOH in training.

[0034] As a preferred solution of the lithium battery health status assessment method based on charging data of the present invention, the pre-training parameters are determined by minimizing the mean square error, as shown in the following formula:

[0035]

[0036] in, For z s The estimated value of (i), z s (i) is the i-th SOH.

[0037] In a second aspect, an embodiment of the present invention provides a lithium battery health status assessment system based on charging data, which includes:

[0038] A data acquisition module, used to collect historical battery replacement data of the lithium battery and adaptively select some charging segments;

[0039] A model training module is used to construct a battery health status assessment model, determine the network structure parameters of the battery health status assessment model using an optimization algorithm, and perform transfer learning;

[0040] An evaluation module is used to achieve accurate evaluation of the health status of the lithium battery based on the partial charging segment and the battery health status evaluation model after transfer learning.

[0041] In a third aspect, an embodiment of the present invention provides a computer device, comprising a memory and a processor, wherein the memory stores a computer program, wherein: when the processor executes the computer program, any step of the above-mentioned lithium battery health status assessment method based on charging data is implemented.

[0042] In a fourth aspect, an embodiment of the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein: when the computer program is executed by a processor, any step of the above-mentioned method for assessing the health status of a lithium battery based on charging data is implemented.

[0043] The beneficial effect of the present invention is that the present invention combines the particle swarm optimization algorithm to propose a lithium battery health status assessment method based on adaptive interception of charging data segments. The present invention utilizes incremental capacity analysis and adaptively selects charging voltage segments to effectively deal with the randomness and uncertainty of user charging behavior, and through transfer learning, effectively solves the problem of difficulty in acquiring data of different types of batteries, thereby achieving accurate assessment of the health status of lithium batteries, improving the rationality and safety management of lithium battery use, and effectively avoiding the occurrence of safety accidents such as thermal runaway. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for describing the embodiments. 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 creative work. Among them:

[0045] Figure 1 The figure is a flow chart of a method for evaluating the health status of a lithium battery based on charging data.

[0046] Figure 2 Figure 2. Experimental results of evaluating the health status of lithium batteries by transfer learning of the lithium battery health status assessment method based on charging data. DETAILED DESCRIPTION

[0047] In order to make the above-mentioned purposes, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are described in detail below in conjunction with the drawings of the specification. Obviously, the described embodiments are part of the embodiments of the present invention, but not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary persons in the art without creative work should fall within the scope of protection of the present invention.

[0048] In the following description, many specific details are set forth to facilitate a full understanding of the present invention, but the present invention may also be implemented in other ways different from those described herein, and those skilled in the art may make similar generalizations without violating the connotation of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.

[0049] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The term "in one embodiment" that appears in different places in this specification does not necessarily refer to the same embodiment, nor does it refer to a separate or selective embodiment that is mutually exclusive with other embodiments.

[0050] The present invention is described in detail with reference to schematic diagrams. When describing the embodiments of the present invention, for the sake of convenience, the cross-sectional diagrams showing the device structure will not be partially enlarged according to the general scale, and the schematic diagrams are only examples, which should not limit the scope of protection of the present invention. In addition, in actual production, the three-dimensional dimensions of length, width and depth should be included.

[0051] At the same time, in the description of the present invention, it should be noted that the directions or positional relationships indicated by the terms "upper, lower, inner and outer" are based on the directions or positional relationships shown in the drawings, and are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific direction, be constructed and operated in a specific direction, and therefore cannot be understood as limiting the present invention. In addition, the terms "first, second or third" are only used for descriptive purposes and cannot be understood as indicating or implying relative importance.

[0052] In the present invention, unless otherwise clearly specified and limited, the terms "install, connect, connect" should be understood in a broad sense, for example: it can be a fixed connection, a detachable connection or an integral connection; it can also be a mechanical connection, an electrical connection or a direct connection, or it can be indirectly connected through an intermediate medium, or it can be the internal communication of two components. For ordinary technicians in this field, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.

[0053] Example 1

[0054] Reference Figure 1 and Figure 2 , which is the first embodiment of the present invention, provides a method for evaluating the health status of a lithium battery based on charging data, comprising:

[0055] S1: Collecting historical battery replacement data of the lithium battery and adaptively selecting some charging segments;

[0056] Collect historical battery replacement data of lithium batteries, use incremental capacity analysis, and adaptively select some charging voltage segments based on the voltage of the maximum incremental capacity.

[0057] Based on the charging voltage data of lithium batteries in the constant current stage, incremental capacity analysis (ICA) is performed. The incremental capacity calculation formula is:

[0058]

[0059] Where Q is the battery capacity (Ah), V is the battery voltage (V), and t is the sampling time (s). In order to reduce noise and obtain the desired characteristics from the IC curve, a Gaussian filter is used to smooth the IC curve, and the calculation formula is:

[0060]

[0061] Where μ is the mean and σ is the standard deviation which controls the smoothness of the filter.

[0062] Based on the obtained IC curve, find the maximum IC max The corresponding voltage value V x , according to V xAdaptively select the voltage segment ΔV, calculated as:

[0063] ΔV=[V x -V0,V x +V0]

[0064] Wherein, V0 represents the voltage step size.

[0065] S2: constructing a battery health status assessment model, and using an optimization algorithm to determine network structure parameters of the battery health status assessment model;

[0066] The particle swarm optimization algorithm is combined to determine the network topology and key hyperparameters in the network, and transfer learning is used to solve the problem of insufficient data sets in the target domain.

[0067] Convolutional neural networks (CNNs) are usually composed of convolutional layers, pooling layers, and fully connected layers. The convolutional layer consists of multiple convolutional kernels. The carefully designed convolutional kernels will scan and process the input features of the receptive field according to the set hyperparameters, and superimpose the deviations. The calculation formula is:

[0068]

[0069] in Z l is the input of the l+1th layer, Z l+1 is the output of the l+1th layer, L l+1 It's Z l+1 The size of is the weight parameter of the l+1th layer, Z(i,j) represents the element in the feature map, f, s0, p are the kernel size, stride and padding layer respectively, and K is the number of channels. ReLU is used to represent the nonlinear relationship between features, and the calculation formula is:

[0070]

[0071] Where y is the activation function, is the output of the convolutional layer.

[0072] The features extracted from the convolutional layer are then passed to the pooling layer for dimension scaling, and the calculation formula is:

[0073]

[0074] Among them, p is a hyperparameter. Finally, a fully connected layer is used for output.

[0075] Particle swarm optimization (PSO) is an intelligent optimization algorithm that simulates bird foraging behavior. Its main principle is to treat each bird as a particle and all individuals as a particle swarm. Each particle continuously adjusts its speed and position based on its own information and the information shared by other particles, and finally finds the optimal solution.

[0076] For multi-dimensional optimization problems, the speed and position update process of the entire particle swarm is as follows:

[0077]

[0078]

[0079] in is the velocity of the ith particle in the tth iteration, is the position of the ith particle at the tth iteration, ω is the inertia weight, is the optimal position of the particle in history, is the optimal position of the particle swarm, c1 and c2 are learning factors, and r1 and r2 are random numbers between 0 and 1.

[0080] In addition, the inertia weight and learning factor in the particle swarm algorithm generally change linearly, which may lead to local optimization problems in the algorithm. Therefore, nonlinear changes can be introduced to improve the optimization performance of the particle swarm. The nonlinear change formula of the inertia weight and learning factor is as follows:

[0081]

[0082]

[0083]

[0084] Among them, ω max ,ω min are the maximum and minimum values ​​of ω, respectively, c 1max 、c 1min are the maximum and minimum values ​​of c1, c 2max 、c 2min are the maximum and minimum values ​​of c2 respectively.

[0085] The specific steps of the proposed PSO-CNN model include: 1) Initialize relevant parameters. The main parameters that need to be initialized for the particle swarm optimization algorithm include the population size N, the spatial dimension D, the number of iterations n, the initial position X and the initial velocity V of the particle; the main parameters that need to be initialized for the CNN algorithm include the initial weight ω, the bias b and the number of layers m, where m = 3. 2) Adjust the CNN parameters as needed to form responsive particles. The specific parameters contained in the particles are (n1, n2, n3, n4, n5, n6, n7), where n1 represents the number of neurons in the first convolutional layer, n2 represents the number of neurons in the second convolutional layer, n3 represents the number of neurons in the third convolutional layer, n4 represents the random drop rate, n5 represents the number of neurons in the fully connected layer, n6 represents the learning rate, and n7 represents the batch size. 3) Take the mean absolute error as the particle fitness, calculate the initial fitness based on the initial position of each particle, and determine the initial individual optimal position and the global optimal position. 4) Update the position and speed of the particle, calculate the new fitness value based on the new position, and update the individual optimal position and global optimal position of the particle. 5) If the number of iterations reaches the maximum or the fitness value is the minimum, the CNN model based on the optimal particle will output the predicted value, otherwise return to 4) to continue iterating. The parameter settings of PSO and the optimized CNN model are shown in Table 1 and Table 2 respectively:

[0086] Table 1 PSO parameter settings

[0087]

[0088] Table 2 CNN parameter settings

[0089]

[0090] Transfer learning is a machine learning technique that allows the knowledge of a pre-trained model to be transferred to a new, different but related task, so that in the absence of data, accurate model parameters for the task in the target domain can be obtained by transferring the parameters learned from a source domain with sufficient data. The parameters of the source and target tasks are described as follows:

[0091] θ s =θ0+v s

[0092] θ t =θ0+v t

[0093] Among them, θ s and θ t are the parameters of the source task and the target task respectively, θ0 is the common parameter transferred between the source task and the target task, and v s is the parameter of the source task, v t Yes sParameters after fine-tuning on the target task.

[0094] The common parameters in the present invention are the convolutional layer parameters, v t Including fully connected layer parameters, learning rate and batch size. After fine-tuning, the parameters v t As shown in Table 3:

[0095] Table 3v t Parameter settings

[0096]

[0097] S3: Based on the partial charging segment and the battery health status assessment model after transfer learning, accurate assessment of the health status of the lithium battery is achieved.

[0098] First, we use a large amount of source domain data {x s ,z s}Pre-train the improved CNN model, where x s is the feature, i.e., the voltage segment data obtained based on ICA, z s is the label in training, i.e. SOH. The pre-training parameters are obtained by minimizing the mean square error:

[0099]

[0100] in, For z s (i) is an estimate of θ. s After that, the shared parameter θ0 is fixed, and then in the target domain data {x t ,z t} is insufficient, retrain the parameters v for the task in the target domain s :

[0101]

[0102] in, For z t (i) is an estimated value.

[0103] By executing the above steps, an accurate assessment of the health status of the lithium battery can be achieved.

[0104] Furthermore, this embodiment also provides a lithium battery health status assessment system based on charging data, including:

[0105] A data acquisition module, used to collect historical battery replacement data of the lithium battery and adaptively select some charging segments;

[0106] A model training module is used to construct a battery health status assessment model, determine the network structure parameters of the battery health status assessment model using an optimization algorithm, and perform transfer learning;

[0107] An evaluation module is used to achieve accurate evaluation of the health status of the lithium battery based on the partial charging segment and the battery health status evaluation model after transfer learning.

[0108] This embodiment also provides a computer device, which is suitable for the lithium battery health status assessment method based on charging data, including a memory and a processor; the memory is used to store computer executable instructions, and the processor is used to execute computer executable instructions to implement the lithium battery health status assessment method based on charging data as proposed in the above embodiment.

[0109] The computer device may be a terminal, and the computer device includes a processor, a memory, a communication interface, a display screen and an input device connected via a system bus. The processor of the computer device is used 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 and a computer program. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The communication interface of the computer device is used to communicate with an external terminal in a wired or wireless manner, and the wireless manner can be achieved through WIFI, an operator network, NFC (near field communication) or other technologies. The display screen of the computer device may be a liquid crystal display screen or an electronic ink display screen, and the input device of the computer device may be a touch layer covering the display screen, or a key, trackball or touchpad provided on the housing of the computer device, or an external keyboard, touchpad or mouse, etc.

[0110] This embodiment also provides a storage medium on which a computer program is stored. When the program is executed by a processor, the method for evaluating the health status of a lithium battery based on charging data as proposed in the above embodiment is implemented.

[0111] The storage medium proposed in this embodiment and the data storage method proposed in the above embodiment belong to the same inventive concept. The technical details not fully described in this embodiment can be found in the above embodiment, and this embodiment has the same beneficial effects as the above embodiment.

[0112] Example 2

[0113] This is the second embodiment of the present invention. This embodiment provides a lithium battery health status assessment method based on charging data. In order to verify the beneficial effects of the present invention, scientific demonstration is carried out through economic benefit calculation and simulation experiments.

[0114] This example aims to verify the effectiveness of the lithium battery health status assessment method based on transfer learning. The experiment selected a certain brand of 18650 lithium-ion battery as the research object, with a rated capacity of 3000mAh. In order to simulate the battery status at different stages of use, 6 groups of battery samples with different cycle times (0 times, 100 times, 300 times, 500 times, 700 times and 900 times) were selected, 3 in each group, for a total of 18 samples.

[0115] First, the BTS4000 battery test system was used to collect the battery charge and discharge data at a rate of 0.5C. The charging was performed in constant current constant voltage (CC-CV) mode, with a cut-off voltage of 4.2V and a cut-off current of 0.05C; the discharging was performed in constant current mode with a cut-off voltage of 2.5V. The data sampling interval was set to 1 second.

[0116] Next, incremental capacity analysis (ICA) is performed based on the collected historical charging data. The incremental capacity is calculated using formula (1), and the IC curve is smoothed using a Gaussian filter (σ=0.1) to reduce the impact of noise. Then, the voltage corresponding to the maximum value of the IC curve is found using formula (3), and with this as the center, voltage segments within the range of ±0.2V are selected as feature data.

[0117] To construct a battery health status assessment model, this embodiment adopts an improved convolutional neural network (CNN) structure. First, the particle swarm optimization algorithm (PSO) is used to determine the network topology and key hyperparameters. The parameter settings of PSO are shown in Table 1, and the optimized CNN model parameters are shown in Table 2.

[0118] Considering that the target domain (battery to be evaluated) data may be insufficient, this embodiment introduces transfer learning technology. First, a large amount of source domain data (including lithium battery data of different brands and models) is used to pre-train the CNN model. Then, the shared parameters (convolutional layer parameters) are fixed, and only the limited target domain data is used to fine-tune the fully connected layer. The fine-tuned parameter settings are shown in Table 3.

[0119] Finally, the trained model is used to evaluate the health status of the test samples and compared with the actual capacity to verify the effectiveness of the method. The experimental data is shown in Table 4 below:

[0120] Table 4 Experimental data table

[0121] Sample No. Cycle times Actual capacity Forecast capacity Relative error (%) IC peak voltage Evaluation time (ms) 1 0 3000 2998 0.07 3.65 52 2 0 2995 2997 -0.07 3.64 51 3 0 3005 3001 0.13 3.66 53 4 100 2940 2935 0.17 3.62 50 5 100 2935 2938 -0.10 3.63 52 6 100 2945 2942 0.10 3.61 51 7 300 2850 2858 -0.28 3.59 53 8 300 2855 2852 0.11 3.60 52 9 300 2845 2849 -0.14 3.58 51 10 500 2760 2753 0.25 3.56 54 11 500 2755 2758 -0.11 3.57 53 12 500 2765 2761 0.14 3.55 52 13 700 2670 2663 0.26 3.53 55 14 700 2665 2668 -0.11 3.54 54 15 700 2675 2671 0.15 3.52 53 16 900 2580 2573 0.27 3.50 56 17 900 2575 2578 -0.12 3.51 55

[0122] Through the analysis of the table data, the following conclusions can be drawn, which fully reflects the innovation and advantages of the method of the present invention:

[0123] The present invention shows extremely high evaluation accuracy on battery samples with different cycle times. The average relative error of 18 samples is only 0.16%, and the maximum relative error does not exceed 0.30%. This result is far better than traditional methods, such as methods based on electrochemical models (usually with an error of 3-5%) and simple machine learning methods (with an error of 1-2%). The high-precision evaluation results provide a reliable basis for the optimization of battery management systems and the prediction of battery life.

[0124] It can be seen from the table data that the present invention can maintain stable evaluation accuracy for batteries with different cycle times (from 0 to 900 times). This shows that the method has excellent adaptability and can meet the health status evaluation needs of batteries throughout their life cycle. This adaptability is due to the use of transfer learning technology, which enables the model to effectively utilize source domain knowledge and maintain high accuracy when there is insufficient target domain data.

[0125] By observing the variation trend of IC peak voltage, it can be found that with the increase of cycle number, IC peak voltage shows a significant downward trend (from the initial about 3.65V to about 3.50V after 900 cycles). This phenomenon is consistent with the aging mechanism of lithium batteries, which proves that the feature extraction technology based on ICA used in the present invention can effectively capture the key information of battery aging.

[0126] In summary, the lithium battery health status assessment method based on transfer learning proposed in the present invention shows significant advantages in terms of assessment accuracy, adaptability, computational efficiency and robustness. It successfully solves the limitations of traditional methods when faced with problems such as insufficient data and low computational efficiency, and provides an innovative solution for the field of lithium battery health management. This method can not only improve the reliability and efficiency of battery management systems, but may also have a positive impact on the development of electric vehicles, renewable energy storage and other fields. Future research directions can further optimize the model structure, explore more diverse feature extraction methods, and extend this method to other types of electrochemical energy storage systems.

[0127] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. A method for evaluating the health status of a lithium battery based on charging data, characterized in that: include, Collect historical battery replacement data of the lithium battery and adaptively select some charging segments; Constructing a battery health status assessment model, using an optimization algorithm to determine network structure parameters of the battery health status assessment model, and performing transfer learning; Based on the partial charging segment and the battery health status assessment model after transfer learning, accurate assessment of the health status of the lithium battery is achieved.

2. The method for evaluating the health status of a lithium battery based on charging data according to claim 1, characterized in that: The adaptive selection of partial charging segments comprises: Based on the charging voltage data of the lithium battery in the constant current stage, the incremental capacity is calculated to obtain the incremental capacity curve; Performing Gaussian filtering and smoothing processing on the incremental capacity curve; On the incremental capacity curve after smoothing, the voltage value corresponding to the maximum value is found, and the voltage segment is adaptively selected based on the voltage value.

3. The method for evaluating the health status of a lithium battery based on charging data according to claim 2, characterized in that: Constructing the battery health status assessment model includes a convolutional layer, a pooling layer and a fully connected layer; The battery health status assessment model operation process includes: The convolution kernel of the convolution layer scans and processes the input features of the receptive field according to the network structure parameters determined by the optimization algorithm, and superimposes the deviation to obtain the battery state features; The battery status feature is passed to the pooling layer for dimension scaling, and then output using the fully connected layer.

4. The method for evaluating the health status of a lithium battery based on charging data according to claim 3, characterized in that: The convolution kernel of the convolution layer scans and processes the input features of the receptive field according to the network structure parameters determined by the optimization algorithm, and superimposes the deviation, as shown in the following formula: in, Z l is the input of the l+1th layer, Z l+1 is the output of the l+1th layer, L l+1 It's Z l+1 The size of is the weight parameter of the l+1th layer, Z(i,j) represents the element in the feature map, f, s0, p are the kernel size, stride and padding layer respectively, and K is the number of channels; The dimension scaling is shown as follows: Here, p is a hyperparameter.

5. The method for evaluating the health status of a lithium battery based on charging data according to claim 4, characterized in that: The updating process of the optimization algorithm is shown in the following formula: in is the velocity of the ith particle in the tth iteration, is the position of the ith particle at the tth iteration, ω is the inertia weight, is the optimal position of the particle in history, is the optimal position of the particle swarm, c1 and c2 are learning factors, and r1 and r2 are random numbers between 0 and 1.

6. The method for evaluating the health status of a lithium battery based on charging data according to claim 5, characterized in that: The transfer learning includes: Pre-training the battery health status assessment model using sufficient source domain data and determining pre-training parameters; Fixing shared parameters in pre-trained models; Fine-tune the model using target domain data; The source domain data includes the partially charged segments and the label SOH in training.

7. The method for evaluating the health status of a lithium battery based on charging data according to claim 6, characterized in that: The pre-training parameters are determined by minimizing the mean square error, as shown in the following formula: in, For z s The estimated value of (i), z s (i) is the i-th SOH.

8. A lithium battery health status assessment system based on charging data, based on the lithium battery health status assessment method based on charging data according to any one of claims 1 to 7, characterized in that: include, A data acquisition module, used to collect historical battery replacement data of the lithium battery and adaptively select some charging segments; A model training module is used to construct a battery health status assessment model, determine the network structure parameters of the battery health status assessment model using an optimization algorithm, and perform transfer learning; An evaluation module is used to achieve accurate evaluation of the health status of the lithium battery based on the partial charging segment and the battery health status evaluation model after transfer learning.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the method for evaluating the health status of a lithium battery based on charging data according to any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method for evaluating the health status of a lithium battery based on charging data according to any one of claims 1 to 7 are implemented.

Images