Rapid detection method for health state of vehicle lithium iron phosphate battery pack

Abstract

The invention relates to a rapid detection method for the health state of a vehicle lithium iron phosphate battery pack, and belongs to the technical field of electric vehicle detection. The method comprises the following steps: constructing an SOH evaluation model based on preset cyclic aging data of the lithium iron phosphate battery; collecting voltage data of a to-be-detected vehicle battery pack in a charging and discharging stage; and inputting voltage data of a to-be-detected vehicle battery pack into the SOH evaluation model, and carrying out rapid detection on the health state of the battery pack. According to the method, a deep learning method is utilized, and an effective evaluation model for the lithium iron phosphate battery pack is constructed by mining the relationship between the charging capacity and the battery health degree in the differential voltage interval. And the application of the battery SOH detection method based on data driving in the field of vehicle detection is effectively expanded.

Description

Technical Field

[0001] The present invention relates to the technical field of electric vehicle detection, and in particular to a method for quickly detecting the health status of a lithium iron phosphate battery pack for a vehicle. Background Art

[0002] New energy vehicles (NEVs) have become a key development direction for the automotive industry due to their low pollution and high energy efficiency. As a core component of NEVs, the performance and lifespan of battery packs directly impact the reliability, affordability, and user experience of the entire vehicle. Battery pack state of health (SOH) assessment is crucial for ensuring battery safety, extending service life, optimizing energy management, and reducing operation and maintenance costs.

[0003] However, direct methods for measuring the health of vehicle battery packs, particularly lithium iron phosphate (LFP) batteries, are currently lacking. The main reasons are as follows: First, lithium-ion batteries convert chemical energy into electrical energy through a series of electrochemically coupled reactions. Without unpacking the battery pack, it acts as a "black box" to the outside world, making it difficult to detect the microchemical reactions occurring internally and associated with aging. Therefore, in actual measurements, indirect estimates can only be made using physical quantities such as voltage and current during the battery's charge and discharge processes, which presents a challenge. Second, the charge and discharge curves of ternary lithium batteries vary relatively linearly with SOH, and subtle voltage changes can reflect changes in SOC or capacity, facilitating SOH assessment. In contrast, the charge and discharge curves of LFP batteries exhibit a distinct "plateau," meaning that voltage changes are very gradual across much of the SOC range, rendering most traditional detection methods ineffective. Third, the capacity decay of ternary lithium batteries is often accompanied by a significant increase in internal resistance and lithium metal deposition, making the decay process relatively concentrated and easily detectable. However, the degradation mechanisms of LFP batteries include gradual loss of active materials, increased polarization, and changes in the SEI film. These changes do not always significantly affect internal resistance or other characteristic parameters, making them difficult to assess using simple electrochemical methods.

[0004] In summary, in the field of vehicle testing, the development of SOH assessment models and algorithms for vehicles equipped with lithium iron phosphate battery packs is relatively lagging. Many existing SOH assessment methods are more suitable for ternary lithium batteries, while algorithms specifically for lithium iron phosphate batteries still require further research and optimization.

[0005] The continuous advancement of big data science and the widespread application of deep learning technology in the engineering field have provided an advanced operational data-driven approach for vehicle battery pack SOH assessment, meeting new requirements for assessment methods.

[0006] The latest existing technical solutions are based on obtaining characteristic data during the battery charging process, such as current, voltage, temperature, etc. (mostly voltage), and evaluating the health status of the battery by analyzing and modeling this data. This type of method is called a data-driven method. The general process is divided into: 1. Data acquisition. Collect voltage data from the battery system during battery charging; 2. Data preprocessing. Clean, correct, and preprocess the collected data to ensure data consistency and accuracy; 3. Feature extraction. Extract features related to the battery health status from the processed data based on deep learning or machine learning; 4. Modeling. Use the extracted features to build a model, usually a mathematical model that associates the battery SOH and features. 5. Calculation results. Calculate the battery SOH through the model. Figure 1 This is a flowchart of the data-driven method.

[0007] The above-mentioned method has been successfully applied to health monitoring of ternary lithium battery packs, achieving satisfactory assessment results. However, when the assessment targets are changed to lithium iron phosphate battery packs, due to the more complex aging mechanism of lithium iron phosphate and the obvious "plateau" during the charge and discharge process, the assessment accuracy is reduced or the model fails. Therefore, it is difficult to extract the relationship between voltage and capacity changes during the charge and discharge phase using traditional methods.

[0008] Therefore, how to improve the detection model's ability to detect lithium iron phosphate battery packs, especially the ability to detect battery packs inside vehicles without unpacking, is a key issue in ultimately realizing the practical application of data-driven detection methods in the health status assessment of battery packs in vehicles. Summary of the Invention

[0009] The purpose of the present invention is to propose a rapid detection method for the health status of automotive lithium iron phosphate battery packs to solve the problems existing in the above-mentioned prior art. The present invention successfully applies the evaluation model obtained based on partial charging training data to the SOH detection of battery packs inside vehicles. The method of evaluating battery health is achieved by detecting the change in charging capacity in the differential voltage stage. The present invention can sensitively capture the battery capacity within the range of slight voltage changes, thereby indirectly evaluating the degree of degradation of the active material inside the battery, thereby improving the model's evaluation performance for lithium iron phosphate batteries.

[0010] To achieve the above object, the present invention provides the following solutions:

[0011] A rapid detection method for the health status of automotive lithium iron phosphate battery packs, comprising:

[0012] Based on the preset cycle aging data of lithium iron phosphate batteries, a SOH evaluation model is constructed;

[0013] Collect voltage data of the battery pack of the vehicle to be tested during the charge and discharge stages;

[0014] The voltage data of the battery pack of the vehicle to be tested is input into the SOH evaluation model to perform a rapid test of the health status of the battery pack.

[0015] Optionally, based on preset cycle aging data of lithium iron phosphate batteries, constructing an SOH evaluation model includes:

[0016] Collect voltage data of the preset lithium iron phosphate battery during the cycle aging process;

[0017] Standardize the voltage data during all cycle aging processes;

[0018] Based on the standardized voltage data, a ΔQ-ΔU curve is drawn; wherein the ΔQ-ΔU curve is a curve showing the corresponding relationship between the voltage increase by ΔU and the charge capacity ΔQ starting from the starting voltage;

[0019] Build a deep learning model with multi-head attention mechanism as the core;

[0020] The deep learning model is trained using the ΔQ-ΔU curve to obtain an SOH evaluation model.

[0021] Optionally, collecting voltage data of a preset lithium iron phosphate battery during a cycle aging process includes:

[0022] The preset lithium iron phosphate battery is charged and discharged in a constant current cycle at a current rate of 1C until the maximum available capacity of the lithium iron phosphate battery is lower than the preset threshold of the rated available capacity, and the voltage data of all charging stages are collected.

[0023] Optionally, the expression for the normalization process is:

[0024] U 包 '=U 包 ×U 额，单 / U 额，包

[0025] Among them, U 包 ' represents the battery pack voltage data after standardization, U 额，包、 U 额，单 Respectively represent the rated voltage of the battery pack and the rated voltage of the single cell, U 包 Indicates the voltage data during the cycle aging process.

[0026] Optionally, inputting voltage data of the battery pack of the vehicle to be tested into the SOH evaluation model includes:

[0027] Preprocess the voltage data of the battery pack of the vehicle to be tested to obtain a standardized time series;

[0028] The normalized time series is fed into the multi-head attention mechanism, which captures the global contextual dependencies and obtains hidden feature representations by calculating the correlation between time steps.

[0029] The hidden feature representation is input into the feedforward neural network, and the time step dimension of the feedforward neural network output is reduced using the global average pooling layer to extract the global feature representation of the sequence;

[0030] The global feature representation is converted into a scalar output through a fully connected network layer, representing the predicted SOH value.

[0031] Optionally, the multi-head attention mechanism includes several independent attention heads, each of which extracts dependency features in different feature subspaces, and finally concatenates the output results of all attention heads to form a hidden feature representation of a preset dimension.

[0032] Optionally, the feedforward neural network includes a two-layer structure; wherein the first layer includes a number of neurons and adopts a ReLU activation function to increase nonlinear expression capabilities, and the second layer compresses the features to a dimension matching the attention output to ensure the consistency of the feature space.

[0033] Optionally, building the SOH assessment model further includes:

[0034] Calculate the mean square error between the predicted SOH and the actual SOH of the preset lithium iron phosphate battery, and use the Adam optimizer to update the model parameters with the initial learning rate.

[0035] The beneficial effects of the present invention are:

[0036] This paper uses deep learning to build an effective evaluation model for lithium iron phosphate battery packs by exploring the relationship between charge capacity and battery health within the differential voltage range. This effectively expands the application of data-driven battery SOH detection methods in the field of vehicle testing.

[0037] Compared with existing methods, the advantages of the present invention are mainly reflected in the following three aspects: First, the present invention constructs an SOH evaluation model by calculating the charging capacity within the differential voltage range, which effectively solves the inaccuracy problem of existing methods caused by the "platform area" during the charging stage of lithium iron phosphate batteries. Second, the present invention can effectively detect the SOH of electric vehicles equipped with lithium iron phosphate battery packs without unpacking, filling the gap in lithium iron phosphate battery pack detection methods in the field of vehicle detection. Third, a battery SOH detection model is constructed based on a large artificial intelligence model to extract the conversion relationship between battery data and battery SOH, achieving a higher accuracy rate than traditional models. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in 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 paying any creative work.

[0039] Figure 1 This is a schematic diagram of the data-driven approach;

[0040] Figure 2 The figure is a flow chart of a method for quickly detecting the health status of a lithium iron phosphate battery pack for a vehicle according to an embodiment of the present invention. DETAILED DESCRIPTION

[0041] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0042] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the present invention is further described in detail below with reference to the accompanying drawings and specific embodiments.

[0043] This embodiment provides a rapid detection method for the health status of a lithium iron phosphate battery pack for a vehicle, including:

[0044] Based on the preset cycle aging data of lithium iron phosphate batteries, a SOH evaluation model is constructed;

[0045] Collect voltage data of the battery pack of the vehicle to be tested during the charge and discharge stages;

[0046] The voltage data of the vehicle battery pack to be tested is input into the SOH assessment model to quickly detect the health status of the battery pack.

[0047] Specifically, the method of this embodiment is divided into a model construction phase and a battery pack testing phase. In the model construction phase, a SOH assessment model is constructed based on pre-prepared cycle aging data of lithium iron phosphate batteries. In the testing phase, the model is used to evaluate the battery pack of the vehicle to be tested, and the battery pack SOH assessment results are output.

[0048] Furthermore, based on the preset cycle aging data of lithium iron phosphate batteries, the SOH evaluation model is constructed including:

[0049] Collect voltage data of the preset lithium iron phosphate battery during the cycle aging process;

[0050] Standardize the voltage data during all cycle aging processes;

[0051] Based on the standardized voltage data, a △Q-△U curve is drawn. The △Q-△U curve is the corresponding relationship curve between the charging capacity △Q and each voltage increase △U stage starting from the starting voltage.

[0052] Build a deep learning model with multi-head attention mechanism as the core;

[0053] The △Q-△U curve is used to train the deep learning model to obtain the SOH evaluation model.

[0054] Furthermore, collecting voltage data of a preset lithium iron phosphate battery during cycle aging includes:

[0055] The preset lithium iron phosphate battery is charged and discharged in a constant current cycle at a current rate of 1C until the maximum available capacity of the lithium iron phosphate battery is lower than the preset threshold of the rated available capacity, and the voltage data of all charging stages are collected.

[0056] Furthermore, building the SOH assessment model also includes:

[0057] Calculate the mean square error between the predicted SOH and the actual SOH of the preset lithium iron phosphate battery, and use the Adam optimizer to update the model parameters with the initial learning rate.

[0058] Specifically, during the model building phase in this embodiment, the lithium iron phosphate battery is charged at a constant current rate of 1C until the maximum charging voltage of the battery is reached. The voltage curve of this phase is extracted. Then, starting from the starting voltage, the charging capacity ΔQ within each voltage increase ΔU stage is calculated, and a ΔQ-ΔU curve is plotted. Based on the curves obtained from the charging voltage under different health states, an assessment model is constructed to convert the curve characteristics into health states, and the model parameters are updated using the battery charging data.

[0059] Furthermore, inputting the voltage data of the battery pack of the vehicle to be tested into the SOH evaluation model includes:

[0060] Preprocess the voltage data of the battery pack of the vehicle to be tested to obtain a standardized time series;

[0061] The normalized time series is fed into the multi-head attention mechanism, which captures the global contextual dependencies and obtains hidden feature representations by calculating the correlation between time steps.

[0062] The hidden feature representation is input into the feedforward neural network, and the time step dimension of the feedforward neural network output is reduced using the global average pooling layer to extract the global feature representation of the sequence;

[0063] The global feature representation is converted into a scalar output through a fully connected network layer, representing the predicted SOH value.

[0064] Specifically, during the battery pack testing phase of this embodiment, for the vehicle battery pack to be tested, the battery pack is first communicated with the vehicle before evaluation to obtain the rated capacity and maximum charging voltage of the vehicle battery pack. The vehicle to be tested is then charged with a constant current of 1C until the maximum charging voltage Umax of the battery pack is reached. The voltage data from the charging phase is extracted, and the ΔQ-ΔU curve is plotted using the same procedures as in the model building phase. The SOH evaluation model obtained in the model building phase is used to evaluate the ΔQ-ΔU curve characteristics of the battery pack, ultimately determining the SOH of the vehicle battery pack to be tested.

[0065] The following is an example of an EV battery pack inspection station. Figure 2 As shown, the detailed implementation process of the technical solution of this embodiment is as follows:

[0066] Environment: EV power battery testing station;

[0067] Equipment: EV battery pack data reading device, equipped with a high-voltage charging gun, current sensor, voltage sensor, memory, and processor.

[0068] step:

[0069] 1. Data Collection: Prepare lithium iron phosphate (LiFePO4) cells and cycle them at a constant current rate of 1C until their maximum available capacity falls below 80% of the rated available capacity. Collect voltage data from all charging stages to form a training dataset. Use an EV battery pack data reader to read the battery pack information of the vehicle under test, obtaining its rated capacity and maximum charging voltage. Perform a charge and discharge cycle on the vehicle under test at the same rate of 1C, collecting voltage data from the charging phase to form the test data.

[0070] 2. Data standardization: Since the voltage data relationship between the battery pack and the battery cell satisfies: U 额，包 / U 额，单 =U 包 / U 单 , where U 额，包、 U 额，单 Respectively represent the rated voltage of the battery pack and the rated voltage of the single cell, U 额，包 It can be obtained from the data collection in step 1 that the U of the lithium iron phosphate cell 额，单 =3.2V, which is a fixed value. 包 、U 单 Respectively represent the battery pack voltage and single cell voltage collected from step 1 data collection. Therefore, based on the formula U 包 '=U 包 ×U 额，单 / U 额，包， Among them U 包 ' is the normalized battery pack voltage data, which is kept at the same order of magnitude as the voltage data of the single cell.

[0071] 3. Starting from the starting voltage U0 = 2.7V, calculate the battery charging capacity △Q for every voltage increase △U = 60mV, and draw the △Q-△U curve.

[0072] Building a SOH evaluation model: A deep learning model with a multi-head attention mechanism at its core was constructed. First, the input data was preprocessed into a normalized time series with a three-dimensional tensor shape (number of samples, time steps, feature dimensions), where the time step length was set to 100 and the feature dimension was 1. This data was then fed into the multi-head attention mechanism, which captures global contextual dependencies by calculating correlations between time steps. The multi-head attention mechanism consists of eight independent attention heads, each extracting dependency features in a different feature subspace. The outputs of all heads are then concatenated to form a latent feature representation of dimension 64. The extracted time series features are passed to a two-layer feedforward neural network. The first layer contains 128 neurons and uses the ReLU activation function to increase nonlinear representation capability. The second layer compresses the features to 64 dimensions to match the attention outputs and ensure consistency in the feature space. Subsequently, a global average pooling layer is used to reduce the time step dimension, extract a global feature representation of the sequence, and map the time series to a fixed-length vector. Finally, these global features are converted to a scalar output representing the predicted SOH value through a fully connected layer. The mean square error between the predicted SOH and the actual SOH of the experimental cell was calculated, and the model parameters were updated using the Adam optimizer with an initial learning rate of 0.001.

[0073] Application: At EV battery pack testing stations, this method can be used to quickly and effectively assess the SOH of new energy vehicles equipped with lithium iron phosphate battery packs.

[0074] Taking the application in the annual vehicle inspection station as an example, the detailed implementation process of the technical solution of this embodiment is as follows:

[0075] Environment: Annual vehicle inspection station

[0076] Equipment: Fixed automotive testing equipment equipped with a high-voltage charging gun, current sensor, voltage sensor, memory, and processor.

[0077] step:

[0078] Data Collection: At the annual vehicle inspection station, the battery health monitoring equipment is connected to the battery system of the vehicle being inspected. A high-voltage charging gun is used to charge the vehicle at a high rate (80A) to ensure that the battery quickly reaches a stable charging state. During the charging phase, battery pack data, including current, voltage, and temperature, is collected and saved as a test data set.

[0079] Model Construction: Build a deep learning electric vehicle battery health status detection model based on the attention mechanism. Standardize the raw data to ensure it is suitable for model input. Then, input the processed data into the network. The attention mechanism calculates the data's characteristics in different feature subspaces. Following the attention mechanism, a two-layer full neural network is deployed, with 128, 64, and 1 neurons in each layer. The fully connected neural network extracts deep neuron features from the data. Finally, ReLU is used as the activation function to output the final SOH prediction.

[0080] Parameter determination: Prepare lithium iron phosphate battery cells in advance. Divide 80A by the rated capacity of the battery pack to obtain a charge rate of 1C. Then, charge and discharge the battery cells at a constant current rate of 1C until the battery's available capacity is less than 80% of the cell's rated capacity. Extract voltage data from all charging stages. Starting at 2.7V, calculate the battery's charge capacity for each 60mV increase in voltage and plot a △Q-△U curve. Based on the resulting △Q-△U curve and the backpropagation algorithm, adjust the model parameters to minimize the difference between the model output and the actual SOH. After the parameters are determined, the model is deployed in the device memory before the test equipment leaves the factory.

[0081] Battery SOH detection: Input the features in the test data set into the parameterized model to obtain the state of health (SOH) estimate of the test vehicle.

[0082] Application: This technology enables rapid and accurate battery health assessment at annual vehicle inspection stations, ensuring that electric vehicle battery systems meet annual inspection standards. This helps improve inspection station efficiency and ensures road safety. It also provides a convenient way for vehicle owners to monitor their vehicle's battery health status and take necessary maintenance measures.

[0083] This embodiment proposes a novel SOH detection method for new energy vehicles equipped with lithium iron phosphate battery packs. By leveraging deep learning methods and exploring the relationship between charge capacity and battery health within the differential voltage range, an effective evaluation model for lithium iron phosphate battery packs is constructed. This effectively expands the application of data-driven battery SOH detection methods in the vehicle testing field.

[0084] Compared with existing methods, the advantages of this embodiment are mainly reflected in the following three aspects: First, by calculating the charging capacity within the differential voltage range to construct an SOH evaluation model, it effectively solves the inaccuracy problem of existing methods caused by the "platform area" during the charging stage of lithium iron phosphate batteries. Second, without unpacking, it can effectively detect the SOH of electric vehicles equipped with lithium iron phosphate battery packs, filling the gap in lithium iron phosphate battery pack detection methods in the field of vehicle detection. Third, based on the artificial intelligence large model, a battery SOH detection model is constructed to extract the conversion relationship between battery data and battery SOH, achieving a higher accuracy rate than traditional models.

[0085] The embodiments described above are merely descriptions of preferred embodiments of the present invention and are not intended to limit the scope of the present invention. Without departing from the spirit of the present invention, various modifications and improvements made to the technical solutions of the present invention by persons skilled in the art should fall within the scope of protection defined by the claims of the present invention.

Claims

1. A rapid detection method for the health status of a lithium iron phosphate battery pack for a vehicle, characterized in that: include: Based on the preset cycle aging data of lithium iron phosphate batteries, a SOH evaluation model is constructed; Collect voltage data of the battery pack of the vehicle to be tested during the charge and discharge stages; The voltage data of the battery pack of the vehicle to be tested is input into the SOH evaluation model to perform a rapid test of the health status of the battery pack.

2. The rapid detection method for the health status of a lithium iron phosphate battery pack for a vehicle according to claim 1, characterized in that: Based on the preset cycle aging data of lithium iron phosphate batteries, the SOH evaluation model is constructed including: Collect voltage data of the preset lithium iron phosphate battery during the cycle aging process; Standardize the voltage data during all cycle aging processes; Based on the standardized voltage data, a ΔQ-ΔU curve is drawn; wherein the ΔQ-ΔU curve is a curve showing the corresponding relationship between the voltage increase by ΔU and the charge capacity ΔQ starting from the starting voltage; Build a deep learning model with multi-head attention mechanism as the core; The deep learning model is trained using the ΔQ-ΔU curve to obtain an SOH evaluation model.

3. The rapid detection method for the health status of a lithium iron phosphate battery pack for a vehicle according to claim 2, characterized in that: The voltage data collected during the cycle aging process of the preset lithium iron phosphate battery includes: The preset lithium iron phosphate battery is charged and discharged in a constant current cycle at a current rate of 1C until the maximum available capacity of the lithium iron phosphate battery is lower than the preset threshold of the rated available capacity, and the voltage data of all charging stages are collected.

4. The rapid detection method for the health status of a lithium iron phosphate battery pack for a vehicle according to claim 2, characterized in that: The expression of the standardization process is: IN 包 '=U 包 ×U 额，单 / IN 额，包 Among them, U 包 ' represents the battery pack voltage data after standardization, U 额，包、 U 额，单 Respectively represent the rated voltage of the battery pack and the rated voltage of the single cell, U 包 Indicates the voltage data during the cycle aging process.

5. The rapid detection method for the health status of a lithium iron phosphate battery pack for a vehicle according to claim 1, characterized in that: Inputting the voltage data of the battery pack of the vehicle to be tested into the SOH assessment model includes: Preprocess the voltage data of the battery pack of the vehicle to be tested to obtain a standardized time series; The normalized time series is fed into the multi-head attention mechanism, which captures the global contextual dependencies and obtains hidden feature representations by calculating the correlation between time steps. The hidden feature representation is input into the feedforward neural network, and the time step dimension of the feedforward neural network output is reduced using the global average pooling layer to extract the global feature representation of the sequence. The global feature representation is converted into a scalar output through a fully connected network layer, representing the predicted SOH value.

6. The rapid detection method for the health status of a lithium iron phosphate battery pack for a vehicle according to claim 5, characterized in that: The multi-head attention mechanism includes several independent attention heads, each of which extracts dependent features in different feature subspaces. Finally, the output results of all attention heads are spliced together to form a hidden feature representation of a preset dimension.

7. The rapid detection method for the health status of a lithium iron phosphate battery pack for a vehicle according to claim 5, characterized in that: The feedforward neural network includes a two-layer structure; the first layer includes several neurons and adopts the ReLU activation function to increase the nonlinear expression ability, and the second layer compresses the features to a dimension matching the attention output to ensure the consistency of the feature space.

8. The rapid detection method for the health status of a lithium iron phosphate battery pack for a vehicle according to claim 2, characterized in that: Building an SOH assessment model also includes: Calculate the mean square error between the predicted SOH and the actual SOH of the preset lithium iron phosphate battery, and use the Adam optimizer to update the model parameters with the initial learning rate.

Images