Vehicle-mounted power battery charge and health state combined diagnosis method and system

Abstract

The invention relates to the technical field of battery diagnosis, and discloses a vehicle-mounted power battery charge and health state combined diagnosis method and system, and the method comprises the steps: obtaining the initial parameter data of a battery as a first working condition; constructing a derivative working condition generation model; deriving a plurality of second working conditions by using the model based on the first working conditions, ensuring that the second working conditions cover a reasonable parameter range, and keeping correlation with the first working conditions; respectively calculating the battery charge states under the first working condition and all derived second working conditions to obtain a charge state set; and calculating the consistency of the charge state set, evaluating the result accuracy of the battery charge state under the first working condition, and if the result accuracy meets a preset requirement, calculating the battery health state under the first working condition based on the battery charge state and the result accuracy under the first working condition. According to the method, the accuracy of SOC and SOH is improved in a working condition derivation mode, meanwhile, the charge state and the health state of the lithium battery are jointly estimated, and a new thought is provided for prediction of the health state of the lithium battery.

Description

Technical Field

[0001] The present invention relates to the technical field of battery diagnosis, and particularly to a method and system for jointly diagnosing the state of charge and health of in-vehicle power batteries. Background Art

[0002] With the rapid development of the electric vehicle industry, as a core component, the performance and reliability of in-vehicle power batteries directly affect the overall performance of electric vehicles. Accurately evaluating the state of charge (SOC) and state of health (SOH) of power batteries is of crucial significance for improving the safety of electric vehicles, extending the battery service life, and optimizing vehicle energy management.

[0003] Existing SOC and SOH diagnosis methods have many problems. For example, some diagnosis methods based on a single model or algorithm cannot adapt to the dynamic characteristics of batteries under complex working conditions, and the diagnosis accuracy is relatively low. Some methods are highly dependent on battery parameters, and the battery parameters will drift with the change of use time and environmental conditions, resulting in inaccurate diagnosis results. In addition, existing diagnosis methods often diagnose SOC and SOH separately, ignoring the internal connection between the two, and it is difficult to achieve a comprehensive and accurate assessment of the battery state. At the same time, the existing research on SOC and SOH focuses on the accuracy during their independent calculations, but does not judge the result accuracy while diagnosing SOC and SOH. Summary of the Invention

[0004] The present invention aims to provide a method and system for jointly diagnosing the state of charge and health of in-vehicle power batteries to solve the problems of low diagnosis accuracy, strong dependence on battery parameters, and inability to comprehensively evaluate SOC and SOH in the prior art.

[0005] The basic solution provided by the present invention is as follows: A method for jointly diagnosing the state of charge and health of in-vehicle power batteries, the method comprising:

[0006] Obtain the initial parameter data of the battery as the first working condition;

[0007] Construct a working condition derivation model; use the first working condition as the input, and use this model to derive several second working conditions to ensure that the derived second working conditions cover a reasonable parameter range while maintaining the correlation with the first working condition;

[0008] Calculate the state of charge of the battery under the first working condition and all the derived second working conditions respectively to obtain a set of state of charge; calculate the consistency of the set of state of charge and evaluate the result accuracy rate of the state of charge of the battery under the first working condition;

[0009] When the result accuracy of the state of charge of the battery under the first working condition meets the preset requirements, based on the state of charge of the battery and the result accuracy under the first working condition, calculate the state of health of the battery under the first working condition.

[0010] The present invention is based on a joint diagnosis method for the state of charge and health of an on-vehicle power battery, and also provides a joint diagnosis system for the state of charge and health of an on-vehicle power battery. The system includes:

[0011] A data acquisition module, configured to acquire battery initial parameter data as the first working condition;

[0012] A working condition derivation module, configured to construct and run a working condition derivation model. After receiving the first working condition, use the first working condition as the input, and use the model to derive several second working conditions to ensure that the derived second working conditions cover a reasonable parameter range and maintain relevance to the first working condition;

[0013] A first calculation module, configured to calculate the state of charge of the battery under the first working condition and all derived second working conditions respectively to obtain a state of charge set; calculate the consistency of the state of charge set, and evaluate the result accuracy of the state of charge of the battery under the first working condition;

[0014] A second calculation module, configured to determine whether the result accuracy of the state of charge of the battery under the first working condition meets the preset requirements. If it meets, calculate the state of health of the battery under the first working condition based on the state of charge of the battery and the result accuracy under the first working condition.

[0015] The working principle and advantages of the present invention are as follows:

[0016] The charge and discharge process of the power battery is a complex electrochemical change process. For example, the SOC of the power battery is related to many factors such as temperature, discharge current, and the charge and discharge state at the previous moment, and has very strong nonlinearity, making it difficult to perform real-time online estimation of SOC; the calculation of SOH also faces the same problem.

[0017] Conventionally, when calculating SOC and SOH, only the improvement of the SOC and SOH calculation models themselves is considered for their accuracy. However, this solution finds that the accuracy of SOC and SOH lies not only in the calculation models themselves, but more importantly, in the accuracy reflected at the final application moment. The most direct application of the calculated SOC and SOH is the adjustment of battery system-related control strategies. However, there is a corresponding time difference from obtaining battery data to calculating SOC and SOH, and then to using the calculated SOC and SOH for adjusting control strategies. This will result in the adjustment of the final control strategy always being based on the battery data at the corresponding moment when SOC and SOH are calculated, rather than the battery data at the moment of control strategy adjustment. Therefore, from the perspective of the calculation results of SOC and SOH relative to their final application, the calculation results are delayed. This delayed inaccuracy is different from the accuracy of the algorithm model itself. Therefore, using the improvement of the conventional algorithm model cannot solve the problem of determining the accuracy of the calculation results of SOC and SOH relative to their final application.

[0018] Facing the above defects of the existing technology, this solution has different perspectives and improvement strategies for the accuracy of SOC and SOH. It completely breaks through the optimization of the conventional algorithm model itself. Instead, it adopts the method of derivative working conditions to simulate the fluctuation of battery data from the moment of obtaining the original battery data to the moment of the final application of the calculation results of SOC and SOH, constructs a fluctuation scenario, calculates the accuracy of the results through the SOC consistency method while calculating SOC, and corrects SOC based on the result accuracy rate to improve the accuracy of the calculation results of SOC relative to the final application moment. Calculate SOH using the corrected SOC to achieve joint estimation. Joint estimation is to integrate the state of charge and health state of lithium batteries in a specific way, aiming to obtain better results than single-state prediction. This joint estimation method fully considers the internal chemical reactions during the battery operation process and can accurately simulate the dynamic response outside the battery. This method is closer to the actual situation and provides a new idea for predicting the health status of lithium batteries. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] Figure 1 It is a schematic structural diagram of a joint diagnosis system for the state of charge and health of on-vehicle power batteries provided by an embodiment of the present invention;

[0020] Figure 2 It is a schematic flow diagram of a joint diagnosis method for the state of charge and health of on-vehicle power batteries provided by an embodiment of the present invention;

[0021] Figure 3 It is a schematic flow diagram of the calculation of consistency and result accuracy rate provided by an embodiment of the present invention;

[0022] Figure 4Schematic diagram of the process for calculating the battery health state provided by the embodiments of the present invention. Detailed implementation manners

[0023] The following is a further detailed description through specific implementation manners:

[0024] The embodiment is basically as shown in the appendix Figure 1 : A joint diagnosis system for the state of charge and health of an in-vehicle power battery, the system includes:

[0025] A data acquisition module, configured to acquire battery initial parameter data as the first working condition;

[0026] A working condition derivation module, configured to construct and run a working condition derivation model. After receiving the first working condition, using the first working condition as the input, several second working conditions are derived by using the model to ensure that the derived second working conditions cover a reasonable parameter range while maintaining the relevance with the first working condition;

[0027] A first calculation module, configured to calculate the state of charge of the battery under the first working condition and all the derived second working conditions respectively to obtain a state of charge set; calculate the consistency of the state of charge set, and evaluate the result accuracy rate of the state of charge of the battery under the first working condition;

[0028] A second calculation module, configured to determine whether the result accuracy rate of the state of charge of the battery under the first working condition meets a preset requirement. If it meets, based on the state of charge of the battery and the result accuracy rate under the first working condition, calculate the battery health state under the first working condition.

[0029] It can be understood that this system can fully execute the following joint diagnosis method for the state of charge and health of an in-vehicle power battery, and can achieve the same effect as the following method. The specific process is not described here again, and the detailed process is as follows.

[0030] As Figure 2 shown, a joint diagnosis method for the state of charge and health of an in-vehicle power battery, the method includes:

[0031] S100, acquire battery initial parameter data as the first working condition;

[0032] Specifically, the battery initial parameter data includes basic data and operation data. The basic data represents the battery itself information and the battery application scenario information, which is used to assist in completing the corresponding information screening in the subsequent process; the operation data represents the characteristics of the battery during use and participates in the SOC and SOH calculations. Based on different SOC and SOH calculation models or algorithms, targeted selection can be made. In this embodiment, the basic data includes the battery model and the loaded vehicle model; the operation data includes the data acquisition time period, current, voltage, temperature, remaining battery capacity at a moment, relative maximum available capacity, and discharged battery capacity at a moment.

[0033] S200, construct a working condition derivation model; using the first working condition as the input, derive several second working conditions by using this model, ensuring that the derived second working conditions cover a reasonable parameter range and at the same time maintaining the correlation with the first working condition;

[0034] Specifically, construct a working condition derivation model; obtain several battery parameter data, use machine learning technology to learn the fluctuation law of the battery parameter data, establish the mapping relationship between the battery parameter data and the fluctuation law, and use this mapping relationship to perform working condition derivation. In this embodiment, the following machine learning technology is used. In other embodiments, other machine learning corresponding architectures can be used as long as the effect of this solution can be achieved.

[0035] S201, obtain several battery parameter data as the original data set. Each group of battery parameter data includes at least the corresponding battery initial parameter data types, such as battery model, loaded vehicle model, data acquisition time period, current, voltage, temperature, remaining battery power at a moment, relative maximum available capacity, and discharged battery power at a moment. Reasonably select the original data set to ensure that there is a normal relationship between the data. For example, data with serious battery anomalies and no representativeness can be screened to avoid large learning errors in the fluctuation law, so as to ensure that the derived second working conditions cover a reasonable parameter range and at the same time maintain the correlation with the first working condition.

[0036] S202, data preprocessing, including data cleaning: handling missing values and outliers. Missing values can be filled with the mean or median, and outliers can be identified and processed through statistical methods (such as Z-score); it also includes data standardization: standardize numerical features (such as current, voltage, temperature). Common methods include Min-Max scaling or Z-score standardization.

[0037] S203, feature engineering, including feature selection: select features with high correlation with the fluctuation law. Methods such as correlation analysis and recursive feature elimination can be used. In this embodiment, the data types corresponding to the battery initial parameter data are all regarded as features with high correlation with the fluctuation law. It also includes feature extraction: extract more meaningful features from the original data, such as calculating the change rate, standard deviation, etc. of current, voltage, and temperature.

[0038] S204, model selection and architecture. In this embodiment, the Autoencoder is selected. It is an unsupervised learning model that can learn the latent representation of data and is suitable for mining the fluctuation law of data.

[0039] The autoencoder consists of two parts: an encoder and a decoder. The encoder compresses the input data into a low-dimensional latent representation, and the decoder reconstructs the latent representation into the original data. Encoder: h = f enc (x; θenc ), where x is the input data, θ enc are the parameters of the encoder, h is the latent representation, and f enc is the mapping function of the encoder. Decoder: where is the reconstructed data, θ dec are the parameters of the decoder, and f dec is the mapping function of the decoder.

[0040] Use the backpropagation algorithm to train the autoencoder. The goal is to minimize the reconstruction error, and usually the mean squared error (MSE) is used as the loss function:

[0041] S205, Model training, Divide the dataset: Divide the original dataset into a training set, a validation set, and a test set, for example, in the ratio of 70:15:15. Use the training set to train the autoencoder, update the model parameters through the backpropagation algorithm, and minimize the reconstruction error. During the training process, use the validation set to monitor the performance of the model and prevent overfitting.

[0042] Model evaluation, Use the test set to evaluate the performance of the model and calculate the reconstruction error. Metrics such as the mean squared error (MSE) and the root mean squared error (RMSE) can be used.

[0043] S206, Similar data screening, Calculate the latent representation: Input all the data into the trained autoencoder to obtain their latent representations. Calculate the similarity: When inputting a new set of data, calculate the similarity between its latent representation and the latent representations of all the data. Commonly used similarity measurement methods include the Euclidean distance, cosine similarity, etc. Screen similar data: Sort according to the similarity and select multiple groups of data with higher similarity.

[0044] In specific applications, when a set of initial battery parameter data, i.e., the first working condition, is input into the constructed working condition derivation model, through the operation of the model, multiple groups of battery parameters with higher similarity can be obtained as several derived second working conditions to simulate data fluctuations with similarity differences.

[0045] S300, Calculate the state of charge of the battery under the first working condition and all the derived second working conditions respectively to obtain the state of charge set;

[0046] Specifically, the concept of the relative maximum available capacity of lithium iron phosphate power batteries can be used to calculate the state of charge of the battery and calculate the new method of battery SOC.

[0047] In this embodiment, the full charge and full discharge of a 200Ah lithium iron phosphate battery are defined as follows: At room temperature, the battery is charged in a constant current-constant voltage manner. During the constant voltage process, when the battery current drops to the specified current of 0.03C (6A), the battery is fully charged and the SOC is 1. When the battery discharges at a small current of 0.05C (10A) and the battery voltage drops to the cut-off voltage (2.5V), the battery is fully discharged and the SOC is 0. In this way, the discharge capacity during the process of the battery discharging from the fully charged state to the fully discharged state is the relative maximum available capacity Q of the battery. rmax ; And the remaining charge Q l is defined as the discharge capacity during the process of the battery discharging from the current state to the fully discharged state. Therefore, the battery SOC is redefined as the ratio of the remaining charge Q l and the relative maximum available capacity Q rmax , which can be expressed as:

[0048]

[0049] where Q l is the remaining charge of the battery at the calculation moment (i.e., the remaining charge at the moment); Q rmax is the relative maximum available capacity of the battery; Q d is the charge already discharged by the battery at the calculation moment (i.e., the charge already discharged at the moment); η is the Coulomb efficiency of the battery, which is 1; i is the charge and discharge current, with the current being positive during discharge and negative during charge.

[0050] By calculating in the above manner, the decoupling between the battery SOC and the operating conditions, ambient temperature, current, and charge and discharge methods can be achieved, thus effectively avoiding the problems and contradictions existing in the traditional SOC definition method.

[0051] The state of charge of the battery under each condition of this solution can be that of a single battery or a battery pack, without affecting the calculation strategy of the overall solution. For a battery pack used in series, in order to avoid overcharging or over-discharging of any single battery in the battery pack, it is necessary to ensure that the SOC of all single batteries is between 0 and 1, and all single batteries are not overvoltage or undervoltage, so as to ensure the safety and long life of the battery pack during use. Currently, in order to prevent overcharging and over-discharging of single batteries in the battery pack, the SOC of the battery with the smallest SOC in the battery pack is often used as the SOC of the battery pack. Therefore, this method is also adopted in this solution to determine the SOC of the battery pack, but the available capacity of the corresponding single battery is considered during use to facilitate the estimation of variables such as the driving range and health status of electric vehicles.

[0052] Thus, the SOC under various working conditions can be calculated according to the above calculation model. Of course, the estimation of SOC can also be carried out using any existing calculation model, without being limited to a specific method. In this embodiment, according to the improved second-order RC equivalent circuit model, the improved Adaptive Extended Kalman Filter (AEKF) algorithm is used to accurately estimate the SOC of the power battery online. Compared with the existing SOC estimation algorithms, this algorithm makes up for the deficiencies of the conventional SOC estimation algorithms, such as slow convergence speed and low accuracy, improves the estimation accuracy of SOC, and can meet the needs of engineering applications. Among them, for the improved AEKF, a method of introducing a variable gain factor in the AEKF method can be adopted, and the variable gain factor is used to optimize the filtering gain matrix to achieve the purpose of quickly converging the SOC estimated value to the actual value and reducing the SOC estimation error. Through the power battery discharge experiment verification, the above SOC estimation algorithm can accurately estimate the SOC and achieve a good estimation effect.

[0053] As Figure 3 shown, in S400, calculate the consistency of the state of charge set and evaluate the accuracy rate of the result of the state of charge of the battery under the first working condition;

[0054] In this embodiment, the isolation forest algorithm is adopted to determine the anomaly score of the state of charge under each working condition; calculate the mean and standard deviation of all anomaly scores to determine the consistency of the state of charge set.

[0055] Specifically, let the one-dimensional data set X = [x1, x2,..., x i ,..., x n , where n is the number of data points, that is, the number of working conditions, and Xi represents the SOC value of the state of charge of the battery corresponding to the i-th working condition.

[0056] Construct an isolation forest model. The isolation forest algorithm will construct multiple isolation trees. An isolation tree is a binary tree, and the construction process is as follows: randomly select a feature from the data set; randomly select a splitting point within the value range of this feature; divide the data set into two parts according to the splitting point, and recursively continue the above operations on these two parts until each subset only has one data point or reaches the preset maximum depth of the tree.

[0057] Calculate the path length of each data point. For each data point in the data set, calculate its path length from the root node to the leaf node in each isolation tree, and finally take the average value of the path lengths in all trees. Let the path length of the data point x i in the j-th isolation tree be h ij , then its average path length E(h i ) is:

[0058]

[0059] Among them, T is the number of trees in the isolation forest.

[0060] Calculate the anomaly score. The anomaly score S(x i , n) of the data point x i is calculated using the following formula:

[0061]

[0062] Among them, c(n) represents a correction factor related to the size n of the data set; E(h i ) represents the average path length of the data point x i .

[0063] c(n) can be calculated using the following formula:

[0064]

[0065] Among them, H(n - 1) here is the harmonic series, which can be approximately expressed as H(n - 1) = ln(n - 1) + γ; γ ≈ 0.5772 is the Euler constant.

[0066] Judge consistency: Calculate the mean μ and standard deviation σ of all anomaly scores; determine the thresholds, the lower threshold T1 = μ - kσ; the upper threshold T2 = μ + kσ; where, k represents a multiple, usually taking 1, 2, or 3.

[0067] Determine the consistency based on the proportion of the number of data points that satisfy the threshold condition T1 ≤ S(x i , n) ≤ T2 in the total number of data points. When this proportion is greater than 95%, it can be considered that the consistency is good and can be used to evaluate the result accuracy of the state of charge of the battery under the first working condition; when the proportion is less than 95%, it is considered that the consistency does not meet the requirements, return S200, and re - derive the second working condition; of course, the determination of this proportion can be reasonably determined according to the continuous optimization of the model, and the situation of re - derivation should be minimized while ensuring a relatively high consistency.

[0068] Based on all anomaly scores and their mean and standard deviation, determine the consistent data set, and evaluate the result accuracy of the state of charge of the battery under the first working condition based on the average similarity between the anomaly score corresponding to the state of charge of the battery under the first working condition and the anomaly scores corresponding to other states of charge in the consistent data set.

[0069] Specifically, filter out the anomaly scores that meet the threshold conditions to form a consistent data set C.

[0070] Select a suitable similarity algorithm to determine the similarity between the anomaly score corresponding to the battery state of charge (SOC) in the first working condition and the anomaly scores corresponding to other SOCs in the consistent data set C. Since the SOC in this process is one-dimensional data, the reciprocal of the Euclidean distance can be used to determine the similarity.

[0071] Let the anomaly score corresponding to the SOC in the first working condition be S test , S test ∈C, and the Euclidean distance d(S te2t , S i ) = |S test - S i |. The similarity is calculated using the following formula:

[0072]

[0073] Then the result accuracy rate of the SOC in the first working condition is:

[0074]

[0075] where |C| represents the number of data points in the consistent data set C.

[0076] As Figure 4 shown, when the result accuracy rate of the SOC in the first working condition meets the preset requirements, based on the SOC and the result accuracy rate in the first working condition, calculate the state of health (SOH) of the battery in the first working condition.

[0077] In this embodiment, the preset requirement can be that the result accuracy rate is above 95% to ensure high-accuracy input data, making the SOH result calculation more accurate. Of course, this ratio can be determined based on the data obtained after several model runs to ensure a relatively high accuracy rate and minimize the need for re-derivation.

[0078] Utilize the result accuracy rate and introduce a correction function to correct the SOC in the first working condition, and calculate the SOH of the battery in the first working condition using the corrected SOC.

[0079] In this embodiment, SOC1(t) represents the SOC in the first working condition and can be calculated using the aforementioned SOC calculation formula.

[0080] The correction function can be:

[0081] SOC2(t) = λ × SOC1(t) + (1 - λ)SOC1(t) / w

[0082] where SOC1(t) represents the SOC in the first working condition; SOC2(t) represents the corrected SOC; w represents the result accuracy rate; λ represents the correction weight. According to the foregoing process, w is

[0083] The state of health (SOH) of the battery under the first working condition is calculated using the following formula:

[0084]

[0085] Wherein, SOC2(t) represents the corrected state of charge of the battery; I represents the current, and the integral of I over [0, t] represents the amount of electricity discharged by the battery, C0 represents the rated capacity of the battery at the time of factory shipment; SOH represents the state of health of the battery under the first working condition. Thus, the consistency, accuracy determination and correction of the state of charge under the first working condition have been carried out in the foregoing process, and the accuracy of SOC2(t) has been greatly improved with respect to the final application link, and further the accuracy of SOH has been improved.

[0086] A method and system for joint diagnosis of the state of charge and health of an in-vehicle power battery provided in this embodiment have different perspectives and improvement strategies for the accuracy of SOC and SOH. It completely breaks through the optimization of the conventional algorithm model itself, but adopts the method of derivative working conditions to simulate the fluctuation of battery data from the moment of obtaining the original battery data to the moment of final application of the calculation results of SOC and SOH, constructs a fluctuation scenario, while calculating SOC, calculates the accuracy of the result by the SOC consistency method, and corrects SOC based on the result accuracy rate to improve the accuracy of the calculation result of SOC with respect to the final application moment. Use the corrected SOC to calculate SOH to achieve joint estimation. Joint estimation is to integrate the state of charge and health state of the lithium battery in a specific way, aiming to obtain a better effect than single state prediction. This joint estimation method fully considers the internal chemical reactions during the battery operation process and can accurately simulate the dynamic response outside the battery. This method is closer to the actual situation and provides a new idea for predicting the health status of lithium batteries.

[0087] Embodiment 2

[0088] Different from Embodiment 1, when performing working condition derivation, the fluctuation law is selected using the basic data in the input battery initial parameter data to determine the mapping relationship used for working condition derivation; the basic data includes at least one of the battery model and the loaded vehicle model.

[0089] Specifically, in S206, all data are input into the trained autoencoder to obtain their latent representations, and their basic data are labeled; when calculating the similarity, after inputting the battery initial parameter data, first select the original data with the same basic data according to its basic data, and then perform subsequent similarity calculation according to the selected original data.

[0090] To ensure that the quantity of the original data meeting the above requirements meets the threshold quantity requirement, during selection, first select using all data types in the basic data that meet the criteria. When the quantity of data is less than the threshold quantity, gradually reduce the data types in the basic data for selection until the quantity of the selected original data meets the threshold quantity requirement. For example, first select with the same battery model, loading vehicle type, and data acquisition time period. If the selected original data is less than the threshold quantity, then select with the same battery model and loading vehicle type. If the quantity is still insufficient, then select with only the same battery model. If the quantity is still insufficient, then select with only the same loading vehicle type until the quantity of the selected original data meets the threshold quantity requirement. The specific method can be implemented by the operating condition derivation module.

[0091] A method and system for jointly diagnosing the state of charge and health of an in-vehicle power battery provided in this embodiment. Since the basic data can comprehensively represent the characteristic information of the battery itself and the detailed conditions of the battery application environment, using these basic data to preferentially select data with similar battery itself and application environment for similarity calculation can improve the correctness of the selection of the fluctuation law relationship. This method makes full use of the historical data of the battery and its operating environment, making the selected data better reflect the change trend in the actual situation, thereby significantly improving the accuracy and reliability of the fluctuation prediction. In addition, this method can also enhance the accuracy of the operating condition derivation because it ensures that the newly derived operating conditions are not only theoretically reasonable but also highly feasible and predictable in actual operation, which helps to optimize the design and implementation of the battery management system.

[0092] The above are only embodiments of the present invention. Common general knowledge such as specific structures and characteristics in the solution are not described in detail here. Those of ordinary skill in the art know all the common general knowledge in the technical field to which the invention belongs before the application date or the priority date, can know all the existing technologies in this field, and have the ability to apply the conventional experimental means before this date. Those of ordinary skill in the art can, under the inspiration given in this application, complete and implement this solution in combination with their own abilities. Some typical well-known structures or well-known methods should not be an obstacle for those of ordinary skill in the art to implement this application. It should be noted that for those skilled in the art, without departing from the structure of the present invention, several deformations and improvements can still be made, and these should also be regarded as the protection scope of the present invention, and these will not affect the implementation effect of the present invention and the practicality of the patent.

Claims

1. A method for jointly diagnosing the state of charge and health of an in-vehicle power battery, characterized in that, The method includes: Obtaining the initial battery parameter data as the first operating condition; Constructing an operating condition derivation model; using the first operating condition as the input, and deriving a number of second operating conditions by using this model to ensure that the derived second operating conditions cover a reasonable parameter range while maintaining the relevance to the first operating condition; Calculating the state of charge of the battery under the first operating condition and all the derived second operating conditions respectively to obtain a state of charge set; calculating the consistency of the state of charge set and evaluating the result accuracy of the state of charge of the battery under the first operating condition; When the result accuracy of the state of charge of the battery under the first operating condition meets the preset requirements, calculate the state of health of the battery under the first operating condition based on the state of charge of the battery and the result accuracy under the first operating condition.

2. The combined diagnosis method for the charge and health state of an in-vehicle power battery according to claim 1, characterized in that, The initial battery parameter data includes basic data and operating data, where the basic data includes the battery model and the loaded vehicle model; the operating data includes the data acquisition time period, current, voltage, temperature, remaining battery charge at the moment, relative maximum available capacity, and discharged battery charge at the moment.

3. A combined diagnosis method for the state of charge and health of an in-vehicle power battery according to claim 1, characterized in that The construction of the operating condition derivation model is as follows: obtaining a number of battery parameter data, using machine learning technology to learn the fluctuation law of the battery parameter data, establishing a mapping relationship between the battery parameter data and the fluctuation law, and using this mapping relationship for operating condition derivation.

4. A joint diagnosis method for the state of charge and health of an in-vehicle power battery according to claim 3, characterized in that, When performing operating condition derivation, select the fluctuation law by using the basic data in the input initial battery parameter data to determine the mapping relationship used for operating condition derivation; the basic data includes at least one of the battery model and the loaded vehicle model.

5. A method for jointly diagnosing the state of charge and health of an in-vehicle power battery according to claim 1, characterized in that, Adopt the isolation forest algorithm to determine the anomaly score of the state of charge of each operating condition; calculate the mean and standard deviation of all the anomaly scores to determine the consistency of the state of charge set.

6. The combined diagnosis method for the state of charge and health of an in-vehicle power battery according to claim 5, characterized in that, Determine the threshold value based on the mean and standard deviation of all the anomaly scores; the anomaly score is a data point, and the consistency is determined by the proportion of the number of data points that meet the threshold condition to the total number of data points.

7. A combined diagnosis method for the state of charge and health of an in-vehicle power battery according to claim 5, characterized in that, Based on all the anomaly scores and their mean and standard deviation, determine a consistent data set, and evaluate the result accuracy of the state of charge of the battery under the first operating condition by the average similarity between the state of charge of the battery under the first operating condition and the data points in the consistent data set.

8. The combined diagnosis method for the state of charge and health of an in-vehicle power battery according to claim 1, wherein Use the result accuracy and introduce a correction function to correct the state of charge of the battery under the first operating condition, and calculate the state of health of the battery under the first operating condition by using the corrected state of charge of the battery.

9. A method for jointly diagnosing the state of charge and health of an in-vehicle power battery according to claim 8, characterized in that, The correction function is: SOC2(t) = λ * SOC1(t) + (1 - λ)SOC1(t) / w where, SOC1(t) represents the state of charge of the battery under the first operating condition, SOC2(t) represents the corrected state of charge of the battery, w represents the result accuracy, and λ represents the correction weight; The state of health of the battery under the first operating condition is calculated by the following formula: where, SOC2(t) represents the corrected state of charge of the battery, I represents the current, the integral of I on [0, t] represents the discharged battery charge, C0 represents the rated capacity of the battery at the time of factory, and SOH represents the state of health of the battery.

10. A combined diagnosis system for the state of charge and health of in-vehicle power batteries, characterized in that, Execute any one of the on-vehicle power battery state of charge and state of health joint diagnosis methods described in claims 1-9. The system includes: A data acquisition module for obtaining the initial battery parameter data as the first operating condition; The operating condition derivation module is used to construct and run an operating condition derivation model. After receiving the first operating condition, it uses the first operating condition as input and derives several second operating conditions using the model, ensuring that the derived second operating conditions cover a reasonable parameter range while maintaining relevance to the first operating condition; The first calculation module is used to calculate the state of charge of the battery under the first operating condition and all the derived second operating conditions respectively to obtain a set of states of charge; calculate the consistency of the set of states of charge and evaluate the result accuracy of the state of charge of the battery under the first operating condition; The second calculation module is used to determine whether the result accuracy of the state of charge of the battery under the first operating condition meets the preset requirements. If it meets, it calculates the state of health of the battery under the first operating condition based on the state of charge of the battery and the result accuracy under the first operating condition.

Images