A method for predicting state of health of a battery

Abstract

This invention discloses a method for predicting battery health status. The method includes acquiring full lifecycle data of the battery and breaking it down into multiple cycles to obtain the occurrence time and stress influencing factors of each cycle, forming a set T that includes the occurrence time of each cycle; sorting the data in set T and calculating the time interval between adjacent data to obtain multiple time differences to form set T. diff ; set T and T diff The inputs are fed into the time series model for training to obtain model1, and a portion of the stress influence factors and set T are also fed into the time series model for training to obtain model2. temp model soc model dod model pass Using model1 and model2 temp model soc model dod model pass Calculate the predicted values ​​at different times to form a set Temp. p The invention calculates the SOH value of the battery based on the stress values ​​of the set. This method, which uses predicted stress values ​​as input and then calculates SOH, takes into account the influence of various stress factors on SOH, thus improving prediction accuracy.

Description

Technical Field

[0001] This invention relates to the field of battery health status prediction technology, and in particular to a method for predicting battery health status. Background Technology

[0002] Timely maintenance and replacement of lithium-ion battery systems are crucial for rapidly and accurately predicting state of health (SOH). SOH is a dynamic state parameter of a battery, represented by its available capacity compared to its initial conditions. Generally, SOH tends to decrease over the long term as the battery ages during its charge and discharge cycles due to chemical degradation. Accurate prediction of battery SOH is essential for timely replacement of depleted or weakened batteries to ensure they are in sufficiently good condition.

[0003] Industry-standard methods typically focus on the relationship between input parameters and target values, neglecting the crucial dimension of time. Furthermore, subsequent time predictions generally only utilize periodic time points, overlooking factors related to battery charging and discharging behavior. Other methods directly use State of Harmony (SOH) as input to find the relationship between SOH and the target value, but they also ignore the influence of other factors on SOH, resulting in poor accuracy. Summary of the Invention

[0004] The purpose of this invention is to provide a method for predicting battery health status with high accuracy in predicting battery SOH.

[0005] To achieve the above objectives, the technical solution adopted by the present invention is as follows:

[0006] A method for predicting battery health status, the prediction method comprising:

[0007] The battery's entire life cycle data is acquired and broken down into multiple cycles to obtain the time of occurrence and stress influencing factors of each cycle, forming a set T, which includes the time of occurrence of each cycle;

[0008] The data in the set T is sorted, and the time interval between adjacent data is calculated to obtain multiple time differences, which are then used to form the set T. diff ;

[0009] Let the set T and the set T diff The stress influence factors and the set T are input into the time series model for training to obtain model1. Additionally, a portion of the stress influence factors and the set T are input into the time series model for training to obtain model2. temp model soc model dod model pass ;

[0010] Using the aforementioned model1 and model temp model soc model dod model pass Calculate the predicted values ​​at different times to form a set Temp. p , The SOH value of the battery is calculated based on the stress value of the set.

[0011] Furthermore, the method of splitting into multiple cyclic processes includes: first sorting the full life cycle data according to the battery time, and then splitting the full life cycle data according to the battery's SOC.

[0012] Furthermore, the stress influence factors include an average temperature factor, an average SOC factor, a discharge depth factor, and an experience time factor. The average temperature factor is the average temperature detected by all temperature probes during the cycle. The average SOC factor is the average SOC at different times during the cycle. The discharge depth factor is the difference between the discharge start SOC and the discharge end SOC during the discharge process in the cycle. The experience time factor is the time elapsed during the cycle.

[0013] Furthermore, the prediction method further includes: inputting the set Temp, composed of the average temperature factors, and the set T into a time series model for training to obtain the model. temp .

[0014] Furthermore, the prediction method further includes: setting the average SOC factor into a set of SOCs. mean The set T is input into the time series model for training to obtain the model. soc .

[0015] Furthermore, the prediction method further includes: a set of Discharge Depth Factors (DODs) j The set T is input into the time series model for training to obtain the model. dod .

[0016] Furthermore, the prediction method also includes: a set of time factors composed of the experienced time factors. pass The set T is input into the time series model for training to obtain the model. pass .

[0017] Furthermore, the prediction method further includes: in the set T diff Then, calculate the set T. diffThe average value of the data is then added to the set T. diff The first term is used to combine the set T and the set T. diff The input is fed into the time series model for training to obtain model1.

[0018] Furthermore, the method for calculating the predicted values ​​at different times includes: using model1 to make predictions to obtain the sequence. Use the model respectively temp model soc model dod model pass calculate The predicted value at time, thus obtaining the corresponding Temp. p , tn is the time when the prediction begins.

[0019] Furthermore, the method for calculating the SOH value of the battery includes: fitting the battery's degradation curve based on experimental data from cyclic charge-discharge degradation experiments, and utilizing the set Temp... p , The stress value is used to correct the attenuation curve, and the SOH value of the battery is calculated using the corrected attenuation curve.

[0020] The advantages of this invention are: by using the method of first predicting the stress value as input and then calculating the SOH, the influence of various stress factors on the SOH is taken into account, which can improve the prediction accuracy and make it more interpretable. Attached Figure Description

[0021] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0022] Figure 1 This is a flowchart of a battery health state prediction method provided in an exemplary embodiment of the present invention. Detailed Implementation

[0023] To enable those skilled in the art to better understand the present invention and its objectives, technical solutions, and advantages, the technical solutions in the embodiments of the present invention are clearly and completely described below with reference to specific embodiments and accompanying drawings. It should be noted that implementations not illustrated or described in the accompanying drawings are forms known to those skilled in the art. Furthermore, while this document provides examples of parameters containing specific values, it should be understood that the parameters need not be exactly equal to the corresponding values, but can approximate the corresponding values ​​within acceptable error tolerances or design constraints. Obviously, the described embodiments are merely some embodiments of the present invention, and not all embodiments. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without inventive effort should fall within the scope of protection of the present invention. In addition, the terms "comprising" and "having," and any variations thereof, in the specification and claims of this invention are intended to cover non-exclusive inclusion. For example, a process, method, apparatus, product, or device that includes a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to these processes, methods, products, or devices.

[0024] In one embodiment of the present invention, a method for predicting battery health status is provided, such as... Figure 1 As shown, this prediction method includes: acquiring the battery's entire life cycle data and breaking it down into multiple cyclic processes to obtain the occurrence time and stress influencing factors of each cycle, forming a set T that includes the occurrence time of each cycle; sorting the data in set T and calculating the time interval between adjacent data to obtain multiple time differences, which together form set T. diff ; set T and set T diff The inputs are fed into the time series model for training to obtain model1. Additionally, a subset of stress influence factors and the set T are fed into the time series model for training to obtain model2. temp model soc model dod model pass Using model1 and model2 temp model soc model dod model pass Calculate the predicted values ​​at different times to form a set Temp. p , The SOH value of the battery is calculated based on the stress value of the set.

[0025] First, obtain the full life cycle data of the battery, sort it by time, and then split the data into different cycling processes and corresponding stress impact factors according to the battery SOC. In this embodiment, the set of sequences of the time when each cycle occurs is T; the stress impact factors include the average temperature factor temperature avg , the average SOC factor SOC avg , the depth of discharge factor DOD, and the elapsed time factor time pass . Among them, the average temperature factor is the average value of the temperatures detected by all temperature probes during the cycling process, the average SOC factor is the average value of SOC at different times during the cycling process, the depth of discharge factor is the SOC at the start of discharge minus the SOC at the end of discharge during the discharge process of the cycling process, and the elapsed time factor is the time elapsed during the cycling process.

[0026] Secondly, sort the data in the set T, that is, the time when the cycle occurs, and calculate the time intervals between adjacent data to obtain multiple time differences, so as to obtain a set T including multiple time difference data diff , after forming the set T diff , calculate the average value of the data in the set T diff , and add this average value to the set T diff as the first item.

[0027] Then, input the set T and the set T diff into the time series model for training to obtain the model model1; input the set Temp composed of the average temperature factor and the set T into the time series model for training to obtain the model model temp ; input the set SOC mean composed of the average SOC factor and the set T into the time series model for training to obtain the model model soc ; input the set DOD j composed of the depth of discharge factor and the set T into the time series model for training to obtain the model model dod ; input the set time pass composed of the elapsed time factor and the set T into the time series model for training to obtain the model model pass . It should be noted that the time series model in this embodiment can be a Prophet model or other models that meet the requirements.

[0028] Next, use model1 for prediction to obtain a sequence to obtain multiple moments: where tn is the starting time of prediction; then use model tempCalculate the predicted values ​​at multiple times above to obtain the corresponding set Temp. p Similarly, using the model soc model dod model pass The corresponding set is obtained by calculating the predicted values ​​at multiple times above.

[0029] Finally, the battery degradation curve was fitted based on the experimental data from the cyclic charge-discharge degradation experiment, and the Temp ensemble was used. p , The stress value is used to correct the degradation curve, and the SOH value of the battery is then calculated using the corrected degradation curve. Specifically, n batteries of the same model are subjected to cyclic charge-discharge degradation experiments, where n is an integer greater than 3. The degradation curve of the battery model is fitted using the experimental data from the cyclic charge-discharge degradation experiments and used as the degradation reference curve for that model of battery. The degradation curve contains three coefficients, namely α... sei β sei f d,1 Several batteries of this model were subjected to cyclic degradation experiments at different temperatures, average SOCs, and depths of discharge. The experimental data from the cyclic degradation experiments were used to determine f. d,1 The corresponding temperature correction, average SOC correction, depth of discharge correction, and duration correction are then used to adjust f. d,1 The corrected f is obtained by making corrections. d,1 And then use the corrected f d,1 The degradation baseline curve of the battery model is corrected to obtain the corrected degradation curve of the battery model; using the usage data of the battery to be estimated and the corrected degradation curve of the battery model, the degradation amount of the battery to be estimated in each cycle is estimated, thereby estimating the life and SOH of the battery to be estimated.

[0030] The above description is merely a preferred embodiment of the present invention and does not limit its patent scope. Any equivalent structural or procedural transformations made based on the description and drawings of the present invention, whether directly or indirectly applied to other related technical fields, are similarly included within the patent protection scope of the present invention.

Claims

1. A method for predicting battery health status, characterized in that, The prediction method includes: Acquire the battery's entire lifecycle data and break it down into multiple cycles to obtain the timing and stress influencing factors of each cycle, forming a set. The stress influence factors include average temperature factor, average SOC factor, depth of discharge factor, and duration factor. Including the time when each cycle occurs; For the set The data is sorted, and the time intervals between adjacent data are calculated to obtain multiple time differences to form a set. ; The set and the set Input is fed into a time series model for training to obtain the model. Furthermore, a portion of the stress influence factor and the set Input is fed into a time series model for training to obtain the model. , , , This includes: a set of average temperature factors. Temp and set Input into the time series model, and train to obtain The set of average SOC factors SOC mean and set Input into the time series model, and train to obtain The set of discharge depth factors DOD j and set Input into the time series model, and train to obtain The set consisting of time factors. time pass and set Input into the time series model, and train to obtain The time series model is the Prophet model. Using the model , , , , Calculate the predicted values ​​at different times, including: using Perform prediction to obtain the sequence , , ... Thus, multiple moments are obtained. , , ... , wherein The time to begin forecasting; using the model Calculate the predicted values ​​at the above multiple time points to obtain the corresponding set. Using the model Calculate the predicted values ​​at the above multiple time points to obtain the corresponding set. Using the model Calculate the predicted values ​​at the above multiple time points to obtain the corresponding set. Using the model Calculate the predicted values ​​at the above multiple time points to obtain the corresponding set. ; Thus forming a set , , , The battery's state of harmonics (SOH) is calculated based on the stress values ​​of the set, including: fitting the battery's degradation curve based on experimental data from cyclic charge-discharge degradation experiments, and utilizing the set... , , , The stress value is used to correct the attenuation curve, and the SOH value of the battery is calculated using the corrected attenuation curve.

2. The method for predicting battery health status according to claim 1, characterized in that, The method of splitting into multiple cyclic processes includes: first sorting the full life cycle data according to the battery time, and then splitting the full life cycle data according to the battery's SOC.

3. The method for predicting battery health status according to claim 1, characterized in that, The average temperature factor is the average temperature detected by all temperature probes during the cycle; the average SOC factor is the average SOC at different times during the cycle; the discharge depth factor is the difference between the discharge start SOC and the discharge end SOC during the discharge process in the cycle; and the elapsed time factor is the time elapsed during the cycle.

4. The method for predicting battery health status according to claim 1, characterized in that, The prediction method further includes: in the set Then, the set is calculated. The average value of the data is then added to the set. The middle is used as the first item, thereby making the set and the set Input is fed into a time series model for training to obtain the model. .

Images