A method, system and electronic device for monitoring the health of a power battery throughout its life cycle

Abstract

The present invention proposes a method, system and electronic device for monitoring the health of a power battery throughout its life cycle, which divides the battery's life cycle into three stages: degradation stage, potential dive interval and critical stage. Different monitoring methods are used according to the stage the battery is in: in the degradation stage, multiple-step advance prediction of battery capacity is performed based on the battery's historical capacity sequence; in the potential dive interval, advance prediction of battery capacity is performed based on the capacity sequence in the potential dive interval, and the number of advance prediction steps is greater than the number of steps in the degradation stage; in the critical stage, remaining life prediction is performed throughout the entire process. The present invention divides the battery's life cycle into multiple stages based on the battery capacity degradation trend, and then assigns different tasks to each stage, so that each stage of the battery can obtain a high-reliability health status representation, and achieves more detailed and targeted management of each stage of the battery, making up for the inaccurate prediction of the remaining life of the battery in the early and mid-term.

Description

Technical Field

[0001] The present invention belongs to the technical field of power batteries for new energy vehicles, and specifically relates to health management technology for power batteries. Background Art

[0002] As the penetration rate of new energy vehicles gradually increases, there are more and more new energy vehicle accidents caused by battery failures in the market. How to manage the health of new energy vehicle power batteries throughout their life cycle has also received more and more widespread attention. An important part of battery health management is the remaining useful life (RUL) prediction. IEEE Standard 1188-1996 stipulates that the battery should be replaced when the capacity of the battery drops to 80% of the factory capacity. Therefore, the remaining life of the battery generally refers to the time from the current moment that the battery can be used before the capacity is reduced to 80% of the factory capacity. This is the full life cycle of the battery on the vehicle side, and the full life cycle of the power battery referred to below in this invention is this life cycle.

[0003] At present, most of the literature on battery health management also focuses on the prediction of battery remaining life based on capacity, such as: patent document CN202110206262.3 (Lithium battery remaining life prediction method based on multi-core GPR algorithm of MFF), CN201910238129.9 (Lithium-ion battery remaining life prediction method based on gray wolf pack optimized LSTM network), etc. Remaining life prediction plays a vital role in the health management of batteries in the later stage of battery degradation, but it is difficult to support the health management of batteries throughout their life cycle by relying solely on remaining life prediction. This is because the degradation trend of the battery system throughout its life cycle is generally time-varying, and it is difficult to describe the time-varying degradation trend through a single model or a single means. In the early use of power batteries, the degradation is relatively slow; but when the battery capacity declines to a certain extent, the battery will accelerate degradation, and the battery capacity will "dive" and the battery capacity will decline to battery failure in a relatively short period of time, such as Figure 1 As shown. In actual use, since the diving inflection point is difficult to predict, the prediction results of the remaining life of the power battery in the early and mid-term often deviate greatly from the actual remaining life value, and the prediction results are difficult to support the issuance of vehicle maintenance strategies. Therefore, how to set a strategy to obtain a highly reliable health status representation of the battery at each stage is of great significance to the health management of the battery throughout its life cycle. Summary of the invention

[0004] In view of the problems existing in the prior art, the present invention provides a battery health monitoring method, electronic device and automobile for the whole life cycle, which performs health monitoring of the battery, a time-varying system, throughout its life cycle, divides the whole life cycle of the battery into multiple stages based on the battery capacity degradation trend, and then assigns different tasks to each stage, so that the health status of the battery in each stage can be represented with high reliability, and more detailed and targeted management of the battery in each stage can be achieved, thus making up for the inaccurate prediction of the remaining life of the battery in the early and mid-term.

[0005] The technical solution of the present invention is as follows:

[0006] In the first aspect, the present invention provides a method for monitoring the health of a power battery throughout its life cycle. On the one hand, for the battery, a time-varying system, the battery's life cycle is divided into three stages based on the battery capacity degradation trend, and different tasks are assigned to each stage to ensure that the battery can obtain a high-reliability health status representation at each stage, thereby achieving full life cycle health monitoring of the battery. On the other hand, in order to solve the problem that the battery capacity drop point is difficult to predict in advance, a clustering model is established based on the user characteristics of the battery, and the drop points of similar batteries in the cloud are counted as a priori estimates of the current battery drop point, so as to obtain the potential drop interval of the current battery capacity.

[0007] The method specifically divides the entire life cycle of the battery into three stages: degradation stage, potential diving interval and critical stage. Different management methods are adopted according to the stage of the battery:

[0008] In the degradation stage, a multi-step advance forecast of battery capacity is performed based on the battery historical capacity sequence;

[0009] In the potential dive interval, the battery capacity is predicted in advance based on the capacity sequence in the potential dive interval. The number of advance prediction steps should be gradually increased, gradually transitioning to the remaining life prediction, and connecting with the third stage.

[0010] During the critical stage, remaining life prediction is performed throughout the entire process.

[0011] Furthermore, the potential diving interval described here is obtained by counting the diving points of similar batteries in the cloud as a priori estimates of the current battery diving point.

[0012] Specifically, in one embodiment of the present invention, the potential diving interval is obtained by: first, among similar batteries in the cloud, the user's usual driving speed, driving style, electrical appliance information, common SOC driving interval, and common SOC charging and discharging interval features are extracted from the driving behavior and charging habits of the user to whom the battery belongs, and a clustering model is established for the extracted features; the above features are extracted from the current battery, and the user's category is predicted through the clustering model; then, the distribution of the number of charging and discharging cycles when the capacity of the battery of this category dives is counted, and the confidence interval of the distribution is obtained as the potential diving interval of the current battery.

[0013] Furthermore, in the degradation stage, outliers are first removed from the battery historical capacity sequence, and then a state transfer equation and an observation equation are established based on the battery historical capacity sequence, and finally a multi-step advance prediction of the battery capacity is performed.

[0014] Furthermore, in the potential diving interval, outliers are first removed from the battery capacity sequence in the potential diving interval, and then a state transfer equation and an observation equation are established based on the battery capacity sequence in the potential diving interval, and finally a multi-step advance prediction of the battery capacity is performed.

[0015] Furthermore, in the critical stage, outliers are first removed from the potential battery dive interval and the critical stage capacity sequence, and then a state transfer equation and an observation equation are established based on the potential dive interval and the critical stage battery capacity sequence, and finally the remaining life of the whole process is predicted according to the following formula:

[0016] rul k =inf(l:f(l+k)≥γ)

[0017] In the formula, inf(l) represents the lower limit of the number of charge and discharge cycles l; γ represents the failure threshold of the discharge capacity, which is 80% of the initial capacity of the battery; rul k Indicates the remaining life of the battery after the kth charge and discharge cycle.

[0018] In a second aspect, the present invention provides a power battery full life cycle health monitoring system, the system includes three management modules for the battery full life cycle, namely, a degradation stage management module, a potential diving interval management module and a critical stage management module. The three modules adopt different management methods according to the stage of the battery.

[0019] The degradation stage management module is used to perform multi-step advance prediction of battery capacity based on the battery historical capacity sequence during the degradation stage.

[0020] The potential dive interval management module is used to perform advance forecasting of battery capacity in the potential dive interval based on the capacity sequence in the potential dive interval, and the number of advance forecasting steps is greater than the number of steps in the degradation stage.

[0021] The key stage management module is used to predict the remaining life throughout the key stage.

[0022] In a third aspect, the present invention provides an electronic device, comprising:

[0023] one or more processors;

[0024] A storage device is used to store one or more programs. When the one or more programs are executed by the one or more processors, the electronic device implements the power battery full life cycle health monitoring method described in the first aspect above.

[0025] By adopting the above technical solution, the present invention has at least the following advantages:

[0026] 1. The present invention divides the whole life cycle of the battery into multiple stages, specifies different tasks for each stage, and makes advance multi-step predictions of the battery capacity based on the battery historical capacity sequence in the degradation stage, makes advance predictions of the battery capacity based on the capacity sequence in the potential dive interval, and makes full remaining life predictions in the critical stage. Compared with existing solutions such as simple remaining life prediction, the health monitoring method for each stage of the battery proposed by the present invention is more detailed and targeted, and it is easier to realize the health monitoring of the battery throughout its life cycle, making up for the inaccurate prediction of the remaining life of the battery in the early and mid-term.

[0027] 2. In obtaining the battery capacity diving point, the present invention obtains the potential diving range of the current battery by counting the diving points of similar batteries in the cloud as the prior estimation of the current battery diving point. Compared with the prior art solution of obtaining the single-point prediction result of the diving point through mechanism analysis, the present invention gets rid of the complex battery mechanism analysis. The present invention obtains the prior estimation range of the current battery diving point in a data-driven manner, rather than the single-point prediction result. In this way, the expression of uncertainty makes the diving point occurrence point more credible and more fault-tolerant.

[0028] 3. The present invention can make full use of the national standard and enterprise standard data uploaded by the vehicle to the cloud, and is easy to deploy and implement in the cloud. BRIEF DESCRIPTION OF THE DRAWINGS

[0029] Figure 1 It is a schematic diagram of the capacity diving inflection point of the power battery;

[0030] Figure 2 It is a schematic diagram of dividing the whole life cycle of a battery into three stages according to the present invention;

[0031] Figure 3 It is a flowchart of the power battery full life cycle health monitoring method of the present invention.

[0032] It should be noted that the above drawings are used to better understand the present solution and do not constitute a limitation of the present application. DETAILED DESCRIPTION

[0033] The following will illustrate the implementation of the present application through specific examples, and those skilled in the art can easily understand other advantages and effects of the present application from the content disclosed in this specification. The present application can also be implemented or applied through other different specific implementations, and the details in this specification can also be modified or changed in various ways based on different viewpoints and applications without departing from the spirit of the present application. It should be noted that the following embodiments and the features, steps, etc. in the embodiments can be combined with each other without conflict.

[0034] like Figure 2 As shown, in the present invention, battery life cycle health monitoring divides the battery life cycle into three stages, namely, degradation stage, potential diving interval and critical stage.

[0035] Embodiment 1:

[0036] This embodiment is a method for performing battery health monitoring throughout its life cycle based on the above three stages, and its process is as follows: Figure 3 shown.

[0037] First, after the current charge and discharge cycle of the power battery is completed, it is necessary to determine whether the current charge and discharge cycle has entered a potential diving range.

[0038] Since the capacity drop point of the power battery is difficult to predict in advance during the degradation process, the present invention innovatively proposes to divide the potential drop interval, and obtains the potential drop interval by counting the drop points of similar batteries in the cloud as the prior estimate of the current battery drop point.

[0039] In one embodiment of the present invention, the potential dive interval can be obtained by using the following calculation process:

[0040] 1. First, among similar batteries in the cloud, the user's usual driving speed, driving style, electrical appliance information, common SOC driving range, common SOC charging and discharging range and other features are extracted from the driving behavior and charging habits of the battery user, and the extracted features are clustered based on the Gaussian mixture model to obtain a clustering model.

[0041] 2. Then extract the above features from the current battery and predict the user's category through the established clustering model.

[0042] 3. Finally, the distribution of the number of charge and discharge cycles when the capacity of this type of battery drops is counted in the cloud, and the 95% confidence interval of the distribution is obtained, which can be used as the potential drop interval of the current battery.

[0043] The above potential diving range is obtained by using the data of similar batteries in the cloud to obtain the prior estimated range of the current battery diving point in a data-driven way, rather than a single point prediction result. The credibility and fault tolerance of the diving point can be improved by expressing uncertainty. In addition, the national standard and enterprise standard data uploaded by the vehicle to the cloud can be fully utilized, which is easy to deploy and implement in the cloud.

[0044] When it is determined that the current charge and discharge cycle is not in the potential diving interval and is less than the lower limit of the potential diving interval, it means that the current battery is in the degradation stage. It is managed according to the degradation stage method and the battery capacity is predicted multiple steps ahead based on the battery historical capacity sequence.

[0045] When it is determined that the current charge and discharge cycle is greater than the lower limit of the potential dive interval, it is necessary to further determine whether it has entered a critical stage, that is, whether it has exceeded its upper limit. If it has not entered the critical stage, it is managed according to the method of the potential dive interval, and the battery capacity is advanced predicted based on the capacity sequence in the potential dive interval, and the number of advance prediction steps is greater than the number of steps in the degradation stage.

[0046] If it exceeds the upper limit, it means that it has entered the critical stage and the remaining life prediction will be carried out throughout the process.

[0047] 1. Degradation stage: For a brand new battery, when it starts running, the system enters a stable working state, and the period before the capacity dive is the degradation stage. Since the capacity dive point and the subsequent accelerated degradation trend cannot be predicted at this stage, and as the number of advance prediction steps increases, the prediction error gradually accumulates, and the deviation between the predicted value and the actual capacity value gradually increases, it is difficult to accurately predict the remaining life at this stage. However, according to the characteristic that the battery capacity decline will not occur suddenly, it is relatively reliable to make medium- and short-term capacity forecasts based on the existing degradation trend; in addition, the early stage of battery decline is still far from the end of life, and the medium- and short-term capacity forecast can provide sufficient health information to ensure that the battery decline is controllable. In summary, the task set in the degradation stage is mainly to make multi-step advance predictions of battery capacity.

[0048] In one embodiment of the present invention, the process of advanced prediction of battery capacity in the degradation stage is as follows:

[0049] 1. Preprocess the battery historical capacity series, mainly to remove outliers.

[0050] 2. Establish the state transfer equation and observation equation based on the empirical model of the battery degradation process, as follows:

[0051]

[0052] Among them, w k represents the process noise, vk represents observation noise, k=1,2,…represents the cycle of battery charge and discharge.

[0053] 3. Set the current charge and discharge cycle k as the prediction starting point, and use the particle filter algorithm to track the state of the historical capacity sequence to determine the unknown parameters in the state transfer equation.

[0054] 4. Initialize the particle filter algorithm, parameters include: number of particles N, process noise w k covariance R, observation noise v k The covariance Q of .

[0055] 5. Advanced capacity prediction based on particle filter algorithm.

[0056] 6. Set the number of advance forecast steps to m, repeat the above steps in sequence, and finally get the capacity forecast value C of m steps ahead k+1 , C k+2 ,…,C k+m .

[0057] 2. Potential Diving Range:

[0058] The task set for the potential dive interval is also to make multi-step advance predictions of battery capacity. Since the capacity degradation trend of the battery has accelerated after entering the potential dive interval, the degradation trend of the capacity sequence in the degradation stage is no longer of reference significance. Therefore, only the capacity sequence in the potential dive interval is used for advance capacity prediction. In addition, the battery entering the potential dive interval means that it is closer to the end of the battery life. Therefore, the number of prediction steps should be gradually increased during the advance prediction, gradually transitioning to the remaining life prediction.

[0059] In one embodiment of the present invention, the process of advanced prediction of battery capacity in a potential drop interval is as follows:

[0060] 1. Preprocess the capacity sequence of the battery's potential diving range, mainly to remove outliers.

[0061] 2. According to the empirical model of the battery degradation process, the state transfer equation and observation equation are established as follows:

[0062]

[0063] Among them, w k represents the process noise, v k represents observation noise, k=1,2,…represents the cycle of battery charge and discharge.

[0064] 3. Set the current charge and discharge cycle k as the prediction starting point, and use the particle filter algorithm to track the state of the historical capacity sequence to determine the unknown parameters in the state transfer equation.

[0065] 4. Initialize the particle filter algorithm, parameters include: number of particles N, process noise w k covariance R, observation noise v k The covariance Q of .

[0066] 5. Advanced capacity prediction based on particle filter algorithm.

[0067] 6. Set the initial advance forecast step number to m0, and repeat the above steps in sequence to obtain

[0068] 7. After the current charge-discharge cycle enters k+1 from k, the number of advance prediction steps becomes m0+1, and the above steps are repeated in sequence. After that, the current charge-discharge cycle increases, and the number of advance prediction steps also gradually increases by 1. Here, the initial number of advance prediction steps m0 is not necessarily set to a positive integer, for example, it can be set to -10, then the number of prediction steps is gradually increased, which means that the prediction will not start until the potential dive interval has experienced more than 10 discharge cycles, because the sequence is too short to see the trend and the prediction is inaccurate.

[0069] 3. Key Stages:

[0070] When the battery enters this stage, that is, after crossing the battery capacity diving point, the capacity degradation trend is greatly accelerated and reaches the failure threshold in a short time. Since the degradation trend is basically fixed after entering this stage, the remaining life prediction should be carried out throughout this stage.

[0071] The remaining life prediction rule proposed by the present invention is as follows:

[0072] rul k =inf(l:f(l+k)≥γ)

[0073] In the formula, inf(l) represents the lower limit of the number of charge and discharge cycles l; γ represents the failure threshold of the discharge capacity, which is 80% of the initial capacity of the battery; rul k Indicates the remaining life of the battery under the kth charge and discharge cycle (unit: number of charge and discharge cycles).

[0074] In one embodiment of the present invention, the process of predicting the remaining battery life at a critical stage is as follows:

[0075] 1. When the current charge and discharge cycle of the battery enters a critical stage, determine the first prediction time (First Predicting Time, FPT).

[0076] 2. Pre-process the capacity sequence of the battery's potential diving range and key stages, mainly to remove outliers.

[0077] 3. According to the empirical model of the battery degradation process, the state transfer equation and observation equation are established as follows:

[0078]

[0079] Among them, w k represents the process noise, v k represents observation noise, k=1,2,…represents the cycle of battery charge and discharge.

[0080] 4. Set the current charge and discharge cycle k as the prediction starting point, and use the particle filter algorithm to track the state of the historical capacity sequence of the battery's potential diving interval and key stages, so as to determine the unknown parameters in the state transfer equation.

[0081] 5. Initialize the particle filter algorithm, parameters include: number of particles N, process noise w k covariance R, observation noise v k The covariance Q of .

[0082] 6. Advanced capacity prediction based on particle filter algorithm.

[0083] 7. Determine the capacity advance forecast value C k+1 Whether the battery failure threshold is reached, that is, 80% of the initial capacity. If it is not reached, the above steps are repeated in sequence. k+n If the battery failure threshold is exceeded, the remaining life of the battery under the current charge and discharge cycle is n-1 charge and discharge cycles.

[0084] In one embodiment of the present invention, the advance prediction of capacity based on the particle filter algorithm used in the above stages can be performed in the following manner, and the specific steps include:

[0085] a) Update particles, based on N particles of discharge cycle k The prior values ​​of N particles in discharge cycle k+1 are obtained by updating the state transfer equation

[0086] b) Combine the measurement equation and likelihood function to calculate the weight set of N particles' prior values ​​for discharge cycle k+1 Normalize the particle weights to get the weight set

[0087] c) Using weight sets Resample the particles to obtain the resampled particle set

[0088] d) Calculate the state estimate of battery capacity and recorded in the sequence.

[0089] Embodiment 2:

[0090] This embodiment is a power battery full life cycle health monitoring system, which is used to implement the method described in the above embodiment. The system includes three management modules for the battery full life cycle, namely, degradation stage management module, potential diving interval management module and key stage management module. The three modules adopt different management methods according to the stage of the battery.

[0091] The degradation stage management module is used to perform multi-step ahead battery capacity forecasting based on the battery historical capacity sequence in the degradation stage. Specifically, the outliers of the battery historical capacity sequence are first removed, and then the state transfer equation and observation equation are established based on the battery historical capacity sequence, and finally the battery capacity is forecasted multi-step ahead.

[0092] The potential dive interval management module is used to perform advance forecasting of battery capacity in the potential dive interval based on the capacity sequence in the potential dive interval, and the number of advance forecasting steps is greater than the number of steps in the degradation stage. Specifically, outliers are first removed from the battery potential dive interval capacity sequence, and then a state transfer equation and an observation equation are established based on the potential dive interval battery capacity sequence, and finally a multi-step advance forecast of battery capacity is performed.

[0093] The key stage management module is used to predict the remaining life throughout the key stage. Specifically, outliers are first removed from the potential battery drop interval and the key stage capacity sequence, and then the state transfer equation and observation equation are established based on the potential drop interval and the key stage battery capacity sequence. Finally, the remaining life is predicted throughout the process according to the following formula:

[0094] rul k =inf(l:f(l+k)≥γ)

[0095] In the formula, inf(l) represents the lower limit of the number of charge and discharge cycles l; γ represents the failure threshold of the discharge capacity, which is 80% of the initial capacity of the battery; rul k Indicates the remaining life of the battery after the kth charge and discharge cycle.

[0096] The above is only a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in the technical field, several improvements and modifications can be made without departing from the principle of the present invention. These improvements and modifications should be regarded as the scope of protection of the present invention.

Claims

1. A method for monitoring the health of a power battery throughout its life cycle, characterized in that: The method divides the whole life cycle of the battery into three stages: degradation stage, potential diving interval and critical stage. Different treatment methods are adopted according to the stage of the battery: In the degradation stage, a multi-step advance prediction of the battery capacity is performed based on the historical capacity sequence of the current battery; specifically, the outliers of the historical capacity sequence of the battery are first removed, and then a state transfer equation and an observation equation are established based on the historical capacity sequence of the battery, and a state estimation value of the battery capacity, i.e., a capacity prediction value, is calculated based on a particle filter algorithm, and the number of advance prediction steps is set, and the capacity prediction value of the advance number of steps is calculated, and finally, a multi-step advance prediction of the battery capacity is performed; In the potential dive interval, the battery capacity is predicted in advance based on the capacity sequence in the potential dive interval. As the current charge and discharge cycle increases, the number of advance prediction steps also gradually increases, gradually transitioning to the remaining life prediction; During the critical stages, remaining life prediction is performed throughout the process.

2. The method for monitoring the health of a power battery throughout its life cycle according to claim 1, characterized in that: The potential diving interval is obtained by collecting diving point data of similar batteries in the cloud as a priori estimation of the current battery diving point.

3. The method for monitoring the health of a power battery throughout its life cycle according to claim 2, characterized in that: The potential diving interval is obtained by: firstly, extracting the user's usual driving speed, driving style, electrical appliance information, common SOC driving interval, and common SOC charging and discharging interval features from the driving behavior and charging habits of the user to whom the battery belongs among similar batteries in the cloud, and establishing a clustering model for the extracted features; extracting the above features from the current battery, and predicting the category to which the user belongs through the clustering model; then counting the distribution of the number of charging and discharging cycles when the capacity of the battery of this category dives, and obtaining the confidence interval of the distribution as the potential diving interval of the current battery.

4. The method for monitoring the health of a power battery throughout its life cycle according to claim 1, characterized in that: The battery capacity is predicted in advance during the degradation stage, including: (1) Preprocess the battery historical capacity series to remove outliers; (2) Based on the empirical model of the battery degradation process, the state transfer equation and observation equation are established as follows: Among them, w k represents the process noise, v k represents the observation noise, k = 1, 2, ... represents the cycle of battery charge and discharge; (3) Assuming the current charge-discharge cycle k as the prediction starting point, the particle filter algorithm is used to track the state of the historical capacity sequence to determine the unknown parameters in the state transfer equation; (4) Initialize the particle filter algorithm. The parameters include: number of particles N, process noise w k covariance R, observation noise v k The covariance Q of (5) Based on the particle filter algorithm, the capacity is predicted in advance and the state estimation value of the battery capacity, i.e., the capacity prediction value C, is calculated. k+1 ; (6) Set the number of advance forecast steps to m, repeat the above steps in sequence, and finally obtain the capacity forecast value C of m steps ahead k+1、 C k+2、 …、C k+m .

5. The method for monitoring the health of a power battery throughout its life cycle according to claim 1, 2 or 3, characterized in that: In the potential diving interval, the outliers of the battery capacity sequence in the potential diving interval are first removed, and then the state transfer equation and observation equation are established based on the battery capacity sequence in the potential diving interval, and finally the battery capacity is predicted multiple steps ahead.

6. The method for monitoring the health of a power battery throughout its life cycle according to claim 5, characterized in that: Conduct multi-step advance forecast of battery capacity in the potential dive interval, including: (1) Preprocess the capacity sequence of the battery's potential drop interval to remove outliers; (2) Based on the empirical model of the battery degradation process, the state transfer equation and observation equation are established as follows: Among them, w k represents the process noise, v k represents the observation noise, k = 1, 2, ... represents the cycle of battery charge and discharge; (3) Assuming the current charge-discharge cycle k as the prediction starting point, the particle filter algorithm is used to track the state of the historical capacity sequence to determine the unknown parameters in the state transfer equation; (4) Initialize the particle filter algorithm. The parameters include: number of particles N, process noise w k covariance R, observation noise v k The covariance Q of (5) Based on the particle filter algorithm, the capacity is predicted in advance and the state estimation value of the battery capacity, i.e., the capacity prediction value C, is calculated. k+1 ; (6) Set the initial advance forecast step number to m0, and repeat the above steps in sequence to obtain the capacity forecast value (7) After the current charge-discharge cycle enters k+1 from k, the number of advance prediction steps becomes m0+1. The above steps are repeated sequentially, after which the current charge and discharge cycle increases and the number of advance prediction steps gradually increases by +1.

7. The method for monitoring the health of a power battery throughout its life cycle according to claim 1, 2 or 3, characterized in that: In the critical stage, outliers are first removed from the potential battery drop interval and the critical stage capacity sequence, and then the state transfer equation and observation equation are established based on the potential drop interval and the critical stage battery capacity sequence. Finally, the remaining life of the whole process is predicted according to the following formula: NUMBER k =inf(l:f(l+k)≥γ) In the formula, inf(l) represents the lower limit of the number of charge and discharge cycles l; γ represents the failure threshold of the discharge capacity, which is 80% of the initial capacity of the battery; rul k Indicates the remaining life of the battery after the kth charge and discharge cycle.

8. The method for monitoring the health of a power battery throughout its life cycle according to claim 7, characterized in that: The remaining battery life prediction at the key stages includes: (1) Determine the first prediction time FPT; (2) Preprocess the capacity sequence of the battery's potential diving interval and key stage to remove outliers; (3) Based on the empirical model of the battery degradation process, the state transfer equation and observation equation are established as follows: Among them, w k represents the process noise, v k represents the observation noise, k = 1, 2, ... represents the cycle of battery charge and discharge; (4) Assume the current charge and discharge cycle k as the prediction starting point and use the particle filter algorithm to The battery's potential diving range and historical capacity sequence status at key stages Tracking,determining unknown parameters in the state transfer equation; (5) Initialize the particle filter algorithm. The parameters include: number of particles N, process noise sound w k covariance R, observation noise v k The covariance Q of (6) Based on the particle filter algorithm, the capacity is predicted in advance and the estimated value C of the battery capacity is calculated. k+1 ; (7) Determine the capacity advance forecast value C k+1 Whether the battery failure threshold is reached, that is, 80% of the initial capacity, if not, repeat the above steps in sequence. Prediction value C k+n If the battery failure threshold is exceeded, the current charge and discharge cycle can be obtained. The remaining life of the battery is n-1 charge and discharge cycles.

9. The method for monitoring the health of a power battery throughout its life cycle according to claim 4, 6 or 8, characterized in that: The advance prediction of capacity based on the particle filter algorithm specifically includes: initializing a particle set and generating particles; performing importance sampling; calculating weights and normalizing particle weights; performing resampling to obtain a resampled particle set and weights; and finally calculating a state estimation value C of the battery capacity. k+1 and recorded in the sequence.

10. A power battery full life cycle health monitoring system, characterized in that: The system includes three management modules for the entire life cycle of the battery, namely, the degradation stage management module, the potential diving interval management module and the key stage management module. The three modules adopt different management methods according to the stage of the battery: The degradation stage management module is used to perform multi-step advance prediction of battery capacity based on the battery historical capacity sequence in the degradation stage; specifically, firstly, outliers are removed from the battery historical capacity sequence, and then a state transfer equation and an observation equation are established based on the battery historical capacity sequence, and a state estimation value of the battery capacity, i.e., a capacity prediction value, is calculated based on a particle filter algorithm, and the number of advance prediction steps is set, and the capacity prediction value of the advance number of steps is calculated, and finally, multi-step advance prediction of the battery capacity is performed; The potential dive interval management module is used to perform advance prediction of battery capacity in the potential dive interval based on the capacity sequence in the potential dive interval, and the number of advance prediction steps is greater than the number of steps in the degradation stage. As the current charge and discharge cycle increases, the number of advance prediction steps also gradually increases; The key stage management module is used to predict the remaining life throughout the key stage.

11. The power battery full life cycle health monitoring system according to claim 10, characterized in that: The potential diving interval is obtained by counting the diving point data of similar batteries in the cloud as a priori estimates of the current battery diving point. Specifically, firstly, among similar batteries in the cloud, the user's usual driving speed, driving style, electrical appliance information, common SOC driving interval, and common SOC charging and discharging interval features are extracted from the driving behavior and charging habits of the user to whom the battery belongs, and a clustering model is established for the extracted features; the above features are extracted from the current battery, and the user's category is predicted by the clustering model; then, the distribution of the number of charging and discharging cycles when the capacity of the battery of this category dives is counted, and the confidence interval of the distribution is obtained as the potential diving interval of the current battery.

12. The power battery full life cycle health monitoring system according to claim 10 or 11, characterized in that: The potential diving interval management module first removes outliers from the battery potential diving interval capacity sequence, then establishes a state transfer equation and an observation equation based on the potential diving interval battery capacity sequence, and finally performs a multi-step advance forecast of the battery capacity.

13. The power battery full life cycle health monitoring system according to claim 10 or 11, characterized in that: The key stage management module first removes outliers from the potential battery diving interval and the key stage capacity sequence, then establishes a state transfer equation and an observation equation based on the potential diving interval and the key stage battery capacity sequence, and finally performs the full remaining life prediction according to the following formula: NUMBER k =inf(l:f(l+k)≥γ) In the formula, inf(l) represents the lower limit of the number of charge and discharge cycles l; γ represents the failure threshold of the discharge capacity, which is 80% of the initial capacity of the battery; rul k Indicates the remaining life of the battery after the kth charge and discharge cycle.

14. An electronic device, characterized in that: include: one or more processors; A storage device for storing one or more programs, which, when executed by the one or more processors, enables the electronic device to implement the steps of the power battery full life cycle health monitoring method described in any one of claims 1 to 9.

Images