阅读说明：本技术 一种应用数字孪生技术的锂离子电池寿命预测方法 (Lithium ion battery life prediction method applying digital twinning technology ) 是由 熊瑞 田金鹏 卢家欢 于 2020-06-01 设计创作，主要内容包括：一种应用数字孪生技术的锂离子电池寿命预测方法,其通过建立电池的数字孪生体,生成电池在不同工作条件下的老化轨迹,能够有效地应对电池不一致性、环境及工作条件的变化。结合机器学习模型,可以建立快速电池寿命预测模型,并实现其定期更新以应对工况及环境变化,明显克服了现有技术中所存在的不足。(A lithium ion battery service life prediction method applying a digital twinning technology is characterized in that aging tracks of a battery under different working conditions are generated by establishing a digital twinning body of the battery, and the aging tracks can effectively cope with the changes of battery inconsistency, environment and working conditions. By combining a machine learning model, a rapid battery life prediction model can be established, and the model can be updated regularly to cope with working conditions and environmental changes, so that the defects in the prior art are obviously overcome.)

1. A lithium ion battery service life prediction method applying a digital twinning technology is characterized in that: the method specifically comprises the following steps:

I. constructing a battery charging process digital twin body:

the method comprises the following steps that firstly, signal acquisition is carried out on a specific entity battery i, data used for representing the running state of the specific entity battery i are obtained and stored in a battery data storage platform;

establishing a theoretical model for simulating the service life of the battery aiming at the structure and the material of the battery i to be predicted, determining and verifying parameters influencing the service life of the battery in the model by combining an offline battery aging test, and establishing a universal digital twin body based on the parameters;

step three, based on a correction algorithm and by utilizing stored historical running state data, correcting the universal digital twin body, and establishing a service life digital twin body of the entity battery i;

II, constructing a battery life rapid prediction model:

analyzing statistical information of the operation working condition and the environmental change of the battery i from the battery operation historical data, and generating probability distribution of different battery operation working conditions and environmental conditions based on the statistical information;

inputting the operating conditions and the environments generated in the fourth step into the service life digital twin body, and simulating battery aging paths under various operating conditions and environments to obtain a simulated battery aging data set;

establishing a battery life rapid prediction model based on a machine learning algorithm, taking battery temperature, voltage and pressure signals as input, and taking the battery life as output; training the battery life rapid prediction model by utilizing the simulation battery aging data set; and predicting the service life of the entity battery i by using the trained battery service life rapid prediction model.

And seventhly, acquiring and storing the working condition of the battery, the environmental change and the user habit in real time, and updating the battery life rapid prediction model regularly.

2. The method of claim 1, wherein: the data for characterizing the operating state of the device in the first step comprises: voltage, current, power, battery internal temperature distribution, pressure, ambient temperature, humidity of the battery.

3. The method of claim 1, wherein: in the second step, a theoretical model for simulating the service life of the battery is established from the aspects of electrochemistry, thermodynamics, mechanics and the like; the mechanisms according to which universal digital twins are established include: decomposing electrolyte, growing passive film, separating lithium, dissolving transition metal and changing battery structure.

4. The method of claim 1, wherein: and in the third step, the universal digital twin body is corrected, and the capacity loss and the impedance increase data of the entity battery i are specifically compared with the simulation result of the universal digital twin body to realize the correction.

5. The method of claim 1, wherein: the statistical information in the fourth step is obtained by extracting current and power requirements from historical data; generating the probability distribution is based on a Monte Carlo algorithm, a countermeasure generation network, and an auto-encoder algorithm.

6. The method of claim 1, wherein: the machine learning algorithm in the sixth step can adopt an algorithm based on a support vector machine, Gaussian process regression, a deep neural network and the like.

Technical Field

The present invention relates to the field of battery systems, and more particularly to estimating remaining life of lithium ion batteries.

Background

The performance of the lithium ion battery can gradually decline in the using process, and the lithium ion battery has important influence on the safety and reliability of a battery system and a carrying tool. The method for predicting the service life of the battery obtained by the method has some obvious defects, such as incapability of considering inconsistency among batteries and incapability of updating aiming at working condition and environmental change. Therefore, how to provide a lithium battery life prediction method which can effectively adapt to inconsistency among different batteries and can also perform quick response aiming at different working conditions is a problem to be solved in the field.

Disclosure of Invention

The digital twin technology can effectively overcome the defects of the traditional battery control management, has strong adaptability to different battery types and different working conditions, and is beneficial to improving the control of the full life cycle of the battery and the service life prediction effect. In view of this, the present invention provides a method for predicting a lifetime of a lithium ion battery, which is implemented based on a digital twin technology, and specifically includes the following steps:

I. constructing a battery charging process digital twin body:

the method comprises the following steps that firstly, signal acquisition is carried out on a specific entity battery i, data used for representing the running state of the specific entity battery i are obtained and stored in a battery data storage platform;

establishing a theoretical model for simulating the service life of the battery aiming at the structure and the material of the battery i to be predicted, determining and verifying parameters influencing the service life of the battery in the model by combining an offline battery aging test, and establishing a universal digital twin body based on the parameters;

step three, based on a correction algorithm and by utilizing stored historical running state data, correcting the universal digital twin body, and establishing a service life digital twin body of the entity battery i;

II, constructing a battery life rapid prediction model:

analyzing statistical information of the operation working condition and the environmental change of the battery i from the battery operation historical data, and generating probability distribution of different battery operation working conditions and environmental conditions based on the statistical information;

inputting the operating conditions and the environments generated in the fourth step into the service life digital twin body, and simulating battery aging paths under various operating conditions and environments to obtain a simulated battery aging data set;

establishing a battery life rapid prediction model based on a machine learning algorithm, taking signals such as battery temperature, voltage and pressure as input, and taking the battery life as output; training the battery life rapid prediction model by utilizing the simulation battery aging data set; and predicting the service life of the entity battery i by using the trained battery service life rapid prediction model.

And seventhly, acquiring and storing the working condition of the battery, the environmental change and the user habit in real time, and updating the battery life rapid prediction model regularly.

Further, the data for characterizing the operation state in the first step includes: voltage, current, power of the battery, temperature distribution inside the battery, pressure, ambient temperature, humidity, etc.

Further, in the second step, a theoretical model for simulating the service life of the battery is established from the aspects of electrochemistry, thermodynamics, mechanics and the like; the mechanisms according to which universal digital twins are established include: electrolyte decomposition, passive film growth, lithium precipitation, transition metal dissolution, battery structure change and the like.

Further, in the third step, the general digital twin is corrected, and specifically, the data of capacity loss, impedance increase and the like of the entity battery i is compared with the simulation result of the general digital twin.

Further, the statistical information in the fourth step is obtained by extracting current and power requirements from historical data; the probability distribution is generated based on algorithms such as a Monte Carlo algorithm, a countermeasure generation network, an autoencoder, and the like.

Further, the machine learning algorithm in the sixth step adopts an algorithm based on a support vector machine, gaussian process regression, a deep neural network and the like.

According to the method provided by the invention, the aging tracks of the battery under different working conditions are generated by establishing the digital twin body of the battery, so that the method can effectively cope with the changes of the inconsistency of the battery, the environment and the working conditions. By combining a machine learning model, a rapid battery life prediction model can be established, and the model can be updated regularly to cope with working conditions and environmental changes, so that the defects in the prior art are obviously overcome.

Drawings

FIG. 1 is a schematic overall view of the method provided by the present invention;

FIG. 2 is a schematic diagram of a process for constructing a life digital twin in the method of the present invention;

FIG. 3 is a statistical distribution of the ambient temperature distribution and the required power of a certain power battery;

fig. 4 shows the prediction result of the remaining life of a certain power battery.

Detailed Description

The technical solutions of the present invention will be described clearly and completely with reference to the accompanying drawings, and it should be understood that the described embodiments are some, but not all embodiments of the present invention. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.

The lithium ion battery service life prediction method provided by the invention is realized based on a digital twin technology, and as shown in figure 1, the method specifically comprises the following steps:

I. constructing a battery charging process digital twin body, as shown in fig. 2:

the method comprises the following steps that firstly, signal acquisition is carried out on a specific entity battery i, data used for representing the running state of the specific entity battery i are obtained and stored in a battery data storage platform; the data for characterizing the operating state thereof, comprising: voltage, current, power of the battery, temperature distribution inside the battery, pressure, ambient temperature, humidity, etc.

Step two, establishing a theoretical model for simulating the service life of the battery from the aspects of electrochemistry, thermodynamics, mechanics and the like according to the structure and the material of the battery i to be predicted, determining and verifying parameters influencing the service life of the battery in the model by combining an offline battery aging test, and establishing a universal digital twin body based on the parameters, wherein the mechanism comprises the following steps: electrolyte decomposition, passive film growth, lithium precipitation, transition metal dissolution, battery structure change and the like;

thirdly, correcting the general digital twin body based on a correction algorithm and by using stored historical running state data, and establishing a life digital twin body of the entity battery i by comparing data of capacity loss, impedance increase and the like of the entity battery i with a simulation result of the general digital twin body;

II, constructing a battery life rapid prediction model:

analyzing statistical information of the operation working condition and the environmental change of the battery i from the battery operation historical data, for example, extracting current and power requirements from the historical data, and generating probability distribution of different battery operation working conditions and environmental conditions based on the statistical information in combination with algorithms such as a Monte Carlo algorithm, a countermeasure generation network and a self-encoder; for example, fig. 3 shows statistical information of required power and ambient temperature of a certain power battery. Possible combinations of different powers and temperatures of the battery can be obtained by sampling the distribution by monte carlo. Or by learning the distribution through algorithms such as a countermeasure generation network, a self-encoder, and the like, a large number of possible combinations of different operating conditions and environmental conditions having the same distribution can be generated.

Inputting the operating conditions and the environments generated in the fourth step into the service life digital twin body, and simulating battery aging paths under various operating conditions and environments to obtain a simulated battery aging data set;

establishing a battery life rapid prediction model based on machine learning algorithms such as a support vector machine, Gaussian process regression, a deep neural network and the like, taking signals such as battery temperature, voltage, pressure and the like as input, and taking the battery life as output; training the battery life rapid prediction model by utilizing the simulation battery aging data set; and predicting the service life of the entity battery i by using the trained battery service life rapid prediction model. For example, a gaussian process regression algorithm is used to establish a relationship between battery charging voltage and battery life, and the life of a certain power battery is predicted, and the result is shown in fig. 4.

And seventhly, acquiring and storing the working condition of the battery, the environmental change and the user habit in real time, and updating the battery life rapid prediction model regularly.

It should be understood that, the sequence numbers of the steps in the embodiments of the present invention do not mean the execution sequence, and the execution sequence of each process should be determined by the function and the inherent logic of the process, and should not constitute any limitation on the implementation process of the embodiments of the present invention.

Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that changes, modifications, substitutions and alterations can be made in these embodiments without departing from the principles and spirit of the invention, the scope of which is defined in the appended claims and their equivalents.