CN110161414B - A method and system for online prediction of power battery thermal runaway - Google Patents

Abstract

The invention discloses an on-line prediction method and system for thermal runaway of a power battery. The method includes: calculating a voltage deviation matrix at each time according to the voltage value of each battery cell in the power battery; the voltage value includes the voltage data of the battery cell from the time T-M to the current time T; according to the voltage deviation matrix, the battery cell The rated voltage of the battery and the voltage offset increment matrix of the previous moment corresponding to each moment are calculated, and the voltage offset increment matrix at each moment is calculated; the voltage at the current moment T is calculated according to the voltage offset increment matrix at each moment. Offset growth rate matrix; input the voltage offset growth rate of each cell corresponding to the current vehicle mileage, the current temperature average temperature probe, and the voltage offset growth rate matrix at the current time T into the thermal runaway cell prediction In the model, the thermal runaway prediction result of the power battery is obtained. The invention realizes the online prediction of the thermal runaway of the power battery in the real vehicle environment, and improves the prediction accuracy.

Description

A method and system for online prediction of power battery thermal runaway

technical field

The invention relates to the technical field of battery thermal runaway prediction, in particular to a power battery thermal runaway online prediction method and system.

Background technique

Lithium-ion batteries are widely used in electric vehicles due to their high specific energy, high specific power and long service life. However, with the improvement of specific energy of power batteries and the wide application of ternary lithium-ion batteries, the safety issues of lithium-ion batteries have become increasingly prominent. In 2018, there were 50 new energy vehicle safety accidents, of which thermal runaway of the battery was the main cause of the accident. The battery thermal runaway accident involves a large number of casualties and property losses. Therefore, the battery thermal runaway is the core problem that needs to be solved in the development of electric vehicles.

At present, the research on battery thermal runaway is mainly to explore the internal reaction mechanism and external characteristics of the battery when thermal runaway occurs through experiments, and then propose measures to prevent thermal runaway. Existing methods can accurately diagnose the abnormal state of the battery through the battery voltage, temperature and other characterization parameters measured in the laboratory. However, in the real vehicle environment, the characteristics of the battery are affected by many factors, which is a complex process coupled with multiple factors. Therefore, the existing methods are difficult to apply to real electric vehicles.

In order to ensure driving safety and avoid potential failures of electric vehicles, in recent years, some scholars have proposed methods for battery failure prediction and state of health assessment. These methods are mostly based on SOH for one-dimensional evaluation. SOH refers to the state of health of the battery. Calculated by the ratio of the maximum available usage to the rated capacity, using SOH for prediction can well reflect the battery's health status, aging degree and remaining life, but cannot diagnose and predict short-term battery thermal runaway, overcharge, overdischarge, battery short circuit, etc. failure.

SUMMARY OF THE INVENTION

Based on this, it is necessary to provide an online prediction method and system for the thermal runaway of the power battery, so as to realize the online prediction of the thermal runaway of the power battery in the real vehicle environment and improve the prediction accuracy.

For achieving the above object, the present invention provides the following scheme:

An online prediction method for thermal runaway of a power battery, comprising:

Obtain the current mileage of the car, the temperature average value of the current temperature probe, and the voltage value of each battery cell in the power battery; the voltage value includes the voltage data of the battery cell from time T-M to the current time T; one time corresponds to one frame data;

Calculate the voltage deviation matrix at each moment according to the voltage value of each battery cell in the power battery; the voltage deviation matrix is composed of a plurality of voltage deviation values; one battery cell corresponds to one voltage deviation value;

Calculate the voltage offset increment matrix at each moment according to the voltage offset matrix, the rated voltage of the battery cell and the voltage offset increment matrix at the previous moment corresponding to each moment; the voltage offset increment matrix It consists of multiple voltage offset increments; one battery cell corresponds to one voltage offset increment;

Calculate the voltage offset growth rate matrix at the current time T according to the voltage offset increment matrix at each moment; the voltage offset growth rate matrix is composed of multiple voltage offset growth rates; one battery cell corresponds to one voltage offset growth rate;

Input the current vehicle mileage, the temperature average value of the current temperature probe, and the voltage offset growth rate of each cell corresponding to the voltage offset growth rate matrix at the current time T into the thermal runaway cell prediction model to obtain Prediction results of power battery thermal runaway.

Optionally, the current mileage of the vehicle, the temperature average value of the current temperature probe, and the voltage offset growth rate of each cell corresponding to the voltage offset growth rate matrix at the current time T are input into the thermal runaway list. In the volume prediction model, the thermal runaway prediction results of the power battery are obtained, including:

Input the current mileage of the car, the temperature average value of the current temperature probe, and the voltage offset growth rate of each cell corresponding to the voltage offset growth rate matrix at the current time T into the thermal runaway cell prediction model, and output The thermal runaway prediction matrix at the current time T;

Judging whether there is a potential thermal runaway battery cell according to the thermal runaway prediction matrix at the current time T;

If there is a potential thermal runaway battery cell, transmit the serial number of the potential thermal runaway battery cell to the dashboard of the car, the big data monitoring platform for new energy vehicles and the vehicle maintenance platform for monitoring and early warning;

If there is no potential thermal runaway battery cell, determine whether to generate new data;

If new data is generated, set T=T+1, and return the obtained current vehicle mileage, the temperature average value of the current temperature probe, and the voltage value of each battery cell in the power battery;

If no new data is generated, it ends.

Optionally, calculating the voltage deviation matrix at each moment according to the voltage value of each battery cell in the power battery specifically includes:

Calculate the median value of the voltage of each battery cell at each moment according to the voltage value of each battery cell in the power battery;

Calculate the voltage deviation matrix at each time according to the voltage value of each battery cell in the power battery and the median voltage value of the battery cell at each time

Among them, M _t represents the voltage deviation matrix at time t, t∈[TM,T], ΔV _1,t represents the voltage deviation value of the first battery cell at time t, ΔV _n,t represents the nth battery cell The voltage deviation value at time t, V _1,t represents the voltage value of the first battery cell at time t, V _n,t represents the voltage value of the nth battery cell at time t, and n represents the voltage value of the battery cell The total number, V _m,t represents the median value of the cell voltage at time t.

Optionally, according to the voltage deviation matrix, the rated voltage of the battery cell, and the voltage offset increment matrix at the previous moment corresponding to each moment, the voltage offset increment matrix at each moment is calculated, specifically: :

Among them, N _t represents the voltage offset increment matrix at time t, F _1,t represents the voltage offset increment of the first battery cell at time t, and F _{n, t} represents the nth battery cell at time t The voltage offset increment of , F _1,t-1 represents the voltage offset increment of the first battery cell at time t-1, F _n,t-1 represents the nth battery cell at time t-1 The voltage offset increment, V ₀ represents the rated voltage of the battery cell.

Optionally, calculating the voltage offset growth rate matrix at the current moment T according to the voltage offset increment matrix at each moment, specifically:

K _T =(k _1,T ,...,k _n,T ),

Among them, K _T represents the voltage offset growth rate matrix at the current time T, k _{1, T} represents the voltage offset growth rate of the first battery cell at the current time T, and k _{n, T} represents the nth battery cell at the current time T. The voltage offset growth rate of the current time T, the voltage offset growth rate of the i-th battery cell at time t

The present invention also provides an online prediction system for thermal runaway of a power battery, including:

The data acquisition module is used to acquire the current mileage of the car, the temperature average value of the current temperature probe and the voltage value of each battery cell in the power battery; the voltage value includes the voltage of the battery cell from the time T-M to the current time T data; one moment corresponds to one frame of data;

The first matrix calculation module is used to calculate the voltage deviation matrix at each moment according to the voltage value of each battery cell in the power battery; the voltage deviation matrix is composed of a plurality of voltage deviation values; one battery cell corresponds to one voltage Deviation;

The second matrix calculation module is configured to calculate the voltage offset increment matrix at each moment according to the voltage deviation matrix, the rated voltage of the battery cell, and the voltage offset increment matrix at the previous moment corresponding to each moment; The voltage offset increment matrix is composed of a plurality of voltage offset increments; one battery cell corresponds to one voltage offset increment;

The third matrix calculation module is used to calculate the voltage offset growth rate matrix at the current time T according to the voltage offset increment matrix at each moment; the voltage offset growth rate matrix is composed of a plurality of voltage offset growth rates; a The battery cell corresponds to a voltage offset growth rate;

The prediction module is used to input the voltage offset growth rate of each cell corresponding to the current vehicle mileage, the temperature average value of the current temperature probe and the voltage offset growth rate matrix at the current time T into the thermal runaway cell In the prediction model, the thermal runaway prediction result of the power battery is obtained.

Optionally, the prediction module specifically includes:

The prediction matrix acquisition unit is used to input the current vehicle mileage, the current temperature average value of the temperature probe, and the voltage offset growth rate of each cell corresponding to the voltage offset growth rate matrix at the current time T into the thermal runaway In the single prediction model, the thermal runaway prediction matrix at the current time T is output;

a first judging unit, configured to judge whether there is a potential thermal runaway battery cell according to the thermal runaway prediction matrix at the current time T;

The serial number transmission unit is used to transmit the serial number of the potential thermal runaway battery cell to the dashboard of the car, the new energy vehicle big data monitoring platform and the vehicle maintenance platform if there is a potential thermal runaway battery cell, so as to realize monitoring and early warning;

a second judging unit for judging whether to generate new data if there is no potential thermal runaway battery cell;

A return unit is used to make T=T+1 if new data is generated, and return to the data acquisition module;

The end unit is used to end if no new data is generated.

Optionally, the first matrix calculation module specifically includes:

a median calculation unit, configured to calculate the median value of the voltage of each battery cell at each moment according to the voltage value of each battery cell in the power battery;

The first matrix calculation unit is used to calculate the voltage deviation matrix at each time according to the voltage value of each battery cell in the power battery and the voltage median value of the battery cell at each time

Among them, M _t represents the voltage deviation matrix at time t, t∈[TM,T], ΔV _1,t represents the voltage deviation value of the first battery cell at time t, ΔV _n,t represents the nth battery cell The voltage deviation value at time t, V _1,t represents the voltage value of the first battery cell at time t, V _n,t represents the voltage value of the nth battery cell at time t, and n represents the voltage value of the battery cell The total number, V _m,t represents the median value of the cell voltage at time t.

Optionally, the second matrix calculation module is specifically:

Among them, N _t represents the voltage offset increment matrix at time t, F _1,t represents the voltage offset increment of the first battery cell at time t, and F _{n, t} represents the nth battery cell at time t The voltage offset increment of , F _1,t-1 represents the voltage offset increment of the first battery cell at time t-1, F _n,t-1 represents the nth battery cell at time t-1 The voltage offset increment, V ₀ represents the rated voltage of the battery cell.

Optionally, the third matrix calculation module is specifically:

K _T =(k _1,T ,...,k _n,T ),

Among them, K _T represents the voltage offset growth rate matrix at the current time T, k _{1, T} represents the voltage offset growth rate of the first battery cell at the current time T, and k _{n, T} represents the nth battery cell at the current time T. The voltage offset growth rate of the current time T, the voltage offset growth rate of the i-th battery cell at time t

Compared with the prior art, the beneficial effects of the present invention are:

The invention provides an online prediction method and system for thermal runaway of a power battery. This method analyzes the difference between the voltage curve of a potential thermal runaway cell and a normal cell based on a time series, and then couples the historical data with the online data by accumulating the absolute value of the voltage deviation. Thermal runaway monomers are predicted. Compared with the laboratory research method, the state of the real vehicle is predicted after analyzing the data of the actual vehicle operation, which is closer to the actual engineering application. The method or system of the present invention can realize the online prediction of the thermal runaway of the power battery in the real vehicle environment, and can accurately perform the real-time online prediction of the thermal runaway potential cells a few days before the thermal runaway occurs, and the prediction accuracy is high.

Description of drawings

In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the accompanying drawings required in the embodiments will be briefly introduced below. Obviously, the drawings in the following description are only some of the present invention. In the embodiments, for those of ordinary skill in the art, other drawings can also be obtained according to these drawings without creative labor.

FIG. 1 is a flowchart of an online prediction method for thermal runaway of a power battery according to Embodiment 1 of the present invention;

2 is a distribution diagram of the voltage offset growth rate of each cell at a certain moment in Embodiment 2 of the present invention;

3 is a frequency distribution diagram of the number of cells in different voltage offset growth rate intervals in Embodiment 2 of the present invention;

4 is a schematic diagram of a thermal runaway monomer prediction model in Example 2 of the present invention;

FIG. 5 is a graph of prediction results of different values of M when M≥3800 according to Embodiment 2 of the present invention.

Detailed ways

The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, but not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

In order to make the above objects, features and advantages of the present invention more clearly understood, the present invention will be described in further detail below with reference to the accompanying drawings and specific embodiments.

Example 1

FIG. 1 is a flowchart of an online prediction method for thermal runaway of a power battery according to Embodiment 1 of the present invention.

The power battery thermal runaway online prediction method of the embodiment includes:

Step S1: Acquire the current mileage of the car, the average temperature of the current temperature probe, and the voltage value of each battery cell in the power battery; the voltage value includes the voltage data of the battery cell from time T-M to the current time T.

Among them, one time corresponds to one frame of data.

Step S2: Calculate the voltage deviation matrix at each moment according to the voltage value of each battery cell in the power battery.

The voltage deviation matrix is composed of a plurality of voltage deviation values; one battery cell corresponds to one voltage deviation value.

The step S2 specifically includes:

21: Calculate the median value of the voltage of each battery cell at each moment according to the voltage value of each battery cell in the power battery.

22: Calculate the voltage deviation matrix at each moment according to the voltage value of each battery cell in the power battery and the voltage median value of the battery cell at each moment

Among them, M _t represents the voltage deviation matrix at time t, t∈[TM,T], ΔV _1,t represents the voltage deviation value of the first battery cell at time t, ΔV _n,t represents the nth battery cell The voltage deviation value at time t, V _1,t represents the voltage value of the first battery cell at time t, V _n,t represents the voltage value of the nth battery cell at time t, and n represents the voltage value of the battery cell The total number, V _m,t represents the median value of the cell voltage at time t.

Step S3: Calculate the voltage offset increment matrix at each moment according to the voltage deviation matrix, the rated voltage of the battery cell, and the voltage offset increment matrix at the previous moment corresponding to each moment.

The voltage offset increment matrix is composed of a plurality of voltage offset increments; one battery cell corresponds to one voltage offset increment.

The step S3 is specifically:

Among them, N _t represents the voltage offset increment matrix at time t, F _1,t represents the voltage offset increment of the first battery cell at time t, and F _{n, t} represents the nth battery cell at time t The voltage offset increment of , F _1,t-1 represents the voltage offset increment of the first battery cell at time t-1, F _n,t-1 represents the nth battery cell at time t-1 The voltage offset increment, V ₀ represents the rated voltage of the battery cell.

Step S4: Calculate the voltage offset growth rate matrix at the current moment T according to the voltage offset increment matrix at each moment.

The voltage offset growth rate matrix is composed of a plurality of voltage offset growth rates; one battery cell corresponds to one voltage offset growth rate.

The step S4 is specifically:

K _T =(k _1,T ,...,k _n,T ),

Among them, K _T represents the voltage offset growth rate matrix at the current time T, k _{1, T} represents the voltage offset growth rate of the first battery cell at the current time T, and k _{n, T} represents the nth battery cell at the current time T. The voltage offset growth rate of the current time T, the voltage offset growth rate of the i-th battery cell at time t

Step S5: Input the current vehicle mileage, the temperature average value of the current temperature probe, and the voltage offset growth rate of each cell corresponding to the voltage offset growth rate matrix at the current time T into the thermal runaway cell prediction model , the thermal runaway prediction results of the power battery are obtained.

The step S5 specifically includes:

51: Input the current vehicle mileage, the current temperature average value of the temperature probe, and the voltage offset growth rate of each cell corresponding to the voltage offset growth rate matrix at the current time T into the thermal runaway cell prediction model , output the thermal runaway prediction matrix at the current time T.

52: Determine whether there is a potential thermal runaway battery cell according to the thermal runaway prediction matrix at the current time T.

If there is a potential thermal runaway battery cell, the serial number of the potential thermal runaway battery cell is transmitted to the dashboard of the car, the new energy vehicle big data monitoring platform and the vehicle maintenance platform to realize monitoring and early warning.

If there is no potential thermal runaway battery cell, then judge whether to generate new data; if new data is generated, set T=T+1, and return to the method of obtaining the current mileage of the car and the temperature of the current temperature probe The average value and the voltage value of each battery cell in the power battery; if no new data is generated, it ends.

The online prediction method for power battery thermal runaway in this embodiment can realize online prediction of power battery thermal runaway in a real vehicle environment, and can accurately predict thermal runaway potential cells in real time a few days before thermal runaway occurs. , and the prediction accuracy is high.

A more detailed example is provided below.

Example 2

1. Data selection

The data in this embodiment comes from the national big data platform for new energy vehicles, which can collect and store various data during the operation of new energy vehicles, including online data and offline data. The data in the platform covers aspects of vehicle location, speed, and battery system status. The data of electric vehicles in the platform includes vehicle driving status data, vehicle position data, vehicle battery system status data, vehicle motor system status data, vehicle fault alarm data, etc. This embodiment analyzes the data of a number of vehicles that are overheating and out of control and normal vehicles on the national big data platform for new energy vehicles. The time interval between each two frames of data is 10s.

The data preprocessing steps are: (1) The data of one month before the thermal runaway of the thermal runaway car and the data of one month of the normal car are retrieved from the platform. (2) Transcode and segment the data to obtain a readable table file. (3) According to the real-time information collection items in GB/T32960, the dimensions related to thermal runaway prediction are extracted, including battery cell voltage, probe temperature, and current vehicle mileage, and various data distributed in time series are obtained. (4) The outliers and null values are removed by the threshold method, and the valid data of the car is obtained.

2. Thermal runaway monomer prediction model

There are many reasons for thermal runaway of car batteries, mainly mechanical abuse and electrical abuse. Mechanical abuse includes battery collision, extrusion, puncture, etc., while electrical abuse mainly includes internal short circuit, external short circuit, overcharge, overdischarge, etc. . Most thermal runaway starts from one or several cells and gradually spreads. The battery management system BMS can alarm the thermal runaway phenomenon by monitoring the temperature, voltage and other parameters of the battery. However, due to the temperature and voltage of the battery when thermal runaway occurs With the rapid rise, it is difficult to avoid accidents in real-time online monitoring of BMS. Therefore, the identification and prediction of potential thermal runaway monomers is crucial. Fortunately, the method based on big data can analyze the data for a period of time before the thermal runaway of the battery occurs, so that the potential thermal runaway cell of the battery can be predicted, and the thermal runaway can be predicted before the occurrence of thermal runaway, avoiding accidents. happened.

1) Construction of thermal runaway monomer prediction model

The thermal runaway of the battery is affected by many factors, and the battery cell voltage is a comprehensive manifestation of the failure. In order to study the potential thermal runaway fault of the battery, a statistical analysis was carried out on the voltage of each battery cell one month before the occurrence of thermal runaway. It can be seen from the analysis that in the month before the thermal runaway occurs, the fluctuation of the cell voltage of the thermal runaway potential fault is higher than that of the normal battery. The cell is larger, and the voltage is too low for many times. At the end of the discharge, the voltage of the thermal runaway potential fault cell is lower than 3.3V many times, and the terminal voltage can reflect the size of the battery SOC, so the thermal runaway potential single The body produces an overdischarge at the end of the discharge. In addition, at the end of charging, the potential cell voltage for thermal runaway is slightly higher than the normal cell voltage, indicating that the potential cell for thermal runaway has a slight overcharge at the end of charging. The electrochemical properties gradually deteriorated.

In order to quantitatively describe the fluctuation degree of the battery cell voltage, the voltage deviation of the battery cell in each frame of data is calculated, and the calculation formula is as follows:

ΔV _i,t =V _i,t -V _median,t

In the formula, ΔV _i,t is the voltage deviation of the data of the t-th frame of the i-th cell, the unit is V; V _i,t is the voltage of the t-th frame of the i-th cell, the unit is V, V _median,t is the median of all cell voltages in the t-frame data, in V.

The voltage deviation of the thermal runaway cell and the normal cell within one month is counted. According to the statistical data, the voltage deviation of the normal cell is generally kept between -0.1V and +0.1V in the month before the thermal runaway occurs. , has a good consistency, and the voltage deviation of the potential thermal runaway cell exceeds the range of -0.1V—+0.1V many times, which is obviously larger than that of other normal cells, and the voltage deviation of the cell even exceeds -0.5 many times. V, and the voltage deviation shows a phenomenon of sudden positive and negative.

From the above analysis of the cell voltage, it can be inferred that the thermal runaway of the cell can be reflected by the deviation of the voltage, and there is a gradually deteriorating accumulation process. The absolute value of the voltage deviation of each battery cell at the current moment and the historical moment is accumulated as the current voltage offset increment of the cell, so as to couple the current data of the car with the historical data, and identify the potential thermal runaway cell. Assuming that the current moment data is the T-th frame, the voltage offset increment of each battery cell is defined as follows:

In the formula, F _{i, T} is the voltage offset increment of the data of the i-th cell in the T-th frame, V ₀ is the rated voltage of the battery cell, and the unit is V.

From the above formula, the voltage offset increment of each battery cell at the time corresponding to the T-1 frame data can be obtained:

In the formula, F _{i, T-1} is the voltage offset increment of the T-1 frame data of the i-th cell

From the above two formulas, we can finally get:

Every time the car generates a new frame of data, F i,T can be directly calculated by the relationship between F _{i ,T} and F _i,T-1 in the above formula, instead of accumulating the absolute values of all individual deviations way, so there will be no increase in the amount of computation due to the increase in the amount of car data.

According to the above formula, the voltage offset increment of each battery cell at each moment is calculated. It can be seen from the analysis that the voltage offset increment of each battery cell at each moment varies roughly linearly with time, and the potential thermal runaway cell ratio The normal cell voltage offset increment increases faster.

According to the above analysis, the least squares line fitting can be performed on the voltage shift increment curve of the battery cell, and the potential thermal runaway cell can be identified by the slope of the fitted line. However, with the continuous increase of the amount of car data, the calculation time of the least squares line fitting is also increasing, so the calculation step M is defined to limit the amount of historical data involved in the fitting.

The calculation step size M represents the number of historical data frames involved in the voltage offset incremental fitting each time the vehicle generates a new frame of data. The calculation step size is a constant, which can be determined according to the computing power of the big data platform and BMS when designing the BMS control strategy, to ensure that the calculation time of data fitting is less than the time interval between every two frames of data of the car, so as to realize online real-time prediction.

Every time a new frame of data is generated by the car, the least-squares linear fitting is performed on the voltage offset increments of the current moment data and the previous M frames of data, and the slope K _i of the voltage offset increment curve of each cell is calculated, which is defined as is the voltage offset growth rate, and its calculation formula is as follows:

In the formula, Ki is the voltage offset growth rate of the _ith cell, t is the number of data frames, and F _i,t is the voltage offset increment of the ith cell in the t-th frame of data.

FIG. 2 is a distribution diagram of the voltage offset growth rate of each cell at a certain time in Embodiment 2 of the present invention, and FIG. 3 is a frequency distribution diagram of the number of cells in different voltage offset growth rate intervals in Embodiment 2 of the present invention. It can be seen from the figure that the voltage excursion growth rate of No. 125 thermal runaway cell is larger than that of other cells, and the voltage excursion growth rate of most cells is below 0.375*10(-5). The voltage excursion growth rate can identify potential cells for thermal runaway.

Although the battery cell voltage is a comprehensive manifestation of the fault, the battery thermal runaway and other faults are also related to the aging degree of the battery. The aging degree of the battery is represented by the current mileage. In addition, the temperature is the most obvious characterization parameter of thermal runaway, so each The probe temperature is also used as one of the criteria for judging potential thermal runaway monomers. The voltage excursion growth rate of each cell, the current mileage, and the average temperature of the current temperature probe are used as the input of the neural network.

Since this model is based on the actual vehicle data of the vehicle that has occurred thermal runaway, it is more in line with the actual situation of the vehicle. , should be maintained in time. Therefore, this model only predicts whether a certain monomer is a potential monomer of thermal runaway, but does not predict the probability of thermal runaway. Whether it is a potential thermal runaway cell is used as the output of the neural network to train a potential thermal runaway cell prediction model. The neural network training model is shown in Figure 4.

2) Thermal runaway monomer prediction algorithm flow

According to the thermal runaway prediction model established for battery cells, a data-driven battery thermal runaway prediction method is proposed. Let the current moment of the car be the T-th frame data, the steps are:

(1) Set the number M of online fitting data frames.

(2) Extract the T-th frame data (data at the current moment) of the car, and calculate the median of all battery cell voltages in the T-th frame data.

(3) Find the absolute value of the difference between the voltage of each cell and the median of the voltage, as the voltage deviation matrix at the current moment:

M _T =(|V _1,T -V _m,T |,...,|V _n,T -V _m,T |)

In the formula, M _T is the voltage deviation matrix at the current moment, V _m,T is the median of all battery cell voltages at the current moment, V _1,T ,V _2,T ,…,V _i,T ,…,V _{n , T} is the voltage of each battery cell.

(4) Calculate the voltage offset increment matrix at the current moment according to the formula:

In the formula, N _T is the voltage offset increment matrix at the current moment.

(5) Perform a least-squares linear fitting on the data from (T-M) to T frames in the voltage offset increment matrix to obtain the voltage offset growth rate matrix of each cell at the current moment.

K _T =(k _1,T ,...,k _n,T )

In the formula, K _T is the voltage offset growth rate matrix at the current moment, and k _1,T ,k _2,T ,…,ki _,T ,…,k _n,T is the voltage offset growth rate of each cell.

(6) Input the voltage offset growth rate of each cell, the current mileage, and the average temperature of the current temperature probe into the thermal runaway prediction model to determine the potential thermal runaway cell.

(7) Every time the car generates a new frame of data, it repeats (1)-(6), and loops continuously.

3. Real vehicle data verification

The total sample data, training data and test data selected for real vehicle verification are shown in Table 1.

Table 1 Real vehicle verification data

Set the prediction matrix P to record the prediction of battery thermal runaway:

In the formula, p _ij (i=1,2,...,t)(j=1,2,...,n) represents the predicted value of the thermal runaway of the battery cell j at time i calculated according to the model. If the thermal runaway is satisfied, For the condition of out-of-control warning, p _ij is recorded as 1, otherwise it is recorded as 0.

The thermal runaway battery cell of this vehicle is No. 125 cell, and the calculation step M is set to 5000. From the simulation verification results, it can be seen that the data of the first warning of No. 125 potential thermal runaway cell is the 49111th frame. After that, it has been in the early warning state, that is, the thermal runaway fault can be warned before 7 days; and the other monomers have no early warning records. The accuracy of the algorithm is verified.

In order to study the influence of the calculation step size M on the prediction results, the prediction results of the thermal runaway potential monomers in the case of M=100, 200, 300, ..., 9900, 10000 were compared. Through the analysis of the corresponding prediction results when M=100, 100, 4000, and 10000, it can be seen that for this vehicle, when M<3800, the prediction cannot be accurately performed, and the amount of data is too small, which leads to the fitted voltage offset growth rate Affected by voltage fluctuations, some normal cells are wrongly judged as potential cells for thermal runaway; when M≥3800, the potential cells for thermal runaway can be accurately predicted.

5 is a graph of prediction results for different values of M when M ≥ 3800 in Example 2 of the present invention, where the ordinate represents the number of frames n where the potential thermal runaway monomer is predicted for the first time, and the smaller n is, the more timely the prediction is. It can be seen from the figure that n increases with the increase of M, and the growth trend becomes slower and slower. This is because with the increase of M, the proportion of newly generated data in the data participating in the fitting decreases, resulting in the prediction of The time for the potential monomer to be released from thermal runaway is relatively delayed. In addition, in the process of M increasing from 3800 to 10000, n increased from 48135 to 51031, and the interval between two adjacent frames of data is 10s, so the lag is 8 hours, compared with 7 days in advance to predict potential thermal runaway orders However, the prediction lag caused by increasing M is not very obvious. However, since increasing M will increase the calculation time of the model, M should not be selected too large. From the formula of the voltage offset growth rate, it can be seen that the calculation amount of the multiplication operation is 2M+2, and the calculation amount of the addition operation is 5M+5.

The on-line prediction method for thermal runaway of a power battery in this embodiment analyzes the data of the power battery cells in the previous month of the car that has thermal runaway. First, based on the time series, the difference between the voltage curves of the potential cells of thermal runaway and the normal cells is analyzed. Then, the influence of voltage deviation on battery overcharge and overdischarge is analyzed, and then the historical data and current data are coupled by the method of accumulating the absolute value of voltage deviation, and a thermal runaway cell prediction model is established based on the neural network method to prevent potential thermal runaway. The single unit is predicted, and finally the model is verified with real vehicle data. It has been verified that the power battery thermal runaway online prediction method of this embodiment can accurately perform real-time online prediction on potential thermal runaway cells, and provide certain design ideas and reference for the power battery online thermal runaway diagnosis.

The online prediction method for thermal runaway of a power battery in this embodiment has the following characteristics:

(1) Coupling current vehicle data with historical data to predict potential thermal runaway cells in the battery pack.

(2) Real-time online fault prediction, iterative with the update of automobile data, as the amount of data increases, the difference between the voltage offset increment of the potential thermal runaway cell and the voltage offset increment of other normal cells becomes more and more Therefore, the prediction accuracy is getting higher and higher.

(3) Since this embodiment only needs to consider the deviation of the voltage of each cell, it is not necessary to identify the current state of the vehicle, and the influence of data frame loss on the prediction result can also be avoided, thereby improving the prediction accuracy.

(4) The battery cells with good and poor consistency can be quantitatively judged at the same time by the magnitude of the voltage offset growth rate, which provides a method for the evaluation of battery consistency.

(5) This embodiment predicts whether a certain monomer is a potential monomer of thermal runaway, and does not predict the probability of thermal runaway.

(6) This embodiment can more accurately predict thermal runaway faults such as battery overcharge and overdischarge caused by battery inconsistency.

The present invention also provides an online prediction system for thermal runaway of a power battery, including:

The data acquisition module is used to acquire the current mileage of the car, the temperature average value of the current temperature probe and the voltage value of each battery cell in the power battery; the voltage value includes the voltage of the battery cell from the time T-M to the current time T Data; a moment corresponds to a frame of data.

The first matrix calculation module is used to calculate the voltage deviation matrix at each moment according to the voltage value of each battery cell in the power battery; the voltage deviation matrix is composed of a plurality of voltage deviation values; one battery cell corresponds to one voltage Deviation.

The second matrix calculation module is configured to calculate the voltage offset increment matrix at each moment according to the voltage deviation matrix, the rated voltage of the battery cell, and the voltage offset increment matrix at the previous moment corresponding to each moment; The voltage offset increment matrix is composed of a plurality of voltage offset increments; one battery cell corresponds to one voltage offset increment.

The third matrix calculation module is used to calculate the voltage offset growth rate matrix at the current time T according to the voltage offset increment matrix at each moment; the voltage offset growth rate matrix is composed of a plurality of voltage offset growth rates; a A battery cell corresponds to a voltage offset growth rate.

The prediction module is used to input the voltage offset growth rate of each cell corresponding to the current vehicle mileage, the temperature average value of the current temperature probe and the voltage offset growth rate matrix at the current time T into the thermal runaway cell In the prediction model, the thermal runaway prediction result of the power battery is obtained.

As an optional implementation manner, the prediction module specifically includes:

The prediction matrix acquisition unit is used to input the current vehicle mileage, the current temperature average value of the temperature probe, and the voltage offset growth rate of each cell corresponding to the voltage offset growth rate matrix at the current time T into the thermal runaway In the single prediction model, the thermal runaway prediction matrix at the current time T is output;

a first judging unit, configured to judge whether there is a potential thermal runaway battery cell according to the thermal runaway prediction matrix at the current time T;

The serial number transmission unit is used to transmit the serial number of the potential thermal runaway battery cell to the dashboard of the car, the new energy vehicle big data monitoring platform and the vehicle maintenance platform if there is a potential thermal runaway battery cell, so as to realize monitoring and early warning;

a second judging unit for judging whether to generate new data if there is no potential thermal runaway battery cell;

A return unit is used to make T=T+1 if new data is generated, and return to the data acquisition module;

The end unit is used to end if no new data is generated.

As an optional implementation manner, the first matrix calculation module specifically includes:

a median calculation unit, configured to calculate the median value of the voltage of each battery cell at each moment according to the voltage value of each battery cell in the power battery;

The first matrix calculation unit is used to calculate the voltage deviation matrix at each time according to the voltage value of each battery cell in the power battery and the voltage median value of the battery cell at each time

Among them, M _t represents the voltage deviation matrix at time t, t∈[TM,T], ΔV _1,t represents the voltage deviation value of the first battery cell at time t, ΔV _n,t represents the nth battery cell The voltage deviation value at time t, V _1,t represents the voltage value of the first battery cell at time t, V _n,t represents the voltage value of the nth battery cell at time t, and n represents the voltage value of the battery cell The total number, V _m,t represents the median value of the cell voltage at time t.

As an optional implementation manner, the second matrix calculation module is specifically:

Among them, N _t represents the voltage offset increment matrix at time t, F _1,t represents the voltage offset increment of the first battery cell at time t, and F _{n, t} represents the nth battery cell at time t F _1,t-1 represents the voltage offset increment of the first battery cell at time t-1, F _n,t-1 represents the nth battery cell at time t-1 The voltage offset increment, V ₀ represents the rated voltage of the battery cell.

As an optional implementation manner, the third matrix calculation module is specifically:

K _T =(k _1,T ,...,k _n,T ),

Among them, K _T represents the voltage offset growth rate matrix at the current time T, k _{1, T} represents the voltage offset growth rate of the first battery cell at the current time T, and k _{n, T} represents the nth battery cell at the current time T. The voltage offset growth rate of the current time T, the voltage offset growth rate of the i-th battery cell at time t

The power battery thermal runaway online prediction system in this embodiment can realize the online prediction of the power battery thermal runaway in the real vehicle environment, and can accurately perform real-time online prediction on the potential cells of thermal runaway a few days before the occurrence of thermal runaway. , and the prediction accuracy is high.

The various embodiments in this specification are described in a progressive manner, and each embodiment focuses on the differences from other embodiments, and the same and similar parts between the various embodiments can be referred to each other. For the system disclosed in the embodiment, since it corresponds to the method disclosed in the embodiment, the description is relatively simple, and the relevant part can be referred to the description of the method.

In this paper, specific examples are used to illustrate the principles and implementations of the present invention. The descriptions of the above embodiments are only used to help understand the methods and core ideas of the present invention; meanwhile, for those skilled in the art, according to the present invention There will be changes in the specific implementation and application scope. In conclusion, the contents of this specification should not be construed as limiting the present invention.

Claims (10)

1. A power battery thermal runaway online prediction method is characterized by comprising the following steps:

acquiring the current driving mileage of the automobile, the temperature average value of the current temperature probe and the voltage value of each battery monomer in the power battery; the voltage value comprises voltage data of the battery monomer from T-M moment to current moment T; one moment corresponds to one frame data; m is more than 0 and less than T, and M is an integer;

calculating a voltage deviation matrix at each moment according to the voltage value of each battery monomer in the power battery; the voltage deviation matrix is composed of a plurality of voltage deviation values; one battery monomer corresponds to one voltage deviation value;

calculating a voltage offset increment matrix at each moment according to the voltage deviation matrix, the rated voltage of the battery monomer and the voltage offset increment matrix at the last moment corresponding to each moment; the voltage offset delta matrix is composed of a plurality of voltage offset deltas; one cell corresponds to one voltage offset increment;

calculating a voltage offset increase rate matrix of the current time T according to the voltage offset increment matrix of each time; the voltage offset growth rate matrix is composed of a plurality of voltage offset growth rates; one cell corresponds to one voltage offset increase rate;

and inputting the current driving mileage of the automobile, the temperature average value of the current temperature probe and the voltage offset increase rate of each monomer corresponding to the voltage offset increase rate matrix at the current time T into a thermal runaway monomer prediction model to obtain a thermal runaway prediction result of the power battery.

2. The power battery thermal runaway online prediction method according to claim 1, wherein the step of inputting the current driving mileage of the automobile, the temperature average value of the current temperature probe, and the voltage deviation increase rate of each cell corresponding to the voltage deviation increase rate matrix at the current time T into a thermal runaway cell prediction model to obtain a power battery thermal runaway prediction result specifically comprises the steps of:

inputting the current driving mileage of the automobile, the temperature average value of the current temperature probe and the voltage offset growth rate of each monomer corresponding to the voltage offset growth rate matrix at the current time T into a thermal runaway monomer prediction model, and outputting the thermal runaway prediction matrix at the current time T;

judging whether a potential thermal runaway battery monomer exists according to the thermal runaway prediction matrix at the current moment T;

if the potential thermal runaway single battery exists, transmitting the serial number of the potential thermal runaway single battery to an instrument panel of the automobile, a new energy automobile big data monitoring platform and a vehicle maintenance platform so as to realize monitoring and early warning;

if no potential thermal runaway battery monomer exists, judging whether new data are generated;

if new data are generated, enabling T to be T +1, and returning to the step of obtaining the current driving mileage of the automobile, the average temperature value of the current temperature probe and the voltage value of each battery cell in the power battery;

and if no new data is generated, ending the process.

3. The method for online predicting the thermal runaway of the power battery according to claim 1, wherein the calculating the voltage deviation matrix at each moment according to the voltage values of the battery cells in the power battery specifically comprises:

calculating the voltage median value of each battery monomer at each moment according to the voltage value of each battery monomer in the power battery;

calculating a voltage deviation matrix at each moment according to the voltage value of each battery monomer in the power battery and the voltage median value of each battery monomer at each moment

M_t＝(ΔV_1,t,…,ΔV_n,t)

＝(|V_1,t-V_m,t|,…,|V_n,t-V_m,t|)，

Wherein M is_tA voltage deviation matrix representing time T, T ∈ [ T-M, T]，ΔV_1,tIndicates the voltage deviation value, DeltaV, of the first battery cell at the moment t_n,tIndicates the voltage deviation value, V, of the nth battery cell at the time point t_1,tRepresenting the voltage value, V, of the first cell at time t_n,tRepresenting the voltage value of the nth battery cell at the moment t, n representing the total number of battery cells, V_m,tThe median value of the voltage of the battery cell at the time t is shown.

4. The online prediction method for the thermal runaway of the power battery according to claim 3, wherein the voltage offset increment matrix at each moment is calculated according to the voltage offset matrix, the rated voltage of the battery cell and the voltage offset increment matrix at the last moment corresponding to each moment, and specifically:

wherein N is_tA delta matrix of voltage offsets representing time t, F_1,tRepresenting the increment of the voltage offset of the first cell at time t, F_n,tIndicates the increment of voltage deviation of the nth cell at the time t, F_1,t-1Representing the increment of the voltage offset of the first cell at time t-1, F_n,t-1Indicates the nth cellIncrement of voltage offset of cell at time t-1, V₀Indicating the rated voltage of the battery cell.

5. The online prediction method for the thermal runaway of the power battery according to claim 4, wherein the voltage offset increase rate matrix of the current time T is calculated according to the voltage offset increment matrix of each time, and specifically comprises the following steps:

K_T＝(k_1,T,…,k_n,T)，

wherein, K_TA voltage offset growth rate matrix, k, representing the current time T_1,TRepresents the voltage offset increase rate, k, of the first cell at the current time T_n,TThe voltage deviation growth rate of the nth battery cell at the current time T and the voltage deviation growth rate of the ith battery cell at the time T are shown

6. An online prediction system for thermal runaway of a power battery is characterized by comprising:

the data acquisition module is used for acquiring the current driving mileage of the automobile, the temperature average value of the current temperature probe and the voltage value of each battery monomer in the power battery; the voltage value comprises voltage data of the battery monomer from T-M moment to current moment T; one moment corresponds to one frame data; m is more than 0 and less than T, and M is an integer;

the first matrix calculation module is used for calculating a voltage deviation matrix at each moment according to the voltage value of each battery monomer in the power battery; the voltage deviation matrix is composed of a plurality of voltage deviation values; one battery monomer corresponds to one voltage deviation value;

the second matrix calculation module is used for calculating a voltage offset increment matrix at each moment according to the voltage deviation matrix, the rated voltage of the battery monomer and the voltage offset increment matrix at the last moment corresponding to each moment; the voltage offset delta matrix is composed of a plurality of voltage offset deltas; one cell corresponds to one voltage offset increment;

the third matrix calculation module is used for calculating a voltage offset increase rate matrix of the current time T according to the voltage offset increment matrix of each time; the voltage offset growth rate matrix is composed of a plurality of voltage offset growth rates; one cell corresponds to one voltage offset increase rate;

and the prediction module is used for inputting the current driving mileage of the automobile, the temperature average value of the current temperature probe and the voltage offset increase rate of each monomer corresponding to the voltage offset increase rate matrix at the current time T into a thermal runaway monomer prediction model to obtain a thermal runaway prediction result of the power battery.

7. The power battery thermal runaway online prediction system of claim 6, wherein the prediction module specifically comprises:

the prediction matrix obtaining unit is used for inputting the driving mileage of the current automobile, the temperature average value of the current temperature probe and the voltage offset increase rate of each monomer corresponding to the voltage offset increase rate matrix at the current time T into the thermal runaway monomer prediction model and outputting the thermal runaway prediction matrix at the current time T;

the first judgment unit is used for judging whether a potential thermal runaway battery monomer exists according to the thermal runaway prediction matrix at the current time T;

the serial number transmission unit is used for transmitting the serial number of the potential thermal runaway battery monomer to an instrument panel of an automobile, a new energy automobile big data monitoring platform and a vehicle maintenance platform if the potential thermal runaway battery monomer exists so as to realize monitoring and early warning;

the second judgment unit is used for judging whether new data are generated or not if no potential thermal runaway single battery exists;

a returning unit, configured to, if new data is generated, make T equal to T +1, and return to the data obtaining module;

and an ending unit for ending if no new data is generated.

8. The power battery thermal runaway online prediction system of claim 6, wherein the first matrix calculation module specifically comprises:

the median calculating unit is used for calculating the voltage median of each battery monomer at each moment according to the voltage value of each battery monomer in the power battery;

a first matrix calculation unit, configured to calculate a voltage deviation matrix at each time according to the voltage value of each battery cell in the power battery and the voltage median value of each battery cell at each time

M_t＝(ΔV_1,t,…,ΔV_n,t)

＝(|V_1,t-V_m,t|,…,|V_n,t-V_m,t|)，

Wherein M is_tA voltage deviation matrix representing time T, T ∈ [ T-M, T]，ΔV_1,tIndicates the voltage deviation value, DeltaV, of the first battery cell at the moment t_n,tIndicates the voltage deviation value, V, of the nth battery cell at the time point t_1,tRepresenting the voltage value, V, of the first cell at time t_n,tRepresenting the voltage value of the nth battery cell at the moment t, n representing the total number of battery cells, V_m,tThe median value of the voltage of the battery cell at the time t is shown.

9. The power battery thermal runaway online prediction system of claim 8, wherein the second matrix calculation module is specifically:

wherein N is_tA delta matrix of voltage offsets representing time t, F_1,tRepresenting the increment of the voltage offset of the first cell at time t, F_n,tIndicates the increment of voltage deviation of the nth cell at the time t, F_1,t-1Representing the increment of the voltage offset of the first cell at time t-1, F_n,t-1Indicating the increment of voltage deviation of the nth battery cell at the time point of t-1，V₀Indicating the rated voltage of the battery cell.

10. The power battery thermal runaway online prediction system of claim 9, wherein the third matrix calculation module specifically is:

K_T＝(k_1,T,…,k_n,T)，

wherein, K_TA voltage offset growth rate matrix, k, representing the current time T_1,TRepresents the voltage offset increase rate, k, of the first cell at the current time T_n,TThe voltage deviation growth rate of the nth battery cell at the current time T and the voltage deviation growth rate of the ith battery cell at the time T are shown

Images