CN117388564A - Power battery consistency abnormality detection method based on real vehicle operation data - Google Patents

Abstract

The invention relates to the technical field of power batteries, in particular to a power battery consistency abnormality detection method based on real vehicle operation data, which comprises the following steps: analyzing corresponding battery signal data through the original message, so as to construct a voltage difference matrix and an abnormal threshold upper limit; traversing the voltage difference values in the voltage difference matrix, judging whether each difference value is larger than the upper limit of the abnormal threshold, if so, judging whether the voltage difference value of the adjacent time points is larger than the upper limit of the abnormal threshold, if so, marking the time, acquiring the marking time, judging whether the frequency in each period of preset time of the preset times meets the preset frequency, if so, judging that the consistency variation occurs, and returning to the marking time and the abnormal cell number. The scheme can improve the accuracy, timeliness and sustainability of detection, reduce the detection cost and complexity, so as to dynamically adjust and optimize, thereby improving the safety, service life and energy utilization efficiency of the battery.

Description

Power battery consistency anomaly detection method based on actual vehicle operating data

Technical field

The invention relates to the technical field of power batteries, and is specifically a power battery consistency anomaly detection method based on actual vehicle operating data.

Background technique

Battery consistency abnormality detection refers to detecting and evaluating the performance and status of the battery in the battery pack through certain methods and means to determine whether there are batteries with abnormal consistency in the battery pack.

Abnormal battery consistency usually refers to abnormalities in battery performance, capacity, internal resistance and other parameters, which may lead to a decline in the overall performance of the battery pack, even cause safety issues, and affect the service life and energy utilization efficiency of the battery pack.

Although the existing battery consistency anomaly detection method can detect battery consistency anomalies, there are some problems:

1. The detection accuracy is low. The existing battery consistency anomaly detection method relies on the data extracted by the BMS system. However, the data reported by the BMS system may be affected by factors such as the characteristics of the battery itself, the vehicle operating environment and working conditions, etc. There are many false positive issues, which may lead to less accurate test results;

2. Detection costs are high. Consistency anomaly detection requires a large amount of data collection, processing and analysis, as well as dedicated test equipment and battery management systems. These will increase costs, especially for mass-produced electric vehicle manufacturers, which may Will face higher cost pressure, and electric vehicles use various types of batteries, such as lithium-ion batteries, nickel-metal hydride batteries, etc.; and different types of batteries have different characteristics and consistency requirements, so consistency testing needs to be based on Adaptation and optimization of different battery types further increases detection costs;

3. High complexity. The consistency of the battery pack is affected by many factors, including battery characteristics, usage environment, temperature, etc. Therefore, consistency anomaly detection involves complex data processing and analysis, and requires comprehensive consideration of multiple factors. , increasing the difficulty of technical implementation and high complexity;

4. Low timeliness. For real-time monitoring of battery pack consistency, an efficient data transmission and processing system needs to be established to ensure real-time performance. However, during vehicle operation, data transmission and processing may be affected by various factors. Reducing the timeliness of monitoring;

5. Low long-term stability. During long-term use, the consistency of the battery pack may change. The existing detection methods have low long-term stability and reliability and cannot ensure the sustainability of the detection and the accuracy of the detection results. .

Therefore, there is an urgent need for a power battery consistency anomaly detection method based on actual vehicle operating data, which can improve the accuracy, timeliness and sustainability of detection, reduce detection costs and complexity, facilitate dynamic adjustment and optimization, and thereby improve Battery safety, lifespan and energy efficiency.

Contents of the invention

The present invention is intended to provide a method for detecting power battery consistency anomalies based on actual vehicle operating data, which can improve the accuracy, timeliness and sustainability of detection, reduce detection costs and complexity, and facilitate dynamic adjustment and optimization. Improve battery safety, lifespan and energy efficiency.

The present invention provides the following basic solution: a power battery consistency anomaly detection method based on actual vehicle operation data, including the following content: S1. Obtain the original message of the actual vehicle operation, and parse the corresponding battery signal data;

S2. Preprocess the battery signal data;

S3. Select features from the preprocessed battery signal data and extract the battery signal data in the discharge state;

S4. Calculate the average voltage value between each battery cell based on the battery signal data in the discharge state;

S5. Calculate the difference between the voltage value of each cell and the average voltage value of its row, and take the absolute value of each difference as the voltage difference to obtain the voltage difference matrix;

S6. Traverse the row vectors in the voltage difference matrix, find the 25th quantile and the 75th quantile of the vector, and obtain the upper limit of the abnormal threshold based on the 25th quantile and the 75th quantile. The upper limit of the abnormal threshold is based on the battery cell. Changes in real time due to changes in voltage;

S7. Traverse the voltage differences in the voltage difference matrix and determine whether each difference is greater than the upper limit of the abnormal threshold. If so, execute S8;

S8. Determine whether the difference between the voltages at adjacent time points is greater than the upper limit of the abnormal threshold. If so, execute S9; if not, execute S7;

S9. Mark this moment, obtain the marked time, and determine whether the frequency within each preset period of the preset number of times meets the preset frequency. If so, determine that a consistent mutation has occurred, and return the marked time and the abnormal battery cell. No.; if not, execute S7; where the frequency is the number of times the voltage difference is greater than the upper limit of the abnormal threshold.

Further, the battery signal data includes: Time, Charge_Status, Sum_Current and V.

Further, the preprocessing includes: cleaning the battery signal data and deleting invalid data; if there are abnormal values in the battery signal data, cleaning the battery signal data by a sliding average, and cleaning the battery signal data with a voltage greater than 6V or less than 1V. Data is deleted.

Further, the S4 includes: V has a total of N columns, each column represents a battery cell, and a total of N battery cells; the number of rows of V represents time, and the time unit is seconds;

Calculate the average voltage value V _{m mean} between each cell in the m-th row = (V _m1 + V _m2 +...+V _mN )/N.

Further, the S5 includes:

Calculate the difference between the voltage value of each cell and the average voltage value of its row: D _m = (V _m1 -V _{m mean} , V _m2 -V _{m mean} ,..., V _mN -V _{m mean} ), where D _m represents the difference between the voltage value of each cell in the m-th row and the average voltage value in the m-th row;

And the absolute value of each difference is taken as the voltage difference, and the voltage difference matrix D is obtained.

Further, the S6 includes:

Sort the row vectors in D and calculate the number of digits j=C×75%, where C is the number of columns. If C×75% is not an integer, round up;

The 75th quantile q ₇₅ is the average of the j-th item and the (j+1)-th item: q ₇₅ = (d _j +d _(j+1) )/2;

In the same way, q ₂₅ = (d _j' + d _{(j' + 1)} )/2, where j' = C × 25%, if C × 25% is not an integer, round up;

According to the 25th percentile and the 75th percentile, the upper limit of the abnormal threshold thre _up =q ₇₅ +C×(q ₇₅ -q ₂₅ ) is obtained, where the upper limit of the abnormal threshold changes in real time according to the change of the cell voltage.

Beneficial effects of this solution: This solution is based on a series of analysis and detection of actual vehicle operation data. The actual vehicle operation data, that is, the original message of the actual vehicle operation, is obtained directly under the actual vehicle operation conditions. According to the original message The analysis, preprocessing, feature extraction, analysis and detection can also be carried out under actual vehicle operation conditions. The data obtained are more real and accurate, can truly reflect the performance status of the battery pack in actual use, and solve the problem of false alarms. , improving the detection accuracy; and all electric vehicles will have original messages. By obtaining the analysis of the original messages, the detection cost can be effectively reduced without the need to establish additional efficient data transmission and processing systems, which improves timeliness and sustainability. sex;

Specifically, this solution obtains the original messages of actual vehicle operation and parses the corresponding battery signal data. The obtained data is more real and accurate, and can truly reflect the performance status of the battery pack in actual use; the battery signal data is processed Preprocessing further ensures the accuracy of the data, avoids the interference of invalid data during subsequent analysis, and reduces the amount of calculations; selects features from the preprocessed battery signal data, and extracts the battery signal data in the discharge state; then based on the battery signal data in the discharge state Signal data is used to detect power battery consistency anomalies. The upper limit of the abnormal threshold changes in real time according to changes in cell voltage. The difference values of voltages in the voltage difference matrix are traversed to determine whether each difference is greater than the upper limit of the abnormal threshold. If so, Then determine whether the difference between the voltages at adjacent time points is greater than the upper limit of the abnormal threshold. If so, mark this time t _i, and perform detection at the preset time, and determine whether consistent variation occurs through frequency. Since the battery signal data is Dynamically changing, this solution can monitor the changes in battery pack performance in real time, promptly discover potential consistency problems, such as voltage differences, capacity differences, etc., to avoid battery overheating, overcharging, over-discharging and other failures due to inconsistency, and Dynamically adjust and optimize to improve the safety of the battery pack. At the same time, problems in the battery pack can be dealt with in a timely manner to avoid premature failure of some battery modules, thus extending the life of the entire battery pack and helping to detect problems in the battery pack. Modules with low energy utilization efficiency, and take measures to optimize them to improve the energy utilization efficiency of the battery pack.

In addition, this solution can analyze the consistency problem by analyzing the voltage data, avoiding the comprehensive consideration of multiple factors, increasing the difficulty of technical implementation, and resulting in high complexity.

In summary, this solution can improve the accuracy, timeliness and sustainability of detection, reduce detection costs and complexity, facilitate dynamic adjustment and optimization, and thereby improve battery safety, lifespan and energy utilization efficiency.

Description of the drawings

Figure 1 is a schematic flowchart of an embodiment of the power battery consistency anomaly detection method based on actual vehicle operating data according to the present invention.

Detailed ways

The following is further detailed through specific implementation methods:

The embodiment is basically as shown in Figure 1: a power battery consistency anomaly detection method based on actual vehicle operating data, including the following content:

S1. Obtain the original message of the actual vehicle operation and parse out the corresponding battery signal data. The battery signal data includes: Time (time), Charge_Status (charge and discharge status), Sum_Current (current) and V (voltage matrix); The original message is a message that complies with the national standard GB32960;

S2. Preprocess the battery signal data; the preprocessing includes: cleaning the battery signal data and deleting invalid data, such as NAN, spaces, etc.; if there are abnormal values in the battery signal data, perform a preprocessing on the battery signal data. Cleaning of the sliding average, deletes data with a voltage greater than 6V or less than 1V, and performs consistency anomaly detection through Time, Charge_Status and V;

S3. From the preprocessed battery signal data, select features and extract the battery signal data of the discharge state, specifically: extract the corresponding Time and V when Charge_status==3;

S4. Calculate the average voltage value V _{m mean} between each cell according to the battery signal data in the discharge state;

Specifically, V has a total of N columns, each column represents a battery cell, and there are N battery cells in total; the number of rows of V represents time, and the time unit is seconds;

Calculate the average voltage value V _{m mean} between each cell in the m-th row = (V _m1 + V _m2 +...+V _mN )/N;

S5. Construct a voltage difference matrix D, including: calculating the difference between the voltage value of each cell and the average voltage value of its row, that is, calculating the difference between the voltage value of each column of each row and the average voltage value of its row. :

D _m = (V _m1 -V _{m mean} , V _m2 -V _{m mean} ,..., V _mN -V _{m mean} ); where D _m represents the voltage value of each cell in the m-th row and the average voltage value of the m-th row the difference between;

And take the absolute value of each difference as the voltage difference, and obtain the voltage difference matrix D;

S6. Obtain the upper limit of the abnormal threshold thre _up , including: traversing the row vector in D and obtaining the 25th and 75th percentiles of the vector;

Specifically, sort the row vectors in D and calculate the number of digits j=C×75%, where C is the number of columns. If C×75% is not an integer, round up;

The 75th quantile q ₇₅ is the average of the j-th item and the (j+1)-th item, that is, q ₇₅ = (d _j +d _(j+1) )/2;

In the same way, q ₂₅ =(d _j' +d _(j'+1) )/2, where j' = C×25%, if C×25% is not an integer, round up;

According to the 25th percentile and the 75th percentile, obtain the upper limit of the abnormal threshold thre _up =q ₇₅ +C×(q ₇₅ -q ₂₅ ), where the upper limit of the abnormal threshold changes in real time according to the change of the cell voltage;

S7. Traverse the difference values of the N voltages in each row in D, and determine whether each difference value is greater than the upper limit of the abnormal threshold. If so, execute S8; if not, continue executing S7;

That is: d _ij > thre _up , i represents time, j represents the j-th cell, then determine whether the difference in voltage between adjacent time points is greater than the upper limit of the abnormal threshold; for the N cell voltage values in each row in D Traverse the difference with the average voltage value to determine whether there is a difference greater than the upper limit of the abnormal threshold. If so, find the cell voltage and execute S8 for the cell voltage;

S8. Determine whether the difference between voltages at adjacent time points is greater than the upper limit of the abnormal threshold. If so, execute S9; if not, execute S7; where the adjacent time points are i+ 1, that is, the difference in voltage in step D of S7 is greater than the difference in the next voltage at which the upper limit of the abnormal threshold is exceeded;

S9. Mark this time t _i , t _i is the marking time, and determine whether the frequency within each preset period of the preset number of times meets the preset frequency. If so, determine that a consistent mutation has occurred, and return the marking time and occurrence Abnormal cell number; if not, continue to execute S7;

The frequency is the number of times the voltage difference is greater than the upper limit of the abnormal threshold.

Specifically, if the difference between the voltages at adjacent time points is greater than the upper limit of the abnormal threshold, mark this time t _i , where t _i is the marking time, that is, the N cell voltage values and the average voltage of each row in D in S7 Traverse the difference of values, among which there is a difference that is greater than the upper limit of the abnormal threshold, the time of the cell voltage, and determine that the difference in the voltage of the cell within a period of time (preset time) meets the judgment conditions in S7 Whether the frequency meets the preset frequency, if it meets the condition of S7 at least once every 2 hours after the marked time, the conditions of S7 at least once every 2 hours will be accumulated, and so on; when the accumulation occurs more than 4 times and If the time condition is met, it will be determined that a consistency abnormality has occurred, and the marked time and the cell number where the abnormality occurred will be returned; that is, the preset number of times in this embodiment is 4, the preset time is 2 hours, and the preset frequency is 1, that is, after the marked time Within eight hours, if S7 is met at least once every two hours starting from the marked time, it is determined that a consistent mutation has occurred;

If the above conditions are not met, S7 will continue to be executed, and then the next cell (voltage difference) that satisfies the judgment condition of S7 last time will continue to be traversed, and if the judgment condition of S7 is satisfied, S8 will be executed.

The above are only embodiments of the present invention. Common knowledge such as the specific structures and characteristics of the solutions are not described in detail here. Those of ordinary skill in the art are aware of all common knowledge in the technical field to which the invention belongs before the filing date or priority date. Technical knowledge, being able to know all the existing technologies in the field, and having the ability to apply conventional experimental methods before that date. Persons of ordinary skill in the field can, under the inspiration given by this application, combine their own abilities to perfect and implement this plan, Some typical well-known structures or well-known methods should not be an obstacle for those of ordinary skill in the art to implement the present application. It should be pointed out that for those skilled in the art, several modifications and improvements can be made without departing from the structure of the present invention. These should also be regarded as the protection scope of the present invention and will not affect the implementation of the present invention. effectiveness and patented practicality. The scope of protection claimed in this application shall be based on the content of the claims, and the specific implementation modes and other records in the description may be used to interpret the content of the claims.

Claims (6)

1. The power battery consistency abnormality detection method based on the real vehicle operation data is characterized by comprising the following steps of:

s1, acquiring an original message of the running of a real vehicle, and analyzing corresponding battery signal data;

s2, preprocessing battery signal data;

s3, selecting characteristics from the preprocessed battery signal data, and extracting battery signal data in a discharging state;

s4, calculating an average voltage value among each cell according to the battery signal data of the discharging state;

s5, calculating the difference value between the voltage value of each cell and the average voltage value of the current row of the cell, taking the absolute value of each difference value as a voltage difference, and obtaining a voltage difference matrix;

s6, traversing the row vectors in the voltage difference matrix, solving 25 quantiles and 75 quantiles of the vectors, and acquiring an abnormal threshold upper limit according to the 25 quantiles and the 75 quantiles, wherein the abnormal threshold upper limit changes in real time according to the change of the voltage of the battery cell;

s7, traversing the voltage difference values in the voltage difference matrix, judging whether each difference value is larger than the upper limit of the abnormal threshold, and if so, executing S8;

s8, judging whether the difference value is larger than the voltage of the upper limit of the abnormal threshold value or not, if so, executing S9; if not, executing S7;

s9, marking the time, obtaining the marking time, judging whether the frequency in each section of preset time of the preset times meets the preset frequency, if so, judging that the consistency variation occurs, and returning to the marking time and the abnormal cell number; if not, executing S7; wherein the frequency is the number of times the difference in voltage is greater than the upper threshold of abnormality.

2. The method for detecting the consistency abnormality of the power battery based on the real vehicle operation data according to claim 1, wherein the battery signal data includes: time, charge_status, sum_current, and V.

3. The method for detecting the consistency abnormality of the power battery based on the real vehicle operation data according to claim 2, wherein the preprocessing includes: cleaning the battery signal data and deleting invalid data; if there is an abnormal value in the battery signal data, the battery signal data is cleaned by one time of the sliding average value, and the data with the voltage more than 6V or less than 1V is deleted.

4. The method for detecting the consistency abnormality of the power battery based on the real vehicle operation data according to claim 2, wherein the S4 includes: v is N columns, each column represents one cell, and N cells are all arranged; the number of rows of V represents time in seconds;

calculating the average voltage value V between each cell of the mth row _mmean ＝(V _m1 +V _m2 +…+V _mN )/N。

5. The method for detecting the consistency anomaly of the power battery based on the real vehicle operation data according to claim 4, wherein the step S5 comprises:

calculating the difference between the voltage value of each cell and the average voltage value of the current row: d (D) _m ＝(V _m1 -V _mmean ，V _m2 -V _mmean ，…，V _mN -V _mmean ) Wherein D is _m Representing the difference between the voltage value of each cell in the mth row and the average voltage value of the mth row;

and taking the absolute value of each difference value as the voltage difference to obtain a voltage difference matrix D.

6. The method for detecting the consistency anomaly of the power battery based on the real vehicle operation data according to claim 5, wherein S6 comprises:

sorting the row vectors in D, and calculating the number of bits j=c×75%, wherein C is the number of columns, and rounding up if c×75% is not an integer;

75 quantiles q ₇₅ The average value of the j-th and (j+1) -th items: q ₇₅ ＝(d _j +d _(j+1) )/2；

Similarly, q ₂₅ ＝(d _j’ +d _(j’+1) ) 2, wherein j' =cx 25%, and rounding up if cx 25% is not an integer;

obtaining the upper limit thre of the abnormal threshold according to the 25 quantiles and the 75 quantiles _up ＝q ₇₅ +C×(q ₇₅ -q ₂₅ ) Wherein the abnormal upper threshold varies in real time according to the variation of the cell voltage.