A method and system for diagnosing abnormality of an electric vehicle battery

Abstract

The present invention provides a method and system for diagnosing abnormal battery in electric vehicles. The method comprises the following steps: collecting battery data of X vehicles during the charging stage, the battery data comprising the battery voltage or temperature, battery current and date collected at each preset time interval; dividing the battery current into multiple current intervals, and grouping the battery voltage or temperature according to the three dimensions of vehicle-date-current interval; constructing a statistical model based on the vehicle-date-current interval grouping data to calculate the abnormal voltage or temperature ratio of X vehicles; arranging the abnormal voltage or temperature ratio of X vehicles in descending order, and issuing a battery abnormality warning signal to the vehicles ranked in the top n. The method of the present invention can promptly issue a warning to the owner of a vehicle with a potential battery failure.

Description

Technical Field

[0001] The present invention relates to the technical field of electric vehicle battery safety, and in particular to a method and system for diagnosing abnormality of an electric vehicle battery. Background Art

[0002] Electric vehicles can catch fire due to battery failure, which endangers the life and property safety of the owner, causes huge economic losses to the owner, and causes casualties to the society. Therefore, by monitoring the battery performance indicators of electric vehicles, identifying potential problems and failures of electric vehicle batteries in advance, and promptly warning the owner to enter the station for maintenance, a series of social hazards caused by abnormal batteries in electric vehicles can be avoided.

[0003] Generally speaking, electric vehicles are in two states, namely charging state and discharging state. In the discharging state, electric vehicles convert electrical energy into kinetic energy to provide power for the vehicle. While the vehicle is driving, it is easily affected by personal and road conditions such as the driver's driving habits and the environment around the road. In this case, monitoring battery performance indicators to identify potential battery failures is easily affected by various external variables, resulting in reduced accuracy and stability of the analysis results.

[0004] The following method exists in the prior art: predicting the performance indicators of the battery through a time series prediction model, and comparing the predicted value with the abnormal threshold of the performance indicator set in advance. If the abnormal value is exceeded, it is considered that the battery may fail. This method has certain defects. On the one hand, it is very difficult to model and predict the historical values ​​of the performance indicators only through the time series model prediction, because the performance indicators of the battery, such as voltage, fluctuate greatly, and its peaks and troughs, and periodicity are difficult to accurately fit only by the time series model. On the other hand, it is still difficult to set a reasonable abnormal threshold for the performance indicator. Summary of the invention

[0005] One purpose of the present invention is to identify potential failures of a car battery in advance and issue a warning to the car owner in time.

[0006] A further object of the present invention is to improve the accuracy of identifying potential faults.

[0007] In particular, the present invention provides a method for diagnosing abnormality of an electric vehicle battery, comprising the following steps:

[0008] Collecting battery data of X vehicles during the charging stage, the battery data including battery voltage or temperature, battery current and date collected at preset intervals;

[0009] Dividing the battery into d current intervals, and grouping the voltage or temperature of the battery according to three dimensions of vehicle-date-current interval;

[0010] According to the vehicle-date-current interval grouping data, the voltage or temperature abnormality ratio of X vehicles is counted;

[0011] The voltage or temperature abnormality ratios of X vehicles are arranged in descending order, and the first n vehicles are issued with battery abnormality warning signals.

[0012] Optionally, the counting of the abnormal voltage or temperature ratios of X vehicles comprises the following steps:

[0013] The voltage or temperature abnormality ratio r of vehicle i in current interval j is calculated according to the following formula: j , where 0<i<X:

[0014]

[0015] Among them, q j represents the number of abnormal voltage or temperature records of vehicle i in current interval j, and W represents the total number of voltage or temperature records of vehicle i in d current intervals;

[0016] According to the voltage or temperature abnormality ratio r j The calculation formula is used to calculate the voltage or temperature anomaly ratio in other current intervals, thereby obtaining all voltage or temperature anomaly ratios r in d current intervals. 1 、r 2 ,...,r d , and then get the voltage or temperature abnormality ratio of vehicle i:

[0017]

[0018] Optionally, the voltage or temperature abnormality ratio r of vehicle i in current interval j is calculated according to the following formula: j , where 0<i<X:

[0019] In the steps,

[0020] Determine the voltage or temperature anomaly of vehicle i on a certain date within current interval j as follows:

[0021] According to the quartile principle, the mean or quantile value of the voltage or temperature of vehicle i on a certain date is recorded as z. If the following conditions are met, the mean or quantile value of the voltage or temperature of vehicle i on that date is considered abnormal:

[0022] z>Z 0.75 +1.5*(Z0.7 5-Z 0.25 )

[0023] Where Z is a random variable. In the current interval j, the mean or quantile value of the voltage or temperature of all vehicles on each date is the sample of the random variable Z. 0.75 and Z 0.25 is the quantile of the random variable Z;

[0024] Records an abnormality in the voltage or temperature of vehicle i.

[0025] Optionally, the voltage or temperature abnormality ratio r of vehicle i in current interval j is calculated according to the following formula: j , where 0<i<X:

[0026] In the steps,

[0027] Determine the voltage or temperature anomaly of vehicle i on a certain date within current interval j as follows:

[0028] According to the 3σ principle, the mean or percentile value of the voltage or temperature of vehicle i on a certain date is recorded as z. If the following conditions are met, the mean or percentile value of the voltage or temperature of vehicle i on that date is considered abnormal:

[0029] z>Z β

[0030] Where Z is a random variable. In the current interval j, the mean or quantile value of the voltage or temperature of all vehicles on each date is the sample of the random variable Z. β is the quantile of the random variable Z, P(Z>Z β )=1-β, β≥0.9, P represents the probability distribution function;

[0031] Records an abnormality in the voltage or temperature of vehicle i.

[0032] Optionally, in the step of counting the voltage or temperature anomaly ratios of X vehicles, an unsupervised anomaly detection algorithm in machine learning is used to directly identify the voltage or temperature anomaly ratios.

[0033] Optionally, the counting of the abnormal voltage or temperature ratios of X vehicles comprises the following steps:

[0034] Group vehicle i by date, merge the mean voltage or temperature of all current intervals, and various quantile values ​​into an h-dimensional vector Q(c 1 ,c 2 ,...,c h ), and Q(c 1 ,c 2 ,...,ch ) as the feature of vehicle i on a certain date;

[0035] Use the unsupervised anomaly detection algorithm to train and learn samples of all dates of X vehicles. Let the number of dates for each vehicle be g, then the total number of samples is X*g Q vectors;

[0036] Output the result of whether the voltage or temperature of vehicle i on all dates within g dates is abnormal, and obtain the total number of records of abnormal voltage or temperature of vehicle i t;

[0037] According to the formula ratio=t / g, the voltage or temperature abnormality ratio of vehicle i is obtained.

[0038] Optionally, collecting battery data of X vehicles during the charging phase includes the following steps:

[0039] Collect the cell voltage or cell temperature of each vehicle in the charging stage among X vehicles, the battery of the vehicle is composed of m cells, and the cell voltage or cell temperature at time k is recorded as U k1 , U k2 , ..., U km , and the highest voltage is U k,max , the minimum voltage is U k,min , the median voltage is U k,median ;

[0040] Use one of the following formulas to calculate the pressure difference or temperature difference The pressure difference or temperature difference As the battery voltage or temperature:

[0041]

[0042]

[0043]

[0044] Among them, U k,a％ Indicates P(U k >U k,a％ )=a%U k,b％ It means P(U k >U k,b％ )=b%, 0<a<100, 0<b<100, and a>b, U k is a random variable, and the voltage or temperature of m cells is a random variable U k Sample, P(U k ) is the probability distribution function;

[0045] Optionally, the pressure difference or temperature difference is calculated using one of the following formulas: In the step , determine which formula to choose to calculate the pressure difference or temperature difference as follows

[0046] The pressure difference or temperature difference calculated using the above three formulas Substitute this into the subsequent steps to verify the effect of battery abnormality diagnosis;

[0047] The pressure or temperature difference obtained using the best formula And use it as the voltage or temperature of the battery.

[0048] Optionally, dividing the battery current into d current intervals, and grouping the voltage or temperature of the battery according to three dimensions of vehicle-date-current interval includes the following steps:

[0049] The collected voltage or temperature of the battery is grouped according to vehicle and date, and the voltage or temperature of a vehicle that has no record at a certain time is filled with a null value;

[0050] Delete the battery voltage or temperature data of the vehicle-date with a small sample size to obtain the corrected vehicle-date grouping data;

[0051] The corrected vehicle-date grouped data are further grouped according to the current interval in the charging stage, thereby obtaining grouped data in three dimensions of vehicle-date-current interval.

[0052] Optionally, the step of deleting the battery voltage or temperature data of vehicle-dates with a small sample size to obtain the corrected vehicle-date grouping data comprises the following steps:

[0053] Assume that for X vehicles, the number of data records for each vehicle counted on a certain date is Cnt 1 ,Cnt 2 ,...,Cnt X , empty value records are not counted, Cnt is a random variable;

[0054] Set the alpha quantile of Cnt to Cnt α , that is, Cnt α It means P(Cnt>Cnt α )=1-α, where P(Cnt) is the probability distribution function;

[0055] Delete data records less than Cnt α Vehicle-date data;

[0056] Optionally, in the step of dividing the battery into d current intervals, the current intervals are determined according to the following method:

[0057] Setting multiple possible values ​​for the interval of the current;

[0058] Calculate the voltage or temperature of the vehicle-date in each interval Standard Deviation

[0059] will meet the standard deviation Possible values ​​that are less than or equal to the first preset threshold and greater than or equal to the second preset threshold are used as the current interval.

[0060] In particular, the present invention provides a system for diagnosing abnormalities in electric vehicle batteries, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the steps of the aforementioned method when executing the computer program.

[0061] According to the solution of the present invention, the data collected is the battery data of the vehicle in the charging stage. Compared with the battery data collected in the discharging stage, this avoids the influence of personal and road conditions such as the driver's driving habits and the environment around the driving road, which causes the analysis results to be less stable and accurate. In addition, the voltage or temperature of the battery is grouped according to the three dimensions of vehicle-date-current interval, and the voltage or temperature abnormality ratio of X vehicles is counted, and the voltage or temperature abnormality ratio of X vehicles is arranged in descending order, and the battery abnormality warning signal is issued to the first n vehicles, so that the warning can be issued to the vehicle owner with potential battery failure in time.

[0062] The present invention proposes to segment the charging current of the battery using some machine learning or statistical methods. Different current segmentation intervals are regarded as various operating conditions of the battery. This operating condition classification method is more refined than the traditional simple extensive classification such as charging and discharging, and prevents the performance indicators of the battery under different operating conditions from being affected by the battery process and physical laws on the big data model.

[0063] In addition, by using the method of counting the abnormal voltage or temperature ratios of X vehicles in the embodiment of the present invention, battery failures can be identified more accurately.

[0064] Based on the following detailed description of specific embodiments of the present invention in conjunction with the accompanying drawings, those skilled in the art will become more aware of the above and other objects, advantages and features of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0065] Hereinafter, some specific embodiments of the present invention will be described in detail in an exemplary and non-limiting manner with reference to the accompanying drawings. The same reference numerals in the accompanying drawings indicate the same or similar components or parts. It should be understood by those skilled in the art that these drawings are not necessarily drawn to scale. In the accompanying drawings:

[0066] Figure 1A schematic flow chart showing a method for diagnosing abnormality of an electric vehicle battery according to an embodiment of the present invention;

[0067] Figure 2 Shows Figure 1 The schematic flow chart of collecting battery data of X vehicles during the charging stage is shown;

[0068] Figure 3 Shows Figure 1 The schematic flow chart of the method for grouping the voltage or temperature of the battery according to the three dimensions of vehicle-date-current interval in step S200 is shown;

[0069] Figure 4 A schematic flow chart of a method for determining an interval of current according to an embodiment of the present invention is shown;

[0070] Figure 5 Shows Figure 1 A schematic flow chart of a method for calculating the abnormal voltage or temperature ratio of X vehicles in step S300 is shown;

[0071] Figure 6 A schematic flow chart of a method for determining abnormal voltage or temperature of a vehicle i on a certain date within a current interval j according to an embodiment of the present invention is shown;

[0072] Figure 7 A schematic flow chart of a method for determining abnormal voltage or temperature of a vehicle i on a certain date within a current interval j according to another embodiment of the present invention is shown;

[0073] Figure 8 The figure shows the ranking of the first 16 vehicles whose voltage or temperature abnormality ratio ratio of X vehicles is arranged in descending order according to an embodiment of the present invention;

[0074] Fig. 9 shows a pressure difference fluctuation diagram of five accident vehicles according to an embodiment of the present invention;

[0075] Fig.10 A schematic block diagram of a system for diagnosing abnormality of an electric vehicle battery according to an embodiment of the present invention is shown;

[0076] Fig.11 A schematic flow chart of calculating the abnormal voltage or temperature ratio of X vehicles according to another embodiment of the present invention is shown;

[0077] In the figure: 100-data acquisition module, 200-message parsing module, 300-battery indicator analysis module, 400-abnormal detection algorithm module, 500-warning signal distribution module. DETAILED DESCRIPTION

[0078] Figure 1 FIG. 1 is a schematic flow chart of a method for diagnosing abnormality of an electric vehicle battery according to an embodiment of the present invention. Figure 1 As shown, the method includes:

[0079] Step S100, collecting battery data of X vehicles during the charging stage, the battery data including the battery voltage or temperature, battery current and date collected at each preset time interval;

[0080] Step S200, dividing the battery into d current intervals, and grouping the battery voltage or temperature according to the three dimensions of vehicle-date-current interval;

[0081] Step S300, calculating the voltage or temperature abnormality ratio of X vehicles according to the vehicle-date-current interval grouping data;

[0082] Step S400, the voltage or temperature abnormality ratios of the X vehicles are arranged in descending order, and a battery abnormality warning signal is issued to the first n vehicles.

[0083] According to the solution of the present invention, the data collected is the battery data of the vehicle in the charging stage. Compared with the battery data collected in the discharging stage, this avoids the influence of personal and road conditions such as the driver's driving habits and the environment around the driving road, which causes the analysis results to be less stable and accurate. In addition, the voltage or temperature of the battery is grouped according to the three dimensions of vehicle-date-current interval, and the voltage or temperature abnormality ratio of X vehicles is counted, and the voltage or temperature abnormality ratio of X vehicles is arranged in descending order, and the battery abnormality warning signal is issued to the first n vehicles, so that the warning can be issued to the vehicle owner with potential battery failure in time.

[0084] The following is a detailed description with specific embodiments:

[0085] Embodiment 1:

[0086] Figure 2 Shows Figure 1 The schematic flow chart of collecting battery data of X vehicles during the charging phase is shown in FIG. Figure 2 As shown, the method includes:

[0087] Step S110, collecting the cell voltage or cell temperature of each vehicle in the charging stage among the X vehicles, the battery of the vehicle is composed of m cells, and the cell voltage or cell temperature at time k is recorded as U k1 , U k2 , ..., U km , and the highest voltage is U k,max , the minimum voltage is U k,min, the median voltage is U k,median ;

[0088] Step S120, calculate the pressure difference or temperature difference using the following formula: The pressure difference or temperature difference As the battery voltage or temperature:

[0089]

[0090]

[0091]

[0092] Among them, U k,a％ It means P(U k >U k,a％ )=a%U k,b％ It means P(U k >U k,b％ )=b%, 0<a<100, 0<b<100, and a>b, U k is a random variable, and the voltage or temperature of m cells is a random variable U k Sample, P(U k ) is the probability distribution function.

[0093] In step S120, the formula to be selected to calculate the pressure difference or temperature difference is determined in the following manner: The pressure difference or temperature difference calculated using the above three formulas Substitute it into the subsequent steps to verify the effect of battery abnormality diagnosis, and use the pressure difference or temperature difference obtained by the best formula And use it as the voltage or temperature of the battery.

[0094] In step S110, for each electric vehicle, the pressure difference or temperature difference can be used As a state of its battery at time k.

[0095] In step S120, This formula uses the difference between the highest voltage and the lowest voltage at time k as the pressure difference or temperature difference. This calculation method is simple, but there is a problem that it is easily affected by abnormal values ​​of the voltage or temperature of m cells. When the T-box uploads voltage or temperature data, it may cause errors in the voltage or temperature data of one or several cells due to various reasons, such as T-box failure, data acquisition system parsing errors, etc. If this erroneous data happens to be the lowest voltage, it will cause a large deviation in the pressure difference calculation, making the subsequent analysis results unreliable.

[0096] This formula uses the difference between the maximum voltage and the median voltage as the voltage difference, which can avoid the influence of abnormal battery cell voltage or battery cell temperature data to a certain extent.

[0097] This formula can completely avoid the influence of abnormal cell voltage data by further using the voltage difference of different quantiles as the voltage difference at time k. Theoretically, this formula is the most reasonable way to calculate the voltage difference. In the formula, a and b are parameters, which can be set as a=95 and b=5 according to experience. You can also set different values ​​according to the actual battery voltage data for testing and select a set of reasonable values. The formula is actually A specific form of the formula, that is, the formula when a=100 and b=50.

[0098] In the actual implementation of the electric vehicle fault warning solution, the above three formulas can be used for calculation respectively, and then the calculation method that is verified by the formula with the best final effect can be adopted. The evaluation method is to calculate the pressure difference or temperature difference obtained by the above formula. Substitute it into the subsequent steps and finally verify the effect of battery abnormality diagnosis.

[0099] Figure 3 Shows Figure 1 The schematic flow chart of the method for grouping the voltage or temperature of the battery according to the three dimensions of vehicle-date-current interval in step S200 is shown. The method includes:

[0100] Step S210, grouping the collected battery voltage or temperature according to vehicle and date, and filling the voltage or temperature of a vehicle with no record at a certain time with a null value;

[0101] Step S220, deleting the battery voltage or temperature data of the vehicle-date with a small sample size to obtain the corrected vehicle-date grouping data;

[0102] Step S230 , further grouping the corrected vehicle-date grouped data according to the current interval in the charging stage, thereby obtaining grouped data in three dimensions of vehicle-date-current interval.

[0103] In step S210, the collected voltage or temperature of the battery is grouped according to vehicle and date, where the date is accurate to the day. For vehicle i, data is collected every 10 seconds, and the moment every 10 seconds is recorded as moment k. The total number of data collections per day is K. Therefore, for a certain day, the pressure difference or temperature difference of the i-th vehicle is If there is no record at a certain time, the pressure difference at that time is filled with a null value. This method treats the pressure difference or temperature difference of a vehicle in a day as a random variable. The number of pressure difference or temperature difference records of vehicle X on a certain day will be counted to ensure It is a random variable, and its sample size must reach a certain number. If the sample size is small, the distribution function it constitutes may not accurately represent This random variable. Therefore, it is necessary to delete the vehicle-date pressure difference values ​​with a small sample size.

[0104] The present invention uses the finest data collection granularity, collects battery data of electric vehicle batteries every 10 seconds, and then constructs analysis indicators through statistical methods. This processing method can consider both the microscopic characteristics of the analysis indicators and the macroscopic characteristics of the analysis indicators. It comprehensively excavates the hidden value of the data from the perspective of macro and micro.

[0105] In step S220, for X vehicles, the number of data records for each vehicle counted on a certain date is Cnt 1 , Cnt 2 , ..., Cnt X In this method, empty value records are not counted, and Cnt is considered to be a random variable, and P(Cnt) is its probability distribution function. Delete the vehicle-date data with fewer data records than Cntα. α can be set to 0.1, 0.2, etc. according to experience. α can also be set to a range, such as α∈[0.05,0.4], and then conduct multiple test analyses based on specific battery data to find a reasonable value within this range.

[0106] In step S230, the current of the battery of the electric vehicle will fluctuate greatly during the charging process, for example, it will fluctuate from -200 to 0, or fluctuate within other larger ranges. If the pressure difference data in the range of [-200,0] is modeled, the change in the pressure difference will be affected by the large fluctuation of the current. Therefore, this method will divide the current into multiple intervals. For example, the current [-200,0] can be divided into 20 small current intervals of [-200,-190], [-190,180], ..., [-10,0] at intervals of 10 amperes. Each current interval is marked with serial numbers 1, 2, ..., 20, so that the vehicle-date data can be further grouped according to the three dimensions of vehicle-date-current interval. Assume that the current interval is finally divided into d current intervals, and generally d>5.

[0107] The current can be divided into small current intervals according to experience and at certain intervals, such as directly dividing it by 10 amperes. Of course, it can also be divided by 5 amperes, 15 amperes or other values.

[0108] Figure 4 FIG. 2 shows a schematic flow chart of a method for determining a current interval according to an embodiment of the present invention. Figure 4As shown, the method includes:

[0109] Step S231, setting a plurality of possible values ​​for the current interval;

[0110] Step S232, calculate the voltage or temperature of the vehicle-date in each interval Standard Deviation

[0111] Step S233: Possible values ​​that are less than or equal to the first preset threshold and greater than or equal to the second preset threshold are used as the current interval.

[0112] In step S231, possible values ​​may be, for example, 3, 5, 8, 10, 12, 15 amperes, etc. In step S233, the first preset threshold may be, for example, 0.02, and the second preset threshold may be, for example, 0.01. For example, when 8 amperes is selected as the interval, the pressure difference or temperature difference variable between the vehicle and the date is counted. The standard deviation meets the requirements, and the subsequent data processing divides the current interval into 8 amperes.

[0113] Figure 5 Shows Figure 1 FIG. 1 is a schematic flow chart of a method for calculating the voltage or temperature abnormality ratio of X vehicles in step S300. Figure 5 As shown, the method includes:

[0114] Step S310: Calculate the voltage or temperature abnormality ratio r of vehicle i in current interval j according to the following formula: j , where 0<i<X:

[0115]

[0116] Among them, q j represents the number of abnormal voltage or temperature records of vehicle i in current interval j, and W represents the total number of voltage or temperature records of vehicle i in d current intervals;

[0117] Step S320: according to the voltage or temperature abnormality ratio r j The calculation formula is used to calculate the voltage or temperature anomaly ratio in other current intervals, thereby obtaining all voltage or temperature anomaly ratios r in d current intervals. 1 、r 2 ,...,r d , and then get the voltage or temperature abnormality ratio of vehicle i:

[0118]

[0119] Before step S310, the following steps are also included:

[0120] 1) For a certain current interval j, delete the records where the pressure difference or temperature difference data of the vehicle-current interval j is null;

[0121] 2) For a vehicle i, the total number of data in its current interval j is pj;

[0122] 3) According to a certain statistical method, filter out the number of abnormal pressure difference or temperature difference records q of vehicle i in current interval j j ;

[0123] 4) Calculate the abnormal proportion q of vehicle i in current interval j j / pj;

[0124] 5) Multiply the abnormal proportion by the weight coefficient w j , assuming that the total number of records of vehicle i in d current intervals is W, then w j =p j / W.

[0125] In step S310,

[0126] In step S320,

[0127] Figure 6 FIG. 1 is a schematic flow chart of a method for determining abnormal voltage or temperature of a vehicle i on a certain date within a current interval j according to an embodiment of the present invention. Figure 6 As shown, the voltage or temperature anomaly of vehicle i on a certain date within current interval j is determined according to the following method:

[0128] In step S311, according to the quartile principle, the mean or quantile value of the voltage or temperature of vehicle i on a certain date is recorded as z. If the following conditions are met, the mean or quantile value of the voltage or temperature of vehicle i on this date is considered abnormal:

[0129] z>Z 0.75 +1.5*(Z 0.7 5-Z 0.25 )

[0130] Where Z is a random variable. In the current interval j, the mean or quantile value of the voltage or temperature of all vehicles on each date is the sample of the random variable Z. 0.75 and Z 0.25 is the quantile of the random variable Z;

[0131] Step S312, recording an abnormality in the voltage or temperature of vehicle i.

[0132] In step S311, the percentile value of the voltage or temperature of vehicle i on a certain date may be, for example, the 10% percentile, the 20% percentile, the 30% percentile, the 40% percentile, the 50% percentile, the 60% percentile, the 70% percentile, the 80% percentile, or the 90% percentile.

[0133] Figure 7 FIG. 2 shows a schematic flow chart of a method for determining abnormal voltage or temperature of a vehicle i on a certain date within a current interval j according to another embodiment of the present invention. Figure 7 As shown, the voltage or temperature anomaly of vehicle i on a certain date within current interval j is determined according to the following method:

[0134] Step S313: According to the 3σ principle, the mean or percentile value of the voltage or temperature of vehicle i on a certain date is recorded as z. If the following conditions are met, the mean or percentile value of the voltage or temperature of vehicle i on this date is considered abnormal:

[0135] z>Z β

[0136] Where Z is a random variable. In the current interval j, the mean or quantile value of the voltage or temperature of all vehicles on each date is the sample of the random variable Z. β is the quantile of the random variable Z, P(Z>Z β )=1-β, β≥0.9, P represents the probability distribution function;

[0137] Step S314, recording an abnormality in the voltage or temperature of vehicle i.

[0138] In step S313, β is generally set to 0.95 or 0.99. The smaller the value of β is set, the higher the probability of identifying the accident vehicle is, but the set value cannot be too small. If it is set too small, it will violate the 3σ principle, so generally β≥0.9.

[0139] In step S314, for example, if the pressure difference or temperature difference value of the current interval 1 of vehicle i on a certain day (such as 2020-10-25) is z and satisfies the formula in the above step S311 or step S313, then it is considered that the vehicle i is in the current interval 1, and the average pressure difference or temperature difference value on 2020-10-25 is an abnormal value, and the battery pressure difference of vehicle i is recorded as abnormal once.

[0140] In step S400, n can be 10, 20, etc., or can be calculated as a proportion of the total number of vehicles X, such as n=ceil(X*0.01), where ceil is a rounding function, such as ceil(1.2)=2.

[0141] In order to verify the accuracy of the method of the embodiment of the present invention, in the embodiment of the present invention, the X vehicles in the above method include Y vehicles that were accident vehicles caused by battery failure. When using the above method for analysis and modeling, the data on the accident date of the Y vehicles that are accident vehicles and all data after the accident are deleted. In the specific implementation, for example, X is 1000 and Y is 5. In the specific implementation, the specific use of step S100 is Formula, in step S230, the current is divided into small current intervals according to 10 amperes. Figure 8 As shown, it shows the ranking of the first 16 vehicles in descending order of the voltage or temperature abnormality ratio ratio of X vehicles in an embodiment of the present invention, wherein the vin column is the VIN number of the vehicle, and the ratio is the calculated abnormal pressure difference or temperature difference record ratio. Figure 8 It can be seen that according to the method described in this article, it can be considered that the first ceil (1000*0.01) = 10 vehicles have the risk of battery failure. From the figure, it can be seen that the top 3 vehicles among the first 10 vehicles are actual accident vehicles, and the other 7 vehicles are non-accident vehicles.

[0142] For the five accident vehicles with potential battery failures, the changes in pressure difference are as follows Fig. 9 As shown in the figure, the framed curves in the figure are the pressure difference fluctuation curves of the accident vehicle with vin=4 on the top and the pressure difference fluctuation curves of the accident vehicle with vin=5 on the bottom. Vin=4 and vin=5 are two accident vehicles that are not identified by the method described in this article. From the pressure difference change chart, it can be seen that the fluctuation of the pressure difference data is smaller than that of the other three vehicles. The pressure difference fluctuations of these two vehicles are basically stable, so the method in this article cannot detect that they are accident vehicles. From the figure, it can be roughly assessed that the vehicles with vin=4 and vin=5 were not identified because their battery failures were not reflected by the pressure difference fluctuations.

[0143] In particular, the present invention also provides a system for diagnosing abnormalities in electric vehicle batteries, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the steps of the aforementioned method when executing the computer program.

[0144] The following briefly describes the structure of the electric vehicle battery abnormality diagnosis system. Fig.10As shown, the system also includes a data acquisition module 100, which collects the message data uploaded by the electric vehicle T-Box in real time. The processor includes a message parsing module 200, a battery indicator analysis module 300, an abnormality detection algorithm module 400 and an early warning signal distribution module 500. The message parsing module 200 is connected to the data acquisition module 100, and is used to parse the battery current and the real-time voltage and temperature data of all cells or other performance indicator data according to the message specification.

[0145] Due to occasional failures of the T-Box device and errors in the transmission of uploaded messages resulting in messages not meeting specifications, some indicators obtained by parsing the messages may be abnormal values ​​or erroneous values. Therefore, it is necessary to set certain abnormal filtering rules to correct or directly delete the abnormal data. Through data cleaning and data processing of the battery indicator analysis module 300, the collected data is basically regular and neat battery performance indicator data.

[0146] The abnormality detection algorithm module 400 analyzes the battery performance indicators (such as pressure difference, temperature difference) in steps according to the method proposed above, and uses the algorithm described above to model and analyze its abnormal data, obtain the ratio of abnormal data of indicators of each vehicle, and then arrange them in descending order according to the abnormal data ratio to determine the high risk of battery failure of the top n vehicles. The early warning signal distribution module 500 is used to distribute risk warning signals to electric vehicle owners with high risk of battery failure through data distribution technology based on the analysis results of the abnormality detection algorithm module, reminding the owners to check the battery or enter the station for maintenance. The other features of the system correspond to the above methods one by one and will not be repeated here.

[0147] Embodiment 2:

[0148] The difference between the second embodiment and the first embodiment lies in the difference in step S300. In the second embodiment of the present invention, the method for counting the voltage or temperature abnormality ratio of X vehicles is to directly identify the voltage or temperature abnormality ratio by using the unsupervised anomaly detection algorithm in machine learning. The unsupervised anomaly detection algorithm can be, for example, one-class SVM, iForest, etc. For example, Fig.11 FIG. 4 is a schematic flow chart of calculating the abnormal voltage or temperature ratio of X vehicles according to an embodiment of the present invention. Fig.11 As shown, step S300 includes:

[0149] Step S321: group vehicle i by date, merge the mean values ​​of voltage or temperature in all current intervals, and various quantile values ​​into an h-dimensional vector Q(c 1 ,c 2 ,...,c h ), and Q(c 1 ,c2 ,...,c h ) as the feature of vehicle i on a certain date;

[0150] Step S322, using an unsupervised anomaly detection algorithm to train and learn samples of all dates of X vehicles, denoting the number of dates of each vehicle as g, then the total number of samples is X*g Q vectors;

[0151] Step S323, outputting the result of whether the voltage or temperature of vehicle i on all dates within g dates is abnormal, and obtaining the total number of records t of abnormal voltage or temperature of vehicle i;

[0152] Step S324 , according to the formula ratio=t / g, the voltage or temperature abnormality ratio of vehicle i is obtained.

[0153] The other technical solutions are the same as those in Example 1 and will not be described in detail here. This invention is funded by the National Key R&D Program (No. 2020YFB1711803).

[0154] At this point, those skilled in the art should recognize that, although multiple exemplary embodiments of the present invention have been shown and described in detail herein, many other variations or modifications of the common principles of the present invention can still be directly determined or derived from the contents disclosed in the present invention without departing from the spirit and scope of the present invention. Therefore, the scope of the present invention should be understood and recognized as covering all such other variations or modifications.

Claims

1. A method for diagnosing abnormality of an electric vehicle battery, It is characterized in that The steps include: Collecting battery data of X vehicles during the charging stage, the battery data including battery voltage or temperature, battery current and date collected at preset intervals; Dividing the battery current into d current intervals, and grouping the voltage or temperature of the battery according to three dimensions of vehicle-date-current interval; A statistical model is built based on the vehicle-date-current interval grouping data to calculate the proportion of abnormal voltage or temperature in X vehicles; Arrange the voltage or temperature abnormality ratios of X vehicles in descending order, and send battery abnormality warning signals to the first n vehicles; The method of calculating the voltage or temperature abnormality ratio of X vehicles includes the following steps: The voltage or temperature abnormality ratio r of vehicle i in current interval j is calculated according to the following formula: j , where 0<i<X: Among them, q j represents the number of abnormal voltage or temperature records of vehicle i in current interval j, and W represents the total number of voltage or temperature records of vehicle i in d current intervals; According to the voltage or temperature abnormality ratio r j The calculation formula is used to calculate the voltage or temperature anomaly ratio in other current intervals, thereby obtaining all voltage or temperature anomaly ratios r in d current intervals. 1 、r 2 ,...,r d , and then get the voltage or temperature abnormality ratio of vehicle i:

2. The method according to claim 1, It is characterized in that The voltage or temperature abnormality ratio r of vehicle i in current interval j is calculated according to the following formula: j , where 0<i<X: In the steps, Determine the voltage or temperature anomaly of vehicle i on a certain date within current interval j as follows: According to the quartile principle, the mean or quantile value of the voltage or temperature of vehicle i on a certain date is recorded as z. If the following conditions are met, the mean or quantile value of the voltage or temperature of vehicle i on that date is considered abnormal: from>from 0.75 +1.5*(From 0.75 -WITH 0.25 ) Where Z is a random variable. In the current interval j, the mean or quantile value of the voltage or temperature of all vehicles on each date is the sample of the random variable Z. 0.75 and Z 0.25 is the quantile of the random variable Z; Records an abnormality in the voltage or temperature of vehicle i.

3. The method according to claim 1, It is characterized in that The voltage or temperature abnormality ratio r of vehicle i in current interval j is calculated according to the following formula: j , where 0<i<X: In the steps, Determine the voltage or temperature anomaly of vehicle i on a certain date within current interval j as follows: According to the 3σ principle, the mean or percentile value of the voltage or temperature of vehicle i on a certain date is recorded as z. If the following conditions are met, the mean or percentile value of the voltage or temperature of vehicle i on that date is considered abnormal: from>from β Where Z is a random variable. In the current interval j, the mean or quantile value of the voltage or temperature of all vehicles on each date is the sample of the random variable Z. β is the quantile of the random variable Z, P(Z>Z β )=1-β, β≥0.9, P represents the probability distribution function; Records an abnormality in the voltage or temperature of vehicle i.

4. The method according to any one of claims 1 to 3, It is characterized in that The method of collecting battery data of X vehicles during the charging phase includes the following steps: Collect the cell voltage or cell temperature of each vehicle in the charging stage among X vehicles, the battery of the vehicle is composed of m cells, and the cell voltage or cell temperature at time k is recorded as U k1 , U k2 , ..., U km , and the highest voltage or temperature is recorded as U k,max , the minimum voltage or minimum temperature is U k,min , the median voltage or temperature is U k,median ; Use one of the following formulas to calculate the pressure difference or temperature difference The pressure difference or temperature difference As the battery voltage or temperature: Among them, U k,a％ Indicates P(U k >U k,a％ )=a%U k,b％ It means P(U k >U k,b％ )=b%, 0<a<100, 0<b<100, and a>b, U k is a random variable, and the voltage or temperature of m cells is a random variable U k Sample, P(U k ) is the probability distribution function; The pressure difference or temperature difference is calculated using one of the following formulas In the step , determine which formula to choose to calculate the pressure difference or temperature difference as follows The pressure difference or temperature difference calculated using the above three formulas Substitute this into the subsequent steps to verify the effect of battery abnormality diagnosis; The pressure or temperature difference obtained using the best formula And use it as the voltage or temperature of the battery.

5. The method according to claim 4, It is characterized in that The battery current is divided into d current intervals, and the voltage or temperature of the battery is grouped according to the three dimensions of vehicle-date-current interval, including the following steps: The collected voltage or temperature of the battery is grouped according to vehicle and date, and the voltage or temperature of a vehicle that has no record at a certain time is filled with a null value; Delete the battery voltage or temperature data of the vehicle-date with a small sample size to obtain the corrected vehicle-date grouping data; The corrected vehicle-date grouped data are further grouped according to the current interval in the charging stage, thereby obtaining grouped data in three dimensions of vehicle-date-current interval.

6. The method according to claim 5, It is characterized in that The method of deleting the battery voltage or temperature data of the vehicle-date with a small sample size to obtain the corrected vehicle-date grouping data includes the following steps: Assume that for X vehicles, the number of data records for each vehicle counted on a certain date is Cnt 1 ,Cnt 2 ,...,Cnt X , empty value records are not counted, Cnt is a random variable; Set the alpha quantile of Cnt to Cnt α , that is, Cnt α It means P(Cnt>Cnt α )=1-α, where P(Cnt) is the probability distribution function; Delete data records less than Cnt α Vehicle-date data; In the step of dividing the battery into d current intervals, the current intervals are determined according to the following method: Setting multiple possible values ​​for the interval of the current; Calculate the voltage or temperature of the vehicle-date in each interval Standard Deviation will meet the standard deviation Possible values ​​that are less than or equal to the first preset threshold and greater than or equal to the second preset threshold are used as the current interval.

7. A method for diagnosing abnormality of an electric vehicle battery, It is characterized in that The steps include: Collecting battery data of X vehicles during the charging stage, the battery data including battery voltage or temperature, battery current and date collected at preset intervals; Dividing the battery current into d current intervals, and grouping the voltage or temperature of the battery according to three dimensions of vehicle-date-current interval; A statistical model is built based on the vehicle-date-current interval grouping data to calculate the proportion of abnormal voltage or temperature in X vehicles; Arrange the voltage or temperature abnormality ratios of X vehicles in descending order, and send battery abnormality warning signals to the first n vehicles; In the step of counting the voltage or temperature anomaly ratios of the X vehicles, the voltage or temperature anomaly ratios are directly identified using an unsupervised anomaly detection algorithm in machine learning; Counting the abnormal voltage or temperature ratio of X vehicles includes the following steps: Group vehicle i by date, merge the mean voltage or temperature of all current intervals, and various quantile values ​​into an h-dimensional vector Q(c 1 ,c 2 ,...,c h ), and Q(c 1 ,c 2 ,...,c h ) as the feature of vehicle i on a certain date; Use the unsupervised anomaly detection algorithm to train and learn samples of all dates of X vehicles. Let the number of dates for each vehicle be g, then the total number of samples is X*g Q vectors; Output the result of whether the voltage or temperature of vehicle i on all dates within g dates is abnormal, and obtain the total number of records of abnormal voltage or temperature of vehicle i t; According to the formula ratio=t / g, the voltage or temperature abnormality ratio of vehicle i is obtained.

8. The method according to claim 7, It is characterized in that The method of collecting battery data of X vehicles during the charging phase includes the following steps: Collect the cell voltage or cell temperature of each vehicle in the charging stage among X vehicles, the battery of the vehicle is composed of m cells, and the cell voltage or cell temperature at time k is recorded as U k1 , U k2 , ..., U km , and the highest voltage or temperature is recorded as U k,max , the minimum voltage or minimum temperature is U k,min , the median voltage or temperature is U k,median ; Use one of the following formulas to calculate the pressure difference or temperature difference The pressure difference or temperature difference As the battery voltage or temperature: Among them, U k,a％ It means P(U k >U k,a％ )=a%U k,b％ It means P(U k >U k,b％ )=b%, 0<a<100, 0<b<100, and a>b, U k is a random variable, and the voltage or temperature of m cells is a random variable U k Sample, P(U k ) is the probability distribution function; The pressure difference or temperature difference is calculated using one of the following formulas In the step , determine which formula to choose to calculate the pressure difference or temperature difference as follows The pressure difference or temperature difference calculated using the above three formulas Substitute this into the subsequent steps to verify the effect of battery abnormality diagnosis; The pressure or temperature difference obtained using the best formula And use it as the voltage or temperature of the battery.

9. The method according to claim 8, It is characterized in that The battery current is divided into d current intervals, and the voltage or temperature of the battery is grouped according to the three dimensions of vehicle-date-current interval, including the following steps: The collected voltage or temperature of the battery is grouped according to vehicle and date, and the voltage or temperature of a vehicle that has no record at a certain time is filled with a null value; Delete the battery voltage or temperature data of the vehicle-date with a small sample size to obtain the corrected vehicle-date grouping data; The corrected vehicle-date grouped data are further grouped according to the current interval in the charging stage, thereby obtaining grouped data in three dimensions of vehicle-date-current interval.

10. The method according to claim 9, It is characterized in that The method of deleting the battery voltage or temperature data of the vehicle-date with a small sample size to obtain the corrected vehicle-date grouping data includes the following steps: Assume that for X vehicles, the number of data records for each vehicle counted on a certain date is Cnt 1 ,Cnt 2 ,...,Cnt X , empty value records are not counted, Cnt is a random variable; Set the alpha quantile of Cnt to Cnt α , that is, Cnt α It means P(Cnt>Cnt α )=1-α, where P(Cnt) is the probability distribution function; Delete data records less than Cnt α Vehicle-date data; In the step of dividing the battery into d current intervals, the current intervals are determined according to the following method: Setting multiple possible values ​​for the interval of the current; Calculate the voltage or temperature of the vehicle-date in each interval Standard Deviation will meet the standard deviation Possible values ​​that are less than or equal to the first preset threshold and greater than or equal to the second preset threshold are used as the current interval.

11. A system for diagnosing abnormality of an electric vehicle battery, It is characterized in that The method comprises a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the steps of the method according to any one of claims 1 to 10 when executing the computer program.

Images