Power battery detection method, terminal and storage medium

Abstract

The invention discloses a power battery detection method, a terminal and a storage medium, and the detection method comprises the steps: obtaining state information corresponding to a power battery of a current vehicle when a preset condition is satisfied; inputting the state information into a first attribution model to obtain a working condition level attribution result; the first attribution model is obtained by pre-training according to first historical state information of a plurality of power batteries under different endurance working conditions and a first label corresponding to the first historical state information, and the content of the first label is used for representing a main reason for generating the historical state information; inputting the state information and the working condition level attribution result into a second attribution model to obtain a single vehicle level attribution result; and displaying the working condition level attribution result and the single vehicle level attribution result. According to the scheme, the accuracy of power battery detection is improved.

Description

Technical Field

[0001] The present application relates to the technical field of power batteries, and in particular to a detection method, terminal and storage medium for a power battery. Background Art

[0002] As the number of pure electric vehicles continues to increase, the endurance of electric vehicles has also received widespread attention. Given that the endurance of electric vehicles mainly depends on the power source, which is the power battery, how to determine the cause of the endurance problem has become an urgent problem to be solved in the electric vehicle field.

[0003] In the related art, static testing or manual inspection is often used, but due to the small amount of battery data obtained, it is difficult to effectively determine the problems existing in the power battery. Summary of the invention

[0004] The present application provides a power battery detection method, terminal and storage medium.

[0005] A technical solution adopted in this application is to provide a power battery detection method, the method comprising:

[0006] Under the preset conditions, obtain the status information corresponding to the power battery of the current vehicle;

[0007] The state information is input into a first attribution model to obtain a working condition-level attribution result; the first attribution model is pre-trained based on first historical state information of a plurality of power batteries under different endurance working conditions and first labels corresponding to the first historical state information, wherein the content of the first label is used to indicate the main reason for generating the first historical state information;

[0008] The state information and the working condition-level attribution result are input into the second attribution model to obtain the single-vehicle-level attribution result; the second attribution model is pre-trained based on the second historical state information of the current vehicle under different cruising conditions and the second label corresponding to the second historical state information; the content of the second label is obtained based on the working condition-level attribution result output by the first attribution model during the training process;

[0009] Displays the working condition-level attribution results and the vehicle-level attribution results.

[0010] Optionally, before the step of training the first attribution model, the method includes:

[0011] Acquire multiple target operating condition indicators corresponding to the first historical state information; the target operating condition indicators at least include energy consumption indicators, battery indicators, aging indicators, environmental indicators, and endurance indicators;

[0012] Using the fluctuation contribution method, the main causes are determined from multiple target operating indicators;

[0013] A first label corresponding to the first historical status information is constructed using the main reason.

[0014] Optionally, the fluctuation contribution method is used to determine the main causes from multiple target operating conditions indicators, including:

[0015] Acquire multiple benchmark operating condition indicators corresponding to the benchmark state information;

[0016] Calculate the target indicator score corresponding to each target operating condition indicator, and determine the target total score corresponding to the first historical state information; and calculate the benchmark indicator score corresponding to each benchmark operating condition indicator, and determine the benchmark total score corresponding to the benchmark state information;

[0017] Calculate the difference between the target total score and the benchmark total score to get the total score difference;

[0018] Calculate the difference between the target indicator score and the corresponding benchmark indicator score to obtain the indicator score difference;

[0019] Calculate the quotient of the indicator score difference and the total score difference to obtain the indicator proportion corresponding to the target operating condition indicator, and determine the main cause from multiple target operating condition indicators based on the indicator proportion.

[0020] Optionally, the main cause is determined from multiple target operating condition indicators based on the indicator proportion, including:

[0021] In response to the fact that the indicator proportion corresponding to the first candidate operating condition indicator is greater than or equal to a preset threshold, and the first candidate operating condition indicator is an atomic indicator, the first candidate operating condition indicator is taken as the main reason;

[0022] or,

[0023] In response to the fact that the indicator ratio corresponding to any target operating condition indicator is less than a preset threshold, and at least two second candidate operating condition indicators with the largest indicator ratios are atomic indicators, the second candidate operating condition indicator is taken as the main reason.

[0024] Optionally, after the step of calculating the quotient of the indicator score difference and the total score difference to obtain the indicator proportion corresponding to the target operating condition indicator, the following steps are included:

[0025] In response to the fact that the indicator proportion corresponding to the third candidate operating condition indicator is greater than or equal to a preset threshold, and the third candidate operating condition indicator is not an atomic indicator, the third candidate operating condition indicator is disassembled to obtain a plurality of first sub-operating condition indicators;

[0026] According to the proportion of indicators corresponding to the first sub-operating condition indicators, the main cause is determined from the multiple first sub-operating condition indicators and the remaining target operating condition indicators.

[0027] Optionally, after the step of calculating the quotient of the indicator score difference and the total score difference to obtain the indicator proportion corresponding to the target operating condition indicator, the following steps are included:

[0028] In response to the fact that the proportion of the indicator corresponding to any target operating condition indicator is less than a preset threshold, and at least two fourth candidate operating condition indicators with the largest proportion of the indicators are not atomic indicators, the fourth candidate operating condition indicator is disassembled to obtain a plurality of second sub-operating condition indicators;

[0029] According to the proportion of indicators corresponding to the second sub-operating condition indicators, the main cause is determined from the multiple second sub-operating condition indicators and the remaining target operating condition indicators.

[0030] Optionally, display the working condition level attribution results and the single vehicle level attribution results, including:

[0031] Determine a plurality of operating condition indicators from the state information, and generate a first structured data table according to the plurality of operating condition indicators;

[0032] Generate a second structured data table according to the state information and the working condition level attribution result, and generate a third structured data table according to the state information and the single vehicle level attribution result;

[0033] A first structured data table, a second structured data table, and a third structured data table are displayed.

[0034] Optionally, under the preset conditions, before the step of obtaining the status information corresponding to the power battery of the current vehicle, the method includes:

[0035] Obtain the current vehicle's driving time and average driving speed, as well as the power battery's charge change value;

[0036] In response to the driving duration being greater than or equal to a preset duration, the average driving speed being greater than or equal to a preset speed, and the battery charge change value being greater than or equal to a preset change value, it is determined that the preset conditions are met.

[0037] Another technical solution adopted by the present application is to provide a terminal device, the terminal device comprising a memory and a processor connected to the memory;

[0038] The memory is used to store program data, and the processor is used to execute the program data to implement the power battery detection method as described above.

[0039] Another technical solution adopted by the present application is to provide a computer storage medium, which is used to store program data. When the program data is executed by a computer, it is used to implement the power battery detection method as described above.

[0040] The beneficial effects of the present application are as follows: based on the first historical status information of several power batteries under different endurance conditions, and the first label corresponding to the main reason for generating the historical status information, the first attribution model is pre-trained. Secondly, the second historical status data of the current vehicle under different endurance conditions is input into the trained first attribution model to obtain the second label corresponding to the second historical status data, and the second attribution model is pre-trained using the second historical status data and the second label, and the training speed of the second attribution model is improved through the model cascading method, and the accuracy of attribution is improved. Thirdly, the status information corresponding to the power battery of the current vehicle is input into the first attribution model to obtain the operating condition-level attribution result, and the status information and the operating condition-level attribution result are input into the second attribution model to obtain the single-vehicle-level attribution result. By displaying the operating condition-level attribution result and the single-vehicle-level attribution result, the attribution result corresponding to the power battery can be determined faster and more accurately. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0042] Figure 1 It is a flow chart of an embodiment of a method for detecting a power battery provided by the present application;

[0043] Figure 2 It is a flow chart of another embodiment of the power battery detection method provided by the present application;

[0044] Figure 3 It is a structural diagram of an embodiment of a terminal device provided by the present application;

[0045] Figure 4 It is a structural diagram of an embodiment of a computer storage medium provided by the present application. DETAILED DESCRIPTION

[0046] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0047] The low endurance problem of power batteries refers to the fact that after a power battery is fully charged, the time or driving distance that it can support new energy vehicles is significantly less than its design or expected value. For example, the design value of a power battery is that it can support a new energy vehicle to travel 550 kilometers when fully charged, but in a certain actual scenario, the power battery can only support the new energy vehicle to travel 450 kilometers when fully charged. At this time, the power battery may have a low endurance problem.

[0048] At present, the diagnosis of low battery life problems mostly relies on static testing and manual inspection. These methods usually have the following limitations:

[0049] (1) Insufficient data processing capabilities. Traditional methods often have difficulty processing and analyzing massive amounts of real-time battery data, resulting in the inability to detect potential problems in a timely manner.

[0050] (2) Lack of real-time performance. Static testing cannot monitor the battery status in real time, and users often realize that the battery life has decreased only after a problem occurs.

[0051] (3) Low diagnostic accuracy. Analysis based on a single or small number of data points may not fully reflect the health status of the power battery, which can easily lead to misdiagnosis or missed diagnosis.

[0052] (4) Data island phenomenon. In many cases, power battery data is not centrally stored and shared, resulting in information islands. Battery data between different vehicles is difficult to integrate, and algorithm differences are too large, which hinders the realization of global analysis and intelligent decision-making.

[0053] (5) Lack of predictive maintenance. Existing monitoring systems usually only diagnose low battery life problems that have already occurred. They lack predictive maintenance capabilities and are unable to identify potential battery problems in advance. As a result, users only become aware of the battery life after it has significantly decreased, resulting in a decline in user experience.

[0054] This application mainly involves a method for improving the accuracy of power battery detection results. Unlike traditional methods, this application improves the accuracy of attribution results by means of model cascading. At the same time, by displaying the working condition-level attribution results and the single-vehicle-level attribution results, the accuracy of attribution is further improved, helping technicians to quickly identify the root cause of the low endurance of the current vehicle, and to lock in vehicles with abnormal indicators but not yet low endurance in advance, providing predictive maintenance.

[0055] Please refer to Figure 1 , Figure 1 It is a flow chart of an embodiment of a method for detecting a power battery provided in the present application.

[0056] like Figure 1 As shown, the detection method of the power battery in the embodiment of the present application may specifically include the following steps:

[0057] S1, obtaining status information corresponding to the power battery of the current vehicle under preset conditions.

[0058] The power battery detection method provided in the present application is mainly performed by a power battery detection device. In some embodiments, the power battery detection device may be a battery management system (BMS) of the current vehicle or a battery management control unit itself. In some embodiments, the power battery detection device may be a device that is communicatively connected to the battery management system or the battery management control unit. For example, the device may be a device for monitoring images, a user equipment (UE), a mobile device, a user terminal, a terminal, a cellular phone, a cordless phone, a personal digital assistant (PDA), a handheld device, a computing device, a vehicle-mounted device, a wearable device, and an autonomous vehicle, a robot, a security system, or any one or more products of glasses and helmets for augmented reality or virtual reality. In some possible implementations, the power battery detection method may be implemented by a processor calling a computer-readable instruction stored in a memory. In some possible implementations, the power battery detection device is a cloud server.

[0059] Specifically, the power battery detection device obtains status information corresponding to the power battery of the current vehicle under preset conditions.

[0060] For example, the preset condition may refer to the current vehicle in a certain operation process, the driving distance is greater than the distance threshold, the driving speed is greater than the speed threshold, and the vehicle operation time is greater than the time threshold. This step avoids using some invalid data such as short driving distance, low driving speed, and short vehicle operation time as abnormal detection data by setting the preset condition.

[0061] Optionally, the status information may include at least one of the following data: the initial voltage value and final voltage value of each cell in the power battery, the SOC (State of Charge), SOE (State of Energy), SOH (State of Health), the ambient temperature of the power battery, the total mileage of the current vehicle, etc.

[0062] It should be noted that the current vehicles or new energy vehicles mentioned in this application are pure electric vehicles.

[0063] In some possible embodiments, the power battery detection device determines whether there is abnormal data in the state information, and if so, removes the abnormal data. Exemplarily, the power battery detection device determines candidate abnormal data that exceeds a preset interval from the state information, and then uses corresponding data information in other state information with similar data content to determine whether the candidate abnormal data is abnormal data.

[0064] In some embodiments, the preset condition also includes that the current vehicle is in a low-range running state. If so, the power battery detection device detects the state information corresponding to the power battery of the current vehicle.

[0065] In some possible embodiments, the status information may be real-time status information of the power battery.

[0066] S2, inputting the state information into the first attribution model to obtain the working condition level attribution result.

[0067] The first attribution model is pre-trained based on first historical state information of a plurality of power batteries under different endurance conditions and first labels corresponding to the first historical state information. The content of the first label is used to indicate the main reason for generating the first historical state information.

[0068] In some embodiments, the first attribution model is a CatBoost model, which has good performance (high accuracy and fast training speed) in classification problems. Of course, according to actual needs, the first attribution model can also be other machine learning models or deep learning models.

[0069] Specifically, the first attribution model is pre-trained by the power battery detection device using first historical state information of a number of power batteries under different low-endurance conditions and first labels corresponding to the first historical state information.

[0070] It should be noted that the several power batteries in this step may refer to all power batteries from the same batch, or all power batteries from the same batch and installed on the same pure electric vehicle model, or all power batteries installed on the same pure electric vehicle model.

[0071] In some possible embodiments, the content of the first tag is used to indicate a root cause of generating the first historical state information, wherein the root cause refers to a fundamental reason for generating the first historical state.

[0072] In some embodiments, the power battery detection device uses the state information to determine the operating condition corresponding to the current vehicle. It can be understood that the operating condition is a low-range operating condition. Further, the power battery detection device inputs the state information into the first attribution model to obtain an operating condition-level attribution result, which is used to indicate the main reason (or root cause) for the operating condition.

[0073] S3, inputting the state information and the working condition level attribution result into the second attribution model to obtain the single vehicle level attribution result.

[0074] Among them, the second attribution model is pre-trained based on the second historical state information of the current vehicle under different endurance conditions and the second label corresponding to the second historical state information; the content of the second label is obtained based on the condition-level attribution result output by the first attribution model during the training process.

[0075] In some possible embodiments, the first historical state information contains battery identification information corresponding to the power battery, and the power battery detection device uses the battery identification information to determine the vehicle identification information corresponding to the power battery. Further, the power battery detection device traverses the first historical state information, finds all first historical state information of the vehicle identification information belonging to the current vehicle, and obtains the second historical state information of the current vehicle under different endurance conditions.

[0076] Furthermore, the power battery detection device inputs the second historical state information into the first attribution model to obtain a second label corresponding to the second historical state information. It can be understood that the second label can also represent the main reason for generating the second historical state information.

[0077] In some embodiments, the second attribution model is a LightGBM model, which has good performance in classification problems (with the advantages of high reasoning accuracy, fast training speed, and fast reasoning speed). Of course, according to actual needs, the second attribution model can also be other machine learning models or deep learning models.

[0078] The first attribution model and the second attribution model provided in the embodiment of the present application are trained in a cascade manner. After the first attribution model is trained, the second historical state information is input into the trained first attribution model to obtain a second label. Furthermore, the second historical state information and the second label are used to train the second attribution model, which can avoid the problem of excessive data volume and improve the training speed of the two attribution models, and can also improve the accuracy of the single-vehicle-level attribution results.

[0079] It should be noted that the operating condition-level attribution result obtained in S2 is mainly used to indicate the main reason for a single low-range operating condition. For a single vehicle (such as the current vehicle), the corresponding operating condition-level attribution results may be different under different low-range operating conditions. Relying solely on a number of operating condition-level attribution results for detection may not be able to more accurately identify the main reason for the low-range operating condition of the power battery. Based on this, this step obtains a single-vehicle-level attribution result by inputting the status information and the operating condition-level attribution information into the second attribution model, which can more carefully determine the main reason for the low range of the current vehicle and achieve more accurate fault detection and troubleshooting.

[0080] S4, displays the working condition level attribution results and the single vehicle level attribution results.

[0081] In some embodiments, the power battery detection device displays the operating condition level attribution results and the single vehicle level attribution results in the form of a data table.

[0082] In some possible embodiments, the power battery detection device displays the operating condition-level attribution results, the vehicle-level attribution results, and the reference state information corresponding to the power batteries of other vehicles. Furthermore, by comparing and verifying the above-mentioned attribution results and the reference state information, the accuracy of attribution can be further improved, and the root cause of the low endurance problem of the current vehicle can be determined more quickly.

[0083] The above scheme pre-trains the first attribution model based on the first historical status information of several power batteries under different endurance conditions and the first label corresponding to the main reason for generating the historical status information. Secondly, the second historical status data of the current vehicle under different endurance conditions is input into the trained first attribution model to obtain the second label corresponding to the second historical status data. The second attribution model is pre-trained using the second historical status data and the second label. By means of model cascading, the training speed of the second attribution model is improved, and the accuracy of attribution is improved. Thirdly, the status information corresponding to the power battery of the current vehicle is input into the first attribution model to obtain the operating condition-level attribution result, and the status information and the operating condition-level attribution result are input into the second attribution model to obtain the single-vehicle-level attribution result. By displaying the operating condition-level attribution result and the single-vehicle-level attribution result, the attribution result corresponding to the power battery can be determined faster and more accurately.

[0084] In some embodiments, before training the first attribution model, the following steps may be included:

[0085] S11, obtaining a plurality of target operating condition indicators corresponding to the first historical state information.

[0086] Among them, the target operating condition indicators include at least energy consumption indicators, battery indicators, aging indicators, environmental indicators, and endurance indicators.

[0087] Specifically, the power battery detection device obtains a plurality of target operating condition indicators corresponding to the first historical state information.

[0088] In some embodiments, the energy consumption index is composed of comprehensive energy consumption Composition, further, comprehensive energy consumption Drive energy consumption and non-drive energy consumption Composition, not drive energy consumption Energy consumption by air conditioning Other energy consumption And so on.

[0089] In some possible embodiments, the comprehensive energy consumption satisfies the following relationship:

[0090] C a =E out -E in

[0091] Among them, E out E is the output energy of the power battery under a single operation condition (kW·h). un It is the input energy (kW·h) of the power battery under a single operation condition. It can be understood that when the vehicle is performing regenerative braking, the power motor inputs energy to the power battery.

[0092] In some embodiments, the non-driving energy consumption C n It can be the energy consumption generated by the vehicle's sensors, controllers, and other on-board equipment.

[0093] In some embodiments, the battery indicator can be obtained by the cell voltage difference V d , Voltage consistency V uinf , SOC jump maximum value SOC jump times SOE jump maximum value SOE transition times The estimated error M of the battery cell SOC e And so on.

[0094] Among them, voltage consistency V uinf The voltage dispersion ratio can be used for quantitative analysis. Specifically, the voltage dispersion ratio satisfies the following relationship:

[0095]

[0096] in, is the average voltage of all cells in the power battery, V min is the minimum voltage among all cells in the power battery, V maxIt is the maximum voltage of all cells in the power battery.

[0097] In some embodiments, the estimated error M of the battery cell SOC is e Specifically, the following relations are satisfied:

[0098]

[0099] in, The maximum value of the SOC of the battery cell; It is the minimum value of the SOC of the battery cell.

[0100] In some embodiments, the aging indicator can be a current SOH value SOH of the power battery. now , calendar time T from power battery off-line to detection (data collection) c , the total vehicle mileage corresponding to the power battery M a And so on.

[0101] In some embodiments, the environmental indicator may be the ambient temperature T of the power battery. e And so on.

[0102] In some embodiments, the battery life indicator may be represented by a battery life achievement rate E t 、Estimated range E e Among them, the battery life achievement rate E t , which can specifically satisfy the following relationship:

[0103]

[0104] Among them, M r is the actual mileage change under a single operating condition, M t It is the change in displayed mileage under a single operating condition.

[0105] Furthermore, the estimated range E e , the following relationship can be satisfied:

[0106]

[0107] Among them, M r is the actual mileage change under a single operating condition, and ΔSOC is the SOC change of the power battery under a single operating condition.

[0108] S12, using the fluctuation contribution method, determine the main cause from multiple target operating condition indicators.

[0109] Specifically, the power battery detection device uses a fluctuation contribution method to determine the main reason for generating the first historical state information from energy consumption indicators, battery indicators, aging indicators, environmental indicators, and endurance indicators.

[0110] This embodiment adopts the fluctuation contribution method and chain-replaces abnormal indicators, which can directly locate the core features that cause low endurance conditions, improve the interpretability of the model, make the decision-making process more transparent, and distinguish the signals that truly affect the target variable, as well as exclude unnecessary noise, thereby optimizing the production of the first label and improving the accuracy and stability of the trained first attribution model.

[0111] In some possible application scenarios, the use of the fluctuation contribution method may require a large amount of computing power. This embodiment may also use the interquartile range method (IQR), by determining the first quartile and the third quartile corresponding to all the first historical state information, determining the IQR threshold, and then using the IQR threshold to determine the upper and lower bounds of the abnormal indicator, thereby determining the main cause. Compared with the fluctuation contribution method, the interquartile range method may have the problem of lower accuracy.

[0112] S13: construct a first label corresponding to the first historical status information by using the main reason.

[0113] Specifically, the power battery detection device uses the main reason as the first label of the corresponding first historical status information.

[0114] In the above steps, the first attribution model is pre-trained based on the first historical status information of several power batteries under different endurance conditions and the first label corresponding to the main reason for generating the historical status information. Secondly, by using the fluctuation contribution method and chain replacement of abnormal indicators, the core features that cause low endurance can be directly located, the interpretability of the model can be improved, the decision-making process can be made more transparent, and the signals that truly affect the target variable can be distinguished from unnecessary noise, thereby optimizing the production of training labels and improving the accuracy and stability of the model.

[0115] Furthermore, although the embodiments of the present application reveal that the fluctuation contribution method can be used to determine the main cause, the interquartile range method also has the ability to determine the main cause, but the accuracy of the main cause determined by this method may drop significantly, resulting in the first attribution model obtained by subsequent training having a low accuracy problem.

[0116] In some possible embodiments, S12 may include the following steps:

[0117] S21, obtaining a plurality of reference operating condition indicators corresponding to the reference state information.

[0118] Among them, the baseline status information refers to the status information of the power battery under non-low endurance conditions (normal endurance conditions).

[0119] In some possible embodiments, the reference state information may be historical state information of the power battery of the current vehicle under non-low cruising range conditions. In some possible embodiments, the reference state information may be historical state information of the power battery of the same model under non-low cruising range conditions.

[0120] In some embodiments, the benchmark operating condition indicator and the target operating condition indicator are of the same indicator type, that is, the benchmark operating condition indicator includes at least an energy consumption indicator, a battery indicator, an aging indicator, an environmental indicator, and a battery life indicator.

[0121] S22, calculating the target indicator score corresponding to each target operating condition indicator, and determining the target total score corresponding to the first historical state information.

[0122] Among them, the target operating condition indicators include at least energy consumption indicators, battery indicators, aging indicators, environmental indicators, and endurance indicators.

[0123] In some embodiments, the energy consumption index is composed of comprehensive energy consumption Composition, further, comprehensive energy consumption Drive energy consumption and non-drive energy consumption Composition, not drive energy consumption Energy consumption by air conditioning Other energy consumption The target index scores corresponding to the energy consumption index can satisfy the following relationship:

[0124]

[0125] Among them, C r is the target index score corresponding to the energy consumption index, m1, m2, and m3 are the weight coefficients corresponding to driving energy consumption, air conditioning energy consumption, and other energy consumption, respectively.

[0126] In some embodiments, the battery indicator is a battery cell voltage difference Voltage consistency SOC jump maximum value SOC jump times SOE jump maximum value SOE transition times Estimation error of battery cell SOC The target indicator scores corresponding to the battery indicators can satisfy the following relationship:

[0127]

[0128] Among them, B ris the target indicator score corresponding to the battery indicator, and n1, n2, n3, n4, n5, n6, and n7 are the weight coefficients corresponding to the cell voltage difference, voltage consistency, SOC jump maximum value, SOC jump number, SOE jump maximum value, SOE jump number, and cell SOC estimation error respectively.

[0129] In some embodiments, the aging indicator is a current SOH value of the power battery. Calendar time from power battery off-line to detection (data collection) Total vehicle mileage corresponding to the power battery The target index scores corresponding to the aging index can satisfy the following relationship:

[0130]

[0131] Among them, O r is the target indicator score corresponding to the aging indicator, p1, p2, and p3 are the current SOH value of the power battery, the calendar time from the power battery being offline to the test (when data is collected), and the weight coefficient corresponding to the total vehicle mileage corresponding to the power battery.

[0132] In some embodiments, the environmental indicator is the ambient temperature of the power battery. The target indicator scores corresponding to the environmental indicators satisfy the following relationship:

[0133]

[0134] Among them, E r is the target index score corresponding to the environmental index, and t1 is the weight coefficient corresponding to the ambient temperature.

[0135] Furthermore, the target total score satisfies the following relationship:

[0136]

[0137] Among them, C r is the target index score corresponding to the energy consumption index, B r is the target index score corresponding to the battery index, O r is the target index score corresponding to the aging index, E r is the target indicator score corresponding to the environmental indicator, and m, n, p, and t are the weight coefficients corresponding to the energy consumption indicator, battery indicator, aging indicator, and environmental indicator, respectively.

[0138] In short, the power battery testing device calculates the weighted sum of the target indicator scores corresponding to the energy consumption index, battery index, aging index, environmental index, and endurance index to obtain the target total score.

[0139] S23, calculating the benchmark index score corresponding to each benchmark operating condition index, and determining the benchmark total score corresponding to the benchmark state information.

[0140] Since the index type of the benchmark operating condition index is the same as the type of the target operating condition index, the benchmark operating condition index score C corresponding to the energy consumption index under the benchmark operating condition is obtained by using the method of S22. 0 , the benchmark operating condition index score corresponding to the battery index B 0 , the benchmark operating condition index score corresponding to the aging index is O 0 , the benchmark operating condition index score E corresponding to the environmental index r .

[0141] Likewise, the benchmark total score satisfies the following relationship:

[0142]

[0143] S24, calculating the difference between the target total score and the benchmark total score to obtain the total score difference.

[0144] In some embodiments, the total score difference satisfies the following relationship:

[0145]

[0146] in, is the target total score, is the benchmark total score, ΔE t The total score difference.

[0147] S25, calculating the difference between the target indicator score and the corresponding benchmark indicator score to obtain the indicator score difference.

[0148] In some embodiments, the difference in index scores corresponding to the energy consumption index may satisfy the following relationship:

[0149] ΔC=C r -C 0

[0150] In some embodiments, the difference in the indicator scores corresponding to the battery indicators may satisfy the following relationship:

[0151] ΔB=B r -B 0

[0152] In some embodiments, the difference in the index scores corresponding to the aging index may satisfy the following relationship:

[0153] ΔO=O r -O 0

[0154] In some embodiments, the difference in the indicator scores corresponding to the environmental indicators may satisfy the following relationship:

[0155] ΔN=N r -CN 0

[0156] S26, calculating the quotient of the index score difference and the total score difference to obtain the index proportion corresponding to the target working condition index.

[0157] In some embodiments, the indicator ratio corresponding to the energy consumption indicator satisfies the following relationship:

[0158]

[0159] Among them, N c It is the proportion of indicators corresponding to energy consumption indicators.

[0160] In some embodiments, the indicator ratio corresponding to the battery indicator satisfies the following relationship:

[0161]

[0162] Among them, N b The indicator ratio corresponding to the battery indicator.

[0163] In some embodiments, the indicator ratio corresponding to the aging indicator satisfies the following relationship:

[0164]

[0165] Among them, N o is the proportion of indicators corresponding to the aging indicators.

[0166] In some embodiments, the indicator ratio corresponding to the environmental indicator satisfies the following relationship:

[0167]

[0168] Among them, N e It is the proportion of indicators corresponding to environmental indicators.

[0169] S27, determining the main cause from multiple target operating condition indicators according to the indicator proportion.

[0170] In some possible embodiments, the power battery detection device selects the main cause from multiple target operating condition indicators according to the proportion of the indicators.

[0171] In some possible embodiments, the power battery detection device takes the target operating condition index with the largest index proportion as the main reason.

[0172] The above steps pre-train the first attribution model based on the first historical status information of several power batteries under different endurance conditions and the first label corresponding to the main reason for generating the historical status information. Secondly, by using the fluctuation contribution method and chain replacement of abnormal indicators, the core features that lead to low endurance can be directly located, the interpretability of the model can be improved, the decision-making process can be made more transparent, and the signals that truly affect the target variables can be distinguished from unnecessary noise, thereby optimizing the production of training labels and improving the accuracy and stability of the model. Thirdly, the label data in this embodiment is a layer-by-layer processing of the original data, and therefore has the characteristics of lightweight, fewer features, fast model training speed, and less resource usage, which is suitable for rapid iteration or even replacement of training models.

[0173] In some embodiments, S27 may specifically include the following sub-steps:

[0174] S31, in response to the fact that the indicator proportion corresponding to the first candidate operating condition indicator is greater than or equal to a preset threshold, and the first candidate operating condition indicator is an atomic indicator, the first candidate operating condition indicator is taken as the main reason.

[0175] In some embodiments, in response to the fact that the proportion of indicators corresponding to the first candidate operating condition indicator is greater than or equal to a preset threshold, and the first candidate operating condition indicator is an atomic indicator, the power battery detection device uses the first candidate operating condition indicator as the main cause.

[0176] Optionally, the preset threshold may be greater than or equal to 50%. Exemplarily, the preset threshold is 50%.

[0177] Among them, the atomic indicator refers to an indicator that cannot be further decomposed. In other words, in response to the indicator proportion corresponding to the first candidate operating condition indicator being greater than or equal to the preset threshold, and the first candidate operating condition indicator cannot be decomposed further, the power battery detection device uses the first candidate operating condition indicator as the main reason.

[0178] S32: In response to the fact that the proportion of the indicator corresponding to any target operating condition indicator is less than a preset threshold, and at least two second candidate operating condition indicators with the largest proportion of the indicators are atomic indicators, the second candidate operating condition indicator is taken as the main reason.

[0179] In some embodiments, in response to the indicator ratio corresponding to any target operating condition indicator being less than a preset threshold, and at least two of the second candidate operating condition indicators with the largest indicator ratios are both atomic indicators, the power battery detection device takes at least two of the second candidate operating condition indicators with the largest indicator ratios as the main cause.

[0180] Exemplarily, in response to the indicator ratio corresponding to any target operating condition indicator being less than a preset threshold, and the top three second candidate operating condition indicators with the largest indicator ratios are all atomic indicators, the power battery detection device takes these three second candidate operating condition indicators as the main reasons.

[0181] In some embodiments, after step S26, the following steps may also be included:

[0182] S41, in response to the fact that the indicator proportion corresponding to the third candidate operating condition indicator is greater than or equal to a preset threshold, and the third candidate operating condition indicator is not an atomic indicator, the third candidate operating condition indicator is disassembled to obtain a plurality of first sub-operating condition indicators.

[0183] In some embodiments, in response to the proportion of indicators corresponding to the third candidate operating condition indicator being greater than or equal to a preset threshold, and the third candidate operating condition indicator is not an atomic indicator, the third candidate operating condition indicator is disassembled according to the composition of the third candidate operating condition indicator to obtain multiple first sub-operating condition indicators.

[0184] Taking the third candidate operating condition indicator as the energy consumption indicator as an example for explanation, the energy consumption indicator is composed of comprehensive energy consumption, driving energy consumption, non-driving energy consumption, air-conditioning energy consumption, and other energy consumptions. The indicator proportion corresponding to the energy consumption indicator is greater than or equal to the preset threshold. The power battery detection device splits the energy consumption indicator into comprehensive energy consumption, driving energy consumption, non-driving energy consumption, air-conditioning energy consumption, and other energy consumptions. Among them, the comprehensive energy consumption, driving energy consumption, non-driving energy consumption, air-conditioning energy consumption, and other energy consumption correspond to the first sub-operating condition indicator mentioned above.

[0185] S42, determining a main cause from a plurality of first sub-operating condition indicators and the remaining target operating condition indicators according to the indicator ratio corresponding to the first sub-operating condition indicator.

[0186] In some embodiments, the power battery detection device calculates the index ratio corresponding to each first sub-operating condition index, and determines the main cause by using the size of the index ratio corresponding to each first sub-operating condition index and the remaining target operating condition indexes.

[0187] Taking the third candidate operating condition indicator as the energy consumption indicator as an example, the corresponding first sub-operating condition indicator includes comprehensive energy consumption, driving energy consumption, non-driving energy consumption, air conditioning energy consumption, and other energy consumption. The remaining target operating condition indicators include battery indicators, aging indicators, environmental indicators, and endurance indicators.

[0188] Furthermore, the power battery detection device obtains the proportion of indicators corresponding to the comprehensive energy consumption, driving energy consumption, non-driving energy consumption, air-conditioning energy consumption, and other energy consumption respectively.

[0189] For example, the proportion of indicators corresponding to air conditioning energy consumption satisfies the following relationship:

[0190]

[0191] Among them, c air r is the target index score corresponding to the air conditioning energy consumption under the first historical state information, cair 0 It is the benchmark index score corresponding to the air conditioning energy consumption under the benchmark status information.

[0192] Similarly, the proportions of indicators corresponding to comprehensive energy consumption, driving energy consumption, non-driving energy consumption, and other energy consumption can be determined through relationships similar to the above.

[0193] Furthermore, the power battery detection device determines the main cause according to the proportion of indicators corresponding to the comprehensive energy consumption, driving energy consumption, non-driving energy consumption, air-conditioning energy consumption, other energy consumption, battery indicators, aging indicators, environmental indicators, and endurance indicators, with reference to S31 or S32.

[0194] The above steps pre-train the first attribution model based on the first historical status information of several power batteries under different endurance conditions and the first label corresponding to the main reason for generating the historical status information. Secondly, by using the fluctuation contribution method and chain replacement of abnormal indicators, the core features that lead to low endurance can be directly located, the interpretability of the model can be improved, the decision-making process can be made more transparent, and the signals that truly affect the target variables can be distinguished from unnecessary noise, thereby optimizing the production of training labels and improving the accuracy and stability of the model. Thirdly, the label data in this embodiment is a layer-by-layer processing of the original data, and therefore has the characteristics of lightweight, fewer features, fast model training speed, and less resource usage, which is suitable for rapid iteration or even replacement of training models.

[0195] In some other embodiments, after step S26, the following steps may also be included:

[0196] S51, in response to the fact that the indicator ratio corresponding to any target operating condition indicator is less than a preset threshold, and at least two fourth candidate operating condition indicators with the largest indicator ratios are not atomic indicators, the fourth candidate operating condition indicator is disassembled to obtain multiple second sub-operating condition indicators.

[0197] In some embodiments, in response to the indicator ratio corresponding to any target operating condition indicator being less than a preset threshold, and at least two fourth candidate operating condition indicators with the largest indicator ratios are not atomic indicators, the power battery detection device respectively disassembles at least two fourth candidate operating condition indicators to obtain multiple second sub-operating condition indicators.

[0198] Taking the fourth candidate operating condition indicator as the energy consumption indicator and the cruising range indicator as an example, the energy consumption indicator is composed of comprehensive energy consumption, driving energy consumption, non-driving energy consumption, air-conditioning energy consumption, and other energy consumptions; the cruising range indicator is composed of the displayed cruising range achievement rate and the estimated cruising range; the energy consumption indicator and the cruising range indicator are the fourth candidate operating condition indicators with the top two indicator proportions, and the indicator proportions corresponding to all target operating condition indicators are less than the preset threshold value. The power battery detection device splits the energy consumption indicator into comprehensive energy consumption, driving energy consumption, non-driving energy consumption, air-conditioning energy consumption, and other energy consumptions, and splits the cruising range indicator into the displayed cruising range achievement rate and the estimated cruising range, among which the comprehensive energy consumption, driving energy consumption, non-driving energy consumption, air-conditioning energy consumption, other energy consumption, displayed cruising range achievement rate, and estimated cruising range correspond to the second sub-operating condition indicator mentioned above.

[0199] S53, determining a main cause from a plurality of second sub-operating condition indicators and the remaining target operating condition indicators according to the indicator proportion corresponding to the second sub-operating condition indicator.

[0200] In some embodiments, the power battery detection device calculates the index ratio corresponding to each second sub-operating condition index, and determines the main cause by using the size of the index ratio corresponding to each second sub-operating condition index and the remaining target operating condition indexes.

[0201] Taking the fourth candidate operating condition indicators as energy consumption indicators and endurance indicators as an example, the corresponding second sub-operating condition indicators include comprehensive energy consumption, driving energy consumption, non-driving energy consumption, air conditioning energy consumption, other energy consumption, displayed endurance achievement rate, and estimated endurance. The remaining target operating condition indicators include battery indicators, aging indicators, and environmental indicators.

[0202] Furthermore, the power battery detection device respectively obtains the comprehensive energy consumption, driving energy consumption, non-driving energy consumption, air-conditioning energy consumption, other energy consumption, displayed cruising range achievement rate, and the proportion of indicators corresponding to the estimated cruising range.

[0203] Furthermore, the power battery detection device determines the main cause according to the proportion of indicators corresponding to the comprehensive energy consumption, driving energy consumption, non-driving energy consumption, air-conditioning energy consumption, other energy consumption, displayed cruising range achievement rate, and estimated cruising range, with reference to S31 or S32.

[0204] In the above steps, the first attribution model is pre-trained based on the first historical status information of several power batteries under different endurance conditions and the first label corresponding to the main reason for generating the historical status information. Secondly, by using the fluctuation contribution method and chain replacement of abnormal indicators, the core features that cause low endurance can be directly located, the interpretability of the model can be improved, the decision-making process can be made more transparent, and the signals that truly affect the target variable can be distinguished from unnecessary noise, thereby optimizing the production of training labels and improving the accuracy and stability of the model.

[0205] Another embodiment of the power battery detection method provided in the present application may specifically include the following steps:

[0206] S61, obtaining status information corresponding to the power battery of the current vehicle under the condition that a preset condition is met.

[0207] S62, inputting the state information into the first attribution model to obtain the operating condition level attribution result.

[0208] The first attribution model is pre-trained based on first historical state information of a plurality of power batteries under different endurance conditions and first labels corresponding to the historical state information. The content of the first label is used to indicate the main reason for generating the historical state information.

[0209] S63, inputting the state information and the working condition level attribution result into the second attribution model to obtain the single vehicle level attribution result.

[0210] Among them, the second attribution model is pre-trained based on the second historical state information of the current vehicle under different endurance conditions and the second label corresponding to the second historical state information; the content of the second label is obtained based on the condition-level attribution result output by the first attribution model during the training process.

[0211] S64, determining a plurality of operating condition indicators from the state information, and generating a first structured data table according to the plurality of operating condition indicators.

[0212] In some embodiments, the operating condition index may include at least an energy consumption index, a battery index, an aging index, an environmental index, and a battery life index. Further, the power battery detection device generates a first structured data table including the energy consumption index, the battery index, the aging index, the environmental index, and the battery life index.

[0213] In some possible embodiments, the power battery detection device determines the indicator scores and corresponding indicator proportions corresponding to the energy consumption indicator, battery indicator, aging indicator, environmental indicator and endurance indicator from the status information, and generates a first structured data table containing the energy consumption indicator, battery indicator, aging indicator, environmental indicator and endurance indicator, indicator scores and indicator proportions.

[0214] S65, generating a second structured data table according to the working condition level attribution result and the state information, and generating a third structured data table according to the single vehicle level attribution result and the state information.

[0215] In some embodiments, the power battery detection device generates a second structured data table according to the operating condition level attribution result and the status information.

[0216] In some possible embodiments, the power battery detection device generates a second structured data table according to a number of status information belonging to the current vehicle and an operating condition level attribution result corresponding to each status information.

[0217] In some embodiments, the power battery detection device generates a third structured data table based on the vehicle-level attribution results and status information.

[0218] In some possible embodiments, the power battery detection device generates a third structured data table based on a number of status information belonging to the current vehicle and a single-vehicle-level attribution result corresponding to each status information.

[0219] S66, displaying the first structured data table, the second structured data table and the third structured data table.

[0220] Specifically, the power battery display device displays a first structured data table, a second structured data table, and a third structured data table.

[0221] In this embodiment, by simultaneously displaying three structured data tables, interactive automatic attribution analysis of low-range conditions can be provided to technicians, thereby achieving real-time monitoring of the vehicle and automatic attribution of low range. By comparing and verifying the low-range attribution results with other battery performance data in the cloud database, the accuracy of the attribution is further improved, helping technicians to quickly identify the root cause of the low range problem of the current vehicle.

[0222] The above scheme pre-trains the first attribution model based on the first historical status information of several power batteries under different endurance conditions and the first label corresponding to the main reason for generating the historical status information. Secondly, the second historical status data of the current vehicle under different endurance conditions is input into the trained first attribution model to obtain the second label corresponding to the second historical status data. The second attribution model is pre-trained using the second historical status data and the second label. By means of model cascading, the training speed of the second attribution model is improved, and the accuracy of attribution is improved. Thirdly, the status information corresponding to the power battery of the current vehicle is input into the first attribution model to obtain the operating condition-level attribution result, and the status information and the operating condition-level attribution result are input into the second attribution model to obtain the single-vehicle-level attribution result. By displaying the operating condition-level attribution result and the single-vehicle-level attribution result, the attribution result corresponding to the power battery can be determined faster and more accurately.

[0223] Another embodiment of the power battery detection method provided in the present application may specifically include the following steps:

[0224] S71, obtaining the current driving time and average driving speed of the vehicle, and the power change value of the power battery.

[0225] Specifically, the power battery detection device obtains the driving time, average driving speed, and power change value (SOC change value) of the current vehicle under the operating condition.

[0226] In some embodiments, the power battery detection device obtains the current driving distance of the vehicle under the operating condition, and determines the average driving speed using the driving distance and driving time.

[0227] S72, in response to the driving duration being greater than or equal to the preset duration, the average driving speed being greater than or equal to the preset speed, and the battery charge change value being greater than or equal to the preset change value, it is determined that the preset conditions are met.

[0228] In some embodiments, in response to a driving duration being greater than or equal to a preset duration, an average driving speed being greater than or equal to a preset speed, and a charge change value being greater than or equal to a preset change value, the power battery device determines that the operating condition meets preset conditions.

[0229] Exemplarily, the preset duration may be 3600 seconds, the preset speed may be 10 km / h, and the preset change value may be 30%.

[0230] S73, obtaining status information corresponding to the power battery of the current vehicle under the condition that a preset condition is met.

[0231] S74, inputting the state information into the first attribution model to obtain the operating condition level attribution result.

[0232] The first attribution model is pre-trained based on first historical state information of a plurality of power batteries under different endurance conditions and first labels corresponding to the historical state information. The content of the first label is used to indicate the main reason for generating the historical state information.

[0233] S75, inputting the state information and the working condition level attribution result into the second attribution model to obtain the single vehicle level attribution result.

[0234] Among them, the second attribution model is pre-trained based on the second historical state information of the current vehicle under different endurance conditions and the second label corresponding to the second historical state information; the content of the second label is obtained based on the condition-level attribution result output by the first attribution model during the training process.

[0235] S76, displays the working condition level attribution results and the single vehicle level attribution results.

[0236] The above scheme pre-trains the first attribution model based on the first historical status information of several power batteries under different endurance conditions and the first label corresponding to the main reason for generating the historical status information. Secondly, the second historical status data of the current vehicle under different endurance conditions is input into the trained first attribution model to obtain the second label corresponding to the second historical status data. The second attribution model is pre-trained using the second historical status data and the second label. By means of model cascading, the training speed of the second attribution model is improved, and the accuracy of attribution is improved. Thirdly, the status information corresponding to the power battery of the current vehicle is input into the first attribution model to obtain the operating condition-level attribution result, and the status information and the operating condition-level attribution result are input into the second attribution model to obtain the single-vehicle-level attribution result. By displaying the operating condition-level attribution result and the single-vehicle-level attribution result, the attribution result corresponding to the power battery can be determined faster and more accurately.

[0237] Furthermore, the power battery detection method provided in the present application reduces the amount of status data step by step by continuously constructing and fusing the original battery status data, which can greatly improve the training speed and iteration efficiency of the model, and improve the accuracy of low battery life attribution through model cascading.

[0238] See also Figure 2 , Figure 2 It is a flow chart of another embodiment of the power battery detection method provided in the present application.

[0239] like Figure 2 As shown, another embodiment of the power battery detection method provided by the present application may further include the following steps:

[0240] S101, low battery life portrait structure.

[0241] Specifically, the power battery detection device constructs the power battery operating condition index of the current vehicle during driving from five aspects: energy consumption index, battery index, aging index, environmental index and endurance index, and forms an operating condition data table, wherein the operating condition data table corresponds to the first structured data table mentioned above.

[0242] In some embodiments, the energy consumption index is composed of the comprehensive energy consumption C a Composition, further, comprehensive energy consumption The driving energy consumption C d and non-driving energy consumption C n Composition, not driving energy consumption C n Energy consumption of air conditioner C air 、Other energy consumption C o And so on.

[0243] Among them, the comprehensive energy consumption C a Specifically, the following relations are satisfied:

[0244] C a =E out -E in

[0245] Among them, E out is the output energy of the power battery under single working condition (kwh), E in It is the input energy of the power battery under single working condition (kwh).

[0246] In some embodiments, the battery indicator is mainly composed of the cell voltage difference V d , Voltage consistency V uinf , SOC jump maximum value SOC jump times SOE jump maximum value SOE transition times The estimated error of the single SOC M e Etc. sub-indicators composition.

[0247] Among them, the voltage consistency adopts the discrete ratio V uinf , specifically satisfying the following relationship:

[0248]

[0249] in, is the average voltage of all cells in the battery pack (i.e., power battery), V min is the minimum voltage among all cells in the battery pack, V max It is the maximum voltage of all cells in the battery pack.

[0250] Among them, the estimated error of single SOC M e , specifically satisfying the following relationship:

[0251]

[0252] in, The maximum value of the SOC of the battery cell; It is the minimum value of the SOC of the battery cell.

[0253] In some embodiments, the aging indicator is mainly composed of the current SOH value SOH of the vehicle now , calendar time from offline to detection T c , total mileage M a Etc. sub-indicators composition.

[0254] In some embodiments, the environmental indicator is mainly composed of the ambient temperature T e Etc. sub-indicators composition.

[0255] In some embodiments, the battery life indicator is mainly represented by the battery life achievement rate Et 、Estimated range E e Etc. sub-indicators composition.

[0256] Among them, the displayed endurance rate is E t The following relations are satisfied:

[0257]

[0258] Among them, M r is the actual mileage change of a single working condition, M t Displays the mileage change for a single operating condition.

[0259] Furthermore, the estimated range E e The following relations are satisfied:

[0260]

[0261] Among them, M r is the actual mileage change in a single operating condition, and ΔSOC is the change in SOC in a single operating condition.

[0262] S102, extracting effective working conditions.

[0263] In order to ensure the effectiveness of the low-range test conditions of the power battery, the power battery test device needs to extract the effective test condition data from the cloud database. First, filter out the abnormal data in the conditions, and then select the discharge conditions of all pure electric models (excluding hybrid models). In addition, it is necessary to ensure that the vehicle's driving time is not less than 1 hour (3600 seconds) (t≥3600) and the average speed is not less than 10kwh / h. The change of SOC is not less than 30% (ΔSOC ≥ 30), and the mileage of a single operation is not less than 2km (M r ≥2).

[0264] S103, obtaining a low battery life problem label of a single working condition.

[0265] In some embodiments, S103 includes the following sub-steps:

[0266] S131, use DuPont analysis method to disassemble abnormal indicators and make a root cause tree of low battery life problem.

[0267] S132, volatility contribution method, chain substitution of abnormal indicators to directly locate core indicators.

[0268] It should be noted that each low endurance condition is not caused by just one factor. There is often a main cause and multiple auxiliary factors that jointly lead to the occurrence of low endurance. The above indicators are used for quantitative description, and specifically meet the following relationship:

[0269] The energy consumption index satisfies the following relationship:

[0270] C=m1C d +m2C air +m3C o

[0271] The battery indicators satisfy the following relationship:

[0272]

[0273] The aging index satisfies the following relationship:

[0274] O=p1SOH now +p2T c +p3M a

[0275] Environmental indicators satisfy the following relationship:

[0276] E=t1T e

[0277] Furthermore, the range score corresponding to a single low range condition satisfies the following relationship:

[0278] E t =mC+nB+pO+tE

[0279] Furthermore, the power battery detection device uses the endurance score of the current vehicle under normal endurance conditions as the reference period. And obtain the endurance score of the current vehicle under a certain low endurance condition In this way, the endurance score difference between the two working conditions is obtained, and the endurance score difference satisfies the following relationship:

[0280]

[0281] The power battery testing device calculates the impact ratio of each major category of indicators. Taking the energy consumption indicator as an example, the energy consumption indicator satisfies the following relationship:

[0282]

[0283] Next, the power battery detection device calculates the battery index N b , aging index N o , Environmental index N e .

[0284] In some embodiments, if the indicator ratio of a certain indicator exceeds 50%, and the indicator can be broken down, the power battery detection device breaks down the indicator and determines the influence ratio of the sub-indicators again.

[0285] In some embodiments, if the index proportion of a certain index exceeds 50% and the index cannot be broken down, the power battery detection device takes the influencing factors of the index as the core factors (corresponding to the main reasons above).

[0286] In some embodiments, if the proportion of any indicator does not exceed 50%, and the top three indicators cannot be disassembled, the power battery detection device determines that the low endurance condition is caused by the top three factors.

[0287] In some embodiments, if the indicator proportion of any indicator does not exceed 50%, and the top three factors can be further broken down, the power battery detection device continues to break down and determines the indicator proportion of the top three factors again.

[0288] In some embodiments, if this indicator is not the root cause of low battery life, the power battery detection device continues to calculate the impact ratio of each sub-category indicator, taking the air conditioning energy consumption in the energy consumption indicator as an example: Then drive the energy consumption indicator Other energy consumption indicators Other indicators will not be elaborated.

[0289] Furthermore, the power battery detection device repeats the above steps until the root cause of the low endurance in a single operating condition is found.

[0290] It should be noted that S103 is used to label the battery data in order to produce training data for the automatic attribution model of low battery life at the working condition level described below.

[0291] S104, low endurance attribution at working condition level.

[0292] In some embodiments, the power battery detection device inputs the relevant battery data under the low-range working condition and the working condition-level low-range attribution label obtained in S103 into the CatBoost model for training to obtain a working condition-level battery low-range automatic attribution model. The input of the working condition-level battery low-range automatic attribution model is the battery data, and the output is the working condition-level low-range attribution result.

[0293] In some embodiments, the power battery detection device takes battery data under different low-range operating conditions as input, obtains several operating condition-level low-range attribution results, and further forms an operating condition-level low-range attribution result table.

[0294] S105, summarizing the working condition data.

[0295] In some embodiments, the power battery detection device summarizes the operating condition level low endurance attribution result table output in S104 according to the vehicle unique identification code, aggregates all relevant indicators, and obtains a single vehicle low endurance attribution label.

[0296] S106, low battery life attribution at the vehicle level.

[0297] In some embodiments, the power battery detection device inputs the single-vehicle low-range attribution label obtained in S105 into the LightGBM model again for secondary modeling (training) to obtain a single-vehicle-level battery low-range automatic attribution model. The input of the single-vehicle-level battery low-range automatic attribution model is battery data, and the output is the single-vehicle-level low-range attribution result.

[0298] In some embodiments, the power battery detection device takes battery data of the same vehicle under different low-range conditions as input, obtains a number of single-vehicle-level low-range attribution results, and further forms a single-vehicle-level low-range attribution result table.

[0299] S107, display the attribution result.

[0300] In some embodiments, the power battery testing device links the operating condition level low endurance attribution result table of S104 and the single vehicle level low endurance attribution result table of S105, and displays them on the front-end big data analysis platform for use by battery testers.

[0301] The above scheme pre-trains the first attribution model based on the first historical status information of several power batteries under different endurance conditions and the first label corresponding to the main reason for generating the historical status information. Secondly, the second historical status data of the current vehicle under different endurance conditions is input into the trained first attribution model to obtain the second label corresponding to the second historical status data. The second attribution model is pre-trained using the second historical status data and the second label. By means of model cascading, the training speed of the second attribution model is improved, and the accuracy of attribution is improved. Thirdly, the status information corresponding to the power battery of the current vehicle is input into the first attribution model to obtain the operating condition-level attribution result, and the status information and the operating condition-level attribution result are input into the second attribution model to obtain the single-vehicle-level attribution result. By displaying the operating condition-level attribution result and the single-vehicle-level attribution result, the attribution result corresponding to the power battery can be determined faster and more accurately.

[0302] Please continue to see Figure 3 , Figure 3 The terminal device 500 of the embodiment of the present application includes a processor 51 and a memory 52 .

[0303] The processor 51 and the memory 52 are connected to the bus. The memory 52 stores program data. The processor 51 is used to execute the program data to implement the power battery detection method described in the above embodiment.

[0304] In the embodiment of the present application, the processor 51 may also be referred to as a CPU (Central Processing Unit). The processor 51 may be an integrated circuit chip having the ability to process signals. The processor 51 may also be a general-purpose processor, a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field programmable gate array (FPGA), or other programmable logic devices, discrete gates or transistor logic devices, or discrete hardware components. A general-purpose processor may be a microprocessor or the processor 51 may also be any conventional processor, etc.

[0305] This application also provides a computer storage medium, please continue to refer to Figure 4 , Figure 4 It is a schematic diagram of the structure of an embodiment of a computer storage medium provided in the present application. The computer storage medium 600 stores program data 61. When the program data 61 is executed by a processor, it is used to implement the power battery detection method of the above embodiment.

[0306] When the embodiments of the present application are implemented in the form of software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application is essentially or the part that contributes to the prior art or all or part of the technical solution can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including several instructions to enable a computer device (which can be a personal computer, a server, or a network device, etc.) or a processor to perform all or part of the steps of the method described in each embodiment of the present application. The aforementioned storage medium includes: various media that can store program codes, such as a USB flash drive, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a disk or an optical disk.

[0307] The above description is only an implementation method of the present application, and does not limit the patent scope of the present application. Equivalent structures or equivalent process changes made by utilizing the contents of the present application specification and drawings, or directly or indirectly applied in other related technical fields, are also included in the patent protection scope of the present application.

Claims

1. A method for detecting a power battery, characterized in that: The method comprises: Under the preset conditions, obtain the status information corresponding to the power battery of the current vehicle; The state information is input into a first attribution model to obtain a working condition-level attribution result; the first attribution model is pre-trained based on first historical state information of a plurality of power batteries under different endurance working conditions and a first label corresponding to the first historical state information, wherein the content of the first label is used to indicate the main reason for generating the historical state information; The state information and the operating condition-level attribution result are input into a second attribution model to obtain a single-vehicle-level attribution result; the second attribution model is pre-trained based on the second historical state information of the current vehicle under different cruising conditions and a second label corresponding to the second historical state information; the content of the second label is obtained based on the operating condition-level attribution result output by the first attribution model during the training process; The working condition level attribution results and the single vehicle level attribution results are displayed.

2. The method according to claim 1, characterized in that Before the step of training the first attribution model, the method includes: Acquire a plurality of target operating condition indicators corresponding to the first historical state information; the target operating condition indicators at least include an energy consumption indicator, a battery indicator, an aging indicator, an environmental indicator, and a battery life indicator; Determining the main cause from the plurality of target operating condition indicators using a fluctuation contribution method; The first label corresponding to the first historical status information is constructed using the main reason.

3. The method according to claim 2, characterized in that The method of using the fluctuation contribution method to determine the main cause from the plurality of target operating condition indicators includes: Acquire multiple benchmark operating condition indicators corresponding to the benchmark state information; Calculating the target indicator score corresponding to each of the target operating condition indicators, and determining the target total score corresponding to the first historical state information; and calculating the reference indicator score corresponding to each of the reference operating condition indicators, and determining the reference total score corresponding to the reference state information; Calculating the difference between the target total score and the benchmark total score to obtain a total score difference; Calculate the difference between the target indicator score and the corresponding benchmark indicator score to obtain the indicator score difference; The quotient of the indicator score difference and the total score difference is calculated to obtain the indicator proportion corresponding to the target operating condition indicator, and the main cause is determined from the multiple target operating condition indicators based on the indicator proportion.

4. The method according to claim 3, characterized in that Determining the main cause from the plurality of target operating condition indicators according to the indicator proportions includes: In response to the fact that the proportion of the indicator corresponding to the first candidate operating condition indicator is greater than or equal to the preset threshold, and the first candidate operating condition indicator is an atomic indicator, taking the first candidate operating condition indicator as the main reason; or, In response to the fact that the indicator ratio corresponding to any target operating condition indicator is less than the preset threshold, and at least two second candidate operating condition indicators with the largest indicator ratios are atomic indicators, the second candidate operating condition indicator is used as the main reason.

5. The method according to claim 4, characterized in that After the step of calculating the quotient of the index score difference and the total score difference to obtain the index proportion corresponding to the target operating condition index, the method further comprises: In response to the fact that the proportion of indicators corresponding to the third candidate operating condition indicator is greater than or equal to the preset threshold, and the third candidate operating condition indicator is not an atomic indicator, the third candidate operating condition indicator is disassembled to obtain a plurality of first sub-operating condition indicators; According to the proportion of indicators corresponding to the first sub-operating condition indicators, the main cause is determined from multiple first sub-operating condition indicators and other target operating condition indicators.

6. The method according to claim 4, characterized in that After the step of calculating the quotient of the index score difference and the total score difference to obtain the index proportion corresponding to the target operating condition index, the method further comprises: In response to the fact that the indicator ratio corresponding to any target operating condition indicator is less than the preset threshold, and at least two fourth candidate operating condition indicators with the largest indicator ratios are not atomic indicators, the fourth candidate operating condition indicator is disassembled to obtain a plurality of second sub-operating condition indicators; According to the proportion of indicators corresponding to the second sub-operating condition indicators, the main cause is determined from multiple second sub-operating condition indicators and other target operating condition indicators.

7. The method according to claim 1, characterized in that The display of the working condition level attribution result and the single vehicle level attribution result includes: Determine a plurality of operating condition indicators from the state information, and generate a first structured data table according to the plurality of operating condition indicators; Generate a second structured data table according to the state information and the working condition level attribution result, and generate a third structured data table according to the state information and the single vehicle level attribution result; The first structured data table, the second structured data table, and the third structured data table are displayed.

8. The method according to claim 1, characterized in that Before the step of obtaining the status information corresponding to the power battery of the current vehicle under the preset conditions, the method includes: Obtaining the driving time and average driving speed of the current vehicle, and the power change value of the power battery; In response to the driving duration being greater than or equal to a preset duration, the average driving speed being greater than or equal to a preset speed, and the battery charge change value being greater than or equal to a preset change value, it is determined that the preset condition is met.

9. A terminal device, characterized in that: The terminal device includes a processor and a memory connected to the processor, wherein: The memory stores program instructions; The processor is configured to execute program instructions stored in the memory to implement the method according to any one of claims 1 to 8.

10. A computer-readable storage medium, characterized in that: The storage medium stores program instructions, and when the program instructions are executed, the method according to any one of claims 1 to 8 is implemented.

Images