Method and device for identifying cause of vehicle insulation failure

Abstract

This invention belongs to the field of insulation fault analysis, specifically relating to a method and apparatus for identifying the causes of vehicle insulation faults. This method and apparatus, tailored to different insulation fault situations and their characteristics, uses manually set judgment methods to assist machine learning, thereby establishing a combined algorithm. This algorithm includes both expert-based identification methods and data-driven identification methods. In other words, insulation problems are treated as multi-dimensional issues, and multiple algorithms are designed to identify the causes of insulation faults from various angles, performing multi-dimensional in-depth analysis. This compensates for the limited number of fault samples. Furthermore, it identifies water-related insulation problems and insulation problems caused by excessive capacitance in the insulation detection module through data feature identification, considering insulation faults caused by water or excessive capacitance in the insulation detection module, making fault cause identification more comprehensive. The continuity algorithm also uses a comprehensive judgment of instantaneous and statistical aspects, improving both the sensitivity and reliability of the judgment.

Description

Technical Field

[0001] This invention belongs to the field of insulation fault analysis, and specifically relates to a method and device for identifying the causes of vehicle insulation faults. Background Technology

[0002] Among the various faults in new energy vehicles, insulation problems are not only a type of safety-related issue but also the most frequently reported. However, during on-site troubleshooting, it is often difficult to reproduce the fault, pinpoint the problem, and summarize the patterns, a problem that has plagued the automotive industry for many years.

[0003] Existing technologies often employ a self-learning network. This network is trained using insulation fault data from vehicle feedback and analyzed fault causes to generate a fault analysis model. Subsequent input of the fault information into the model yields the analyzed fault cause. However, this approach has several drawbacks: First, fault cause analysis relies on the limited experience of problem handlers, as there are too few cases where the specific cause can be pinpointed, leading to insufficient depth of understanding. Second, training data with misjudgments can cause model inaccuracies, resulting in incorrect fault cause feedback. Therefore, using machine learning methods cannot lead to practically effective products. Furthermore, from a machine learning perspective, insulation fault cause identification is a weak form of supervised learning, relying on only a small portion of effective labels, and the accuracy of machine-judged labels cannot be guaranteed. Therefore, the reliability of machine-based self-learning networks for identification is relatively low. Summary of the Invention

[0004] The purpose of this invention is to provide a method and apparatus for identifying the causes of vehicle insulation faults, in order to solve the problem of low reliability of existing insulation fault identification methods that use machine self-learning.

[0005] To achieve the above objectives, the present invention provides a method for identifying the causes of vehicle insulation faults. The specific method for identifying the causes of vehicle insulation faults is as follows:

[0006] 1) First, use statistical methods to determine whether the vehicle insulation fault is a vehicle-side insulation fault or a charger-side insulation fault; if it is determined to be a vehicle-side insulation fault, then proceed to step 2).

[0007] The statistical method determines whether a vehicle insulation fault is a vehicle-end insulation fault or a charger-end insulation fault by setting up a fault table, and directly finds and determines the corresponding vehicle-end insulation fault cause or charger-end insulation fault cause.

[0008] 2) Obtain the battery management system status of the vehicle with the insulation alarm on the day of the alarm, and determine whether the vehicle's insulation fault is an insulation fault in the charging state or an insulation fault in the non-charging state; if the insulation alarm occurs when the battery management system status is in the charging state, it is determined that the vehicle has an insulation fault in the charging state; otherwise, it is determined that it is an insulation fault in the non-charging state, and continue to the judgment in step 3).

[0009] 3) Use the continuity algorithm to determine which electrical accessories are related to the vehicle insulation fault, thereby obtaining the cause of the vehicle insulation fault; if the cause of the vehicle insulation fault cannot be obtained through the continuity algorithm, then continue to the judgment in step 4).

[0010] 4) Identify the data pattern of vehicle insulation faults using data pattern algorithms to determine the corresponding cause of the insulation fault; if the cause of the vehicle insulation fault cannot be obtained through data pattern algorithms, continue to step 5).

[0011] The data pattern algorithm identifies the cause of vehicle insulation failure by recognizing the data patterns of different vehicle insulation failures corresponding to different insulation failure causes.

[0012] 5) Determine the cause of the vehicle insulation failure based on the vehicle parameter variables using a correlation algorithm; if the cause of the vehicle insulation failure cannot be determined by the correlation algorithm, proceed to step 6).

[0013] The correlation algorithm determines which vehicle parameter variables are related to the vehicle insulation fault by judging whether different vehicle parameter variables and vehicle insulation faults are correlated.

[0014] 6) Using the fault concurrency algorithm, determine the cause of the vehicle insulation fault based on the occurrence of other types of faults in the vehicle at the time the vehicle insulation fault occurs; if the cause of the vehicle insulation fault cannot be determined by the fault concurrency algorithm, then continue to the judgment in step 7).

[0015] The fault concurrency algorithm determines the cause of the vehicle insulation fault by correlating the other types of vehicle faults with the vehicle insulation fault, and by considering the occurrence of other types of vehicle faults at the time the vehicle insulation fault occurs.

[0016] 7) Using statistical methods, search the set fault table. If the search result is not empty, directly determine the corresponding vehicle-end insulation fault cause or charger-end insulation fault cause based on the set fault table. If the search result is empty, output the judgment result in step 2) that the vehicle insulation fault belongs to the charging state insulation fault or the non-charging state insulation fault.

[0017] The beneficial effects of the above technical solution are as follows: For different insulation fault situations and their characteristics, a manually set judgment method assists the machine in learning, thereby establishing a combined algorithm. This algorithm includes both identification methods based on expert experience and those established from a data perspective. The insulation fault cause identification method of this invention treats insulation problems as multi-dimensional problems, designs multiple sub-algorithms, and identifies the causes of insulation faults from various angles. Through multi-dimensional in-depth analysis of each algorithm, it compensates for the problem of insufficient fault samples. Furthermore, by considering the data characteristics of water-related insulation problems and insulation problems caused by excessive capacitance in the insulation detection module, it makes the fault cause identification more comprehensive.

[0018] Furthermore, if step 1) determines that the vehicle insulation fault belongs to the charger end insulation fault, or step 2) determines that the vehicle insulation fault belongs to the charging state insulation fault, then the data pattern algorithm is used to determine whether the data pattern of the vehicle insulation fault corresponds to the fault cause of excessive insulation detection module capacitance. If it corresponds to the fault cause of excessive insulation detection module capacitance, then the insulation fault cause is determined to be excessive insulation detection module capacitance. If it does not correspond, then the corresponding vehicle end insulation fault cause or charger end insulation fault cause is directly determined by statistical method to look up the set fault table.

[0019] Furthermore, if the cause of the vehicle insulation fault can be obtained through the on / off algorithm in step 3), then the water-related algorithm is used to determine whether the vehicle insulation fault is related to water. If it is related to water, then the cause of the vehicle insulation fault is determined to include not only the cause of the vehicle insulation fault obtained by the on / off algorithm, but also the vehicle insulation fault is related to water. If it is not related to water, then the cause of the vehicle insulation fault is determined to include only the cause of the vehicle insulation fault obtained by the on / off algorithm.

[0020] If the cause of the vehicle insulation failure can be obtained through the correlation algorithm in step 5), then the water correlation algorithm is used to determine whether the vehicle insulation failure is related to water. If it is related to water, then the cause of the vehicle insulation failure is determined to include not only the cause of the vehicle insulation failure obtained by the correlation algorithm, but also the vehicle insulation failure is related to water. If it is not related to water, then the cause of the vehicle insulation failure is determined to include only the cause of the vehicle insulation failure obtained by the correlation algorithm.

[0021] If the cause of the vehicle insulation fault can be obtained through the fault concurrency algorithm in step 6), then the water-related algorithm is used to determine whether the vehicle insulation fault is related to water. If it is related to water, then the cause of the vehicle insulation fault is determined to include not only the cause of the vehicle insulation fault obtained by the fault concurrency algorithm, but also the cause of the vehicle insulation fault being related to water. If it is not related to water, then the cause of the vehicle insulation fault is determined to include only the cause of the vehicle insulation fault obtained by the fault concurrency algorithm.

[0022] The beneficial effects of the above technical solution are: in addition to determining the causes of failures in the charger and the vehicle itself, it also considers water-related failures, making the identified causes of failures more comprehensive and accurate.

[0023] Furthermore, if in step 2) it is determined that the vehicle insulation fault is an insulation fault in a non-charging state, then the data missing algorithm is first used to identify whether the vehicle with the insulation fault has missing vehicle insulation resistance data.

[0024] If data is missing, the cause of the vehicle insulation fault is determined by the fault concurrency algorithm; if the cause of the vehicle insulation fault cannot be determined by the fault concurrency algorithm, it is determined that the vehicle insulation resistance data is missing and the cause of the vehicle insulation fault cannot be determined.

[0025] If there is no missing data, proceed to step 3) for judgment.

[0026] Furthermore, the on / off algorithm is as follows:

[0027] The system acquires the following information for vehicles with insulation alarms on the day of the alarm (when not charging / not plugged in): the times when the insulation fault occurs / disappears, when electrical accessories are turned on, when electrical accessories are turned off, and when the insulation resistance increases significantly, increases slightly, decreases significantly, and decreases slightly. Combining these information with the times when the insulation fault occurs / disappears and when the insulation resistance increases significantly, increases slightly, decreases significantly, and decreases slightly, the system obtains the time period during which the vehicle's insulation fault exists. The system defines the time period from the first activation to the next activation of an electrical accessory as a group for that accessory. By analyzing the activation / deactivation times of each accessory group and the duration of the vehicle's insulation fault, the system determines the fault ratio of the electrical accessory under operating conditions. If the fault ratio of an electrical accessory under operating conditions exceeds a set threshold, the system determines that the vehicle's insulation fault is related to that electrical accessory.

[0028] The beneficial effects of the above technical solution are as follows: through a comprehensive judgment of instantaneous and statistical aspects, it is required that the occurrence and disappearance of insulation problems not only have a sensitive response to the on / off state of each accessory (it is required that the opening and closing must coexist simultaneously, and the time interval between opening and the occurrence of insulation problems must be extremely short), but also that the insulation problem persists for a large proportion of the time during the time the accessory is open, and persists for a very small proportion of the time during the time the accessory is closed; thus, the reliability is improved while the judgment sensitivity is enhanced.

[0029] Furthermore, the fault concurrency algorithm is as follows:

[0030] Acquire and statistically analyze which other faults occurred simultaneously with the insulation fault and the frequency of each fault; determine the cause of the vehicle insulation fault by analyzing the occurrence of other types of faults in the vehicle at the time the insulation fault occurred.

[0031] Furthermore, the water-related algorithm is as follows:

[0032] Extract the insulation resistance values ​​of the positive and negative terminals of the battery on the day the insulation alarm is triggered, when the battery is not charging / not plugged in. Determine whether the vehicle's insulation fault is related to water by analyzing the fault recovery time of the positive and negative terminal insulation resistance values ​​and the proportion of rising groups to the total number of groups. The fault recovery time refers to the time interval between the moment before the insulation resistance value of a sampling point that was less than the insulation abnormality threshold at one moment and greater than the insulation abnormality threshold at another moment, and the moment after that moment when the insulation resistance value first exceeds the insulation normal threshold. The rising group refers to setting the insulation resistance values ​​of every two adjacent sampling points as a group, comparing the insulation resistance values ​​within all groups, and identifying the group where the insulation resistance value of the later sampling point is greater than the insulation resistance value of the previous sampling point.

[0033] Furthermore, the correlation algorithm is as follows:

[0034] Calculate and statistically analyze the correlation coefficients between various vehicle parameters and insulation resistance values. If a parameter variable has a correlation coefficient greater than a set correlation coefficient threshold, it is determined that there is a correlation between the parameter variable and the insulation resistance value. Based on this, it is determined whether different parameter variables are correlated with vehicle insulation faults, and which parameter variables are related to vehicle insulation faults.

[0035] Furthermore, the data format algorithm is as follows:

[0036] Obtain the insulation resistance value of the vehicle on the day of the insulation alarm and the number of days prior to the alarm. Identify the cause of the vehicle insulation fault by analyzing the data format of the insulation resistance value.

[0037] The present invention also provides a vehicle insulation fault cause identification device, which is used to implement the above-described vehicle insulation fault cause identification method.

[0038] The vehicle insulation fault cause identification device can achieve the same beneficial effects as the vehicle insulation fault cause identification method described above. Attached Figure Description

[0039] Figure 1 This is a block diagram of the first half of the logic of the vehicle insulation fault cause identification method in an embodiment of the present invention;

[0040] Figure 2 This is a block diagram of the latter half of the logic of the vehicle insulation fault cause identification method in an embodiment of the present invention.

[0041] Figure 3 This is an example diagram of setting up a fault table in an embodiment of the vehicle insulation fault cause identification method of the present invention;

[0042] Figure 4 This is a schematic diagram of the data format of the negative electrode insulation resistance value of the system when the insulation detection module capacitor is too large, causing an insulation fault, in an embodiment of the vehicle insulation fault cause identification method of the present invention. Detailed Implementation

[0043] To make the objectives, technical solutions, and advantages of the present invention clearer, the present invention will be further described in detail below with reference to the accompanying drawings and embodiments.

[0044] Example of a method for identifying the causes of vehicle insulation faults:

[0045] This embodiment provides a technical solution for identifying the causes of vehicle insulation faults, referring to... Figure 1 and Figure 2 The specific method for identifying the cause of vehicle insulation faults using this identification method is as follows:

[0046] 1) First, use statistical methods to determine whether the vehicle insulation fault is a vehicle-side insulation fault or a charger-side insulation fault; if it is determined to be a vehicle-side insulation fault, continue to step 2); if it is determined to be a charger-side insulation fault, continue to step 3).

[0047] 2) Determine whether the vehicle insulation fault belongs to the charging state or the non-charging state through the overall algorithm. If it belongs to the charging state, continue to step 3); if it belongs to the non-charging state, continue to step 4).

[0048] 3) Determine whether the data pattern of the vehicle insulation fault corresponds to the fault cause of excessive insulation detection module capacitance through the data pattern algorithm; if it corresponds to the fault cause of excessive insulation detection module capacitance, then determine that the insulation fault is caused by excessive insulation detection module capacitance; if it does not correspond, then use statistical methods to look up the set fault table and directly determine the corresponding vehicle-end insulation fault cause or charger-end insulation fault cause.

[0049] 4) Use the data missing data algorithm to identify whether there is missing vehicle insulation resistance data in vehicles with insulation faults; if so, continue to step 5); if not, continue to step 6).

[0050] 5) By using the fault concurrency algorithm, the cause of the vehicle insulation fault is determined based on the occurrence of other types of faults in the vehicle at the time the vehicle insulation fault occurs; if the cause of the vehicle insulation fault cannot be obtained by the fault concurrency algorithm, it is determined that the vehicle insulation resistance data is missing and the cause of the vehicle insulation fault cannot be determined.

[0051] 6) Using a continuity algorithm, determine which electrical accessories are related to the vehicle insulation fault; if the cause of the vehicle insulation fault can be found through the continuity algorithm, proceed to step 7); if the cause of the vehicle insulation fault cannot be found through the continuity algorithm, proceed to step 8), that is... Figure 2 The subsequent decision logic for the on / off method in the middle;

[0052] 7) Use the water-related algorithm to determine whether the vehicle insulation fault is related to water; if it is related to water, then determine that the cause of the vehicle insulation fault includes not only the cause of the vehicle insulation fault obtained by the on / off algorithm, but also that the vehicle insulation fault is related to water; if it is not related to water, then determine that the cause of the vehicle insulation fault only includes the cause of the vehicle insulation fault obtained by the on / off algorithm.

[0053] 8) Identify the data pattern of vehicle insulation faults using data pattern algorithms to determine the corresponding cause of the insulation fault; if the cause of the vehicle insulation fault cannot be obtained through data pattern algorithms, continue to step 9).

[0054] 9) Determine the cause of the vehicle insulation failure based on the vehicle parameter variables using the correlation algorithm; if the cause of the vehicle insulation failure can be obtained through the correlation algorithm, continue to step 10); if the cause of the vehicle insulation failure cannot be obtained through the correlation algorithm, continue to step 11).

[0055] 10) Use the water correlation algorithm to determine whether the vehicle insulation fault is related to water; if it is related to water, then determine that the cause of the vehicle insulation fault includes not only the cause of the vehicle insulation fault obtained by the correlation algorithm, but also that the vehicle insulation fault is related to water; if it is not related to water, then determine that the cause of the vehicle insulation fault only includes the cause of the vehicle insulation fault obtained by the correlation algorithm.

[0056] 11) Using the fault concurrency algorithm, determine the cause of the vehicle insulation fault based on the occurrence of other types of vehicle faults at the time the vehicle insulation fault occurs; if the cause of the vehicle insulation fault can be obtained through the fault concurrency algorithm, then continue to the judgment in step 12); if the cause of the vehicle insulation fault cannot be obtained through the fault concurrency algorithm, then continue to the judgment in step 13).

[0057] 12) Use the water-related algorithm to determine whether the vehicle insulation fault is related to water; if it is related to water, then determine that the cause of the vehicle insulation fault includes not only the cause of the vehicle insulation fault obtained by the fault concurrency algorithm, but also that the vehicle insulation fault is related to water; if it is not related to water, then determine that the cause of the vehicle insulation fault only includes the cause of the vehicle insulation fault obtained by the fault concurrency algorithm.

[0058] 13) By using statistical methods, search the set fault table. If the search result is not empty, directly determine the cause of the corresponding vehicle-end insulation fault or charger-end insulation fault. If the search result is empty, determine the cause of the vehicle insulation fault through the overall algorithm.

[0059] The following section provides a detailed analysis of the specific implementation process of each algorithm involved in the above identification method;

[0060] The statistical method primarily involves setting up a fault table to obtain the battery management system (BMS) status of vehicles experiencing insulation alarms on the day of the alarm. This determines whether the insulation fault is at the vehicle end or the charger end (if the insulation alarm occurs while the vehicle is in BMS status, it's considered a charging insulation fault; otherwise, it's considered a non-charging insulation fault). The corresponding vehicle-end or charger-end insulation fault cause is then directly identified. The BMS status is mainly categorized as driving, charging, and self-wake-up. If the BMS detects a charging gun signal, it's considered charging; if it detects the vehicle's ignition is off and the BMS is on, it's in self-wake-up mode; otherwise, it's driving mode. The BMS status is reported to the vehicle via codes, such as 4 for driving and 6 for charging.

[0061] In this embodiment, the specific steps of the statistical method are as follows:

[0062] ① Determine the power system status of the insulation alarm vehicle based on the protocol version of the vehicle from the previous day.

[0063] Typically, the VIN number of insulation alarm vehicles with alarm codes of 244, *, 8, 244, *, 28, or 244, *, 48 from the previous day is extracted from the faulty vehicle table to determine the specific insulation alarm vehicle to be analyzed. The faulty vehicle table name is: gdm_bsp_vehicle_terminal_fault; the extracted field is: device_vin vehicle VIN number.

[0064] Obtain the minimum and average values ​​of the old version of the insulation alarm vehicle under the power system states of 3 (self-wake-up), 4 (driving), and 6 (charging) on ​​the day before the alarm, and obtain the minimum and average values ​​of the new version of the insulation alarm vehicle under the power system states of 6 (self-wake-up), 8 (driving), and 12 (charging) on ​​the day before the alarm.

[0065] Because the new protocol's power system status is 6 (self-wake-up), 8 (driving), and 12 (charging), while the old protocol's is 3 (self-wake-up), 4 (driving), and 6 (charging), it's necessary to first determine which power supply is being used. Since the meaning of status 6 is different, it's crucial to distinguish between new and old protocol vehicles before obtaining the corresponding power system status for vehicles with insulation alarms. Distinguishing between new and old protocol vehicles is primarily achieved through the field "newold," where 1 represents the new version and 0 represents the old version; alternatively, if the vehicle's power system status shows 8 or 12, it's determined to be a new protocol vehicle.

[0066] ② Obtain the maximum, minimum, and average values ​​of the ratios of the positive, negative, and system insulation resistance values ​​to the total voltage for vehicles with insulation alarms under a certain power system condition on the day of the alarm.

[0067] ③ Based on the above maximum, minimum and average values, determine whether the insulation alarm vehicle has insulation problems under different power system states on a single day.

[0068] First, by using the maximum, minimum, and average values ​​mentioned above, we can determine whether there are insulation problems with the positive and negative terminals and the system under different power system states of the vehicle during a single day. Based on different conditions, the judgment results should be converted into four cases: 1 (insulation problem), 0 (insulation problem), -1 (no value), and -100 (error), which will facilitate subsequent lookup in the fault table.

[0069] Taking vehicles using the old protocol as an example, the specific conditions for judging based on the maximum, minimum, and average values ​​in this embodiment are as follows: In states 4 and 6, if the minimum value is less than 1 kΩ / V, it is determined that there is an insulation problem; if the minimum value is greater than or equal to 1 kΩ / V, it is determined that there is no insulation problem. In state 3, if the minimum value is less than 1 kΩ / V and the average value is less than or equal to 1.5 kΩ / V, it is determined that there is an insulation problem; if the minimum value is less than 1 kΩ / V and the average value is greater than 1.5 kΩ / V, it is determined that there is no insulation problem. The purpose of the above conditions is to eliminate false alarms. In this embodiment, the above conditions eliminate single false alarms in state 3. The same applies to vehicles using the new protocol; states 8 and 12 are equivalent to states 4 and 6 of the old protocol vehicles, and state 6 is equivalent to state 3 of the old protocol vehicles.

[0070] ④ Based on the insulation problems of the vehicle under different power system states on the day of the alarm, find the set fault table to determine whether the vehicle insulation fault is a vehicle-end insulation fault or a charger-end insulation fault, and the corresponding cause of the vehicle-end insulation fault or the charger-end insulation fault.

[0071] The example diagram for setting up the fault table is shown below. Figure 3The table's positioning principle is based on the fact that the battery negative terminal and the vehicle load negative terminal cannot be disconnected during driving and charging. In self-wake-up mode, the vehicle positive load is disconnected from the insulation detection circuit. This fault table allows for direct location and determination of the fault position and cause, thus distinguishing between vehicle-side insulation faults and charger-side insulation faults, and their specific causes. The "statistical method result is empty" situation specifically refers to a situation where the battery management system status has no value due to data quality issues, in which case the statistical method cannot identify the result.

[0072] To facilitate practical use, the overall algorithm categorizes insulation problems into broad types (insulation problems occurring during charging, insulation problems occurring during non-charging, etc.). Specifically, if an insulation alarm occurs when the vehicle is not charging, it is considered a vehicle insulation problem. If an insulation alarm only occurs during charging, it is not a charger problem. The algorithm also statistically analyzes the amount of available data for each state to avoid misjudgments due to insufficient data. The overall algorithm primarily determines whether a vehicle insulation fault is related to the charging state by judging whether it occurs during charging or non-charging. Determining whether a fault is vehicle-side or charger-side is mainly achieved through statistical methods. Even when statistical methods are insufficient, the overall algorithm can still provide a rough estimate of the cause of the insulation fault.

[0073] In this embodiment, the overall algorithm first extracts the vehicle number of the vehicle that has an insulation alarm and the battery management system status of the vehicle with the insulation alarm on the day of the fault; if the vehicle has an insulation alarm while in the battery management system status, it is determined to be a charging state insulation fault; remove the vehicles with charging state insulation faults that are determined to be vehicle insulation abnormalities from all the vehicles with insulation alarms, and the remaining vehicles are determined to be non-charging insulation faults.

[0074] The on / off algorithm determines which electrical accessories are associated with vehicle insulation faults by analyzing the relationship between the on / off state of various electrical accessories and vehicle insulation faults.

[0075] In this embodiment, the on / off algorithm first acquires and marks the times when the insulation fault occurs / disappears, electrical accessories are on, electrical accessories are off, and insulation resistance values ​​increase significantly, increase slightly, decrease significantly, and decrease slightly on the day the insulation alarm occurs, when the vehicle is not charging / not plugged in. Combining the times when the insulation fault occurs / disappears and the times when insulation resistance values ​​increase significantly, increase slightly, decrease significantly, and decrease slightly, the time period in which the vehicle insulation fault exists is obtained. The times when the insulation fault occurs / disappears can be determined by the relative insulation resistance value, which is defined by dividing the system insulation resistance value by the total voltage. If the relative insulation resistance is less than 1 kΩ / V, it is defined as an insulation abnormality; otherwise, it is considered normal insulation. By comparing the insulation status at adjacent sampling times, if the insulation is normal at the previous sampling time and abnormal at the next sampling time, it is determined that an insulation fault has occurred; if the insulation is abnormal at the previous sampling time and normal at the next sampling time, it is determined that the insulation fault has disappeared. The relative insulation resistance value is set so that the change in insulation resistance value can be compared with the insulation resistance value change level of other devices. In other embodiments, if the change in insulation resistance value is only compared with that of the vehicle, the occurrence / disappearance time of the insulation fault corresponding to the insulation alarm can also be directly determined by the insulation resistance value.

[0076] The specific method for determining the time period of a vehicle insulation fault is as follows: if a certain period of time meets either of two conditions, then that period of time is determined to be the time period of a vehicle insulation fault.

[0077] Condition 1: This period is the time interval between the moment when the insulation resistance value drops significantly and the moment when the insulation resistance value first increases significantly after the moment when it drops significantly; in this embodiment, a significant increase / decrease in insulation resistance value means that the increase / decrease in insulation resistance value is greater than 25% of the system insulation resistance value under normal circumstances.

[0078] Condition 2: This time period is the time interval between the occurrence time of the insulation fault corresponding to the insulation alarm and the disappearance time of the insulation fault. The occurrence time of the insulation fault corresponding to the insulation alarm coincides with the time when the insulation resistance value slightly decreases, and the disappearance time of the insulation fault corresponding to the insulation alarm coincides with the time when the insulation resistance value slightly increases. In this embodiment, a slight increase / decrease in insulation resistance value means that the increase / decrease in insulation resistance value is greater than 20% but less than 25% of the system insulation resistance value under normal conditions.

[0079] Normally, vehicles have a threshold for determining whether to issue an insulation fault alarm. For example, if the system insulation resistance is 10000Ω under normal conditions, the threshold for determining an insulation fault is 1000Ω. That is, if the resistance is below 1000Ω, the vehicle will determine that an insulation fault has occurred and issue an alarm. In other words, the occurrence or disappearance of an insulation fault is based on a specific numerical standard. However, this has a drawback. For example, if the system insulation resistance drops from 10000Ω to 2000Ω, the insulation level has deteriorated, meaning an insulation fault may exist, but it has not reached the threshold, so it will not be detected. Therefore, this embodiment combines several amplitude changes (significant increase, slight increase, significant decrease, and slight decrease) with the vehicle alarm status, setting two different judgment conditions. This allows for more flexible judgment of whether an insulation fault has occurred when the switch is turned on or off, and also avoids false alarms for insulation faults. The time interval from the first activation to the next activation of an electrical accessory is defined as a group for that accessory. The fault ratio of the electrical accessory in its operating state is determined by the activation / deactivation time and the time period during which a vehicle insulation fault exists within each accessory group. This ratio is the ratio of the number of faults occurring in the electrical accessory during its operating state to the total number of faults. If the fault ratio of an electrical accessory in its operating state exceeds a set threshold, the vehicle insulation fault is determined to be related to that electrical accessory. The set threshold value must ensure that, even with good on / off and fault response, the insulation fault ratio is not too low when the accessory is activated. In this embodiment, the set threshold is 30%. The data obtained for the on / off algorithm must meet the following conditions: the time of insulation problem occurrence minus the electrical accessory activation time is less than A seconds and greater than or equal to B seconds; the time of insulation problem disappearance minus the electrical accessory deactivation time is less than C seconds and greater than or equal to D seconds; and the insulation resistance value of the first data entry in the first group of the electrical accessory is normal. In this embodiment, the data obtained for the on / off algorithm determination must meet the following conditions: the time when the insulation problem occurs minus the time when the electrical accessory is turned on is less than 120 seconds and greater than or equal to -20 seconds; the time when the insulation problem disappears minus the time when the electrical accessory is turned off is less than 120 seconds and greater than or equal to -20 seconds.

[0080] Insulation problems are essentially on / off problems. Simple instantaneous algorithms can make judgments based on short-term, small amounts of data, showing high sensitivity, but are prone to false positives due to data quality issues. Statistical methods, on the other hand, are less sensitive but more reliable. If the insulation problem diminishes when a switch in a certain accessory is opened, and reappears when the switch is closed, and this occurs repeatedly, it can be determined that the insulation problem originates in that accessory. Therefore, to improve the reliability of the judgment, based on the theoretical characteristic that the insulation problem should always be present (or occur frequently) when the switch is closed, and should disappear when the switch is open, the on / off algorithm not only requires a sensitive response to the on / off states of each accessory (requiring that opening and closing must occur simultaneously, and the time interval between opening and the occurrence of the insulation problem must be extremely short), but also requires that the insulation problem persists for a large proportion of the time during the accessory's open period and a very small proportion of the time during the accessory's closed period. This balances both sensitivity and reliability in the judgment.

[0081] The concurrent fault algorithm determines the cause of a vehicle insulation fault by correlating other types of vehicle faults with the vehicle insulation fault and by observing the occurrence of other types of vehicle faults at the time the insulation fault occurs. When the insulation resistance is missing, the algorithm can provide another perspective for localization, namely, locating the cause of the insulation fault based on the type and location of other vehicle faults. Specifically, the vehicle's battery management system reports fault codes, including corresponding fault codes for insulation faults. The fault concurrency algorithm statistically analyzes the reported fault codes on a daily basis. If an insulation fault occurs on the same day that the vehicle triggers an insulation alarm (i.e., the vehicle's battery management system reports a fault code corresponding to the insulation fault), and an air compressor overcurrent fault also occurs on the same day (i.e., the vehicle's battery management system reports a fault code corresponding to the air compressor overcurrent fault, such as codes 29,1,2), then it can be determined that the cause of the insulation fault is very likely to include the air compressor overcurrent fault. It should be noted that the fault concurrency algorithm does not statistically analyze any fault codes reported by the battery management system on the day the insulation alarm occurs, but only statistically analyzes the fault codes corresponding to several pre-set faults that coexist with insulation faults, which are determined from principles and experience, to locate the cause of the insulation fault.

[0082] Therefore, a data missing data algorithm is also included in the fault concurrency algorithm. This algorithm can identify whether vehicles with insulation faults have missing data on vehicle insulation resistance values ​​(including positive, negative, and system insulation resistance values, with the system insulation resistance value being the battery system insulation resistance value mentioned earlier). Here, positive and negative refer to the insulation resistance values ​​of the vehicle battery's positive terminal to ground and negative terminal to ground, respectively, while the system insulation resistance is the result of both calculations, calculated as: System insulation resistance value = Positive insulation resistance value * Negative insulation resistance value / (Positive insulation resistance value + Negative insulation resistance value). If missing vehicle insulation resistance data is detected, the fault concurrency algorithm is used to determine the cause of the insulation fault. Only if the fault concurrency algorithm cannot determine the cause of the fault is it determined that the cause of the insulation fault cannot be obtained. Thus, the fault concurrency algorithm is essentially a remedial measure when missing vehicle insulation resistance data occurs. In fact, in this embodiment, the fault concurrency algorithm will perform calculations and judgments and obtain the fault cause judgment result regardless of whether the data is missing. However, in the final output, if the data is not missing and other algorithms that are ahead of the fault algorithm in the overall calculation logic have fault cause judgment results, the fault concurrency algorithm will not output its judgment result.

[0083] The water-related algorithm determines whether a vehicle insulation fault is water-related by identifying the data patterns of water-related vehicle insulation faults. In this embodiment, the insulation resistance values ​​of the positive and negative terminals of the battery are extracted when the vehicle is not charging / not plugged in on the day of the insulation alarm. It also statistically analyzes whether there was rainfall in the area on the day the insulation fault occurred and the two days prior. Since vehicle insulation levels are directly related to air humidity, there may be many insulation alarms caused by increased humidity. Therefore, statistically analyzing whether there was rainfall in the area on the day the insulation fault occurred and the two days prior is to provide more information to troubleshooting personnel. If there was rainfall in the area on the day the insulation fault occurred and the two days prior, the insulation fault may have been caused by rainfall. In this embodiment, the information on whether there was rainfall in the area on the day the insulation fault occurred and the two days prior is mainly used for judgment. In fact, it still determines whether the vehicle insulation fault is water-related by judging whether the conditions for water-related vehicle insulation faults are met.

[0084] The data acquired for water-related algorithm determination must meet the following conditions: data after the first insulation fault occurs; data is extracted with the time when the first data greater than A kΩ appears as the end time; data is extracted with the latest time when the data less than B kΩ appears as the start time. In this embodiment, data after the first insulation fault occurs is acquired, and data is extracted with the first data greater than 2500 kΩ as the end time; data is extracted with the latest time when the data less than 650 kΩ appears as the start time, thereby achieving the extraction of the process of the system insulation resistance value from low to normal.

[0085] The conditions for determining that a vehicle insulation fault is related to water are: fault recovery time is greater than C minutes, there are more than D data points, and the percentage increase is greater than E%. Fault recovery time refers to the time interval between the time before the insulation resistance value of a sampling point that was less than the insulation abnormality threshold at one time and greater than the insulation abnormality threshold at another time, and the time after that time when the insulation resistance value first exceeds the insulation normal threshold. For example, if the insulation abnormality threshold is 400Ω and the insulation normal threshold is 2500Ω, then the fault recovery time refers to the time interval (t2-t1) between the time before t1 and the time after t1 when the insulation resistance value first exceeds 2500Ω at the sampling point that was less than 400Ω at one time and greater than 400Ω at another time. In this embodiment, if the insulation fault recovery time is greater than 5 minutes, and the number of data sampling points (i.e., the insulation resistance data points obtained by sampling) within the insulation fault recovery time is greater than 15, and the number of rising groups accounts for more than 60% of the total number of groups, then the vehicle insulation fault is determined to be related to water. In this embodiment, the insulation resistance values ​​of every two adjacent sampling points are set as a group, and the insulation resistance values ​​within all groups are compared. If the insulation resistance value of the later sampling point in the group is greater than the insulation resistance value of the previous sampling point, then the group is determined to be a rising group. These judgment criteria are primarily based on the slow self-recovery characteristics of insulation faults related to water. The recovery time being longer than the set time is to distinguish between slow self-recovery and rapid recovery data caused by data fluctuations. When the number of data sampling points within the insulation fault recovery time is greater than the set number, it is to prevent situations where the vehicle is powered off, resulting in a long calculated recovery time but with sampling intervals between individual data points accounting for a large portion of the recovery time. For example, although the obtained recovery time is 10 minutes, if the power-off time within these 10 minutes is as long as 8 minutes, then this period only includes sampling points within a short period before and after the power-off time, which is obviously unusable for judging whether the sampling point data has self-recovery characteristics. The condition that the percentage increase is greater than the set proportion is to exclude situations where the sampling point data fluctuates significantly up and down within the recovery time. Such situations clearly do not conform to the self-recovery characteristics and cannot be used to judge whether the vehicle insulation fault is related to water.

[0086] The correlation algorithm determines which vehicle parameter variables are related to a vehicle insulation fault by judging whether different vehicle parameter variables and vehicle resistance values ​​are correlated. In this embodiment, the correlation algorithm first calculates and statistically analyzes the correlation coefficients between each vehicle parameter variable and the insulation resistance value. The correlation coefficient used is the Pearson correlation coefficient, calculated according to the Pearson correlation coefficient formula. If the correlation coefficient between a certain vehicle parameter variable and the insulation resistance value is greater than a set correlation coefficient threshold, then a correlation is considered to exist between that vehicle parameter variable and the insulation resistance value. In this embodiment, the judgment that there is a correlation between vehicle parameter variables and insulation resistance values ​​is directly equated with the judgment that there is a correlation between vehicle parameter variables and vehicle insulation faults. Therefore, by judging whether different vehicle parameter variables and vehicle insulation resistance values ​​are correlated, it is possible to directly determine which vehicle parameter variables are related to a vehicle insulation fault. In practical use, if the correlation coefficient is large (set to be greater than 0.7 in this embodiment), it is defined as a correlation between the vehicle parameter variable and the insulation resistance value. Since the correlation coefficient needs to be calculated separately, the parameter variables involved in calculating the correlation coefficient in the correlation algorithm only include two items: motor speed and air conditioner on / off status. However, there is no limit to the number. In other embodiments, the parameter variables can be flexibly set according to the actual situation. In this embodiment, the selected parameter variables are motor speed and air conditioner on / off status.

[0087] The data pattern algorithm identifies the cause of a vehicle's insulation fault by recognizing the data patterns corresponding to different insulation fault causes in different vehicles. The algorithm primarily acquires the insulation resistance values ​​of the vehicle experiencing the insulation alarm on the day of the alarm and for a set number of days prior to the alarm date. It then determines the data pattern of the insulation resistance values ​​for the vehicle's insulation fault, thereby identifying the cause of the fault. The data pattern includes many identification scenarios, each using a different data range, i.e., a different set number of days prior to the alarm date. For example, an insulation problem caused by "excessive Y capacitance" uses data from 7 days prior to the alarm date, while a "water-related" insulation problem uses data from 1 day prior to the alarm date.

[0088] The following analysis uses the case of an insulation fault caused by excessive capacitance in the insulation detection module as an example to illustrate the data pattern:

[0089] First, the insulation resistance sampling data of the vehicle on the day of the insulation alarm and the 7 days prior to the alarm are extracted. The data is grouped according to the time interval. If there are sampling data points with insulation resistance values ​​that are continuously greater than the set resistance threshold for a duration exceeding the set duration threshold (5 minutes in this embodiment), the duration of the sampling data points with continuous insulation resistance values ​​greater than the set resistance threshold is set as the insulation normal time period. Similarly, if there are sampling data points with insulation resistance values ​​that are continuously less than the set resistance threshold for a duration exceeding the set duration threshold (5 minutes in this embodiment), the duration of the sampling data points with continuous insulation resistance values ​​less than the set resistance threshold is set as the insulation abnormal time period or insulation fault time period.

[0090] If the insulation resistance of the positive or negative terminal of the vehicle battery to ground simultaneously meets the following conditions:

[0091] (1) When the vehicle is powered off for more than 1.5 hours, the insulation resistance decreases when it is powered on again;

[0092] (2) The insulation resistance does not increase when the vehicle is powered off for less than or equal to 1.5 hours and then powered on again.

[0093] (3) After the vehicle has been powered on for 4 hours or more, the insulation resistance is greater than 1,000 kΩ (i.e., the vehicle itself has no insulation problems).

[0094] The power-on and power-off times can be determined through sampling. Since the vehicle only samples and uploads data in real time when it is powered on, if the sampling data is blank for a certain period, it indicates that the vehicle is powered off. Similarly, if the sampling frequency is normal for a certain period, it indicates that the vehicle is powered on. Figure 4 The image shows the data format of the negative insulation resistance value of the system corresponding to an insulation fault caused by an excessively large capacitor in the insulation detection module.

[0095] Vehicle insulation fault cause identification device embodiment

[0096] This embodiment provides a technical solution for a vehicle insulation fault cause identification device. This device is used to implement the aforementioned vehicle insulation fault cause identification method and can achieve the same beneficial effects as the method. The specific working principle and steps of this device have been described in detail in the embodiments of the aforementioned vehicle insulation fault cause identification method, and will not be repeated here.

[0097] This invention addresses different insulation fault conditions and their characteristics by employing manually set judgment methods to assist machine learning, thereby establishing a combined algorithm. This algorithm incorporates both expert-based and data-driven identification methods. The insulation fault cause identification method of this invention treats insulation problems as multi-dimensional issues, designing multiple sub-algorithms to identify the causes of insulation faults from various angles. Through multi-dimensional in-depth analysis of each algorithm, it compensates for the limited number of fault samples. Furthermore, by considering the data characteristics of water-related insulation problems and insulation problems caused by excessive capacitance in the insulation detection module, it makes fault cause identification more comprehensive. The continuity algorithm also utilizes a comprehensive judgment combining instantaneous and statistical aspects, improving both the sensitivity and reliability of the judgment.

[0098] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit it. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that modifications or equivalent substitutions can still be made to the specific implementation of the present invention. Any modifications or equivalent substitutions that do not depart from the spirit and scope of the present invention should be covered within the scope of protection of the claims of the present invention.

Claims

1. A method for identifying the cause of insulation faults in vehicles, characterized in that, The specific method for identifying the cause of vehicle insulation faults using the aforementioned identification method is as follows: 1) First, use statistical methods to determine whether the vehicle insulation fault is a vehicle-side insulation fault or a charger-side insulation fault; if it is determined to be a vehicle-side insulation fault, then proceed to step 2). The statistical method determines whether a vehicle insulation fault is a vehicle-end insulation fault or a charger-end insulation fault by setting up a fault table, and directly finds and determines the corresponding vehicle-end insulation fault cause or charger-end insulation fault cause. 2) Obtain the battery management system status of the vehicle with the insulation alarm on the day of the alarm, and determine whether the vehicle's insulation fault is an insulation fault in the charging state or an insulation fault in the non-charging state; if the insulation alarm occurs when the battery management system status is in the charging state, it is determined that the vehicle has an insulation fault in the charging state; otherwise, it is determined that it is an insulation fault in the non-charging state, and continue to the judgment in step 3). 3) Use the continuity algorithm to determine which electrical accessories are related to the vehicle insulation fault, thereby obtaining the cause of the vehicle insulation fault; if the cause of the vehicle insulation fault cannot be obtained through the continuity algorithm, then continue to the judgment in step 4). 4) Identify the data pattern of vehicle insulation faults using data pattern algorithms to determine the corresponding cause of the insulation fault; if the cause of the vehicle insulation fault cannot be obtained through data pattern algorithms, continue to step 5). The data pattern algorithm identifies the cause of vehicle insulation failure by recognizing the data patterns of different vehicle insulation failures corresponding to different insulation failure causes. 5) Determine the cause of the vehicle insulation failure based on the vehicle parameter variables using a correlation algorithm; if the cause of the vehicle insulation failure cannot be determined by the correlation algorithm, proceed to step 6). The correlation algorithm determines which vehicle parameter variables are related to the vehicle insulation fault by judging whether different vehicle parameter variables and vehicle insulation faults are correlated. 6) Using the fault concurrency algorithm, determine the cause of the vehicle insulation fault based on the occurrence of other types of faults in the vehicle at the time the vehicle insulation fault occurs; if the cause of the vehicle insulation fault cannot be determined by the fault concurrency algorithm, then continue to the judgment in step 7). The fault concurrency algorithm determines the cause of the vehicle insulation fault by correlating the other types of vehicle faults with the vehicle insulation fault, and by considering the occurrence of other types of vehicle faults at the time the vehicle insulation fault occurs. 7) Using statistical methods, search the set fault table. If the search result is not empty, directly determine the corresponding vehicle-end insulation fault cause or charger-end insulation fault cause based on the set fault table. If the search result is empty, output the judgment result in step 2) that the vehicle insulation fault belongs to the charging state insulation fault or the non-charging state insulation fault.

2. The vehicle insulation fault cause identification method according to claim 1, characterized in that, If step 1) determines that the vehicle insulation fault is a charger-side insulation fault, or step 2) determines that the vehicle insulation fault is a charging-side insulation fault, then the data pattern algorithm is used to determine whether the data pattern of the vehicle insulation fault corresponds to the fault cause of excessive insulation detection module capacitance. If it corresponds to the fault cause of excessive insulation detection module capacitance, then the insulation fault cause is determined to be excessive insulation detection module capacitance. If it does not correspond, then the corresponding vehicle-side insulation fault cause or charger-side insulation fault cause is directly determined by statistical methods to search the set fault table.

3. The vehicle insulation fault cause identification method according to claim 1, characterized in that, If the cause of the vehicle insulation fault can be obtained through the on / off algorithm in step 3), then the water-related algorithm is used to determine whether the vehicle insulation fault is related to water. If it is related to water, then the cause of the vehicle insulation fault is determined to include not only the cause of the vehicle insulation fault obtained by the on / off algorithm, but also the vehicle insulation fault is related to water. If it is not related to water, then the cause of the vehicle insulation fault is determined to include only the cause of the vehicle insulation fault obtained by the on / off algorithm. If the cause of the vehicle insulation failure can be obtained through the correlation algorithm in step 5), then the water correlation algorithm is used to determine whether the vehicle insulation failure is related to water. If it is related to water, then the cause of the vehicle insulation failure is determined to include not only the cause of the vehicle insulation failure obtained by the correlation algorithm, but also the vehicle insulation failure is related to water. If it is not related to water, then the cause of the vehicle insulation failure is determined to include only the cause of the vehicle insulation failure obtained by the correlation algorithm. If the cause of the vehicle insulation fault can be obtained through the fault concurrency algorithm in step 6), then the water-related algorithm is used to determine whether the vehicle insulation fault is related to water. If it is related to water, then the cause of the vehicle insulation fault is determined to include not only the cause of the vehicle insulation fault obtained by the fault concurrency algorithm, but also the cause of the vehicle insulation fault being related to water. If it is not related to water, then the cause of the vehicle insulation fault is determined to include only the cause of the vehicle insulation fault obtained by the fault concurrency algorithm.

4. The vehicle insulation fault cause identification method according to claim 1, characterized in that, If step 2) determines that the vehicle insulation fault is an insulation fault in a non-charging state, then first use the data missing algorithm to identify whether the vehicle with the insulation fault has missing vehicle insulation resistance data. If data is missing, the cause of the vehicle insulation fault is determined by the fault concurrency algorithm; if the cause of the vehicle insulation fault cannot be determined by the fault concurrency algorithm, it is determined that the vehicle insulation resistance data is missing and the cause of the vehicle insulation fault cannot be determined. If there is no missing data, proceed to step 3) for judgment.

5. The vehicle insulation fault cause identification method according to any one of claims 1-4, characterized in that, The on / off algorithm is as follows: The system acquires the following information for vehicles with insulation alarms on the day of the alarm (when not charging / not plugged in): the times when the insulation fault occurs / disappears, when electrical accessories are turned on, when electrical accessories are turned off, and when the insulation resistance increases significantly, increases slightly, decreases significantly, and decreases slightly. Combining these information with the times when the insulation fault occurs / disappears and when the insulation resistance increases significantly, increases slightly, decreases significantly, and decreases slightly, the system obtains the time period during which the vehicle's insulation fault exists. The system defines the time period from the first activation to the next activation of an electrical accessory as a group for that accessory. By analyzing the activation / deactivation times of each accessory group and the duration of the vehicle's insulation fault, the system determines the fault ratio of the electrical accessory under operating conditions. If the fault ratio of an electrical accessory under operating conditions exceeds a set threshold, the system determines that the vehicle's insulation fault is related to that electrical accessory.

6. The vehicle insulation fault cause identification method according to any one of claims 1-4, characterized in that, The specific fault concurrency algorithm is as follows: Acquire and statistically analyze which other faults occurred simultaneously with the insulation fault and the frequency of each fault; determine the cause of the vehicle insulation fault by analyzing the occurrence of other types of faults in the vehicle at the time the insulation fault occurred.

7. The vehicle insulation fault cause identification method according to claim 3, characterized in that, The water-related algorithm is as follows: Extract the insulation resistance values ​​of the positive and negative terminals of the battery on the day the insulation alarm is triggered, when the battery is not charging / not plugged in. Determine whether the vehicle's insulation fault is related to water by analyzing the fault recovery time of the positive and negative terminal insulation resistance values ​​and the proportion of rising groups to the total number of groups. The fault recovery time refers to the time interval between the moment before the insulation resistance value of a sampling point that was less than the insulation abnormality threshold at one moment and greater than the insulation abnormality threshold at another moment, and the moment after that moment when the insulation resistance value first exceeds the insulation normal threshold. The rising group refers to setting the insulation resistance values ​​of every two adjacent sampling points as a group, comparing the insulation resistance values ​​within all groups, and identifying the group where the insulation resistance value of the later sampling point is greater than the insulation resistance value of the previous sampling point.

8. The vehicle insulation fault cause identification method according to any one of claims 1-4, characterized in that, The specific correlation algorithm is as follows: Calculate and statistically analyze the correlation coefficients between various vehicle parameters and insulation resistance values. If a parameter variable has a correlation coefficient greater than a set correlation coefficient threshold, it is determined that there is a correlation between the parameter variable and the insulation resistance value. Based on this, it is determined whether different parameter variables are correlated with vehicle insulation faults, and which parameter variables are related to vehicle insulation faults.

9. The method for identifying the cause of vehicle insulation faults according to any one of claims 1-4, characterized in that, The specific data format algorithm is as follows: Obtain the insulation resistance value of the vehicle on the day of the insulation alarm and the number of days prior to the alarm. Identify the cause of the vehicle insulation fault by analyzing the data format of the insulation resistance value.

10. A vehicle insulation fault cause identification device, characterized in that, The vehicle insulation fault cause identification device is used to implement the vehicle insulation fault cause identification method according to any one of claims 1-9.

Images