Power battery fault detection method and detection system

Abstract

The invention discloses a power battery fault detection method and system, and belongs to the technical field of batteries and detection, and the method comprises the steps that a data processing center receives power battery operation data collected by a collection device, sets a first threshold value and first delay time, and generates a diagnosis report; secondary diagnosis is carried out according to the diagnosis report, the fault degree is quantified, and at least two fault levels are generated; dynamically adjusting parameters according to the fault level; and performing fault analysis again according to the parameter dynamic adjusting quantity, and outputting an early warning signal. In the implementation process of the technical scheme of the invention, preliminary diagnosis and secondary diagnosis are performed on the power battery, the fault degree is quantized, a plurality of fault levels are generated according to the quantized result, and then parameter dynamic adjustment is performed under different fault levels, so that the accuracy of the fault detection result is improved, and the fault detection efficiency is improved. And the fault misjudgment condition under the influence of the external environment is reduced, various early warning signals are generated after analysis is completed, and the driving safety is improved.

Description

Technical Field

[0001] The present application relates to the field of battery detection technology, and specifically to a power battery fault detection method and detection system. Background Art

[0002] In recent years, with the restriction of carbon emissions, the market share of new energy vehicles has become increasingly higher. In new energy vehicles, power batteries are one of the important components used to provide electrical energy to vehicles and related equipment. Therefore, the safety performance of power batteries is crucial.

[0003] In existing new energy vehicles, a battery management system (BMS) is usually installed on the vehicle. It collects core data of the power battery in real time, such as voltage, current, temperature, SOC (remaining power), etc., and analyzes these data in combination with corresponding algorithms to determine whether the power battery is operating normally. When a power battery fails, a signal is sent to the display screen to display the corresponding fault warning information. When necessary, the power battery will also cut off the power to the motor and other driving devices to prevent the power battery failure from causing driving safety hazards.

[0004] However, when the existing battery management system performs fault detection on the power battery, a fixed threshold and delay time are usually used. After obtaining the operating parameters of the power battery, it is analyzed, and then the analysis result is compared with the pre-set threshold according to the preset delay time to determine whether the power battery has a fault. This method ignores the dynamic changes of the power battery, because the degree of power battery failure will also change under different temperatures and different loads. When the dynamic data changes are large enough, inaccurate detection results may occur, and even false alarms may occur. Fixed thresholds and delay times cannot meet all situations, affecting safe driving.

[0005] Therefore, it is necessary to provide a power battery fault detection method and detection system to solve the above problems.

[0006] It should be noted that the above information disclosed in this background technology section is only for understanding the background technology of the present application concept, and therefore, it may contain information that does not constitute the prior art. Summary of the invention

[0007] Based on the above-mentioned problems existing in the prior art, the problem to be solved by the present application is: to provide a power battery fault detection method and a detection system, by grading the power battery faults and dynamically adjusting the first threshold and the first delay time in the fault detection process, so as to improve the accuracy of the fault detection results.

[0008] The technical solution adopted by the present application to solve the technical problem is: a power battery fault detection method, the method comprising: The data processing center receives the power battery operation data collected by the collection device, sets a first threshold and a first delay time, performs a preliminary diagnosis on the power battery according to the first threshold and the first delay time, and generates a diagnosis report; Perform secondary diagnosis based on the diagnosis report, establish a quantitative model to quantify the fault degree, and classify the fault level according to the quantitative results to generate at least two fault levels; Dynamically adjust parameters according to the fault level, the parameters including a first threshold and a first time delay, and generate a dynamic parameter adjustment amount; The first threshold value and the first delay time are adjusted according to the dynamic adjustment amount of the parameter, and the fault analysis is re-performed according to the adjusted values, and a warning signal is output according to the analysis result.

[0009] During the implementation of the technical solution of the present application, a preliminary diagnosis and a secondary diagnosis are performed on the power battery, and the degree of fault is quantified. Multiple fault levels are generated according to the quantification results, and then parameters are dynamically adjusted at different fault levels to improve the accuracy of the fault detection results and reduce the misjudgment of faults under the influence of the external environment. After the analysis is completed, a variety of warning signals are generated to improve driving safety.

[0010] Furthermore, the secondary diagnosis process includes the following steps: obtaining the first threshold difference in the diagnostic report, and establishing a quantitative model, quantifying the fault degree based on the first threshold difference, and determining the fault level; selecting corresponding fault analysis methods for different fault levels, and further analyzing the power battery; generating secondary fault analysis results, and sending the results to the data processing center.

[0011] Furthermore, the quantification model is a rule-based mapping model. After the mapping rules are set in the quantification model, the first threshold difference is matched with the preset mapping rules to determine the fault level, wherein the mapping rules are as follows: when the first threshold difference is less than the first critical value, it is judged as the first fault level; when the first threshold difference is greater than or equal to the first critical value and less than the second critical value, it is judged as the second fault level; when the first threshold difference is greater than or equal to the second critical value and less than the third critical value, it is judged as the third fault level.

[0012] Furthermore, a deep learning-based abnormality recognition method is used to perform fault diagnosis on the first fault level, and a fault mode tree analysis method is used to perform fault diagnosis on the second fault level and the third fault level.

[0013] Furthermore, the acquisition of the dynamic adjustment amount of the parameter is achieved by adopting a dynamic adjustment method based on online learning.

[0014] Furthermore, the dynamic adjustment method based on online learning further includes: establishing a machine learning model, and inputting the initial diagnosis report into the model as a first training set, and inputting the secondary diagnosis report into the model as a second training set; using the machine learning model to perform online learning and update the model parameters in real time; and dynamically adjusting the first threshold and the first delay time according to the results of online learning.

[0015] Furthermore, the machine learning model is one of a support vector machine, a decision tree, and a random forest.

[0016] Furthermore, when quantifying, a relative difference conversion is performed on the first threshold difference, and the actual difference is converted into a percentage difference. The result after conversion is (actual value-first threshold) / first threshold*100%.

[0017] A power battery fault detection system, the system comprising: The initial diagnosis module is used for the data processing center to receive the power battery operation data collected by the collection device, and to set a first threshold and a first delay time, to perform a preliminary diagnosis on the power battery according to the first threshold and the first delay time, and to generate a diagnosis report; A fault level classification module is used to perform secondary diagnosis according to the diagnosis report, establish a quantitative model to quantify the fault degree, and classify the fault level according to the quantification result to generate at least two fault levels; A parameter dynamic adjustment module, used to dynamically adjust the parameter according to the fault level, the parameter including the first threshold and the first time delay, and generate a parameter dynamic adjustment amount; The early warning module is used to adjust the first threshold value and the first delay time according to the dynamic adjustment amount of the parameter, re-analyze the fault according to the adjusted values, and output an early warning signal according to the analysis result.

[0018] The beneficial effects of the present application are as follows: a power battery fault detection method and detection system provided by the present application, which performs preliminary diagnosis and secondary diagnosis on the power battery, quantifies the degree of fault, generates multiple fault levels according to the quantification results, and then dynamically adjusts parameters under different fault levels, thereby improving the accuracy of the fault detection results and reducing the misjudgment of faults under the influence of the external environment. After the analysis is completed, a variety of warning signals are generated to improve driving safety.

[0019] In addition to the above-described purposes, features and advantages, the present application also has other purposes, features and advantages. The present application will be further described in detail with reference to the drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] The drawings constituting part of the present application are used to provide a further understanding of the present application. The exemplary embodiments and descriptions of the present application are used to explain the present application and do not constitute an improper limitation on the present application. In the drawings: Figure 1 This is a schematic diagram of the overall process of a power battery fault detection method in this application; Figure 2 This is a schematic diagram of the module structure of a power battery fault detection system in this application. DETAILED DESCRIPTION

[0021] It should be noted that, in the absence of conflict, the embodiments and features in the embodiments of the present application can be combined with each other. The present application will be described in detail below with reference to the accompanying drawings and in combination with the embodiments.

[0022] In order to enable those skilled in the art to better understand the solution of the present application, the technical solution in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings 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 should fall within the scope of protection of the present application.

[0023] Embodiment 1: Figure 1 As shown, the present application provides a power battery fault detection method, which is applied to the power battery of a new energy vehicle. By performing fault detection on the power battery and issuing an early warning in time after the detection is completed, driving safety is improved. The power battery fault detection method includes the following steps: Step 10: The data processing center receives the power battery operation data collected by the collection device, sets a first threshold and a first delay time, performs a preliminary diagnosis on the power battery according to the first threshold and the first delay time, and generates a diagnosis report; During the operation of the power battery, it is necessary to use a collection device to collect its operation data and send it to a data processing center. The data processing center can be an on-board data processing platform that integrates data storage, analysis and other functions, and is usually driven by an on-board chip with a certain computing power. It can also be a cloud processing platform that uploads the power battery data to the cloud processing platform by wireless transmission to achieve remote access and data analysis. This method has a low demand on the computing power of the on-board terminal and needs to rely on a relatively stable data transmission capability. In this embodiment, the form of the data processing center is not limited; The acquisition equipment is installed at various key parts of the power battery. The acquisition equipment includes but is not limited to high-precision sensors, such as voltage sensors, current sensors, temperature sensors, etc. After the operation data of the power battery is collected, it is transmitted to the data processing center through wired or wireless means. The data processing center will analyze it and determine whether the power battery has a fault; When making a preliminary diagnosis, it is necessary to set a first threshold and a first delay time. By comparing the analysis result with the first threshold, it is determined whether a fault occurs. The first delay time is set to reduce false alarms of faults, because the operating parameters of the power battery will change at the millisecond level, and some external factors will also affect the operating parameters of the power battery. For example, when a new energy vehicle encounters bumps and vibrations during driving, a series of instantaneous changes will occur. If these instantaneous changes are directly collected without a certain delay, they will be mistakenly determined as power battery failures and an early warning will be issued. Therefore, it is necessary to set the first delay time to smooth these changes and reduce the false alarm rate; The method for the data center to analyze the collected power battery operation data and generate a diagnostic report by comparing it with the first threshold value can refer to the existing technology, such as overvoltage detection, undervoltage detection, overcurrent detection, overtemperature detection, SOH (health state), SOC (charge state), etc., as well as communication fault detection, and generate a diagnostic report, wherein the diagnostic report includes not only conventional fault type, fault occurrence time and other parameters, but also the first threshold difference, that is, the comparison result of the power battery operation data and the first threshold value, so as to facilitate further analysis later; Step 20: Perform secondary diagnosis according to the diagnosis report, establish a quantitative model to quantify the fault degree, classify the fault level according to the quantitative result, and generate at least two fault levels; In the initial diagnosis process, since the first threshold and the first delay time are fixedly set, the dynamic characteristics of the power battery are not considered, which may lead to deviations in the diagnosis results. Therefore, it is necessary to perform a secondary diagnosis based on the diagnosis report of the initial diagnosis, so as to analyze the difference between the first threshold values ​​in the diagnosis report, and establish a quantitative model to quantify the fault degree, and divide it into multiple fault levels according to the quantitative results. Specifically, the secondary diagnosis process includes the following steps: Step 201: obtaining a first threshold difference in a diagnosis report, establishing a quantitative model, quantifying the fault degree based on the first threshold difference, and determining the fault level; After obtaining the first threshold difference in the diagnostic report, it is also necessary to quantify the fault degree according to the first threshold difference. Since the operation data of the power battery includes multiple types, it is necessary to establish a quantification model, which is a rule-based mapping model. After the mapping rule is set in the quantification model, the first threshold difference is matched with the preset mapping rule to determine the fault level, wherein the mapping rule is that when the first threshold difference is less than the first critical value, it is determined to be the first fault level; when the first threshold difference is greater than or equal to the first critical value and less than the second critical value, it is determined to be the second fault level; when the first threshold difference is greater than or equal to the second critical value and less than the third critical value, it is determined to be the third fault level. In this embodiment, the fault level is divided into three levels, and more levels can also be set according to actual needs. It will not be described in detail in this embodiment; For example, taking the voltage difference (|Vactual Vthreshold|) in the power battery operation parameters as an example, the first critical value is set to 3%, the second critical value is set to 5%, and the third critical value is set to 10%. When the voltage difference falls below 3%, it is identified as the first fault level, when the voltage difference falls between 3% and 5%, it is identified as the second fault level, and when it reaches the range of 5% to 10%, it is classified as the third fault level, thereby realizing the classification of fault levels; It should be noted that when quantifying, the first threshold difference needs to be converted into a relative difference, and the actual difference is converted into a percentage difference. The result after conversion is (actual value - first threshold) / first threshold * 100%, thereby obtaining the percentage difference. Compared with the actual difference, the percentage difference can be compared across parameters without considering the data type; Step 202: selecting corresponding fault analysis methods according to different fault levels, and further analyzing the power battery; After different fault levels are classified, the corresponding fault analysis method can be selected according to the fault level, because at different fault levels, the fault priority of the power battery is different, and the final warning method is also different. For example, at a low priority, the power battery may have a minor fault, or the system has a false fault alarm. At this time, only a prompt is needed. And because the data fluctuation is small, it is difficult to distinguish whether the power battery has a fault or a false fault. At a high priority, the power battery may have a major fault. At this time, continuing to supply power will bring safety hazards. Therefore, early warning measures such as power off can be taken to perform targeted fault diagnosis; Specifically, for the first fault level, an abnormality recognition method based on deep learning, such as a deep neural network, is used to perform unsupervised learning on the power battery operation data in the first fault level, and possible misjudgment data is automatically detected and marked, so as to determine the type of data that produces the misjudgment. For the second and third fault levels, a fault mode tree analysis (FTA) method can be used for fault diagnosis. By establishing a fault tree, the possible fault modes and causes under different fault levels are analyzed, so as to identify the source of the fault, such as power battery single cell failure, battery management system failure, etc., and this method can also roughly identify the misjudgment caused by the external environment; Step 203: Generate secondary fault analysis results and send the results to a data processing center.

[0024] After the secondary fault analysis, an analysis result will also be generated, which includes data such as fault type, fault time, fault location, number of misjudgments, etc., and sent to the data center for further processing.

[0025] Step 30: dynamically adjusting parameters according to the fault level, the parameters including a first threshold and a first time delay, and generating a dynamic parameter adjustment amount; After the fault levels are graded and analyzed, the parameters can be dynamically adjusted according to the analysis results, wherein the dynamically adjusted parameters include the first threshold and the first delay time, so as to optimize the operating state of the power battery and reduce false alarms. At the same time, after the first threshold is dynamically adjusted, the recognition accuracy of abnormal pulses and burr data during the sampling process will be improved. When the system state changes (such as battery aging, external temperature changes, etc.), the first threshold is dynamically adjusted to maintain the effective recognition capability of abnormal data. At the same time, when the first delay time is dynamically adjusted, for rapidly changing abnormal data (such as needle tip burrs), a shorter delay time can quickly respond and filter out; and for slowly changing abnormal data, a longer delay time can improve the accuracy of judgment. Therefore, the dynamic adjustment of parameters can enable the entire detection process to better adapt to the changes in sampling data under different working conditions and conditions, thereby enhancing the stability of the system and being more adaptable to the use scenarios of new energy vehicles. The calculation of the dynamic parameter adjustment amount is based on the fault level and fault analysis results to ensure that the operating parameters of the power battery can adapt to its actual working state at different fault levels. For example, when the power battery is at the second or third fault level, only the first threshold needs to be fine-tuned without changing the delay time; while at the first fault level, the first threshold and the delay time need to be adjusted at the same time to reduce the dynamic changes of the power battery caused by subtle changes and environmental factors; The specific method of dynamic parameter adjustment can be a machine learning algorithm based on historical data and real-time data. For example, it can be implemented by using a dynamic adjustment method based on reinforcement learning. The optimal adjustment strategy is obtained by simulating the interaction with the actual environment. According to the current state of the power battery, the control strategy is optimized through continuous simulation and feedback, so as to obtain the dynamic parameter adjustment amount.

[0026] In this embodiment, dynamic adjustment based on online learning is adopted to obtain the dynamic adjustment amount of the parameter, because the dynamic adjustment based on online learning can be continuously updated with the input of data without retraining the entire model, thereby improving the efficiency of data processing. Specifically, the method includes: Step 301: Establish a machine learning model, and input the initial diagnosis report as a first training set into the model, and input the secondary diagnosis report as a second training set into the model; When using dynamic adjustment based on online learning to obtain the dynamic amount of parameter adjustment, it is first necessary to select a suitable machine learning model. In this embodiment, the machine learning model can be one of a support vector machine, a decision tree, a random forest, etc., and the initial diagnosis report is input into the model as the first training set, and the secondary diagnosis report is input into the model as the second training set. The two diagnosis reports are input as different training sets respectively, rather than input at the same time, so that the model can gradually learn and adapt to the changes in the operating state of the power battery; The initial diagnostic report contains the operating data and fault information of the power battery in its initial state, while the secondary diagnostic report contains the latest status data after fault analysis and fault classification. By inputting these two training sets into the model separately, the machine learning model can distinguish the characteristics of power batteries at different stages during the learning process, thereby more accurately predicting and adjusting the operating parameters of the power battery.

[0027] Step 302: Use the machine learning model to perform online learning and update model parameters in real time; After the initial and secondary diagnostic reports are input into the model, the machine learning model will begin the online learning process. Online learning means that the model can update its parameters in real time when receiving new data, without the need to retrain the entire model. This method is particularly suitable for dynamically changing systems such as power batteries, because their operating status and failure modes may change with time and usage conditions. By updating the model parameters in real time, it can be ensured that the model can timely reflect the latest status of the power battery, thereby providing more accurate dynamic parameter adjustment; Step 303: dynamically adjusting the first threshold and the first delay time according to the result of online learning; After the online learning is completed, the model will output the dynamic adjustment amount of the parameters, which will be used to adjust the operating parameters of the power battery. Specifically, the first threshold will be dynamically adjusted according to the fault level of the power battery and the fault analysis results to reduce false alarms and improve the accuracy of fault detection. At the same time, the first delay time will also be adjusted according to the fault level and analysis results to optimize the response speed and safety of the power battery. Through this dynamic adjustment, the operating parameters of the power battery will be more adapted to its actual working state, thereby improving the accuracy of fault detection and eliminating the possibility of false alarms.

[0028] Step 40: adjusting the first threshold value and the first delay time according to the parameter dynamic adjustment amount, re-analyzing the fault according to the adjusted values, and outputting a warning signal according to the analysis result.

[0029] After obtaining the dynamic adjustment amount of the parameters, the first threshold value and the first delay time can be adjusted, and the fault analysis can be performed again after the adjustment is completed. At this time, due to the difference in the first threshold value and the first delay time, the output fault analysis result may be different from the previous one. At this time, it is necessary to output a warning signal based on the fault analysis result. The warning signal includes but is not limited to issuing fault code reminders, power battery output power adjustment, forced power-off, etc., so as to ensure driving safety.

[0030] Embodiment 2: Figure 2 As shown, the present application also proposes a power battery fault detection system, which runs the power battery fault detection method in embodiment 1, and the system includes: The initial diagnosis module is used for the data processing center to receive the power battery operation data collected by the collection device, and to set a first threshold and a first delay time, to perform a preliminary diagnosis on the power battery according to the first threshold and the first delay time, and to generate a diagnosis report; A fault level classification module is used to perform secondary diagnosis according to the diagnosis report, establish a quantitative model to quantify the fault degree, and classify the fault level according to the quantification result to generate at least two fault levels; A parameter dynamic adjustment module, used to dynamically adjust the parameter according to the fault level, the parameter including the first threshold and the first time delay, and generate a parameter dynamic adjustment amount; The early warning module is used to adjust the first threshold value and the first delay time according to the dynamic adjustment amount of the parameter, re-analyze the fault according to the adjusted values, and output an early warning signal according to the analysis result.

[0031] The above description is only the preferred embodiment of the present application and is not intended to limit the present application. For those skilled in the art, the present application may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application shall be included in the protection scope of the present application.

Claims

1. A power battery fault detection method, characterized in that: The method includes: The data processing center receives the power battery operation data collected by the collection device, sets a first threshold and a first delay time, performs a preliminary diagnosis on the power battery according to the first threshold and the first delay time, and generates a diagnosis report; Perform secondary diagnosis based on the diagnosis report, establish a quantitative model to quantify the fault degree, and classify the fault level according to the quantitative results to generate at least two fault levels; Dynamically adjust parameters according to the fault level, the parameters including a first threshold and a first time delay, and generate a dynamic parameter adjustment amount; The first threshold value and the first delay time are adjusted according to the dynamic adjustment amount of the parameter, and the fault analysis is re-performed according to the adjusted values, and a warning signal is output according to the analysis result.

2. A power battery fault detection method according to claim 1, characterized in that: The secondary diagnosis process includes the following steps: obtaining the first threshold difference in the diagnostic report and establishing a quantitative model to quantify the fault degree based on the first threshold difference and determine the fault level; selecting corresponding fault analysis methods for different fault levels and further analyzing the power battery; generating secondary fault analysis results and sending the results to the data processing center.

3. A power battery fault detection method according to claim 2, characterized in that: The quantification model is a rule-based mapping model. After the mapping rules are set in the quantification model, the first threshold difference is matched with the preset mapping rules to determine the fault level, wherein the mapping rules are as follows: when the first threshold difference is less than the first critical value, it is determined to be the first fault level; when the first threshold difference is greater than or equal to the first critical value and less than the second critical value, it is determined to be the second fault level; when the first threshold difference is greater than or equal to the second critical value and less than the third critical value, it is determined to be the third fault level.

4. A power battery fault detection method according to claim 3, characterized in that: For the first fault level, a deep learning-based abnormality recognition method is used for fault diagnosis, and for the second and third fault levels, a fault mode tree analysis method is used for fault diagnosis.

5. A power battery fault detection method according to claim 1, characterized in that: The acquisition of the dynamic adjustment amount of the parameters is achieved by using a dynamic adjustment method based on online learning.

6. A power battery fault detection method according to claim 5, characterized in that: The dynamic adjustment method based on online learning further includes: establishing a machine learning model, and inputting the initial diagnosis report into the model as a first training set, and inputting the secondary diagnosis report into the model as a second training set; using the machine learning model to perform online learning and update the model parameters in real time; and dynamically adjusting the first threshold and the first delay time according to the results of online learning.

7. A power battery fault detection method according to claim 6, characterized in that: The machine learning model is one of a support vector machine, a decision tree, and a random forest.

8. A power battery fault detection method according to claim 2, characterized in that: When quantifying, the first threshold difference is converted into a relative difference, and the actual difference is converted into a percentage difference. The result after conversion is (actual value-first threshold) / first threshold*100%.

9. A power battery fault detection system, characterized in that: The system includes: The initial diagnosis module is used for the data processing center to receive the power battery operation data collected by the collection device, and to set a first threshold and a first delay time, to perform a preliminary diagnosis on the power battery according to the first threshold and the first delay time, and to generate a diagnosis report; A fault level classification module is used to perform secondary diagnosis according to the diagnosis report, establish a quantitative model to quantify the fault degree, and classify the fault level according to the quantification result to generate at least two fault levels; A parameter dynamic adjustment module, used to dynamically adjust the parameter according to the fault level, the parameter including the first threshold and the first time delay, and generate a parameter dynamic adjustment amount; The early warning module is used to adjust the first threshold value and the first delay time according to the dynamic adjustment amount of the parameter, re-analyze the fault according to the adjusted values, and output an early warning signal according to the analysis result.

Images