Energy storage device anomaly detection method, energy storage device, and storage medium

Abstract

The application relates to the field of device control and provides an abnormality detection method for an energy storage device, the energy storage device and a storage medium.The method comprises the following steps: monitoring current state data of the energy storage device; in the case that the current state data is abnormal, obtaining historical state data of the energy storage device according to the current state data; analyzing the historical state data to determine abnormality information of the energy storage device, wherein the abnormality information comprises an abnormality type; and in the case that the abnormality type belongs to a preset type, outputting an abnormality prompt according to the abnormality information. Through analysis of the current state data and the historical state data, the abnormality of the energy storage device can be accurately identified, the abnormality can be classified to obtain the abnormality type, and the abnormality prompt can be differentiated according to the abnormality type. The energy storage device abnormality can be detected and alarmed in a timely manner, and the use experience and safety of the energy storage device are improved.

Description

Technical Field

[0001] This application relates to the field of equipment control, and in particular to an anomaly detection method for energy storage equipment, an energy storage equipment, and a storage medium. Background Technology

[0002] To ensure the safety of energy storage devices and slow down battery degradation, battery management systems (BMS) typically need to continuously monitor various status data of the energy storage devices (such as voltage and temperature) and compare the status data with preset thresholds to determine whether there are any abnormalities in the operation of the energy storage devices. However, this method of simply comparing status data with preset thresholds cannot utilize the large amount of collected status data to analyze and diagnose anomalies, resulting in a lack of interpretability for anomaly alarms. Summary of the Invention

[0003] The main purpose of this application is to provide an anomaly detection method, energy storage device, and storage medium for energy storage devices, aiming to improve the accuracy and interpretability of alarm prompts.

[0004] In a first aspect, this application provides a method for detecting anomalies in an energy storage device, the method comprising the following steps: Monitor the current status data of energy storage devices; If the current status data is abnormal, the historical status data of the energy storage device can be obtained based on the current status data; The historical status data is analyzed to determine the abnormal information of the energy storage device, and the abnormal information includes the type of abnormality. If the exception type belongs to a preset type, an exception prompt will be output based on the exception information.

[0005] Secondly, this application also provides a method for detecting anomalies in an energy storage device, the method comprising the following steps: Monitor the current status data of energy storage devices; If the current status data is abnormal, the historical status data of the energy storage device can be obtained based on the current status data; Feature extraction is performed on the current state data and the historical state data to obtain data feature information; The target anomaly score of the energy storage device is determined based on the data feature information. If the target anomaly score meets the preset conditions, an anomaly prompt corresponding to the target anomaly score will be output.

[0006] Thirdly, this application also provides an energy storage device, which includes a processor, a memory, and a computer program stored in the memory and executable by the processor, wherein when the computer program is executed by the processor, it implements the above-described abnormal detection method for the energy storage device.

[0007] Fourthly, this application also provides a computer-readable storage medium storing a computer program, wherein when the computer program is executed by a processor, it implements the above-described method for detecting anomalies in an energy storage device.

[0008] This application provides a method for detecting anomalies in energy storage devices, an energy storage device, and a storage medium. By analyzing current and historical status data, this application accurately identifies anomalies in the energy storage device, classifies the anomalies into types, and provides differentiated anomaly alerts based on the anomaly type. This enables timely detection and alarming of energy storage device anomalies, improves the accuracy and interpretability of alarm alerts, and ultimately enhances the user experience and safety of the energy storage device. Attached Figure Description

[0009] To more clearly illustrate the technical solutions of the embodiments of this application, the drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0010] Figure 1 A flowchart illustrating an anomaly detection method for an energy storage device provided in an embodiment of this application; Figure 2 A flowchart illustrating an anomaly detection method for an energy storage device provided in an embodiment of this application; Figure 3 This is a schematic diagram illustrating the accumulation of a first number of anomalies according to an embodiment of this application; Figure 4 This is a schematic diagram illustrating the determination of historical state data provided in an embodiment of this application; Figure 5 This is a schematic flowchart of an anomaly detection method for an energy storage device provided in an embodiment of this application; Figure 6 This is a schematic block diagram of the structure of an energy storage device according to an embodiment of this application. Detailed Implementation

[0011] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0012] The flowchart shown in the attached diagram is for illustrative purposes only and does not necessarily include all content and operations / steps, nor does it necessarily have to be performed in the order described. For example, some operations / steps can be broken down, combined, or partially merged, so the actual execution order may change depending on the actual situation.

[0013] It should be understood that, in order to clearly describe the technical solutions of the embodiments of the present invention, the terms "first" and "second" are used in the embodiments of the present invention to distinguish identical or similar items with essentially the same function and effect. Those skilled in the art will understand that the terms "first" and "second" do not limit the quantity or execution order, and the terms "first" and "second" are not necessarily different.

[0014] It should be understood that the terminology used in this application specification is for the purpose of describing particular embodiments only and is not intended to limit the application. As used in this application specification and the appended claims, the singular forms “a,” “an,” and “the” are intended to include the plural forms unless the context clearly indicates otherwise.

[0015] It should also be understood that the term “and / or” as used in this application specification and the appended claims means any combination of one or more of the associated listed items and all possible combinations, and includes such combinations.

[0016] The following detailed description of some embodiments of this application is provided in conjunction with the accompanying drawings. Unless otherwise specified, the following embodiments and features can be combined with each other.

[0017] To ensure the safety of energy storage devices, the role of Battery Management Systems (BMS) is increasingly important in modern industrial and smart device environments, especially in scenarios requiring a stable power supply, such as electric vehicles, energy storage systems, and portable electronic devices. BMS can detect anomalies based on the status data of energy storage devices and output alerts when anomalies are detected, prompting users to pay attention to potential safety risks associated with the energy storage devices.

[0018] While traditional anomaly detection methods can ensure battery safety and lifespan to a certain extent, the conditions for triggering anomaly alerts are either too lenient, leading to a large number of invalid alarms and frequent anomaly alerts that drown out real faults, causing genuine fault signals to be ignored and resulting in insufficient accuracy of anomaly alerts; or too stringent, making it difficult to detect problems in their early stages and only issuing anomaly alerts when they become severe, making it difficult to detect and contain minor anomalies in a timely manner, resulting in insufficient recall of anomaly alerts. Furthermore, the anomaly alerts output by traditional methods lack interpretability and are difficult to differentiate between different types of anomalies.

[0019] Therefore, there is an urgent need for an anomaly detection method for energy storage devices to solve at least one of the above problems.

[0020] To resolve the above issues, please refer to [link / reference]. Figure 1 , Figure 1 This is a schematic flowchart illustrating an anomaly detection method for energy storage devices according to an embodiment of this application. The method monitors the status data of the energy storage device, determines the anomaly type when anomalies are found in the status data, and outputs an anomaly alert based on the anomaly type, thereby enabling timely detection and differentiated alerts for energy storage device anomalies. This anomaly detection method can be implemented by the energy storage device itself, for example, it can be deployed in the battery management system of the energy storage device; alternatively, it can be implemented by a server, whereby the energy storage device sends status data to a single server or server cluster via a communication module, and the single server or server cluster performs anomaly detection and instructs the energy storage device to output an anomaly alert.

[0021] It should be noted that the acquisition of any information mentioned in the method provided in this application, such as current status data and historical status data, is carried out in accordance with relevant regulations and with the consent of the corresponding users, and will not infringe on user privacy or violate relevant laws and regulations.

[0022] To solve the above problem, please refer to Figure 1 , Figure 1 This is a flowchart illustrating an anomaly detection method for an energy storage device according to an embodiment of this application. Specifically, as shown... Figure 1 As shown, the provided anomaly detection method includes steps S101 to S104. Details are as follows: Step S101: Monitor the current status data of the energy storage device.

[0023] An energy storage device typically includes at least a battery pack composed of multiple individual battery cells and a battery management system (BMS) that communicates with the battery pack. During operation, the BMS continuously samples each parameter of the battery pack to obtain a sample value. For example, it continuously samples the voltage of each individual battery cell to obtain a voltage sample value, continuously samples the temperature of the entire battery pack to obtain a temperature sample value, continuously samples the current of the entire battery pack to obtain a current sample value, and continuously samples the state of charge (SOC) to obtain a charge sample value. It's understandable that the current state data of an energy storage device refers to the currently acquired sample value.

[0024] Specifically, the battery management system acquires sampled values ​​of the battery pack at regular intervals. The length of the interval can be set according to the type of indicator; different indicators can have the same or different acquisition intervals, which is not limited here. The current state data is the sampled value acquired by the battery management system in the latest interval.

[0025] Step S102: If the current status data is abnormal, obtain the historical status data of the energy storage device based on the current status data.

[0026] Understandably, historical state data refers to sampled values ​​obtained by the battery management system during historical cycles.

[0027] The anomaly in the current status data could be due to excessively high or low values, or sudden changes. If an anomaly is directly output when it is detected, normal, transient fluctuations in the sampled values ​​might be misinterpreted as anomalies, leading to an excessively high frequency of anomaly alerts. This would prevent the anomaly alerts from effectively and accurately reflecting the abnormal conditions of the energy storage device. To avoid this, this application, upon detecting anomalies in the current status data, obtains historical status data from previous periods for further analysis, improving the accuracy of the anomaly alerts and reducing the possibility of false alarms.

[0028] Step S103: Analyze the historical status data to determine the abnormal information of the energy storage device, including the abnormality type.

[0029] For example, historical status data contains a wealth of information. In addition to determining whether an indicator is abnormal, it can also reflect the type of abnormality. This application has fully explored the historical status data and determined the type of abnormality based on the abnormal information, which makes it easier to output more interpretable abnormality prompts to users.

[0030] Step S104: If the abnormality type belongs to a preset type, output an abnormality prompt based on the abnormality information.

[0031] For example, the exception type can reflect the severity of the exception. If the exception type is of a mild severity, there is no need to output an exception message. Only when the exception type is a preset type of severity will the corresponding exception message be output, thereby reducing the output frequency of exception messages and avoiding alarm fatigue caused by frequent output of exception messages.

[0032] The anomaly detection method for energy storage devices provided in this application accurately identifies anomalies by analyzing current and historical status data, classifies the anomalies into types, and provides differentiated anomaly alerts based on the anomaly type. This achieves a balance between avoiding false alarms and avoiding missed alarms, enabling timely detection and alarming of energy storage device anomalies, thus improving the user experience and safety of energy storage devices.

[0033] Please refer to Figure 2 , Figure 2 This is a flowchart illustrating an anomaly detection method for an energy storage device provided in one embodiment of this application.

[0034] Understandably, if historical state data analysis begins as soon as an anomaly is detected in the current state data—for example, if step S103 is executed as soon as a single cycle's sampled value is detected to be too high, too low, or abruptly changed—the battery management system would need to frequently retrieve historical state data, significantly increasing its computational load. To avoid this, in some embodiments, such as... Figure 2 As shown, step S102, when the current state data is abnormal, obtains the historical state data of the energy storage device based on the current state data, including: step S1021, if the current state data belongs to the first abnormal data, the first abnormal number is accumulated; step S1022, if the current state data does not belong to the first abnormal data, the first abnormal number is cleared to zero; S1023, if the first abnormal number is greater than the first number threshold, it is determined that the current state data is abnormal.

[0035] The first abnormal data can be a sampled value whose value is outside the first numerical range. For example, if the current state data is greater than the upper limit of the first numerical range or less than the lower limit of the first numerical range, then the current state data is determined to be the first abnormal data. Alternatively, the first abnormal data can also be a sampled value whose rate of change is outside the first rate of change range.

[0036] Specifically, assume the first quantity threshold is N, where N is a positive integer greater than 1. Once it is detected that the current state data belongs to the first abnormal data, the current state data continues to be acquired and the subsequent acquired current state data is judged. The number of current state data belonging to the first abnormal data is accumulated. When N consecutive current state data belong to the first abnormal data, it is determined that the current state data is abnormal, and the execution of step S103 is triggered.

[0037] The value of N can be set according to actual needs. Since short-term fluctuations usually only last for 2 to 3 cycles, the value of N can be set to 5 to achieve a balance between avoiding false alarms and avoiding false alarms. Of course, it is not limited to this and is not restricted here.

[0038] If, during the accumulation of the first abnormality count, current state data that does not belong to the first abnormality data appears, the first abnormality count is cleared to zero, and the accumulation is restarted when the first abnormality data appears again. This ensures that step S103 is triggered only when N first abnormality data appear consecutively, thereby reducing the frequency of the battery management system backtracking historical state data and thus reducing the computational load of the battery management system.

[0039] Please refer to Figure 3 , Figure 3 This is a schematic diagram illustrating the accumulation of a first abnormality quantity according to an embodiment of this application.

[0040] like Figure 3 As shown, the arrows indicate the direction of time increment, Tn-3 to Tn+5 represent the timestamps of the sampling periods, and Dn-3 to Dn+5 represent the sampled values. It's understandable that although sampled values ​​Dn-2 and Dn-1 belonging to the first abnormal data appear in periods Tn-2 and Tn-1, if the number of first abnormal data points reaches N (e.g., N=5), and the sampled value Dn returns to normal (i.e., Dn no longer belongs to the first abnormal data), then the current state data is not considered abnormal. Only when the number of consecutive first abnormal data points reaches N is the current state data considered abnormal. For example, if five consecutive Dn+1 to Dn+5 values ​​belonging to the first abnormal data point appear in periods Tn+1 to Tn+5, then the current state data is considered abnormal.

[0041] During the operation of energy storage devices, a large amount of data is accumulated. Analyzing all the data monitored throughout the historical period would result in an enormous amount of data to be analyzed. To avoid this, in some embodiments, step S102, when the current state data is abnormal, involves obtaining the historical state data of the energy storage device based on the current state data. This includes: when the current state data is abnormal, obtaining the historical data sequence collected within a historical time window before the current collection time of the current state data; and determining the historical state data based on the historical data sequence collected within the historical time window.

[0042] For example, all data monitored in the historical period is stored in the battery device or server in the form of time series. In order to reduce the amount of data to be analyzed for historical state data, historical state data is determined only based on the historical data sequence collected within the historical time window in the time series, and only historical state data is analyzed.

[0043] The current collection time refers to the collection time of the current status data. For example, it could be the collection time of the first current status data belonging to the first abnormal data, or the collection time of the Nth current status data belonging to the first abnormal data. Of course, it is not limited to this. It could also be the collection time of the latest current status data. There is no limitation here.

[0044] The historical time window can be a pre-set time window, such as one hour, where historical data sequences collected within one hour before the current collection time are defined as historical state data. Alternatively, the historical data sequences collected within the historical time window can be analyzed to determine the abnormal start time, and the historical data sequences between the abnormal start time and the current collection time can be used as historical state data.

[0045] In some embodiments, the historical data sequence includes multiple historical sampled values ​​arranged according to historical sampling time. Determining the historical state data based on the historical data sequence includes: taking the first historical sampled value closest to the current state data in the historical data sequence as a starting point, and sequentially determining whether the historical sampled values ​​belong to the second abnormal data in a decreasing time direction; when a historical sampled value does not belong to the second abnormal data appears, the historical sampled value that does not belong to the second abnormal data is determined as the second historical sampled value; and the set of the first historical sampled value, the second historical sampled value, and the third historical sampled value between the first historical sampled value and the second historical sampled value is determined as the historical state data.

[0046] Specifically, the historical data sequence is time-series data composed of multiple historical sample values. The historical state data is obtained by tracing back from the latest historical sample value until a historical sample value no longer belongs to the second abnormal data. The latest historical sample value is the first historical sample value, the first historical sample value that does not belong to the second abnormal data is the second historical sample value, and the first historical sample value, the second historical sample value, and the third historical sample value in between are considered as the historical state data.

[0047] Please refer to Figure 4 , Figure 4 This is a schematic diagram illustrating the determination of historical state data according to an embodiment of this application.

[0048] like Figure 4 As shown, when the first quantity threshold is 5 and the current state data Dn+1 to Dn+5 all belong to the first abnormal data, a backtracking of the historical data sequence before Tn-1 is triggered. The first historical sample value Dn closest to the current state data Dn+1 to Dn+5 is taken as the starting point; historical sample values ​​Dn-1, Dn-2, ... are sequentially judged along the time-decreasing direction to determine whether they belong to the second abnormal data; when a historical sample value D1 that does not belong to the second abnormal data appears, the historical sample value D1 that does not belong to the second abnormal data is determined as the second historical sample value; the set of the first historical sample value (i.e., Dn), the second historical sample value (i.e., D1), and the third historical sample values ​​between Dn and D1 (i.e., D2 to Dn-1) is determined as the historical state data.

[0049] Understandably, the first and second anomalous data points have different criteria for judgment. It's also understandable that the first anomalous data point must be the second anomalous data point, but the second anomalous data point is not necessarily the first anomalous data point; otherwise, the continuous generation of second anomalous data would have already triggered the backtracking of historical data. In other words, being "judged as the first anomalous data point" is a sufficient but not necessary condition for being "judged as the second anomalous data point." Assuming the first anomalous data point is a sampled value whose value is outside a first numerical range, and the second anomalous data point is a sampled value whose value is outside a second numerical range, then it's understandable that the second numerical range is within the first numerical range.

[0050] Taking an abnormal situation of excessively high temperature as an example, the first numerical range can be "temperature sample value less than 60℃". If five consecutive temperature sample values ​​are not less than 60℃, historical data backtracking is triggered. The second numerical range can be "temperature sample value less than 58℃". Historical data sequences are backtracked until the temperature sample value returns from not less than 58℃ to less than 58℃, and the backtracked temperature sample value is used as historical data. In other words, when the abnormal situation is an excessively high sample value, the threshold for determining the first abnormal data is greater than the threshold for determining the second abnormal data. Similarly, when the abnormal situation is an excessively low sample value, the threshold for determining the first abnormal data is less than the threshold for determining the second abnormal data.

[0051] In some embodiments, the abnormal information further includes: abnormal time; step S103 analyzes the historical state data to determine the abnormal information of the energy storage device, including: determining the abnormal time of the energy storage device based on the earliest historical sampling value in the historical state data.

[0052] As is understandable, historical state data consists of state data collected from the start of the abnormal event to the present. Therefore, the earliest historical collection time in the historical state data is the time when the abnormal state began, and this earliest historical sampling time is taken as the abnormal time. In other words, the second sampling time of the second historical sample value in the abnormal state data is the abnormal time. In the subsequent step S104, this abnormal time is output to provide users with richer abnormal information, which helps users infer the cause of the abnormality based on the abnormal time.

[0053] Please continue to refer to Figure 4 , Figure 4 If the historical state data is D1~Dn, then the abnormal time is the sampling time T1 corresponding to D1. Although D1 does not belong to the second abnormal data, and strictly speaking, T1 is not the exact time when the abnormal situation started, it is certain that the sampled value began to be abnormal at some point between T1 and T2. Therefore, the error of taking T1 as the abnormal time is within an acceptable range.

[0054] In some embodiments, step S103, which analyzes the historical state data to determine the abnormal information of the energy storage device, further includes: determining whether the abnormality type belongs to a sudden abnormality based on the number of second abnormalities in the third abnormal data in the historical state data; and, if the abnormality type does not belong to a sudden abnormality, determining whether the abnormality type belongs to an intermittent abnormality or a continuous abnormality based on the data characteristics of the historical state data.

[0055] For example, anomaly types include abrupt anomalies, intermittent anomalies, and persistent anomalies. Among them, abrupt anomalies indicate that the state data undergoes a sudden change at a certain moment, producing a small amount of third-party anomalous data, and then returns to normal. Therefore, abrupt anomalies are considered mild anomalies and can be distinguished from other anomaly types based on the amount of third-party anomalous data.

[0056] The criteria for identifying the third outlier and the second outlier are different. It's understandable that a third outlier must be a second outlier, but a second outlier is not necessarily a third outlier; otherwise, all second-state data in the historical state data would belong to the third state, making the calculation of the number of third-state data meaningless. In other words, being "judged as third outlier" is a sufficient but not necessary condition for being "judged as second outlier." Assuming the second outlier is a sampled value whose value is outside the second numerical range, and the third outlier is a sampled value whose value is outside the third numerical range, then it's understandable that the second numerical range falls within the third numerical range.

[0057] Taking an anomaly of excessively high temperature as an example, the second numerical range could be "temperature sample value less than 58℃". Historical data sequences are backtracked until the temperature sample value returns from not less than 58℃ to less than 58℃, and the backtracked temperature sample value is used as historical state data. The third numerical range could be "temperature sample value less than 59℃". Temperature sample values ​​not less than 59℃ in the historical state data are used as the third anomaly data, and the second anomaly quantity of the third anomaly data is determined. In other words, when the anomaly is an excessively high sample value, the threshold for determining the third anomaly data is greater than the threshold for determining the second anomaly data. Similarly, when the anomaly is an excessively low sample value, the threshold for determining the third anomaly data is less than the threshold for determining the second anomaly data. In some embodiments, determining whether the anomaly type belongs to a mutation anomaly based on the number of second anomalies in the third anomaly data in the historical state data includes: if the number of second anomalies in the third anomaly data in the historical state data is less than or equal to a second quantity threshold, determining that the anomaly type belongs to a mutation anomaly; if the number of second anomalies in the third anomaly data in the historical state data is greater than the second quantity threshold, determining that the anomaly type does not belong to a mutation anomaly.

[0058] Specifically, if the number of third-abnormal data in the historical state data is small, that is, the number of second-abnormal data in the third-abnormal data is less than or equal to the second quantity threshold, it indicates that there are few third-abnormal data in the historical state data, and the historical state data only occasionally shows abnormalities. In this case, the abnormality type is determined to be a sudden abnormality. Conversely, if there are many third-abnormal data in the historical state data, that is, the number of second-abnormal data in the third-abnormal data is greater than the second quantity threshold, and the third-abnormal data appears frequently in the historical state data, it indicates that the abnormality type does not belong to the sudden abnormality that occurs occasionally. It is necessary to further determine whether it belongs to the intermittent abnormality or the persistent abnormality.

[0059] In some embodiments, determining whether the anomaly type is an intermittent anomaly or a persistent anomaly based on the data characteristics of the historical state data includes: determining that the anomaly type is a persistent anomaly if the minimum or maximum sampled value in the historical state data belongs to the fourth type of anomalous data; and determining that the anomaly type is an intermittent anomaly if the minimum or maximum sampled value in the historical state data does not belong to the fourth type of anomalous data.

[0060] Understandably, when the current status data shows an anomaly exceeding the normal range, such as excessively high temperature, the anomaly type is determined based on the smallest sample value in the historical status data. The smallest sample value in the historical status data is the lowest value among all sample values ​​in the historical status data. Similarly, when the current status data shows an anomaly below the normal range, such as excessively low voltage, the anomaly type is determined based on the largest sample value in the historical status data.

[0061] Taking the current state data being higher than the normal range as an example, if the minimum sample value in the historical state data also shows anomalies (i.e., it belongs to the fourth type of anomalous data), then it means that all historical state data shows anomalies. In other words, the anomalous event persists throughout the entire historical time window, and the anomaly type is persistent. Conversely, if the minimum sample value in the historical state data does not belong to the fourth type of anomalous data, it means that the anomalous event exists intermittently within the historical time window, and the anomaly type is intermittent.

[0062] In some embodiments, since the second historical sample value is the only historical sample value in the historical state data that does not belong to the second abnormal data, that is, the second historical sample value in the historical state data is the minimum sample value in the historical state data.

[0063] Taking the example of an abnormal event where the state data is too small, if the maximum sample value in the historical state data is also abnormal (i.e., it belongs to the fourth type of abnormal data), then it means that all historical state data contains anomalies. In other words, the abnormal event persists throughout the entire historical time window, and the anomaly type is persistent. Conversely, if the maximum sample value in the historical state data does not belong to the fourth type of abnormal data, it means that the abnormal event exists intermittently within the historical time window, and the anomaly type is intermittent.

[0064] In some embodiments, since the second historical sample value is the only historical sample value in the historical state data that does not belong to the second abnormal data, that is, the second historical sample value in the historical state data is the maximum sample value in the historical state data.

[0065] The fourth abnormal data refers to sampled values ​​whose magnitude falls outside the fourth numerical range. The size of the fourth numerical range can be set according to actual needs. Taking excessively high temperature as an example, the first numerical range could be "temperature sampled value less than 56℃". If the minimum sampled value in the historical data is not less than 56℃, it indicates that the abnormality type is a persistent abnormality. In other words, when the abnormality is an excessively high sampled value, the threshold for determining the fourth abnormal data is lower than the threshold for determining the second abnormal data. Similarly, when the abnormality is an excessively low sampled value, the threshold for determining the fourth abnormal data is higher than the threshold for determining the second abnormal data.

[0066] In some embodiments, the preset type includes intermittent anomalies and persistent anomalies. When the anomaly type belongs to the preset type, outputting an anomaly prompt based on the anomaly information includes: when the anomaly type belongs to intermittent anomalies or persistent anomalies, recording the anomaly information and outputting an anomaly prompt based on the anomaly information.

[0067] Understandably, since mutant anomalies are those that fall back to the normal state after mutation and are of a relatively mild degree, when the anomaly type is mutant, no anomaly prompt may be output. Instead, the anomaly information may be recorded and the anomaly detection method provided in this application embodiment may be continued until the mutant anomaly is detected to have evolved into an intermittent or persistent anomaly before an anomaly prompt is output. Of course, mutant anomalies may also disappear during the continuous detection process.

[0068] Intermittent and persistent anomalies are considered to be more severe types of anomalies. Intermittent and persistent anomalies are designated as preset types for outputting anomaly prompts. When an anomaly of the type of intermittent or persistent anomaly is detected, an anomaly prompt is output based on the anomaly information, and the anomaly information is recorded.

[0069] Specifically, recording the abnormal information includes: storing the abnormal information; more specifically, it includes: storing the abnormal information in a non-volatile storage medium to ensure that the abnormal information is retained after the energy storage device is restarted.

[0070] Specifically, outputting anomaly prompts based on anomaly information includes: outputting the anomaly information; more specifically, it includes: outputting the anomaly information with a first priority when the anomaly type is intermittent, and outputting the anomaly information with a second priority when the anomaly type is persistent, wherein the second priority is higher than the first priority. For example, the anomaly information can be displayed through more channels under the second priority than under the first priority, or the display area of ​​the anomaly information under the second priority is larger than that under the first priority.

[0071] In some embodiments, after analyzing the historical state data and determining the abnormal information of the energy storage device, the method further includes: extracting features from the state data to obtain data feature information of the state data; determining a target abnormality score for the energy storage device based on the data feature information; and outputting an abnormality prompt based on the abnormality information when the abnormality type belongs to a preset type, including: outputting an abnormality prompt based on the abnormality information when the target abnormality score meets preset conditions and the abnormality type belongs to a preset type.

[0072] Understandably, to improve the accuracy of anomaly alerts, the abnormal event is further evaluated using a target anomaly score before outputting the alert. Besides the anomaly type belonging to a preset type, the target anomaly score must also meet preset conditions before an anomaly alert is output, further reducing the possibility of misjudgment. For example, meeting a preset condition for the target anomaly score could be that the target anomaly score is greater than a preset score threshold.

[0073] Specifically, the target anomaly score is determined based on the data feature information of the state data, which includes current state data and historical state data. The data feature information is obtained by extracting features from the dataset composed of current state data and historical state data.

[0074] In some embodiments, determining the target anomaly score of the energy storage device based on the data feature information includes: determining the representative score and anomaly detection score corresponding to the data feature information; and fusing the anomaly detection score corresponding to the data feature information based on the representative score to obtain the target anomaly score of the energy storage device.

[0075] For example, feature extraction is performed on the state data to obtain data feature information in multiple dimensions. The data feature information in different dimensions reflects the degree of abnormality of the state data in each dimension. The higher the degree of abnormality, the higher the anomaly detection score in that dimension.

[0076] To fuse anomaly detection scores from different dimensions, the representativeness score of the data feature information in each dimension is evaluated separately. The representativeness score quantifies the contribution of that dimension to anomaly detection. For example, a high representativeness score for a dimension indicates strong representativeness, meaning that the anomaly detection score for that dimension contributes significantly to the target anomaly score; conversely, a low representativeness score for a dimension indicates strong representativeness, meaning that the anomaly detection score for that dimension contributes less to the target anomaly score. Therefore, the anomaly detection scores corresponding to the data feature information are fused based on the representativeness score. The dimensions of data feature information may include at least one of the following: cumulative statistics, historical backtracking, trend analysis, mutation detection, and continuous anomaly detection.

[0077] In some embodiments, fusing the anomaly detection scores corresponding to the data feature information based on the representative scores to obtain the target anomaly score of the energy storage device includes: determining a representative comprehensive score based on the representative scores of each data feature information; determining a fusion weight for each data feature information based on the ratio of the representative scores of each data feature information to the representative comprehensive score; and fusing the anomaly detection scores corresponding to the data feature information based on the fusion weights to obtain the target anomaly score of the energy storage device.

[0078] For example, suppose the representative scores of the data features in M ​​dimensions are Rep1, Rep2, ..., Rep... M The representative scores from each dimension can be summed to obtain the representative composite score Rep. The ratio of each representative score to the representative composite score is used as the fusion weight for the data feature information of that dimension. Therefore, the fusion weights W1, W2, ..., W... of the data feature information for each dimension are... M They are: Rep1 / Rep, Rep2 / Rep, ..., Rep M / Rep. The anomaly detection scores of each dimension are weighted and fused according to the fusion weights of each dimension to obtain the target anomaly score.

[0079] To prevent the fusion weights from increasing unreasonably, each fusion weight is set to a preset value when the representative comprehensive score is less than a preset threshold. For example, when the representative comprehensive score is 0 or close to 0, each fusion weight is set to 1 / M. However, this is not a limitation; the preset values ​​for different dimensions can be different, and this is not restricted here.

[0080] If the sum of the calculated fusion weights is not 1, for example, the difference from 1 is greater than a preset threshold, then the calculated fusion weights are renormalized. The preset threshold can be, for example, 0.001.

[0081] Understandably, the representative comprehensive score reflects the overall information quality of the data features. If the representative comprehensive score is low, the target anomaly score is not calculated.

[0082] In some embodiments, determining the representative score and anomaly detection score corresponding to the data feature information includes: determining a first representative score based on the third anomaly number and anomaly duration of the fifth anomaly data in the state data; and determining a first anomaly detection score based on the comparison result of the third anomaly number and a third number threshold.

[0083] For example, the first representative score and the first anomaly detection score are the representative score and the anomaly detection score of the cumulative statistics dimension, respectively.

[0084] For example, the number and duration of consecutive occurrences of the fifth abnormal data in the status data are statistically analyzed to obtain the number of third abnormalities and the duration of the abnormality. If there are multiple consecutive sets of fifth abnormal data in the status data, the one with the largest number is determined as the number of third abnormalities and the duration of the abnormality.

[0085] Understandably, the first and fifth outliers have different criteria for being identified. Assuming the first outlier is a sampled value whose value falls outside the first range, and the fifth outlier is a sampled value whose value falls outside the fifth range, then it's understandable that the fifth range is within the first range. In other words, the first outlier must be the fifth outlier, but the fifth outlier does not necessarily belong to the first outlier. Being identified as the first outlier is a sufficient but not necessary condition for being identified as the fifth outlier.

[0086] Specifically, the first representative score is determined based on the number of third anomalies and their duration. For example, if the number of third anomalies is greater than or equal to 5 and the duration is greater than or equal to a preset duration threshold, the first representative score is 0.9; if the number of third anomalies is greater than or equal to 3 and the duration is greater than or equal to half of the preset duration threshold, the first representative score is 0.7; otherwise, the first representative score is 0.5. Of course, this is not the only method; the first representative score can be determined using other methods or other values. This is just an example.

[0087] Specifically, the first anomaly detection score is determined based on the comparison between the third anomaly count and the third count threshold. For example, the first anomaly detection score = min(third anomaly count / third count threshold, 1.0). However, this is not the only method; the first anomaly detection score can also be determined using other methods, which are only examples here.

[0088] In some embodiments, determining the representativeness score and anomaly detection score corresponding to the data feature information includes: determining a data integrity score for the historical state data based on the number of historical sampled values ​​in the historical state data; determining a data freshness score for the historical state data based on the historical sampling time of the historical sampled values ​​in the historical state data; determining a data consistency score for the historical state data based on the historical sampled values ​​and historical sampling time in the historical state data; determining a second representativeness score based on the data integrity score and the data freshness score; and determining a second anomaly detection score based on the data integrity score, the data freshness score, and the data consistency score.

[0089] For example, the second representative score and the second anomaly detection score are the representative score and the anomaly detection score of the historical backtracking dimension, respectively.

[0090] The data integrity score represents the completeness of the statistical items required for historical backtracking, and the percentage of complete statistical items is used as the data integrity score. For example, suppose the statistical items include: maximum value, minimum value, number of data points, duration, and range of change, a total of 5; if all of the above statistical items are complete, the data integrity score is 1; if one of the statistical items is incomplete, the data integrity score is (5-1) / 5=0.8, and so on.

[0091] The data freshness score represents the proximity of the latest data point in the state data to the current time, where the current time represents the moment the calculation occurred. For example, the data freshness score = time interval T1 between the latest data point and the current time / preset duration ΔT.

[0092] The data consistency score represents the degree of consistency of state data in terms of sampling interval, field format, and dimensionality. For example, a weighted fusion of sampling interval consistency, field format consistency, and dimensionality consistency can be used. Specifically, sampling interval consistency can be quantified using the coefficient of variation (CV, standard deviation / mean); field format consistency can be achieved by separately counting the number of each format and calculating the Shannon entropy; and dimensionality consistency can be achieved by performing a logarithmic operation on the state data and calculating the standard deviation, ensuring that the same physical quantity should not have values ​​with excessively large differences in magnitude, or by using IQR outlier detection, etc.

[0093] Specifically, the second representativeness score is determined based on the data integrity score, data freshness score, and data consistency score. For example, if the data integrity score is greater than or equal to 0.8 and the data freshness score is greater than or equal to 0.8, then the second representativeness score is 0.9; if the data integrity score is greater than or equal to 0.6 and the data freshness score is greater than or equal to 0.6, then the second representativeness score is 0.7; otherwise, the second representativeness score is 0.5. Of course, this is not the only method; the second representativeness score can be determined using other methods or other values. This is just an example.

[0094] Specifically, a second anomaly detection score is determined based on the data integrity score, the data freshness score, and the data consistency score. For example, the second anomaly analysis score is obtained by combining the data integrity score, the data freshness score, and the data consistency score based on their respective preset weights.

[0095] In some embodiments, determining the representative score and anomaly detection score corresponding to the data feature information includes: acquiring statistical information of the state data, and determining a statistical item integrity score and a trend consistency score based on the statistical information; determining a third representative score based on the statistical item integrity score and the trend consistency score; and determining a third anomaly detection score based on the anomaly magnitude, anomaly time, and anomaly proportion of the fifth anomaly data in the state data relative to the anomaly determination threshold.

[0096] For example, the third representativeness score and the third anomaly detection score are the representativeness score and anomaly detection score of the trend analysis dimension, respectively.

[0097] The data integrity score represents the completeness of the statistical items required for historical backtracking, and the percentage of complete statistical items is used as the data integrity score. For example, suppose the statistical items include: maximum value, minimum value, number of data points, duration, and range of change, a total of 5; if all of the above statistical items are complete, the data integrity score is 1; if one of the statistical items is incomplete, the data integrity score is (5-1) / 5=0.8, and so on.

[0098] The trend consistency score measures the consistency of the direction of change and the magnitude of change in each period across multiple consecutive time periods. Kendall's tau coefficient can be used, a statistical indicator that measures the strength of the ranking consistency between two variables. It doesn't concern itself with the specific numerical value, but only whether the ranking trends of the two variables are synchronized. Quantifying consistency is not limited to this; other methods can also be used to quantify trend consistency.

[0099] Specifically, if the statistical information is complete and the trend is consistent, the third representativeness score is 0.9; if the statistical information is relatively complete and the trend is basically consistent, the representativeness score is 0.7; otherwise, the third representativeness score is 0.5. Specifically, a threshold can be set for the data completeness score to determine whether the statistical information is complete, and a threshold can be set for the trend consistency score to determine whether the trend is consistent. Of course, this is not the only method; other methods can be used to determine the third representativeness score, and other values ​​can also be used. This is just an example.

[0100] Specifically, the third anomaly detection score is determined based on the anomaly magnitude and duration of the state data relative to the anomaly determination threshold, as well as the anomaly proportion of the fifth anomaly data within the state data. The fifth anomaly data can be a sampled value whose value falls outside a certain range; the fifth range can be set according to actual needs and will not be elaborated upon here.

[0101] In some embodiments, determining the representative score and anomaly detection score corresponding to the data feature information includes: determining a fourth representative score based on the number of third anomalies in the fifth anomaly data in the state data and the anomaly magnitude of the state data relative to the anomaly determination threshold; and determining a fourth anomaly detection score based on whether the anomaly type belongs to the preset type.

[0102] For example, the fourth representative score and the fourth anomaly detection score are the representative score and anomaly detection score of the mutation detection dimension, respectively.

[0103] Specifically, if the number of third anomalies is less than or equal to 2 and the magnitude of the anomaly exceeds the anomaly threshold by more than 50%, the fourth representative score is 0.9; if the number of third anomalies is less than or equal to 2 and the magnitude of the anomaly is within the preset medium range, the fourth representative score is 0.7; otherwise, the fourth representative score is 0.5. Of course, this is not the only method; the method for determining the fourth representative score is not limited to the above method, and other methods or values ​​can also be used. This is just an example.

[0104] Specifically, when the anomaly type is determined to be a mutation-type anomaly, the fourth anomaly detection score is 0.8; when the anomaly type is not a mutation-type anomaly, the fourth anomaly detection score is 0.2. Of course, this is not the only limitation; the fourth anomaly detection score can be determined by other methods and can also be other values. This is just an example.

[0105] In some embodiments, determining the representative score and anomaly detection score corresponding to the data feature information includes: determining a fifth representative score based on the anomaly time, the minimum or maximum sampled value in the state data, and the anomaly proportion and anomaly duration of the fifth anomaly data in the state data; and determining a fifth anomaly detection score based on the anomaly type.

[0106] For example, the fourth representative score and the fourth anomaly detection score are the representative score and anomaly detection score of the persistent anomaly detection dimension, respectively.

[0107] Understandably, when the anomaly in the current status data is above the normal range, such as excessively high temperature, the fifth representative score is determined based on the smallest sampled value in the status data. Similarly, when the anomaly in the current status data is below the normal range, such as excessively low voltage, the fifth representative score is determined based on the largest sampled value in the historical status data.

[0108] Specifically, taking the current state data being higher than the normal range as an example, if the minimum sample value belongs to the fourth abnormal data, and the abnormal duration is greater than or equal to the preset duration threshold, and the abnormal proportion is greater than or equal to 0.8, then the fifth representative score is 0.9; if the minimum sample value belongs to the fourth abnormal data, and the abnormal duration is greater than or equal to half of the preset duration threshold, then the fifth representative score is 0.7; otherwise, the fifth representative score is 0.5. Of course, this is not the only method; the method for determining the fifth representative score is not limited to the above method and can be determined by other methods or other values. This is just an example.

[0109] Specifically, the fifth anomaly detection score is 0.9 when the anomaly type is persistent, 0.6 when the anomaly type is intermittent, and 0.1 otherwise. However, this is not the only possible score; the fifth anomaly detection score can be determined using other methods and can be other values. This is just an example.

[0110] This application also provides a method for detecting anomalies in energy storage devices. Please refer to [link / reference]. Figure 5 , Figure 5 This is a schematic flowchart of an anomaly detection method for an energy storage device provided in another embodiment of this application. Specifically, as... Figure 5 As shown, the provided anomaly detection method includes steps S201 to S205. Details are as follows: Step S201: Monitor the current status data of the energy storage device; Step S202: If the current status data is abnormal, obtain the historical status data of the energy storage device based on the current status data; Step S203: Extract features from the current state data and the historical state data to obtain data feature information; Step S204: Determine the target anomaly score of the energy storage device based on the data feature information; Step S205: If the target anomaly score meets the preset conditions, output the anomaly prompt corresponding to the target anomaly score.

[0111] It is understood that the above steps S201 to S202 can be combined with the embodiments of steps S101 to S102 and the corresponding processes of each embodiment. For example, when the target anomaly score meets the preset conditions, the anomaly type and anomaly time can be determined through the embodiments of steps S101 to S102.

[0112] It should be noted that those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working process of the above steps S201 to S202 can be referred to the embodiments of the above steps S101 to S102 and the corresponding processes of each embodiment, which will not be repeated here.

[0113] For example, the above method can be implemented as a computer program, which can be used in, for example... Figure 6 It operates on the energy storage device shown.

[0114] Please see Figure 6 , Figure 6 This is a schematic block diagram of an energy storage device provided in an embodiment of this application.

[0115] like Figure 6 As shown, the energy storage device includes a processor, a memory, and a network interface connected via a system bus, wherein the memory may include a storage medium and internal memory.

[0116] The storage medium may store the operating system and computer programs. These computer programs include program instructions that, when executed, cause the processor to perform any type of exception detection.

[0117] The processor provides computing and control capabilities to support the operation of the entire energy storage device.

[0118] Internal memory provides an environment for the execution of computer programs stored in the storage medium. When the computer program is executed by the processor, it enables the processor to perform any kind of anomaly detection method.

[0119] This network interface is used for network communication, such as sending assigned tasks. Those skilled in the art will understand that... Figure 6 The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the energy storage device to which the present application is applied. Specific energy storage devices may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0120] It should be understood that the processor can be a Central Processing Unit (CPU), but it can also be other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. Among these, a general-purpose processor can be a microprocessor or any conventional processor.

[0121] In one embodiment, the processor is configured to run a computer program stored in memory to perform the following steps: Monitor the current status data of energy storage devices; If the current status data is abnormal, the historical status data of the energy storage device can be obtained based on the current status data; The historical status data is analyzed to determine the abnormal information of the energy storage device, and the abnormal information includes the type of abnormality. If the exception type belongs to a preset type, an exception prompt will be output based on the exception information.

[0122] In one embodiment, the processor is configured to run a computer program stored in memory to perform the following steps: Monitor the current status data of energy storage devices; If the current status data is abnormal, the historical status data of the energy storage device can be obtained based on the current status data; Feature extraction is performed on the current state data and the historical state data to obtain data feature information; The target anomaly score of the energy storage device is determined based on the data feature information. If the target anomaly score meets the preset conditions, an anomaly prompt corresponding to the target anomaly score will be output.

[0123] It should be noted that those skilled in the art will understand that, for the sake of convenience and brevity, the specific working process of the energy storage device described above can be referred to the corresponding process in the aforementioned anomaly detection method embodiments, and will not be repeated here.

[0124] This application also provides a computer-readable storage medium storing a computer program, the computer program including program instructions, and the method implemented when the program instructions are executed can be referred to various embodiments of the anomaly detection method of this application.

[0125] The computer-readable storage medium can be an internal storage unit of the energy storage device described in the foregoing embodiments, such as the hard drive or memory of the energy storage device. Alternatively, the computer-readable storage medium can be an external storage device of the energy storage device, such as a plug-in hard drive, smart media card (SMC), secure digital card (SD), flash card, etc., equipped on the energy storage device.

[0126] It should be understood that the terminology used in this specification is for the purpose of describing particular embodiments only and is not intended to limit the scope of the application. As used in this specification and the appended claims, the singular forms “a,” “an,” and “the” are intended to include the plural forms unless the context clearly indicates otherwise.

[0127] It should also be understood that the term "and / or" as used in this specification and the appended claims refers to any combination and all possible combinations of one or more of the associated listed items, and includes such combinations. It should be noted that, herein, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or system that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or system. Without further limitation, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or system that includes that element.

[0128] The sequence numbers of the embodiments in this application are for descriptive purposes only and do not represent the superiority or inferiority of the embodiments. The above descriptions are merely specific implementations of this application, but the scope of protection of this application is not limited thereto. Any person skilled in the art can easily conceive of various equivalent modifications or substitutions within the technical scope disclosed in this application, and these modifications or substitutions should all be covered within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A method for detecting anomalies in an energy storage device, characterized in that, The method includes: Monitor the current status data of energy storage devices; If the current status data is abnormal, the historical status data of the energy storage device can be obtained based on the current status data; The historical status data is analyzed to determine the abnormal information of the energy storage device, and the abnormal information includes the type of abnormality. If the exception type belongs to a preset type, an exception prompt will be output based on the exception information.

2. The method according to claim 1, characterized in that, In the event that the current status data is abnormal, obtaining the historical status data of the energy storage device based on the current status data includes: If the current status data belongs to the first abnormal data, the number of the first abnormal data is incremented; If the current status data does not belong to the first abnormal data, the number of the first abnormal data is cleared to zero; If the number of the first anomalies is greater than the first number threshold, it is determined that the current state data is abnormal.

3. The method according to claim 1, characterized in that, In the event that the current status data is abnormal, obtaining the historical status data of the energy storage device based on the current status data includes: In the event that the current status data is abnormal, obtain the historical data sequence collected within the historical time window prior to the current collection time of the current status data; The historical state data is determined based on the historical data sequence collected within the historical time window.

4. The method according to claim 3, characterized in that, The historical data sequence includes multiple historical sample values ​​arranged according to historical sampling time. Determining the historical state data based on the historical data sequence includes: Starting from the first historical sample value in the historical data sequence that is closest to the current state data, determine whether the historical sample values ​​belong to the second abnormal data in the decreasing direction of time. When a historical sample value that does not belong to the second abnormal data appears, the historical sample value that does not belong to the second abnormal data is determined as the second historical sample value; The set of the first historical sample value, the second historical sample value, and the third historical sample value between the first historical sample value and the second historical sample value is determined as the historical state data.

5. The method according to claim 1, characterized in that, The abnormal information also includes: abnormal time; the analysis of the historical state data to determine the abnormal information of the energy storage device includes: The abnormal time of the energy storage device is determined based on the earliest historical sampling value in the historical status data.

6. The method according to claim 1, characterized in that, The step of analyzing the historical status data to determine the abnormal information of the energy storage device further includes: Based on the number of second anomalies in the third anomaly data in the historical state data, determine whether the anomaly type belongs to the mutation anomaly. If the anomaly type is not a sudden anomaly, the anomaly type is determined to be either an intermittent anomaly or a persistent anomaly based on the data characteristics of the historical state data.

7. The method according to claim 6, characterized in that, The step of determining whether the anomaly type belongs to a mutation anomaly based on the number of second anomalies in the third anomaly data in the historical state data includes: If the number of second anomalies in the third anomaly data in the historical state data is less than or equal to the second number threshold, the anomaly type is determined to be a mutation anomaly. If the number of second anomalies in the third anomaly data in the historical state data is greater than the second number threshold, it is determined that the anomaly type does not belong to the mutation anomaly.

8. The method according to claim 6, characterized in that, The step of determining whether the anomaly type is an intermittent anomaly or a persistent anomaly based on the data characteristics of the historical state data includes: If the minimum or maximum sampled value in the historical state data belongs to the fourth abnormal data, it is determined that the anomaly type belongs to a persistent anomaly. If the minimum or maximum sampled value in the historical state data does not belong to the fourth abnormal data, the abnormality type is determined to be an intermittent abnormality.

9. The method according to claim 1, characterized in that, The preset types include intermittent anomalies and persistent anomalies. When the anomaly type belongs to a preset type, the step of outputting an anomaly prompt based on the anomaly information includes: If the anomaly type is an intermittent anomaly or a persistent anomaly, the anomaly information is recorded and an anomaly prompt is output based on the anomaly information.

10. The method according to claim 1, characterized in that, After analyzing the historical state data and determining the abnormal information of the energy storage device, the method further includes: Feature extraction is performed on the state data to obtain the data feature information of the state data; The target anomaly score of the energy storage device is determined based on the data feature information. When the exception type belongs to a preset type, the exception prompt is output based on the exception information, including: If the target anomaly score meets the preset conditions and the anomaly type belongs to the preset type, an anomaly prompt is output based on the anomaly information.

11. The method according to claim 10, characterized in that, Determining the target anomaly score of the energy storage device based on the data feature information includes: Determine the representativeness score and anomaly detection score corresponding to the data feature information; The anomaly detection scores corresponding to the data feature information are fused based on the representative scores to obtain the target anomaly score of the energy storage device.

12. The method according to claim 11, characterized in that, The step of fusing the anomaly detection scores corresponding to the data feature information based on the representative scores to obtain the target anomaly score of the energy storage device includes: A representative comprehensive score is determined based on the representative scores of each of the aforementioned data feature information; The fusion weight of each data feature is determined based on the ratio of the representative score of each data feature to the representative comprehensive score. The anomaly detection scores corresponding to the data feature information are fused according to the fusion weight to obtain the target anomaly score of the energy storage device.

13. The method according to claim 11, characterized in that, The determination of the representativeness score and anomaly detection score corresponding to the data feature information includes: A first representative score is determined based on the number of third anomalies and the duration of anomalies in the fifth anomaly data of the status data. The first anomaly detection score is determined based on the comparison between the third number of anomalies and the third number threshold.

14. The method according to claim 11, characterized in that, The determination of the representativeness score and anomaly detection score corresponding to the data feature information includes: The data integrity score of the historical state data is determined based on the number of historical sample values ​​in the historical state data; the data freshness score of the historical state data is determined based on the historical sampling time of the historical sample values ​​in the historical state data; and the data consistency score of the historical state data is determined based on the historical sample values ​​and historical sampling time in the historical state data. A second representative score is determined based on the data integrity score and the data freshness score; The second anomaly monitoring score is determined based on the data integrity score, the data freshness score, and the data consistency score.

15. The method according to claim 11, characterized in that, The determination of the representativeness score and anomaly detection score corresponding to the data feature information includes: Obtain statistical information of the status data, and determine the statistical item completeness score and trend consistency score based on the statistical information; A third representativeness score is determined based on the statistical item completeness score and the trend consistency score; The third anomaly detection score is determined based on the magnitude of the anomaly relative to the anomaly determination threshold, the anomaly time, and the anomaly proportion of the fifth anomaly data in the state data.

16. The method according to claim 11, characterized in that, The determination of the representativeness score and anomaly detection score corresponding to the data feature information includes: A fourth representative score is determined based on the number of third anomalies in the fifth anomaly data in the state data and the magnitude of the anomaly in the state data relative to the anomaly determination threshold. The fourth anomaly detection score is determined based on whether the anomaly type belongs to the preset type.

17. The method according to claim 11, characterized in that, The determination of the representativeness score and anomaly detection score corresponding to the data feature information includes: The fifth representative score is determined based on the abnormal time, the minimum or maximum sampled value in the status data, the abnormal proportion of the fifth abnormal data in the status data, and the abnormal duration. The fifth anomaly detection score is determined based on the anomaly type.

18. A method for detecting anomalies in an energy storage device, characterized in that, The method includes: Monitor the current status data of energy storage devices; If the current status data is abnormal, the historical status data of the energy storage device can be obtained based on the current status data; Feature extraction is performed on the current state data and the historical state data to obtain data feature information; The target anomaly score of the energy storage device is determined based on the data feature information. If the target anomaly score meets the preset conditions, an anomaly prompt corresponding to the target anomaly score will be output.

19. An energy storage device, characterized in that, The energy storage device includes a processor, a memory, and a computer program stored in the memory and executable by the processor, wherein when the computer program is executed by the processor, it implements the steps of the anomaly detection method as described in any one of claims 1 to 18.

20. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program, wherein when the computer program is executed by a processor, it implements the steps of the anomaly detection method as described in any one of claims 1 to 18.

Images