System risk assessment method and device, and charging equipment operation and maintenance detection method

Abstract

The application discloses a system risk assessment method and device and a charging equipment operation and maintenance detection method. The system risk assessment method comprises the following steps: acquiring system change data of a target system within a target time; the system change data comprises system early warning information and device names corresponding to system changes; based on the system change data, early warning information corresponding to the target system change is determined by using a natural language processing technology; time series data of the target system within the target time is acquired, and a change trend of a monitoring index corresponding to the target system is obtained based on the time series data by using a machine learning algorithm; and the target system is subjected to system risk assessment based on the early warning information and the change trend. According to the application, the risk of system change is assessed by using the early warning information of system change obtained based on the natural language processing and the monitoring index trend anomaly obtained based on the machine learning, and the accuracy of system change risk assessment under a micro-service architecture is improved.

Description

Technical Field

[0001] This invention relates to the field of microservice systems, and in particular to a system risk assessment method and apparatus, and a charging equipment operation and maintenance testing method. Background Technology

[0002] Currently, faced with imperfect business models and the impact of massive traffic surges, many enterprises have abandoned traditional monolithic architectures in favor of microservice architectures. Microservice architecture offers advantages such as independent development, independent deployment, scalability, and reusability. However, the development, iteration, and maintenance of microservice systems are also highly complex. The increased number of services increases the probability of failure in any one service, and a single service failure can cause the entire system to crash, making fault location extremely difficult. Furthermore, the sheer number of services leads to a significant workload in deployment and management, making incidents caused by negligence highly likely. Typically, aside from unforeseen circumstances, 80% of internet product incidents are caused by bugs in the released program or changes to basic resources. Therefore, ensuring the system's original stability and core business logic is crucial during frequent product iterations and releases in complex microservice systems. Thus, the ability to quickly analyze complex systems to identify anomalies caused by system changes, conduct risk assessments of these changes, and promptly notify administrators for attention and action is a prerequisite for ensuring the stability of microservice systems.

[0003] In existing technologies, system stability is maintained through canary releases, which are conducted according to traffic or specific data content. This ensures that problems do not affect all users across the network. By controlling service routing logic and managing traffic flow, the impact of release failures is mitigated. However, this approach cannot fully cover all users or traffic in the production environment, reducing the ability to troubleshoot risks. Existing technologies also utilize version analysis functions like those from Zhuge.io, analyzing data such as new and active users after a release and comparing historical data from different versions to measure the overall effectiveness of the release. However, relying on metric trends over a period after release to evaluate the program fails to promptly identify online issues, further reducing the ability to troubleshoot system risks. Summary of the Invention

[0004] In view of this, the present invention provides a system risk assessment method that can improve the accuracy of system risk assessment under a microservice architecture.

[0005] In another aspect, the present invention provides a method for operation and maintenance testing of charging equipment, which can improve the accuracy and efficiency of operation and maintenance testing of charging equipment.

[0006] In another aspect, the present invention provides a system risk assessment device that can improve the accuracy of system risk assessment under a microservice architecture.

[0007] According to a first aspect of the present invention, this application discloses a system risk assessment method, applied to system risk assessment under a microservice architecture, comprising:

[0008] Acquire system change data of the target system within a target time period; the system change data includes system early warning information and the device name corresponding to the system change.

[0009] Based on the system change data, natural language processing technology is used to determine the early warning information corresponding to the changes in the target system.

[0010] Obtain time-series data of the target system within the target time period, and based on the time-series data, use machine learning algorithms to obtain the changing trends of the monitoring indicators corresponding to the target system;

[0011] A system risk assessment is conducted on the target system based on the early warning information and the changing trend.

[0012] Optionally, the step of determining the early warning information corresponding to the change in the target system based on the system change data using natural language processing technology includes:

[0013] The system warning information is preprocessed with corpus, and the first set of keywords is extracted from the processed system warning information;

[0014] The device name is preprocessed with corpus data, and a second set of keywords is extracted from the processed changed device name.

[0015] Calculate the similarity between the keywords in the first group of keywords and the keywords in the second group of keywords, and determine the early warning information corresponding to the changes in the target system based on the similarity.

[0016] Optionally, the target time includes the system change time of the target system and the time within a preset duration after the change.

[0017] Optionally, obtaining the changing trend of the monitoring indicators corresponding to the target system based on the time series data using machine learning algorithms includes:

[0018] The time series data is classified according to a preset classification standard to obtain corresponding target type data; the target type data includes volatile data, periodic data, and stable data.

[0019] Using a preset detection algorithm corresponding to the target type data, anomaly detection is performed on the target type data to obtain the changing trend of the monitoring indicators corresponding to the target type data, thereby obtaining the changing trend of the monitoring indicators corresponding to the target system.

[0020] Optionally, the step of using a preset detection algorithm corresponding to the target type data to perform anomaly detection on the target type data and obtain the changing trend of the monitoring indicators corresponding to the target type data includes:

[0021] Anomaly detection is performed on the volatile data based on Turkey detection to obtain the changing trend of volatile monitoring indicators.

[0022] Anomaly detection is performed on the periodic data based on the year-on-year and month-on-month comparison algorithm to obtain the changing trend of the periodic monitoring indicators.

[0023] Anomaly detection is performed on the stable data based on the time series ARIMA algorithm to obtain the changing trends of stable monitoring indicators.

[0024] Optionally, classifying the time series data according to a preset classification standard to obtain corresponding target type data includes:

[0025] The time series data is periodically detected based on window data similarity, and classified into volatile and non-volatile data according to a first preset threshold.

[0026] The stability of the non-volatile data is detected based on the STL algorithm, and the periodic data and the stable data are obtained by classification according to the second preset threshold.

[0027] Optionally, the system risk assessment of the target system based on the early warning information and the changing trend includes:

[0028] Based on the early warning information, the changing trend, and the indicator levels of the monitoring indicators, a system risk assessment is performed on the target system to obtain the risk level of the target system;

[0029] A corresponding assessment report is generated based on the risk level;

[0030] The indicator levels include core business level indicators, technical level indicators, and system resource level indicators.

[0031] According to a second aspect of the present invention, this application discloses a method for operation and maintenance testing of charging equipment, comprising:

[0032] Acquire system change data of the target system within a target time period; the target system includes a charging equipment system; the system change data includes the charging equipment operation and maintenance data, which includes system early warning information and the device name corresponding to the system change;

[0033] Based on the system change data, natural language processing technology is used to determine the early warning information corresponding to the changes in the target system.

[0034] Obtain time-series data of the target system within the target time period, and based on the time-series data, use machine learning algorithms to obtain the changing trends of the monitoring indicators corresponding to the target system;

[0035] A system risk assessment is conducted on the target system based on the early warning information and the changing trend.

[0036] Optionally, the step of acquiring time-series data of the target system within the target time period, and using machine learning algorithms based on the time-series data to obtain the changing trends of the monitoring indicators corresponding to the target system, includes:

[0037] Acquire time-series data within the target time period during the operation and maintenance testing of charging equipment;

[0038] The time series data is subjected to periodic detection to determine periodic feature values;

[0039] If the periodic characteristic value is less than a first preset threshold, the time series data is determined to be volatile data.

[0040] If the periodic feature value is greater than or equal to the first preset threshold, the time series data is further subjected to stability detection to determine the stability feature value;

[0041] If the stability characteristic value is greater than the second preset threshold, the time series data is determined to be stable data.

[0042] If the stability characteristic value is less than or equal to the second preset threshold, the time series data is determined to be periodic data.

[0043] Based on the target type of the time series data, anomaly detection is performed on the time series data to obtain the changing trend of the monitoring indicators corresponding to the target type data, and thus obtain the changing trend of the monitoring indicators corresponding to the target system.

[0044] The target data types include volatile data, periodic data, and stable data.

[0045] Optionally, the step of performing periodic detection on the time series data includes:

[0046] Given a reference period T for the time series data;

[0047] Using the reference period T as the dividing point, the time series data is divided into n / T sub-time series units, where n is the length of the time series data;

[0048] For each of the sub-time series units, a similarity coefficient is calculated by comparing each pair of units, and the similarity coefficient is determined as the periodic feature value;

[0049] The further stability detection of the time series data includes:

[0050] The time series data is seasonally decomposed using the moving average algorithm into seasonal periodic components, long-term trend components, and random residual components.

[0051] The stability eigenvalue is determined based on the variance of the random residual components.

[0052] Optionally, the step of performing anomaly detection on the time series data based on the target type of the time series data includes:

[0053] If the time series data is determined to be volatile, the Turkey's Test or 3-Sigema algorithm is used to perform dynamic threshold detection on the time series data in order to obtain the changing trend of the monitoring indicators corresponding to the target system.

[0054] If the time series data is determined to be stable, a baseline threshold detection is performed on the time series data using SARIMAX or moving average algorithms to obtain the changing trend of the monitoring indicators corresponding to the target system.

[0055] If the time series data is determined to be periodic data, machine learning classification and regression models are used to detect anomalies in the time series in order to obtain the changing trends of the monitoring indicators corresponding to the target system.

[0056] According to a third aspect of the present invention, this application discloses a system risk assessment apparatus for system risk assessment under a microservice architecture, comprising:

[0057] The data acquisition module is used to acquire system change data of the target system within a target time period; the system change data includes system early warning information and the name of the change device;

[0058] The early warning information determination module is used to determine the early warning information corresponding to the changes in the target system based on the system change data and using natural language processing technology.

[0059] The trend determination module is used to acquire time series data of the target system within the target time period, and based on the time series data, use machine learning algorithms to obtain the trend of change of the monitoring indicators corresponding to the target system.

[0060] The risk assessment module is used to conduct a system risk assessment of the target system based on the early warning information and the changing trend.

[0061] Fourthly, this application also discloses an electronic device, comprising:

[0062] Memory, used to store computer programs;

[0063] A processor is used to execute the computer program to implement the aforementioned system risk assessment method.

[0064] Fifthly, this application also discloses a computer-readable storage medium for storing a computer program; wherein the computer program, when executed by a processor, implements the aforementioned system risk assessment method.

[0065] In this application, system change data of the target system within a target time period is obtained. This system change data includes system warning information and the names of devices corresponding to the system changes. Then, based on this system change data, natural language processing (NLP) technology is used to determine the warning information corresponding to the changes in the target system. Furthermore, time-series data of the target system within the target time period is obtained, and based on this time-series data, machine learning algorithms are used to obtain the changing trends of the monitoring indicators corresponding to the target system. Finally, a system risk assessment of the target system is performed based on the warning information and the changing trends. It is evident that by using NLP technology to determine the warning information corresponding to the system changes through system warning information within a time range and the names of devices affected by the system changes within a spatial range, and then combining this with the changing trends of the monitoring indicators corresponding to the system determined by time-series data using machine learning algorithms, the current risk of the system can be assessed. This method can quickly and accurately identify risks arising from software patch releases, configuration changes, and basic resource changes, improving the accuracy of system change risk assessment under a microservice architecture. This method is applicable to the operation and maintenance testing of charging equipment, power grid operation and maintenance testing, and / or vehicle-charging-network integration, as well as the operation and maintenance testing of other detection systems under a microservice architecture built on the Internet of Things (IoT), demonstrating strong versatility. Attached Figure Description

[0066] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on the provided drawings without creative effort.

[0067] Figure 1 A flowchart of a system risk assessment method provided in Embodiment 1 of this application;

[0068] Figure 2 A flowchart illustrating a specific system risk assessment method provided in Embodiment 1 of this application;

[0069] Figure 3 A flowchart illustrating a specific system risk assessment method provided in Embodiment 2 of this application;

[0070] Figure 4 A flowchart illustrating a specific system risk assessment method provided in Embodiment 3 of this application;

[0071] Figure 5 A flowchart for classifying time series data is provided in this application;

[0072] Figure 6 This is a flowchart of a charging equipment operation and maintenance testing method provided in Embodiment 4 of this application;

[0073] Figure 7 A diagram of the charging equipment operation and maintenance system architecture;

[0074] Figure 8 A schematic diagram of the specific process for step S43 of the charging equipment operation and maintenance testing method;

[0075] Figure 9 A schematic diagram of a system risk assessment device provided in this application;

[0076] Figure 10 This application provides a structural diagram of an electronic device. Detailed Implementation

[0077] Existing technologies maintain system stability through gray-scale distribution, but this cannot fully cover all users or traffic in the production environment, reducing the ability to investigate risks. While Zhuge.io's version analysis function analyzes data such as new and active users after a version update and compares it with historical data from different versions to measure the overall effect of the release, it cannot promptly identify online issues, further reducing the ability to investigate system risks. To overcome these technical problems, this application provides a system risk assessment method that can improve the accuracy of risk assessment for system changes in a microservice architecture.

[0078] Embodiment 1 of this application discloses a system risk assessment method applied to system risk assessment in a microservice architecture. See [link to documentation]. Figure 1 As shown, the method may include the following steps:

[0079] Step S11: Obtain system change data of the target system within the target time period; the system change data includes system early warning information and the device name corresponding to the system change.

[0080] In this embodiment, system change data of the target system within a target time period is first acquired. This system change data includes system warning information and the names of devices corresponding to the system changes, i.e., the names of devices affected by the system changes. These device names include, but are not limited to, data centers, business units, functional modules, service components, nodes, and hosts. The target time includes the system change time of the target system and the time within a preset duration after the change. It can be understood that the target time includes the period from the start to the end of the change in the target system, and the period after the change ends for a preset duration. Since the impact of system changes does not immediately cease after the change is completed, acquiring system change data within a certain period after the change improves the completeness of the change data. The system changes may include, but are not limited to, software releases, configuration changes, basic setting adjustments, database publish / subscribe adjustments, and middleware maintenance adjustments. Specifically, these may include host expansion, machine restarts, and network adjustments.

[0081] Step S12: Based on the system change data, use natural language processing technology to determine the early warning information corresponding to the change in the target system.

[0082] In this embodiment, after obtaining the aforementioned system change data, natural language processing technology is used to determine the early warning information corresponding to the change in the target system from the aforementioned system change data.

[0083] Understandably, microservice architectures generally have relatively complete monitoring and early warning systems, capable of detecting faults and issuing system alerts as much as possible. However, due to the large number of components in a microservice architecture and the frequent occurrence of alerts, when system changes cause anomalies or accidents, the number of system alerts explodes, making it difficult to find alerts truly related to the system change. Therefore, by combining system alerts with the device names corresponding to the system change, and using corpus preprocessing and feature extraction, alerts related to changes in the target system can be identified from the system change data.

[0084] Step S13: Obtain the time series data of the target system within the target time period, and based on the time series data, use a machine learning algorithm to obtain the changing trend of the monitoring indicators corresponding to the target system.

[0085] In this embodiment, while determining the early warning information corresponding to the changes in the target system through system change data, time-series data of the target system within the target time period is obtained. Then, based on the time-series data, machine learning algorithms are used to obtain the changing trends of the monitoring indicators corresponding to the target system. It is understood that microservice architectures have numerous components (i.e., services), and each component needs to monitor different indicators. However, when a component experiences an anomaly, the trend of its monitored indicators will undergo an abnormal change, either rising or falling overall. Therefore, trend detection of monitoring indicators can be used to determine whether an anomaly has occurred in a component or service. Specifically, different algorithms can be used to detect different types of monitoring indicators to obtain the corresponding changing trends.

[0086] Step S14: Conduct a system risk assessment of the target system based on the early warning information and the changing trend.

[0087] In this embodiment, after obtaining the early warning information corresponding to the changes in the target system and the changing trends of the target system's monitoring indicators, a system risk assessment is performed on the target system based on the early warning information and changing trends to determine the risks and risk levels present in the target system. It is understood that, for example... Figure 2 As shown in this embodiment, after a system change, relevant monitoring data is used to determine the early warning information associated with the change, as well as to detect abnormal trends in the monitoring indicators of the target system. Finally, a comprehensive assessment of the system risk of the target system is conducted. The early warning information can help understand system problems related to the system change, and the changing trends of the monitoring indicators can help determine whether any abnormalities have occurred in the components or services. Thus, the system risk can be accurately obtained through comprehensive analysis.

[0088] As can be seen from the above, this embodiment uses natural language processing technology to determine the warning information corresponding to the system change by using the system warning information within the time range of the system change and the names of the devices affected by the system change within the spatial range. Then, it combines the changing trends of the system's corresponding monitoring indicators determined by the time series data using machine learning algorithms to assess the current risks of the system. This can quickly and accurately identify the risks brought about by software patch releases, configuration changes, and basic resource changes, thereby improving the accuracy of system change risk assessment under the microservice architecture.

[0089] Embodiment 2 of this application discloses a specific system risk assessment method, applied to system risk assessment under a microservice architecture. (See also...) Figure 3 As shown, the method may include the following steps:

[0090] Step S21: Obtain system change data of the target system within the target time; the system change data includes system early warning information and the device name corresponding to the system change.

[0091] Step S22: Perform corpus preprocessing on the system warning information and extract the first set of keywords from the processed system warning information; perform corpus preprocessing on the device name and extract the second set of keywords from the processed device name.

[0092] In this embodiment, after obtaining the aforementioned system warning information and the device names corresponding to the system changes, the system warning information and device names are preprocessed. Then, keywords are extracted from the processed data according to specific business logic, resulting in a first set of keywords and a second set of keywords. The corpus preprocessing includes, but is not limited to, word segmentation, part-of-speech tagging, named entity recognition, and stop word removal. Specifically, firstly, the system warning information text is divided into a data structure based on words using a word segmentation tool. Then, a part-of-speech tagging tool is used to tag the segmented results as verbs, nouns, or adjectives, etc. Next, a named entity recognition tool is used to identify proper nouns after the part-of-speech tagging, such as data center, business unit, host, and node. Finally, stop words in the text are removed according to a Chinese stop word list. The preprocessing tools include, but are not limited to, jieba, Hanlp, NLTK, and Stanford CoreNLP. The Chinese stop word list includes, but is not limited to, the Harbin Institute of Technology stop word list, the Baidu stop word list, and the Sichuan University Machine Intelligence Laboratory stop word library.

[0093] Step S23: Calculate the similarity between the keywords in the first group of keywords and the keywords in the second group of keywords, and determine the early warning information corresponding to the change of the target system based on the similarity.

[0094] In this embodiment, after obtaining the first set of keywords and the second set of keywords, the similarity between the keywords in the first set of keywords and the second set of keywords is calculated. That is, the similarity between the keywords obtained based on the warning information and the keywords obtained based on the device name corresponding to the system change is calculated. Then, keywords with a similarity greater than a preset threshold are selected as the warning information corresponding to the change of the target system. The algorithm for calculating the similarity may include, but is not limited to, Euclidean distance, Manhattan distance, cosine similarity, Pearson similarity, K-means, and DBSACN.

[0095] Step S24: Obtain the time series data of the target system within the target time period, classify the time series data according to a preset classification standard, and obtain the corresponding target type data.

[0096] In this embodiment, time-series data of the target system within the aforementioned target time period is acquired, and then classified according to a preset classification standard to obtain corresponding target type data. The target type data may include volatile data, periodic data, and stable data; that is, time-series data is classified according to data characteristics to obtain corresponding volatile data, periodic data, and stable data. It is understood that monitoring indicators are numerous and their relationships are complex, but the indicators themselves have characteristics such as periodicity, regular spikes, overall rises or falls, and low periods. Factors affecting the indicators include holidays, temporary events, weather, and epidemics. Different types of indicators can be detected using different algorithms to obtain better detection results; therefore, time-series data can be automatically classified according to the characteristics of the indicators.

[0097] Step S25: Using a preset detection algorithm corresponding to the target type data, perform anomaly detection on the target type data to obtain the changing trend of the monitoring indicators corresponding to the target type data, so as to obtain the changing trend of the monitoring indicators corresponding to the target system.

[0098] In this embodiment, after obtaining the target type data, a preset detection algorithm corresponding to the target type data is used to perform anomaly detection on the target type data, thereby obtaining the changing trend of the monitoring indicators corresponding to the target type data, and thus obtaining the changing trend of the monitoring indicators corresponding to the target system. It is understood that using different detection algorithms for different types of data can improve the accuracy of the detection results.

[0099] Step S26: Based on the early warning information, the changing trend, and the indicator level of the monitoring indicators, perform a system risk assessment on the target system to obtain the risk level of the target system.

[0100] In this embodiment, after obtaining the aforementioned early warning information and trend, a system risk assessment is performed on the target system based on the early warning information, trend, and the indicator levels of the aforementioned monitoring indicators to obtain the risk level of the target system. The aforementioned indicator levels may include core business level indicators, technical level indicators, and system resource level indicators.

[0101] Understandably, different monitoring metrics reflect varying degrees of importance to system operation. These metrics can be categorized into three levels of impact: core business-level metrics, technical metrics, and system resource-level metrics, with the impact decreasing in that order. Core business-level metrics reflect the real-time overall health of the business. They provide a clear picture of the extent of damage, market impact, and scope of influence. When core business-level metrics show anomalies after a system change, especially when they are damaged, it indicates a very high risk associated with the system change, impacting the stability of core business operations. Technical metrics reflect the real-time health of components or services. They reflect upstream and downstream call volume (TPS), latency, success rate, and thread count. When technical metrics show abnormal trends after a system change—such as increased latency, decreased success rate, sudden drop in TPS, or sudden increases or decreases in thread count—it can be determined through technical metrics that the component or service has been affected by the system change. System resource level metrics can be real-time health status at the middleware, host, or system level, such as CPU utilization, memory utilization, network traffic, etc. When a system failure occurs, system resource level metrics usually show abnormalities one after another.

[0102] A system risk assessment is conducted on the target system based on early warning information, the changing trends of monitoring indicators, and the indicator levels of these indicators, resulting in a risk level for the target system. These risk levels can include no risk, low risk, medium risk, and high risk. Furthermore, the risk level is determined by weighting the monitoring indicators showing changing trends according to their indicator levels. For example, a situation where no system-related early warning information is detected and no abnormal changing trends are detected can be classified as no risk; a situation where system-related early warning information is detected and abnormal changing trends are detected can be classified as medium risk; and a situation where abnormal changing trends are detected in core business-level indicators can be classified as high risk.

[0103] Step S27: Generate a corresponding assessment report based on the risk level.

[0104] In this embodiment, a corresponding assessment report is generated based on the determined risk level. For example, an assessment report for no risk can be "Change Successful," an assessment report for medium risk can be "Closely Monitor," and an assessment report for high risk can be "Immediately Roll Back." This allows the administrator to be accurately notified of the current status of the target system and the corresponding handling method.

[0105] The specific process of step S21 can be found in the relevant content disclosed in the foregoing embodiments, and will not be repeated here.

[0106] As shown above, by preprocessing the system warning information into a corpus and extracting the first set of keywords from the processed system warning information, and preprocessing the device names into a corpus and extracting the second set of keywords from the processed changed device names, the similarity between the first set of keywords and the second set of keywords is calculated. Based on the similarity, the warning information corresponding to the change in the target system is determined. Based on corpus preprocessing and feature extraction, warning information associated with changes can be determined. Keywords associated with system changes can be extracted from the surge in system warning information during a fault, and then similarity is calculated with the device names obtained from the change spatial range. Based on the degree of association, the warning information corresponding to the change in the target system is determined. Furthermore, by acquiring time-series data of the target system within the target time period, classifying the time-series data according to a preset classification standard, and obtaining corresponding target type data, anomaly detection is performed on the target type data using a preset detection algorithm corresponding to the target type data. The changing trends of the monitoring indicators corresponding to the target type data are obtained, thus revealing the changing trends of the monitoring indicators corresponding to the target system. Machine learning-based monitoring metric trend detection can intelligently select classification models based on the data characteristics of the metrics, make anomaly judgments based on historical data characteristics, and then determine whether changes have affected components or services and whether they have damaged core business.

[0107] Embodiment 3 of this application discloses a specific system risk assessment method, applied to system risk assessment under a microservice architecture. (See also...) Figure 4 As shown, the method may include the following steps:

[0108] Step S31: Obtain system change data of the target system within the target time; the system change data includes system early warning information and the device name corresponding to the system change.

[0109] Step S32: Based on the system change data, use natural language processing technology to determine the early warning information corresponding to the change in the target system.

[0110] Step S33: Obtain the time series data of the target system within the target time period.

[0111] Step S34: Perform periodic detection on the time series data based on window data similarity, and classify the volatile data and non-volatile data according to a first preset threshold.

[0112] In this embodiment, for example Figure 5As shown, the time series data is first subjected to periodicity detection based on window data similarity, and then classified into volatile and non-volatile data according to a first preset threshold. Specifically, the similarity of the time series windows is calculated, which can be based on Pearson correlation and dynamic programming distance. The first preset threshold can be 0.8.

[0113] Step S35: Perform stability detection on the non-volatile data based on the STL algorithm, and classify the data according to the second preset threshold to obtain the periodic data and the stable data.

[0114] In this embodiment, for example Figure 5 As shown, the stability of the aforementioned non-volatile data is tested using the STL algorithm, and the data is classified into periodic and stable types based on a second preset threshold. Specifically, the non-volatile time series data is decomposed using the STL algorithm of the additive model, i.e., time series = periodic component + trend component + residual term. The variance of the residual term is used to determine whether the data is stable. The second preset threshold can be 0.002.

[0115] Step S36: Perform anomaly detection on the volatile data based on Turkey detection to obtain the changing trend of the volatile monitoring indicators.

[0116] In this embodiment, anomaly detection is performed on the aforementioned volatile data based on Turkey's test to obtain the changing trend of the volatile monitoring indicator. Specifically, the anomaly detection algorithm based on Turkey's test may include: firstly, calculating the first quartile (Q1), median, and third quartile (Q3) of the changing trend data, and letting IQR = Q3 - Q1. If IQR is between Q3 + k(IQR) and Q1 - k(IQR), then the volatile data can be considered to have no anomalies; otherwise, the corresponding volatile monitoring indicator is determined to have an abnormal trend. Here, k is a preset threshold, which can be 1.5 or 3.

[0117] Step S37: Perform anomaly detection on the periodic data based on the year-on-year and month-on-month comparison algorithm to obtain the changing trend of the periodic monitoring indicators.

[0118] In this embodiment, anomaly detection is performed on periodic data based on the year-on-year / month-on-month comparison algorithm to obtain the changing trend of periodic monitoring indicators. Specifically, the anomaly detection algorithm based on the year-on-year / month-on-month comparison algorithm may include: extracting 7-day month-on-month data, removing outliers, and filling in missing values; then calculating the mean and standard deviation of the year-on-year data; finally, using the periodic anomaly judgment formula: |tm|>b×σ, determining whether the corresponding periodic monitoring indicator has an abnormal trend; where t is the current value, m is the year-on-year mean, b is a preset threshold, and σ is the standard deviation. If |tm| is greater than b×σ, the corresponding periodic monitoring indicator is determined to have an abnormal trend.

[0119] Step S38: Perform anomaly detection on the stable data based on the time series ARIMA algorithm to obtain the changing trend of the stable monitoring indicators, so as to obtain the changing trend of the monitoring indicators corresponding to the above target system.

[0120] In this embodiment, anomaly detection is performed on stable data based on the time series ARIMA (Autoregressive Integrated Moving Average model) algorithm to obtain the changing trends of stable monitoring indicators. Finally, based on the changing trends of volatile monitoring indicators, periodic monitoring indicators, and stable monitoring indicators, the changing trends of the monitoring indicators corresponding to the target system are obtained. Specifically, the anomaly detection algorithm based on the time series ARIMA algorithm may include: first, performing stability and white noise tests on the first-order differencing of the time series according to preset stability thresholds and white noise thresholds to obtain stable time series data that meets the conditions; then, splitting the stable time series data into: trend component + periodic component + residual term according to the STL algorithm; then, training the ARIMA model on the trend component of historical data, automatically optimizing parameters using a grid search method to obtain the predicted trend; finally, judging whether there is an abnormal trend in the corresponding stable monitoring indicator according to the stable anomaly judgment formula: |rp|>c×l. Where r is the actual data, p is the predicted data, c is the preset threshold, and l is the historical residual; where predicted data = predicted trend + historical period. If |rp| is greater than c×l, then the corresponding stable monitoring indicator is determined to have an abnormal trend.

[0121] Step S39: Conduct a system risk assessment of the target system based on the early warning information and the changing trend.

[0122] The specific processes of steps S31 to S33 and S39 can be found in the relevant content disclosed in the foregoing embodiments, and will not be repeated here.

[0123] As can be seen from the above, anomaly detection is performed on the volatile data based on Turkey detection to obtain the changing trends of volatile monitoring indicators; anomaly detection is performed on the periodic data based on the year-on-year / month-on-month comparison algorithm to obtain the changing trends of periodic monitoring indicators; and anomaly detection is performed on the stable data based on the time series ARIMA algorithm to obtain the changing trends of stable monitoring indicators, thereby obtaining the changing trends of the monitoring indicators corresponding to the target system. By using different anomaly detection algorithms for different types of data, the accuracy of indicator anomaly detection is improved; combined with early warning information of system changes, this embodiment can quickly and accurately identify the risks brought about by system changes such as software patch releases, configuration changes, and basic resource changes, thereby effectively reducing the impact of system changes on the stability of the microservice system.

[0124] The operation and maintenance (O&M) of charging equipment is a crucial aspect of the power system. Daily inspections of charging piles directly impact the safety, reliability, and failure rate of the charging equipment. O&M personnel are tasked with promptly identifying defects and potential safety hazards in charging facilities and repairing them to ensure their stable operation. High-risk areas for charging pile failures are primarily along the energy path, mainly involving charging modules, DC contactors, fuses, charging cables, and charging heads. Damage to charging cables and charging heads is primarily mechanical, leading to electrical faults, which are relatively easy for inspectors to detect. However, potential problems with charging modules, DC contactors, and fuses can only be identified through data acquisition and analysis. This involves installing monitoring and measurement points inside the charging pile to measure the voltage across the DC contactor during charging, storing the voltage drop data in memory. Currently, the O&M of charging facilities suffers from an incomplete fault diagnosis and safety maintenance service system, lacking intelligent fault diagnosis and safety warnings during the charging process, and failing to detect potential equipment defects and accidents in advance.

[0125] According to the embodiments of this disclosure, for the intelligent operation and maintenance of charging equipment, Embodiment 4 of this application provides a method for detecting the operation and maintenance of charging equipment. See [link to previous document]. Figure 6 As shown.

[0126] As an example, this embodiment focuses on the operation and maintenance testing of charging equipment. In this process, electric vehicles serve as the carrier, acting as the on-board terminal for communication with AC / DC charging piles and the charging network. Currently, the on-board terminal of the electric vehicle primarily uses wireless communication methods such as 3G LTE, 4G LTE, and GPRS to upload data to the cloud platform management center in real time. The vehicle-charging-network system architecture is as follows: Figure 7As shown, the system mainly consists of several parts: vehicle-mounted terminals 11, charging piles 12, charging networks 13, energy storage devices 14, and power grids 15. The vehicle-mounted terminals 11 and charging piles 12 communicate wirelessly; the charging piles 12 and charging networks 13 communicate wirelessly; the charging networks 13, energy storage devices 14, and power grids 15 communicate wirelessly; multiple vehicle-mounted terminals 11 communicate wirelessly; and multiple charging piles 12 communicate wirelessly, forming an intelligent operation and maintenance system for the charging shared network. This intelligent operation and maintenance system collects full charging data information from the charging station's power distribution system, charging system, and charging vehicles through charging pile probes, DTUs, image recognition acquisition, and portable diagnostic tools. Data is transmitted through the network layer, relying on vehicle network data and station probe data to support the health status assessment of individual charging piles and charging stations. Taking into account equipment alarms and station health status assessments, it generates operation and maintenance inspection plans as needed.

[0127] See Figure 6 As shown, a method for operation and maintenance testing of charging equipment includes:

[0128] Step S41: Obtain system change data of the target system within the target time period; the target system includes a charging equipment system; the system change data includes the operation and maintenance data of the charging equipment, which includes system early warning information and the device name corresponding to the system change;

[0129] In this embodiment, system change data of the target charging equipment system within a target time period during the operation and maintenance process is first obtained. This system change data includes the charging equipment operation and maintenance data, which includes system warning information and the device names corresponding to the system changes, i.e., the names of the devices affected by the system changes. The target time includes the system change time of the target system and the time within a preset duration after the change. It can be understood that the target time includes the period from the start to the end of the change, and the preset duration after the change ends. Since the impact of the system change does not immediately cease after the change is completed, obtaining system change data within a certain period after the change improves the completeness of the change data.

[0130] Step S42: Based on the system change data, use natural language processing technology to determine the early warning information corresponding to the change in the target system;

[0131] In this embodiment, after obtaining the aforementioned system change data, natural language processing technology is used to determine the early warning information corresponding to the change in the target system from the aforementioned system change data.

[0132] Understandably, intelligent operation and maintenance systems for charging equipment possess relatively complete monitoring and early warning systems, capable of detecting faults and issuing system warnings as much as possible. However, due to the numerous components in the architecture of such systems and the frequent occurrence of warnings, the number of warnings explodes when system changes cause anomalies or accidents, making it difficult to identify warnings truly related to the system change. Therefore, by combining system warnings with the device names corresponding to the system change, and utilizing corpus preprocessing and feature extraction, warnings related to changes in the target system can be determined from the system change data.

[0133] Step S43: Obtain the time series data of the target system within the target time period, and based on the time series data, use a machine learning algorithm to obtain the changing trend of the monitoring indicators corresponding to the target system.

[0134] In specific implementations, such as Figure 8 As shown, this step specifically includes:

[0135] S431: Acquire time-series data within the target time period during the operation and maintenance testing of charging equipment;

[0136] S432: Determine the periodic feature value by performing periodic detection on the time series data;

[0137] In a specific implementation, the step of periodically detecting the time series data includes:

[0138] Given a reference period T for the time series data;

[0139] Using the reference period T as the dividing point, the time series data is divided into n / T sub-time series units, where n is the length of the time series data;

[0140] For each of the sub-time series units, a similarity coefficient is calculated by comparing each pair of units to determine the periodic feature value.

[0141] In some embodiments, optionally, Pearson correlation is used to calculate a similarity coefficient for each of the sub-time series units pairwise to determine the periodic feature value, including:

[0142] The similarity coefficient for each sub-time series unit is calculated by pairwise comparison using Pearson correlation. The Pearson similarity coefficient is expressed as:

[0143]

[0144] in, X T ={x1,...,xT}, Y T ={y1,...,y T} represents two sub-time series units; if X T =Y T Then COR(X) T ,Y T If X = 1, it means that the two sub-time series units are completely identical; if X T =-Y T Then COR(X) T ,Y T -1 indicates that the two sub-time series units are negatively correlated; -1 ≤ COR(X) T ,Y T )≤1.

[0145] Then, the calculated similarity coefficient is determined as the periodic feature value.

[0146] S433: If the periodic characteristic value is less than a first preset threshold, the time series data is determined to be volatile data;

[0147] If the time series data is determined to be volatile, anomaly detection is performed on the time series data to obtain the changing trend of the monitoring indicators corresponding to the target type of data, and thus obtain the changing trend of the monitoring indicators corresponding to the target system.

[0148] In some embodiments, when the time series data is determined to be volatile data, the Turkey's Test or 3-Sigema algorithm is used to perform dynamic threshold detection on the time series data to obtain the changing trend of the monitoring indicators corresponding to the target type data, so as to obtain the changing trend of the monitoring indicators corresponding to the target system.

[0149] S434: If the periodic feature value is greater than or equal to the first preset threshold, further perform stability detection on the time series data to determine the stability feature value;

[0150] In some embodiments, optionally, the further stability detection of the time series data includes:

[0151] The time series data is seasonally decomposed using the moving average algorithm into seasonal periodic components, long-term trend components, and random residual components.

[0152] The stability eigenvalue is determined based on the variance of the random residual components.

[0153] S435: If the stability characteristic value is greater than the second preset threshold, the time series data is determined to be stable data;

[0154] If the time series data is determined to be stable, anomaly detection is performed on the time series data to obtain the changing trend of the monitoring indicators corresponding to the target type of data, and thus obtain the changing trend of the monitoring indicators corresponding to the target system.

[0155] In some embodiments, when the time series data is determined to be stable data, a baseline threshold detection is performed on the time series data using a SARIMAX or moving average algorithm to obtain the changing trend of the monitoring indicators corresponding to the target type data, thereby obtaining the changing trend of the monitoring indicators corresponding to the target system.

[0156] S436: If the stability characteristic value is less than or equal to the second preset threshold, the time series data is determined to be periodic data;

[0157] If the time series data is determined to be periodic data, anomaly detection is performed on the time series data to obtain the changing trend of the monitoring indicators corresponding to the target type of data, and thus obtain the changing trend of the monitoring indicators corresponding to the target system.

[0158] In some embodiments, when the time series data is determined to be periodic data, anomaly detection is performed on the time series using machine learning classification and regression models to obtain the changing trends of monitoring indicators corresponding to the target data type, thereby obtaining the changing trends of monitoring indicators corresponding to the target system. Specifically, this includes:

[0159] The time series data is preprocessed; wherein the time series data includes historical time series data and current time series data.

[0160] Obtain a first anomaly detection result corresponding to the current time series data; wherein, the first anomaly detection result is obtained by extracting various features from the preprocessed time series data and inputting the various features into a pre-built machine learning classification model;

[0161] Obtain a second anomaly detection result corresponding to the current time series data; wherein the second anomaly detection result is obtained by comparing the current time series data with a prediction baseline, and the prediction baseline is obtained by inputting the preprocessed historical time series data into a pre-built regression model;

[0162] The first anomaly detection result and the second anomaly detection result are voted on to obtain the anomaly detection result of the time series data, that is, to obtain the changing trend of the monitoring indicators corresponding to the target type data, and to obtain the changing trend of the monitoring indicators corresponding to the target system.

[0163] This embodiment, based on the above embodiments, employs the Turkey's Test or 3-Sigema algorithm to perform dynamic threshold detection on the time series data when it is determined to be volatile, thereby obtaining anomaly detection results. When the time series data is determined to be stable, it employs the SARIMAX or moving average algorithm to perform baseline threshold detection, thereby obtaining anomaly detection results. When the time series data is determined to be periodic, it performs anomaly detection based on machine learning classification and regression models. This method improves the flexibility and accuracy of anomaly detection in time series data by classifying the time series data through periodicity and stability testing, and then selecting appropriate anomaly detection methods according to the different data categories. The method is also simple to operate.

[0164] Step S44: Conduct a system risk assessment of the target system based on the early warning information and the changing trend.

[0165] In the embodiments of this application, addressing the common problem in charging facility operation and maintenance work—namely, the imperfect fault diagnosis and safe operation and maintenance service system for charging stations, the lack of intelligent fault diagnosis and safety early warning during the charging process, and the inability to detect potential equipment defects and accident hazards in advance—an improvement is made to the anomaly detection of operation and maintenance detection data during the charging equipment operation and maintenance process. This is achieved by periodically detecting time-series data during the charging equipment operation and maintenance process to determine periodic characteristic values. These periodic characteristic values ​​are then compared with a first preset threshold to determine whether the data is volatile. If the data is determined to be non-volatile, further stability detection is performed, comparing the stability characteristic values ​​with a second preset threshold to distinguish between stable and periodic data. Based on the category of the time-series data, different anomaly detection methods are used to detect anomalies in the charging equipment operation and maintenance detection data for periodic data, volatile data, and stable data, respectively, to obtain the anomaly detection results for the time-series data. This method improves the flexibility and accuracy of anomaly detection in charging equipment operation and maintenance by periodically and stably detecting time-series data during the process. It categorizes the time-series data and selects appropriate anomaly detection methods based on the data category, making the process simple to operate. This method comprehensively considers charging equipment alarms, anomaly detection, and equipment health status assessment, and can generate operation and maintenance inspection plans as needed. This method is applicable not only to charging equipment operation and maintenance, grid operation and maintenance, energy storage equipment operation and maintenance, and / or integrated vehicle-charging-grid operation and maintenance, but also to the operation and maintenance of other IoT-based detection systems, demonstrating its strong versatility.

[0166] Furthermore, for the sake of simplicity, the method embodiments are all described as a series of actions. However, those skilled in the art should understand that the embodiments of this application are not limited to the described order of actions, because according to the embodiments of this application, some steps can be performed in other orders or simultaneously. Secondly, those skilled in the art should also understand that the embodiments described in the specification are all preferred embodiments, and the actions involved are not necessarily required by the embodiments of this application.

[0167] Accordingly, this application also discloses a system risk assessment device, applied to system risk assessment under a microservice architecture. (See [link]) Figure 9 As shown, the device includes:

[0168] Data acquisition module 21 is used to acquire system change data of the target system within a target time period; the system change data includes system early warning information and the name of the changing device;

[0169] The early warning information determination module 22 is used to determine the early warning information corresponding to the changes in the target system based on the system change data and using natural language processing technology.

[0170] The trend determination module 23 is used to acquire the time series data of the target system within the target time period, and based on the time series data, use a machine learning algorithm to obtain the trend of the monitoring indicators corresponding to the target system.

[0171] The risk assessment module 24 is used to conduct a system risk assessment of the target system based on the early warning information and the changing trend.

[0172] In this application, system change data of the target system within a target time period is obtained. This system change data includes system warning information and the names of devices corresponding to the system changes. Then, based on this system change data, natural language processing (NLP) technology is used to determine the warning information corresponding to the changes in the target system. Furthermore, time-series data of the target system within the target time period is obtained, and based on this time-series data, machine learning algorithms are used to obtain the changing trends of the monitoring indicators corresponding to the target system. Finally, a system risk assessment of the target system is performed based on the warning information and the changing trends. It is evident that by using NLP technology to determine the warning information corresponding to the system changes through system warning information within a time range and the names of devices affected by the system changes within a spatial range, and then combining this with the changing trends of the monitoring indicators corresponding to the system determined by time-series data using machine learning algorithms, the current risk of the system can be assessed quickly and accurately. This approach can rapidly and accurately identify risks arising from software patch releases, configuration changes, and basic resource changes, improving the accuracy of system change risk assessment in a microservice architecture.

[0173] In some specific embodiments, the warning information determination module 22 may specifically include:

[0174] The first keyword determination unit is used to preprocess the system warning information into a corpus and extract the first set of keywords from the processed system warning information.

[0175] The second set of keyword determination unit is used to perform corpus preprocessing on the device name and extract the second set of keywords from the processed changed device name;

[0176] The similarity determination unit is used to calculate the similarity between the first group of keywords and the keywords in the second group of keywords, and to determine the warning information corresponding to the change of the target system based on the similarity.

[0177] In some specific embodiments, the trend determination module 23 may specifically include:

[0178] A classification unit is used to classify the time series data according to a preset classification standard to obtain corresponding target type data; the target type data includes volatile data, periodic data, and stable data;

[0179] The trend determination unit is used to perform anomaly detection on the target type data using a preset detection algorithm corresponding to the target type data, and obtain the changing trend of the monitoring indicators corresponding to the target type data, so as to obtain the changing trend of the monitoring indicators corresponding to the target system.

[0180] The first trend determination subunit is used to perform anomaly detection on the volatile data based on Turkey detection to obtain the changing trend of the volatile monitoring indicators;

[0181] The second trend determination subunit is used to perform anomaly detection on the periodic data based on the same-year and month-on-month comparison algorithm to obtain the changing trend of the periodic monitoring indicators.

[0182] The third trend determination subunit is used to perform anomaly detection on the stable data based on the time series ARIMA algorithm to obtain the changing trend of stable monitoring indicators;

[0183] The first classification subunit is used to perform periodic detection on the time series data based on window data similarity, and classify the volatile data and non-volatile data according to a first preset threshold.

[0184] The second classification subunit is used to perform stability detection on the non-volatile data based on the STL algorithm, and classify the data into periodic data and stable data according to a second preset threshold.

[0185] In some specific embodiments, the risk assessment module 24 may specifically include:

[0186] The risk level determination unit is used to perform a system risk assessment on the target system based on the early warning information, the changing trend, and the indicator levels of the monitoring indicators, and to obtain the risk level of the target system; wherein, the indicator levels include core business level indicators, technical level indicators, and system resource level indicators;

[0187] An assessment report generation unit is used to generate a corresponding assessment report based on the risk level.

[0188] Furthermore, this application also discloses an electronic device, see [link to relevant documentation]. Figure 10 As shown, the content in the figure should not be considered as any limitation on the scope of use of this application.

[0189] Figure 10This is a schematic diagram of the structure of an electronic device 30 provided in an embodiment of this application. Specifically, the electronic device 30 may include: at least one processor 31, at least one memory 32, a power supply 33, a communication interface 34, an input / output interface 35, and a communication bus 36. The memory 32 stores a computer program, which is loaded and executed by the processor 31 to implement the relevant steps in the system risk assessment method disclosed in any of the foregoing embodiments.

[0190] In this embodiment, the power supply 33 is used to provide operating voltage for each hardware device on the electronic device 30; the communication interface 34 can create a data transmission channel between the electronic device 30 and external devices, and the communication protocol it follows can be any communication protocol applicable to the technical solution of this application, and is not specifically limited here; the input / output interface 35 is used to acquire external input data or output data to the outside world, and its specific interface type can be selected according to specific application needs, and is not specifically limited here.

[0191] In addition, the memory 32, as a carrier for resource storage, can be a read-only memory, random access memory, disk or optical disk, etc. The resources stored thereon include operating system 321, computer program 322 and data 323 including system change data, etc. The storage method can be temporary storage or permanent storage.

[0192] The operating system 321 manages and controls the various hardware devices on the electronic device 30 and the computer program 322 to enable the processor 31 to perform calculations and processing on the massive amount of data 323 in the memory 32. It can be Windows Server, Netware, Unix, Linux, etc. The computer program 322, in addition to including a computer program capable of performing the system risk assessment method executed by the electronic device 30 as disclosed in any of the foregoing embodiments, may further include computer programs capable of performing other specific tasks. The data 323 may include system change data acquired by the electronic device 30.

[0193] Furthermore, this application also discloses a computer storage medium storing computer-executable instructions. When the computer-executable instructions are loaded and executed by a processor, they implement the system risk assessment method steps disclosed in any of the foregoing embodiments.

[0194] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on its differences from other embodiments. Similar or identical parts between embodiments can be referred to interchangeably. For the apparatus disclosed in the embodiments, since it corresponds to the method disclosed in the embodiments, the description is relatively simple; relevant parts can be referred to in the method section.

[0195] The steps of the methods or algorithms described in conjunction with the embodiments disclosed herein can be implemented directly by hardware, a software module executed by a processor, or a combination of both. The software module can be located in random access memory (RAM), main memory, read-only memory (ROM), electrically programmable ROM, electrically erasable programmable ROM, registers, hard disk, removable disk, CD-ROM, or any other form of storage medium known in the art.

[0196] Finally, it should be noted that in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus 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 apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0197] The above provides a detailed description of the system risk assessment method, apparatus, equipment, and medium provided by the present invention. Specific examples have been used to illustrate the principles and implementation methods of the present invention. The descriptions of the above embodiments are only for the purpose of helping to understand the method and core ideas of the present invention. At the same time, for those skilled in the art, there will be changes in the specific implementation methods and application scope based on the ideas of the present invention. Therefore, the content of this specification should not be construed as a limitation of the present invention.

Claims

1. A system risk assessment method applied to system risk assessment under a microservice architecture, characterized in that, The method comprises the following steps: acquiring system change data of a target system at a target time; the system change data comprises system warning information and device names corresponding to system changes; based on the system change data, using natural language processing technology to determine the warning information corresponding to the target system change; acquiring time series data of the target system at the target time, and classifying the time series data according to a preset classification standard to obtain corresponding target type data; the target type data comprises volatile data, periodic data and stable data; using a preset detection algorithm corresponding to the target type data to perform abnormal detection on the target type data to obtain the change trend of the monitoring indicators corresponding to the target type data, so as to obtain the change trend of the monitoring indicators corresponding to the target system; based on the warning information, the change trend and the index level of the monitoring indicators, performing system risk assessment on the target system; wherein the index level comprises a core business level index, a technology level index and a system resource level index, the core business level index is the real-time health status of the overall business, the technology level index is the real-time health status of the component or service, and the system resource level index is the real-time health status of the middleware, host or system level.

2. The system risk assessment method of claim 1, wherein, based on the system change data, using natural language processing technology to determine the warning information corresponding to the target system change, comprising: performing corpus preprocessing on the system warning information, and extracting a first group of keywords from the preprocessed system warning information; performing corpus preprocessing on the device name, and extracting a second group of keywords from the preprocessed change device name; calculating the similarity of the keywords in the first group of keywords and the second group of keywords, and determining the warning information corresponding to the target system change based on the similarity.

3. The system risk assessment method of claim 1, wherein, The target time comprises the system change time of the target system and the time within a preset time period after the change.

4. The system risk assessment method of claim 1, wherein, using a preset detection algorithm corresponding to the target type data to perform abnormal detection on the target type data to obtain the change trend of the monitoring indicators corresponding to the target type data, comprising: based on Turkey detection, performing abnormal detection on the volatile data to obtain the change trend of the volatile monitoring indicators; based on the same ring ratio algorithm, performing abnormal detection on the periodic data to obtain the change trend of the periodic monitoring indicators; based on the time series ARIMA algorithm, performing abnormal detection on the stable data to obtain the change trend of the stable monitoring indicators.

5. The system risk assessment method of claim 1, wherein, based on the preset classification standard, classifying the time series data to obtain corresponding target type data, comprising: based on window data similarity, performing periodic detection on the time series data, and classifying to obtain the volatile data and non-volatile data according to a first preset threshold; based on the STL algorithm, performing stability detection on the non-volatile data, and classifying to obtain the periodic data and the stable data according to a second preset threshold.

6. The system risk assessment method according to any one of claims 1 to 5, characterized in that, perform system risk assessment on the target system based on the early warning information, the change trend and the index level of the monitoring index, to obtain a risk level of the target system; generate a corresponding evaluation report based on the risk level.

7. A charging device operation and maintenance detection method, characterized in that, The method comprises: obtaining system change data of a target system within a target time; the target system comprises a charging device system; the system change data comprises charging device operation and maintenance data, and the charging device operation and maintenance data comprises system early warning information and device names corresponding to system changes; based on the system change data, determining the early warning information corresponding to the target system change by using natural language processing technology; obtaining time series data of the target system within the target time, and classifying the time series data according to a preset classification standard to obtain corresponding target type data; the target type data comprises variable type data, periodic type data and stable type data; the target type data is subjected to abnormality detection by using a preset detection algorithm corresponding to the target type data, to obtain a change trend of a monitoring index corresponding to the target type data, so as to obtain a change trend of the monitoring index corresponding to the target system; perform system risk assessment on the target system based on the early warning information, the change trend and the index level of the monitoring index; wherein the index level comprises a core business level index, a technical level index and a system resource level index, the core business level index is a real-time health status of the whole business, the technical level index is a real-time health status of a component or a service, and the system resource level index is a real-time health status of a middleware, a host or a system level.

8. The method of claim 7, wherein, obtaining time series data within the target time in a charging device operation and maintenance detection process; determining a periodic characteristic value by periodically detecting the time series data; in a case where the periodic characteristic value is less than a first preset threshold, determining that the time series data belongs to the variable type data; in a case where the periodic characteristic value is greater than or equal to the first preset threshold, further performing stability detection on the time series data to determine a stability characteristic value; in a case where the stability characteristic value is greater than a second preset threshold, determining that the time series data belongs to the stable type data; in a case where the stability characteristic value is less than or equal to the second preset threshold, determining that the time series data belongs to the periodic type data.

9. The method of claim 8, wherein the periodically detecting the time series data comprises: giving a reference period T of the time series data; segmenting the time series data into n / T sub-time series units by taking the reference period T as a segmentation point, wherein n is the length of the time series data; calculating a similarity coefficient by comparing each of the sub-time series units with each other, and determining the similarity coefficient as the periodic characteristic value; the further performing stability detection on the time series data comprises: The moving average line algorithm is used for seasonal decomposition of the time series data, and the time series data is decomposed into seasonal periodic component, long-term trend component and random residual component; The stability characteristic value is determined according to the variance of the random residual component.

10. The method according to claim 8 or 9, characterized in that, According to the target type data to which the time series data belongs, the time series data is subjected to abnormality detection, including: In the case of determining that the time series data belongs to the variable type data, the Turkey,s Test or 3-Sigema algorithm is used for dynamic threshold detection of the time series data, so as to obtain the change trend of the monitoring index corresponding to the target system; In the case of determining that the time series data belongs to the stable type data, the SARIMAX or sliding average algorithm is used for baseline threshold detection of the time series data, so as to obtain the change trend of the monitoring index corresponding to the target system; In the case of determining that the time series data belongs to the periodic type data, the machine learning classification model and the regression model are used for abnormality detection of the time series, so as to obtain the change trend of the monitoring index corresponding to the target system. 11.A system risk assessment apparatus applied to system risk assessment under a microservice architecture, characterized in that, Comprising: The data acquisition module is used for acquiring system change data of the target system within a target time; The system change data includes system early warning information and change device name; The early warning information determination module is used for determining the early warning information corresponding to the target system change based on the system change data by using natural language processing technology; The change trend determination module is used for acquiring time series data of the target system within the target time, and classifying the time series data according to a preset classification standard, so as to obtain corresponding target type data; the target type data includes variable type data, periodic type data and stable type data; the target type data is subjected to abnormality detection by using a preset detection algorithm corresponding to the target type data, so as to obtain the change trend of the monitoring index corresponding to the target type data, and the change trend of the monitoring index corresponding to the target system is obtained; The risk assessment module is used for system risk assessment of the target system based on the early warning information, the change trend and the index level of the monitoring index; wherein, the index level includes core business level index, technical level index and system resource level index, the core business level index is the health status of the whole business in real time, the technical level index is the health status of the component or service in real time, and the system resource level index is the health status of the middleware, host or system layer in real time.

Images